Thesis Danique Giesen

89Zr-radiopharmaceuticals to study whole-body distribution and response to antibody-based cancer immunotherapies

Danique Giesen

Printing of this thesis was financially supported by UMCG Graduate School of Medical Sciences, Stichting Werkgroep Interne Oncologie and the University of Groningen.

Cover and layout Danique Giesen

Printed by Ridderprint | www.ridderprint.nl

© 2022 - Danique Giesen

All rights reserved. No part of this thesis may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronically, mechanically, by photo-copying, recording or otherwise, without prior permission of the author.

Proefschrift

ter verkrijging van de graad van doctor aan de Rijksuniversiteit Groningen op gezag van de rector magnificus prof. dr. C. Wijmenga en volgens besluit van het College voor Promoties. De openbare verdediging zal plaatsvinden op maandag 31 oktober 2022 om 14:30 uur door

Danique Giesen

geboren op 7 december 1990 te Zwolle

89Zr-radiopharmaceuticals to study whole-body distribution and response to antibody-based cancer immunotherapies

Promotores

Prof. dr. M. N. Lub-de Hooge

Prof. dr. E. G. E. de Vries

Beoordelingscommissie

Prof. dr. S. de Jong

Prof. dr. Andor Glaudemans

Prof. dr. N. H. Hendrikse

Paranimfen

Francien Talens

Elly van der Veen

Voor papa

Contents

Chapter 1 General introduction

Chapter 2

Theranostics using antibodies and antibody-related therapeutics

J Nucl Med. 2017;58(Suppl 2):83S-90S.

Chapter 3

89Zr-PET imaging to predict tumor uptake of 177Lu-NNV003 antiCD37 radioimmunotherapy in mouse models of B cell lymphoma Sci Rep. 2022;12(1):6286.

Chapter 4 89Zr-pembrolizumab biodistribution is influenced by PD-1mediated uptake in lymphoid organs

J Immunother Cancer. 2020;8(2):e000938.

Chapter 5 89Zr-pembrolizumab imaging as a non-invasive approach to assess clinical response to PD-1 blockade in cancer

Ann Oncol. 2022;33(1):80-88.

Chapter 6 Probody therapeutic design of 89Zr-CX-072 promotes accumulation in PD-L1-expressing tumors compared to normal murine lymphoid tissue

Clin Cancer Res. 2020;26(15):3999-4009.

11 19 37 61 89 113

Chapter 7 First-in-Human study of the biodistribution and pharmacokinetics of 89Zr-CX-072, a novel immunoPET tracer based on an anti-PD-L1 Probody Clin Cancer Res. 2021;27(19):5325-5333.

Chapter 8 Preclinical PET imaging of bispecific antibody ERY974 targeting CD3 and glypican 3 reveals that tumor uptake correlates to T cell infiltrate J Immunother Cancer. 2020;8(1):e000548.

Chapter 9 Whole-body CD8+ T-cell visualization before and during cancer immunotherapy Submitted

Chapter 10 Summary and future perspectives

Chapter 11 Nederlandse samenvatting (Dutch summary)

Appendices Dankwoord (acknowledgements) Publications

Curriculum vitae

139 161 191 231 241 252 256 259

General introduction

Background

Immunotherapy has become an essential pillar of cancer treatment and obtained a clear role in clinical cancer care. This type of therapy has shown impressive results by increasing the survival of patients with advanced stages of several tumor types. Immunotherapeutic drugs are characterized by using components of the immune system to induce an effective antitumor immune response (1). Some cancer immunotherapies consist of (parts of) antibodies that inhibit the function of proteins expressed by cancer cells, mostly immune checkpoints. Other cancer immunotherapies include small immunomodulating molecules, cell-based immunotherapies, vaccines and oncolytic viruses. Currently, the European Medicines Agency (EMA) and Food and Drug Administration (FDA) have approved 8 different immune checkpoint-inhibiting antibodies to treat 19 tumor types. These immune checkpointinhibiting antibodies are directed against the programmed cell death-1 (PD-1) receptor and its ligand, programmed cell death-ligand 1 (PD-L1), as well as cytotoxic T-lymphocyte-associated antigen-4 (CTLA-4) and lymphocyte activation gene-3 (LAG-3).

As of December 2019, the number of immunotherapeutic anticancer agents in the global drug development pipeline showed a 91% increase during the past 2 years: From 2,030 to 3,876 (2). The clinical trial landscape for cancer immunotherapies involves numerous studies with PD-1/PD-L1 immune checkpoint-inhibiting antibodies as a single agent for multiple indications and in combination regimens. CTLA-4 combined with PD-1/PD-L1 improves response rates and overall survival for specific cancers (3–6), but also demonstrates increased side-effects, mostly related to the immune system (7). With many new immunotherapeutic drugs being developed, companion diagnostics to tailor treatment to individual patients are urgently needed.

PD-L1 expression, microsatellite-instability (MSI)/defective mismatch repair (dMMR) and tumor mutational burden (TMB) are clinically used predictive biomarkers for immune checkpoint inhibitors in cancer patients (8). These biomarkers play an important role in assisting patient selection. However, different tumor types require distinct assays with their corresponding limitations. Treatment decisions are often based on immunohistochemistry analysis in a single tumor biopsy and dosing schedules are determined using blood-based pharmacokinetic analyses. Heterogeneity in target expression and variable drug uptake between tumor lesions within one patient are thereby not considered. Molecular imaging with positron emission tomography (PET) can serve to gain real-time information on all tumor lesions within the patient’s body. Radiolabeling immune-targeting antibodies with a PET isotope allows for noninvasive evaluation of drug pharmacokinetics, immune target expression, and immunological responses. When developing these radiopharmaceuticals, the half-life of the applied radioisotope ideally matches the time that antibodies need to distribute and accumulate in

target-expressing tissues. The PET isotope zirconium-89 (89Zr) has a half-life of 3.3 days and is compatible with the serum half-life of most immunotherapeutic antibodies.

The aim of the research described in this thesis is to develop 89Zr-radiopharmaceuticals to advance the development of novel cancer immunotherapies and explore their use as a biomarker of response.

Thesis outline

Radiopharmaceuticals with antitumor activity can be used for therapy and diagnostics, and represent a rapidly expanding group of cancer medicines. In chapter 2, we aim to provide an overview of current research on radiolabeled antibodies and antibody-related therapeutics that may be used for both therapy and diagnostics using PET imaging. We perform a PubMed search using the terms “PET” AND “Cancer” AND “Antibody” OR “ADC” OR “Bispecific” in combination with the most commonly used PET radionuclides: 64Cu, 68Ga, 86Y, 89Zr, and 124I. In addition, we search ClinicialTrials.gov for ongoing studies using the terms “Cancer” AND “PET” NOT “FDG.” A total of 1,448 (pre)clinical studies are reviewed to provide an up-to-date overview. Also, we identify several challenges for translating the use of radiopharmaceuticals to standardized and, ultimately, daily routine patient care.

Radioimmunotherapeutic (RIT) agents are a subclass of radio-pharmaceuticals, which employ antibodies as targeted delivery vehicles for therapeutic α- or β -emitting radionuclides to selectively eradicate tumor cells. 177Lu-NNV003, a RIT agent targeting leukocyte antigen CD37, is developed to potentially treat patients with B cell non-Hodgkin’s lymphoma (NHL) refractory to or relapsed during anti-CD20 radioimmunotherapy (RIT). As CD37 is expressed on malignant and normal B cells, patients are at risk for developing hematological toxicities. Therefore, a tool to non-invasively assess CD37-targeting by 177Lu-NNV003 RIT is of utmost interest and could assist its clinical development and use. In chapter 3, we evaluate the utility of 89Zr-labeled NNV003 PET imaging in predicting whole-body distribution and tumor uptake of 177Lu-NNV003 RIT. NNV003 is radiolabeled with 89Zr and its in vivo distribution is evaluated in immune-compromised mice bearing human CD37-expressing REC1 B cell NHL or RAMOS Burkitt’s lymphoma xenograft tumors. Indium-111 (111In)-labeled IgG served as control antibody. PET imaging is performed at day 5 post 89Zr-NNV003 administration, followed by ex vivo quantification of uptake per organ. Whole-body distribution and tumor-targeting properties of 89Zr-NNV003 are compared to 177Lu-NNV003 in the same mouse model.

To better predict response to immune checkpoint therapy, 89Zr-radiopharmaceuticals can be applied to gain insight in the in vivo behavior of immune checkpoint-targeting antibodies. The PD-1 immune checkpoint its primarily expressed by T cells, while its ligand PD-L1 is expressed

by tumor cells, B-cells, NK-cells, dendritic cells and macrophages. In chapter 4, we study the whole-body distribution of 89Zr-labeled anti-PD-1 antibody pembrolizumab and 89Zr-IgG4 control antibody in humanized mice compared to non-humanized mice, xenografted with A375M human melanoma tumors. We perform PET imaging at day 7 post 89Zr-pembrolizumab administration to visualize distribution to human peripheral mononuclear blood cells (PBMCs) present in the tumor, spleen, lymph nodes, thymus and bone marrow, quantified by ex vivo analysis. Tumor and spleen tissues are analyzed by immunohistochemistry for PD-1, CD3 and CD8 expression and autoradiography.

While PD-L1 expression and TMB are EMA-approved biomarkers for non-small cell lung cancer (NSCLC) and solid tumors respectively, not all patients with high tumor PD-L1 expression and high TMB respond to pembrolizumab treatment. In chapter 5, we used PET imaging to assess 89Zr-pembrolizumab tumor uptake and whole-body distribution before immune checkpoint inhibitor therapy and explored its relationship with patient response. First, 89Zr-pembrolizumab analytical methods and manufacturing procedures are validated under good manufacturing practice (GMP) conditions to enable administration to patients. Eighteen patients, 11 with melanoma and 7 with NSCLC, received 37 MBq 89Zr-pembrolizumab (~2.5 mg) intravenously plus 2.5 or 7.5 mg unlabeled pembrolizumab followed by PET imaging on days 2, 4, and 7. Thereafter, PD-1-targeting antibody treatment per standard of care (pembrolizumab or nivolumab ± ipilimumab) is initiated. 89Zr-pembrolizumab tumor uptake is determined as maximum standardized uptake value (SUVmax), normal organ uptake is determined as mean standardized uptake value (SUVmean). Tumor response is assessed according to (i)RECIST v1.1. Archival tumor tissue or fresh biopsies obtained after the last PET scan were stained immunohistochemically for PD-1, PD-L1, and CD8.

Immune-related adverse events are often observed during treatment with immune checkpoint-inhibiting antibodies. Side-effects experienced by patients treated with singleagent anti-PD-1 or anti-CTLA4 antibody increase by 34% when they are combined. CX-072, a protease-activatable antibody targeting both human and murine PD-L1, is designed to be specifically activated in tumor tissues, thus potentially reducing immune-related toxicities in normal tissues. The in vivo behavior of such a protease-activatable antibody is unknown. In chapter 6, we perform 89Zr-CX072 PET imaging in immune-competent and immunecompromised mice bearing PD-L1-expressing tumors. 89Zr-CX-072 uptake in tumors and immune tissues at day 7 post 89Zr-CX072 administration is quantified ex vivo and compared to 89Zr-labeled normal anti-PD-L1 antibody (not protease-activatable) and 89Zr-labeled control Probody. Tumor, spleen, lymph nodes, bone marrow and thymus are analyzed for PD-L1 expression by immunohistochemistry and flow cytometry. We use autoradiography to correlate 89Zr-CX-072 distribution to PD-L1-expressing tumor areas. Levels of activated CX-072

are measured by Western blot analysis in the tumor and spleen to further explore tumorspecific activation.

To enable administration in patients, a GMP-compliant manufacturing process for 89Zr-CX-072 was developed and validated. In chapter 7, we investigate how CX-072’s Probody therapeutic design affects its whole-body distribution in eight patients with variable types of solid tumors. Patients received 37 MBq 89Zr-CX-072 (~1 mg) plus 0, 4, or 9 mg unlabeled CX-072, followed by PET imaging on days 2, 4, and 7. Thereafter, treatment with CX-072 or CX-072 + ipilimumab is initiated. 89Zr-CX-072 uptake in normal tissues is expressed as SUVmean and tumor uptake as SUVmax. Tumor response is determined according to (ir)RECIST v1.1. We measure PD-L1 expression immunohistochemically in archival tumor tissue. In the blood pool, presence of intact (inactivated) 89Zr-CX-72 was assessed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and autoradiography.

Bispecific antibodies may represent an alternative strategy to increase tumor immunogenicity if they bind T cells with one arm and tumor cells with the other arm, thereby redirecting the T cell response to the tumor. ERY974 is a bispecific antibody that engages CD3 on T cells and glypican 3 (GPC3) on tumors. ERY974 has different binding affinities for CD3 and GPC3, with dissociation constants (Kd) of 207 nM and 1.5 nM respectively, and therefore its in vivo pharmacokinetics are not easily predicted. In chapter 8, PET imaging is used to reveal the whole-body distribution of 89Zr-ERY974, followed by ex vivo quantification of organ uptake in mice bearing xenograft tumors with different levels of GPC3 expression. We use immunecompromised mice and mice reconstituted with human PBMCs, as well as 89Zr-labeled control antibodies targeting CD3/non-mammalian protein keyhole limpet hemocyanin (KLH) or KLH only. The redirection of T cells in ex vivo tumor tissues is evaluated by autoradiography and CD3 immunohistochemistry.

Immune checkpoint-inhibiting antibodies stimulate the distribution of CD8+ T cells from lymph nodes to normal immune tissues and tumors (9–11). Immune checkpoint inhibitor-induced tumor-infiltrating CD8+ T cells are associated with response across multiple tumor types, including melanoma. In chapter 9, we study the whole-body distribution of CD8+ T cells by 89ZED88082A PET imaging in patients with solid tumors before and after immune checkpoint inhibitor therapy. Thirty-eight eligible patients with locally advanced or metastatic solid tumors receive 37 MBq (~1.7 mg) 89ZED88082A followed by PET imaging after 1 hour and on days 2, 4, 7. After selecting the optimal dose, patients receive 89ZED88082A before and after two cycles of atezolizumab treatment and PET scans on days 2 and 4. Normal organ tracer uptake was calculated as SUVmean, tumor lesion and lymph node uptake as SUVmax. Tumor response is determined based on (i)RECIST1.1. Tumor biopsies are collected after the last PET scan and

analyzed by CD8 immunohistochemistry and autoradiography. Blood samples are drawn for 89ZED88082A pharmacokinetic and anti-drug antibody (ADA) analyses.

Finally, a summary of thesis results and future perspectives are described in chapter 10. A Dutch thesis summary is provided in chapter 11.

References

1. Couzin-Frankel J. Breakthrough of the year 2013: Cancer immuno-therapy. Science. 2013;342(12):1432–3.

2. Yu JX, Hubbard-Lucey VM, Tang J. Immuno-oncology drug deve-lopment goes global. Nat Rev Drug Discov. 2019;18(12):899–900.

3. Wolchok JD, Chiarion-Sileni V, Gonzalez R, Rutkowski P, Grob JJ, Cowey CL, et al. Overall survival with combined nivolumab and ipilimumab in advanced melanoma. N Engl J Med. 2017;377(14):1345–56.

4. Hellmann MD, Ciuleanu T-E, Pluzanski A, Lee JS, Otterson GA, Audigier-Valette C, et al. Nivolumab plus ipilimumab in lung cancer with a high tumor mutational burden. N Engl J Med. 2018;378(22):2093–104.

5. Overman MJ, Lonardi S, Wong KYM, Lenz H-J, Gelsomino F, Aglietta M, et al. Durable clinical benefit with nivolumab plus ipilimumab in DNA mismatch repair-deficient/microsatellite instability-high metastatic colorectal cancer. J Clin Oncol. 2018;36(8):773–9.

6. Motzer RJ, Tannir NM, McDermott DF, Arén Frontera O, Melichar B, Choueiri TK, et al. Nivolumab plus ipilimumab versus sunitinib in advanced renal-cell carcinoma. N Engl J Med. 2018;378(14):1277–90.

7. Postow MA, Sidlow R, Hellmann MD. Immune-related adverse events associated with immune checkpoint blockade. N Engl J Med. 2018;378(2):158–68.

8. Wang Y, Tong Z, Zhang W, Zhang W, Buzdin A, Mu X, et al. FDA-approved and emerging next generation predictive biomarkers for immune checkpoint inhibitors in cancer patients. Front Oncol. 2021;11(6):1–15.

9. Hegde PS, Karanikas V, Evers S. The where, the when, and the how of immune monitoring for cancer immunotherapies in the era of checkpoint inhibition. Clin Cancer Res. 2016;22(8):1865–74.

10. Herbst RS, Soria J, Kowanetz M, Fine GD, Hamid O, Kohrt HEK, et al. Predictive correlates of response to the anti-PD-L1 antibody MPDL3280A in cancer patients. Nature. 2014;515(7528):563–7.

11. Tumeh PC, Harview CL, Yearley JH, Shintaku IP, Taylor EJM, Robert L, et al. PD-1 blockade induces responses by inhibiting adaptive immune resistance. Nature. 2014;515(7528):568–71.

Theranostics using antibodies and antibodyrelated therapeutics

Danique Giesen 1 *, Kirsten L Moek 1 *, Iris C Kok 1, Derk Jan A de Groot 1, Mathilde Jalving 1, Rudolf S N Fehrmann 1, Marjolijn N Lub-de Hooge 2, Adrienne H Brouwers 3 , Elisabeth G E de Vries 4

* These authors contributed equally to this work

1 Department of Medical Oncology, 2 Department of Clinical Pharmacy and Pharmacology, 3 Department of Nuclear Medicine and Molecular Imaging, and 4 Department of Medical Oncology, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands. e.g.e.de.vries@umcg.nl.

J Nucl Med. 2017;58(Suppl 2):83S-90S.

Abstract

In theranostics, radiolabeled compounds are used to determine a treatment strategy by combining therapeutics and diagnostics in the same agent. Monoclonal antibodies (mAbs) and antibody-related therapeutics represent a rapidly expanding group of cancer medicines. Theranostic approaches using these drugs in oncology are particularly interesting because antibodies are designed against specific targets on the tumor cell membrane and immune cells as well as targets in the tumor microenvironment. In addition, these drugs are relatively easy to radiolabel. Noninvasive molecular imaging techniques, such as SPECT and PET, provide information on the whole-body distribution of radiolabeled mAbs and antibody-related therapeutics. Molecular antibody imaging can potentially elucidate drug target expression, tracer uptake in the tumor, tumor saturation, and heterogeneity for these parameters within the tumor. These data can support drug development and may aid in patient stratification and monitoring of the treatment response. Selecting a radionuclide for theranostic purposes generally starts by matching the serum half-life of the mAb or antibody-related therapeutic and the physical half-life of the radionuclide. Furthermore, PET imaging allows better quantification than the SPECT technique. This information has increased interest in theranostics using PET radionuclides with a relatively long physical half-life, such as 89Zr. In this review, we provide an overview of ongoing research on mAbs and antibody-related theranostics in preclinical and clinical oncologic settings. We identified 24 antibodies or antibody-related therapeutics labeled with PET radionuclides for theranostic purposes in patients. For this approach to become integrated in standard care, further standardization with respect to the procedures involved is required.

Theranostics is a treatment strategy in which a single agent is used for both diagnostic and therapeutic purposes. Theranostic procedures are based on radiolabeling compounds of interest. This approach potentially enables the evaluation of drug target expression and the actual presence of the drug at the tumor site in vivo in cancer patients using imaging methods such as SPECT or PET. Particularly interesting are theranostic approaches using monoclonal antibodies (mAbs) and antibody-related therapeutics because these agents belong to a rapidly expanding group of effective anticancer drugs. Antibody-related therapeutics include bispecific antibodies (e.g., bispecific T-cell engagers [BiTEs]), engineered antibody structures (e.g., minibodies, diabodies, and nanobodies), antibody–drug conjugates (ADCs), and radiolabeled antibodies for radioimmunotherapy. These drugs have ideal characteristics for theranostic approaches because they are designed against a specific target, often on the cell surface, and are relatively easy to radiolabel.

As of December 2016, 24 mAbs or antibody-related therapeutics had been approved by the

U.S. Food and Drug Administration (FDA) or the European Medicines Agency (EMA) for use in cancer patients. These drugs comprise 20 mAbs, 1 BiTE, 2 ADCs, and 1 radioimmunotherapy antibody. The approved mAbs and antibody-related therapeutics are directed against targets on the tumor cell membrane and immune cells as well as targets in the microenvironment.

mAbs are administered in noncurative and curative settings. In the noncurative setting, these drugs have proven effects on (disease-free) survival (1–3). In the adjuvant setting, the anti–human epidermal growth factor receptor 2 (HER2) antibody trastuzumab and the anti–cytotoxic T-lymphocyte antigen 4 (CTLA-4) antibody ipilimumab increase overall survival in patients with breast cancer and melanoma, respectively (4,5).

In oncology, even when a drug has a proven clinical benefit for a certain patient population, not all patients will benefit. This outcome can potentially be related to heterogeneity in tumor target expression, vascularization of the tumor, or the presence of an immunosuppressive tumor microenvironment. Treatment decisions in both routine practice and drug development are frequently made using information obtained from a biopsy of a single tumor lesion. Furthermore, recommended dosing schedules are mostly determined using blood-based pharmacokinetic analyses. Differences in drug target expression and drug uptake in various tumor lesions within a single patient are almost never considered. In this respect, a theranostic approach is of potential interest because it might provide insight into tumor target heterogeneity and information on whether a drug reaches tumor lesions. For these reasons, molecular antibody imaging can also be a valuable tool in drug development, drug decision making, and patient enrichment strategies.

In this review, we provide an overview of current research on mAbs and antibody-related therapeutics visualized using PET imaging in both preclinical and clinical oncologic settings.

Search strategy

To identify available studies investigating theranostic approaches with mAbs and antibodyrelated therapeutics, we performed a PubMed search on November 21, 2016. The search terms “PET” AND “Cancer” AND “Antibody” OR “ADC” OR “Bispecific” were used in combination with the most commonly used PET radionuclides: 64Cu, 68Ga, 86Y, 89Zr, and 124I. We focused on studies published during the last 5 y to capture most recent developments but included relevant studies published earlier. In addition, we searched ClinicialTrials.gov on November 17, 2016, for ongoing studies over the past 10 y with the search terms “Cancer” AND “PET” NOT “FDG.” Both searches were limited to articles published in English. Case reports, reviews, and books were excluded. In total, 1,448 preclinical and clinical studies were found. All articles and ongoing studies were manually screened for relevance using the following inclusion criteria. First, a full-

Theranostics with antibodies

sized mAb, ADC, bispecific antibody, or fragment with theranostic potential was used. Second, for a study in which humans were included, the subjects were 18 y old or older. Finally, we limited our search to the most commonly used PET radionuclides to provide a comprehensive overview of relevant agents with prime theranostic potential. Articles were excluded if (potential) theranostic applications of those agents were not found.

General aspects of molecular imaging using mAbs and antibodyrelated therpeutics

mAbs and antibody-related therapeutics can be efficiently labeled with a wide range of radionuclides. In general, the different labeling techniques can easily be applied to most mAbs and antibody-related therapeutics. These drugs can therefore be used in studies ranging from mice to humans (6).

Chelation and radiolabeling for molecular antibody imaging

99mTc, 64Cu, 68Ga, 86Y, 89Zr, 111In, 123I, 124I, 131I, and 177Lu are the radionuclides most commonly used for molecular imaging with mAbs and antibody-related therapeutics in the field of oncology (Table 1). Selecting a suitable radionuclide generally starts by matching the serum half-life of the mAb or antibody-related therapeutic and the physical half-life of the radionuclide. This step is essential to ensure that radioactivity can be detected long enough for the drug to reach its target while minimizing the duration of exposure to harmful radiation (6). The serum half-life mainly depends on the structure and size of the mAb or antibodyrelated therapeutic. Generally, the serum half-life is shorter for a smaller mAb construct than for a full-sized mAb because the molecular weight is often below the renal clearance threshold of approximately 70 kDa. For example, the serum half-life of cetuximab (±150 kDa) is 3–4 d, whereas the serum half-life of the BiTE antibody blinatumomab (±60 kDa) is only several hours. In addition, the serum half-life depends on the IgG subtype from which the

Technique Isotope Half life Residualizing

PET 68Ga 67.7 min + 64Cu 12.7 h

86Y 14.7 h

89Zr 78.4 h

124I 100.3 h

SPECT 99mTc 6.0 h + 123I 13.2 h 111In 67.3 h + 177Lu 159.5 h + 131I 192.5 h

TABLE 1: Characteristics of radionuclides used with mAbs or antibody-related theranostics in oncology.

mAb or antibody-related therapeutic is derived and on whether the (constructed) mAb is fully human, humanized murine, or chimeric. The serum half-lives of mAbs and antibody-related therapeutics can vary from 30 min to 30 d.

Furthermore, a chelator is required to link metal-based radionuclides, such as 64Cu, 68Ga, 86Y, 89Zr, 111In, and 177Lu, to an mAb or antibody-related therapeutic. Deciding on a chelator for human use depends on the radionuclide, the most stable chemical link, and the clinical applicability in terms of validation.

Another important consideration in choosing a nuclide for radiolabeling is whether the mAb or antibody-related therapeutic becomes internalized after binding to its target. For example, when radiometal-labeled drugs are metabolized, the metal-based radionuclide is trapped intracellularly in lysosomes through residualization (7). This process results in higher absolute uptake of the tracer and leads eventually to higher tumor-to-blood ratios. Iodine-labeled drugs are characterized by rapid renal clearance of the radionuclide from tumor cells because iodinated mAbs do not residualize. However, methods for increasing the internalization of iodine-labeled drugs are available. For instance, a bivalent peptide consisting of 4 d-amino acids increased the residence time of the 125I radiolabel in renal cell carcinomas (RCCs) significantly over that of the 111In-labeled control peptide (8).

Radionuclides commonly used in molecular antibody imaging

Although radionuclides with different physical half-lives are available for radiolabeling, the clinical use of many nuclides is hampered by the requirement for a cyclotron either on-site or about 1 physical half-life of transport time away from the site. An alternative is to use a generator for which a radionuclide laboratory suffices. In the latter situation, the long-lived “mother” radionuclide allows for instant/constant availability of the “daughter” radionuclide. For example, the 68Ga radionuclide is produced using a generator—containing the cyclotronproduced “mother” radionuclide 68Ge—allowing the radiolabeling of mAbs or antibody-related therapeutics at the site of administration. Unfortunately, the relatively short physical half-life (68 min) of this radionuclide limits its use for imaging full-sized antibodies, which require several days to achieve sufficient tumor-to-blood ratios.

Molecular imaging using the positron emitter 89Zr for antibody labeling has been increasingly used in recent years. This radionuclide has suitable characteristics for molecular antibody imaging; for example, its physical half-life of 78.4 h generally matches the serum half-life of most mAbs and antibody-related therapeutics in vivo and is compatible with the time needed for residualization, generally allowing high tumor-to-background ratios. Furthermore, procedures for the large-scale production of 89Zr have been developed, and mAbs and

with antibodies

antibody-related therapeutics can be stably labeled with this radionuclide (9).

Pharmacokinetics and target visualization of radiolabeled mAbs

Most radiolabeled full-sized antibodies have a relatively long effective half-life (14–21 d). After administration, the drug is distributed throughout the body and taken up by the tumor and other tissues that express its target. Over time, tumor-to-background ratios will generally increase because of tracer binding to the tumor, residualization of the radiolabel in tumor tissue, and clearance of the nonbound tracer from the circulation and background organs/ tissues.

Tumor accumulation of the radiolabeled drug is the consequence of target location, target expression levels, target saturation, and internalization of the drug. In addition, several kinetic features, such as perfusion and vascularization, may influence tumor visualization. For example, tumor uptake of 111In-labeled death receptor 5–targeting antibody CS-1008 was observed in only 63% of 19 patients with metastatic colorectal cancer even though all patients were considered to have death receptor 5–positive lesions (10).

Interestingly, tracer uptake in normal tissues can help explain observed side effects. 111In-trastuzumab scintigraphy revealed an increase in myocardial uptake shortly after anthracycline treatment in a subgroup of patients (11). This observation may explain why trastuzumab-related cardiotoxicity can occur when this drug is combined with anthracyclinebased chemotherapy.

Clinical imaging studies generally start with determination of the optimal protein dose and time point for exploring tumor-to-background ratios and image quality (12). Especially for dose-dependent pharmacokinetics, the optimal protein dose may have to be high. A radioactive dose of 37 MBq and a scan time of 45–60 min allow adequate visualization of a 89Zr-labeled, full-sized mAb at days 4–7 (13,14). The mAb or antibody-related therapeutic is linked to a certain amount of radioactivity per milligram—the so-called specific activity, expressed in MBq/mg. Specific activities for most mAbs and antibody-related therapeutics are generally limited to 750–1,000 MBq/mg because of radiolysis. Unlabeled (naked) antibody is added to the radiolabeled mAb to allow higher tumor uptake of the tracer for adequate tumor visualization. When the total protein dose that can be safely administered to the patient is relatively low—for example, in the microgram range—reaching a sufficient radioactive dose for successful imaging is difficult. To avoid side effects, the protein dose for T-cell–engaging drugs is generally low, making the use of these drugs as theranostics challenging (15).

Use of theranostics in clinical decision making

We identified 6 different antibody structures that are currently used as theranostic agents in patients. Figure 1 shows how these compounds are directed against a specific target located on the tumor cell or in the tumor microenvironment—for example, macrophages, dendritic cells, and T cells. In addition, Figure 1 shows a simplified illustration of the radiolabeled forms of the therapeutics for theranostic purposes. Most molecular imaging clinical trials have been performed using radiolabeled FDA- or EMA-approved drugs, such as trastuzumab for breast cancer, cetuximab for colorectal cancer, and bevacizumab for several indications (Table 2). An example of 89Zr-trastuzumab PET for breast cancer is shown in Figure 2. We identified 14 clinical imaging studies with trastuzumab, making this the most frequently investigated therapeutic mAb in molecular imaging (Fig. 1A).

FIGURE 1: Antibodies and antibodyrelated theranostics.

Six different antibody structures in clinical use and radiolabeled compounds used for theranostics (top right corners). (A) Theranostics using mAbs (e.g., trastuzumab targeting HER2 on tumor cells). (B) Theranostics in angiogenesis (e.g., bevacizumab targeting VEGF-A).

(C) Theranostics using immune checkpoint inhibitors (e.g., anti–PD-L1 antibody targeting PD-L1 on tumor cells and immune cells). (D) Theranostics using BiTEs (e.g., AMG 211 targeting carcinoembryonic antigen on tumor cells and CD3ε on T cells).

(E) Theranostics using ADCs (e.g., trastuzumab emtansine targeting HER2 on tumor cells using radiolabeled naked trastuzumab). (F) Theranostics using radioimmunotherapy (e.g., 90Y-ibritumomab tiuxetan).

Theranostics with antibodies

TABLE 2: Clinical Studies of theranostic uses of antibodies or antibody-related therapeutics labeled with PET radionuclides.

Target Tracer name Tracer structure No. of (ongoing) clinical trials

TUMOR

Patient population No. of centers

A33 124I huA33 mAb 1 CRC 1

CA6 64Cu B Fab Fab fragment 1 Breast or ovarian cancer 1

CA9 124I girentuximab mAb 5 RCC 16

89Zr girentuximab mAb 2 RCC 3

CEA 89Zr AMG 211 BiTE 1 Gastrointestinal adenocarcinoma 2

CD20 89Zr ibritumomab tiuxetan mAb 1 NHL 1

CD44 89Zr RG7356 mAb 1 CD44 positive solid tumor 6

EGFR (HER1) 89Zr cetuximab mAb 4 CRC, HNSCC, stage IV cancer 4

89Zr panitumumab mAb 2 CRC, NSCLC, sarcoma, urothelial carcinoma 1

EphA2 89Zr DS 8895a mAb 1 EphA2 positive cancer 1

HER2 64Cu trastuzumab mAb 7 Breast or gastric cancer 3

68Ga HER2 Nanobody Nanobody 1 Breast cancer 1 68Ga trastuzumab Fab Fab fragment 1 Breast cancer 1

89Zr trastuzumab mAb 7 Breast cancer 7

HER3 64Cu patritumab mAb 1 Solid tumors 1 89Zr GSK2849330 mAb 1 HER3 positive solid tumors 1 89Zr lumretuzumab mAb 1 HER3 positive solid tumors 12

MSLN 89Zr MMOT0530A mAb 1 Ovarian or pancreatic cancer 2

PIGF 89Zr RO5323441 mAb 1 GBM 1

PSCA 124I A11 Minibody 1 Bladder, pancreatic, or prostate cancer 1

PSMA 89Zr J591 mAb 4 GBM, prostate cancer 2 STEAP1 89Zr MSTP2109A mAb 1 Prostate cancer 1

MICROENVIRONMENT

PD 1 89Zr pembrolizumab mAb 1 NSCLC, melanoma 2

PD L1 89Zr atezolizumab mAb 1 Bladder cancer, NSCLC, TNBC 1

TGFβ 89Zr fresolimumab mAb 1 Glioma 1 VEGF A 89Zr bevacizumab mAb 9 Breast cancer, glioma, MM, NET, NSCLC, RCC 3

CRC: colorectal carcinoma; CA6: carbonic anhydrase 6; CA9: carbonic anhydrase 9; NHL: non Hodgkin lymphoma; EGFR: epidermal growth factor receptor; HER1: human epidermal growth factor receptor 1; HNSCC: head and neck squamous cell carcinoma; EphA2: ephrin receptor A2; HER3: human epidermal growth factor receptor 3; MSLN: mesothelin; PIGF: placental growth factor; GBM: glioblastoma multiforme; PSCA: prostate stem cell antigen; PSMA: prostate specific membrane antigen; STEAP1: 6 transmembrane epithelial antigen of prostate family member 1; TNBC: triple negative breast cancer; TGFβ: transforming growth factor β; MM: multiple myeloma; NET: neuroendocrine tumor.

Several lessons can be learned from these studies. First, 111In-trastuzumab SPECT imaging showed new HER2-positive tumor lesions that were not detected using conventional imaging in 13 of 15 metastatic breast cancer patients (16). These data showed that molecular antibody imaging can help identify tumor lesions that are missed by conventional imaging techniques. Second, serial SPECT imaging with 111In-trastuzumab before and after 12 wk of trastuzumab treatment showed persistent uptake in all tumor lesions, with only a 20% decrease in tumor tracer uptake (17). These data indicated that HER2 is constantly available at the tumor cell surface to bind to trastuzumab and that the tumor is not completely saturated by

Theranostics with antibodies

FIGURE 2: 89Zr-trastuzumab PET imaging.

Patient with human epidermal growth factor receptor 2–positive metastatic breast cancer imaged 4 d after injection with 37 MBq of 89Zr-trastuzumab and total protein dose of 50 mg. (A) Maximum-intensity-projection image of 89Zr-trastuzumab PET/CT scan showing tracer present in circulation, uptake in intrahepatic metastases, and intestinal excretion. (B) Transverse plane of fused PET/CT (low-dose CT) of chest showing tracer uptake in cervical lymph node. (C) Transverse plane showing tracer uptake in metastasis (left side) in T7. (D) Transverse plane showing tracer uptake in liver metastases.

trastuzumab treatment. Third, in a study with 89Zr-trastuzumab PET for metastatic breast cancer patients, tumor uptake of 10 mg of naked trastuzumab was compared with that of 50 mg of naked trastuzumab, and the tracer dose was also evaluated (13). In trastuzumabnaive patients, 50 mg of naked trastuzumab was needed for adequate imaging, likely because of the dose-dependent pharmacokinetics of trastuzumab. That study showed the relevance of an adequate naked antibody dose for sufficient accumulation of a radiolabeled antibody in the tumor. Fourth, another study showed the value of adding 89Zr-trastuzumab PET imaging to biopsies for the assessment of intrapatient tumor heterogeneity and for prediction of the treatment outcome in HER2-positive breast cancer patients treated with trastuzumab emtansine (T-DM1) (18). One third of the patients who had HER2-positive breast cancer showed little or no 89Zr-trastuzumab uptake across their metastases and experienced a shorter median time to treatment failure than patients who had more homogeneous HER2-positive PET scan results. These data

illustrated a successful theranostic approach for the assessment of tumor heterogeneity and prediction of the treatment outcome. Finally, 89Zr-trastuzumab PET imaging can provide a functional readout for therapeutics that affect HER2 expression, such as heat shock protein HSP90 inhibitor AUY922 (19). The ongoing IMPACT study is evaluating the clinical utility of 89Zr-trastuzumab PET and 18F-fluoroestradiol PET imaging in 200 newly diagnosed metastatic breast cancer patients (ClinicalTrials.gov identifier NCT01957332).

Another well-known drug target is epidermal growth factor receptor 1, which is targeted by antibodies such as cetuximab and panitumumab. One study demonstrated large differences in tumor 89Zr-cetuximab tracer uptake between intrahepatic tumors and extrahepatic tumors in patients with K-ras wild-type metastatic colorectal cancer (14). Extrahepatic tumor uptake of 89Zr-cetuximab was demonstrated, whereas liver metastases appeared as “cold spots.”

Four of 6 patients with 89Zr-cetuximab uptake in tumor lesions experienced clinical benefit, whereas progressive disease was observed in 3 of 4 patients without 89Zr-cetuximab uptake.

Another study with 89Zr-cetuximab was performed for head and neck squamous cell cancer (20). In both studies, a therapeutic dose of naked cetuximab followed by 89Zr-cetuximab for imaging was used; this protocol may have led to at least partial saturation of the tumor and therefore may have reduced the tumor uptake of 89Zr-cetuximab.

Angiogenesis is a hallmark of cancer and is stimulated by vascular endothelial growth factor A (VEGF-A). Several studies have been performed with anti–VEGF-A antibody 89Zrbevacizumab (Fig. 1B) (21–24). They clearly showed that a drug targeting a growth factor in the microenvironment can be visualized using protein tracer doses as low as 5 mg. For RCCs, 89Zr-bevacizumab PET showed heterogeneous tracer accumulation in tumor lesions (21). Serial 89Zr-bevacizumab PET showed that a therapeutic dose of bevacizumab and interferon-γ reduced tracer uptake. This result suggested that 1 therapeutic dose of the angiogenesis inhibitor reduced access of the antibody to the tumor. A 89Zr-bevacizumab study of advanced non–small cell lung cancer (NSCLC) demonstrated 4-fold-higher tracer uptake in tumor tissue than in nontumor tissue (22). In children with diffuse intrinsic pontine gliomas treated with radiotherapy, heterogeneity of tumor uptake of 89Zr-bevacizumab was shown (23). 89Zrbevacizumab tracer uptake is not limited to malignant disease. In the presence of VEGF-A, benign lesions can also be visualized, as exemplified in patients with von Hippel–Lindau disease (24).

The use of molecular antibody imaging for tumor detection was explored in a large multicenter phase 3 trial in which 14 centers in the United States participated (25). Presurgical 124I-girentuximab PET was compared with CT and histopathologic diagnoses in 195 patients with unclassified renal lesions. Girentuximab targets the membrane protein carbonic anhydrase IX, which is expressed in more than 95% of clear-cell RCCs. 124I-girentuximab PET

had both better sensitivity and better specificity than CT for distinguishing clear-cell RCCs from other renal masses, both benign and malignant. That study showed the possibility of performing a novel molecular imaging study across 14 centers. In a multicenter trial of patients who have metastatic RCCs with good or intermediate prognosis, the value of 89Zr-girentuximab PET combined with 18F-FDG PET is being tested to determine whether it can help in the selection of patients who have relatively indolent disease—for whom the start of treatment can be postponed (ClinicalTrials.gov identifier NCT02228954).

Molecular imaging in immunotherapy

Immune checkpoint inhibitors are immunomodulatory mAbs that block immune checkpoints by targeting CTLA-4, programmed death receptor 1 (PD-1), or programmed death receptor 1 ligand (PD-L1) (Fig. 1C). These drugs show activity across multiple tumor types. The immune checkpoint inhibitors ipilimumab (anti–CTLA-4), nivolumab (anti–PD-1), pembrolizumab (anti–PD-1), and atezolizumab (anti–PD-L1) are FDA- and EMA-approved to treat specific tumor types. However, not all patients benefit from these drugs, and patients may experience major immunity-related toxicities. Moreover, these drugs are extremely expensive. Molecular antibody imaging may provide insight into the immune response and may therefore support better patient and treatment selections.

Five preclinical studies with radiolabeled anti–PD-L1 antibodies showed antibody uptake in PD-L1–overexpressing tumors (26–30). These studies provided data on drug biodistribution and the influence of dose escalation on target saturation in mice. In addition to tumor uptake, high tracer uptake was also observed in organs such as the spleen, thymus, and lymph nodes. These data may reflect the expression of PD-L1 by immune cells, including T cells, dendritic cells, and macrophages.

Three preclinical molecular antibody imaging trials with radiolabeled anti–PD-1 antibodies to visualize T cells in mice and 1 trial in nonhuman primates have been published (29,31,32). All studies showed tracer uptake patterns to be comparable to those of anti–PD-L1 antibodies in healthy mice, with uptake in tumors and secondary lymphoid organs, such as the spleen and lymph nodes.

The first molecular antibody imaging clinical trials with immune checkpoint inhibitors are under way. One study is investigating the 89Zr-labeled anti–PD-L1 antibody atezolizumab in patients with bladder cancer, NSCLC, and triple-negative breast cancer (ClinicalTrials.gov identifier NCT02453984), and another is investigating 89Zr-labeled anti–PD-1 antibody pembrolizumab in patients with melanoma and NSCLC (ClinicalTrials.gov identifier NCT02760225).

Theranostics with antibodies

BiTEs represent a relatively novel approach in immunotherapy (Fig. 1D). These bispecific antibodies consist of 2 linked, single-chain variable fragments directed against a surface target antigen on cancer cells and cluster of differentiation 3ε (CD3ε) on T cells. Simultaneous binding of tumor and T cells mediated tumor-directed T-cell cytotoxicity and cytokine production without the need for costimulatory molecules (33). Blinatumomab, a CD19/CD3ε-directed BiTE, has been approved for the treatment of Philadelphia chromosome–negative relapsed or refractory B-cell precursor acute lymphoblastic leukemia. Two BiTEs have been radiolabeled with 89Zr and studied in mice (34,35). The epithelial cell adhesion molecule–targeting BiTE AMG 110 labeled with 89Zr was used at a 20-μg dose in nude BALB/c mice bearing epithelial cell adhesion molecule–expressing colorectal cancer xenografts (34). The highest 89Zr-AMG 110 uptake was found in the kidneys and then in the liver and tumor. The CEA/CD3ε-directed BiTE AMG 211 radiolabeled with 89Zr showed protein dose–dependent carcinoembryonic antigen (CEA)–specific targeting in mouse tumor xenograft models (35). An ongoing clinical study is exploring the biodistribution of 89Zr-AMG 211 in patients with gastrointestinal adenocarcinomas (ClinicalTrials.gov identifier NCT02760199).

Use of molecular imaging to study antibodies with payload

ADCs are a subclass of antibody-related therapeutics (Fig. 1E). These drugs consist of a tumorspecific mAb conjugated to a cytotoxic payload via a linker. ADCs are designed to improve the potency of chemotherapy by increasing the accumulation of the cytotoxic drug within neoplastic cells, thereby reducing systemic toxic effects. The antibody part of the ADC does not need to exert a therapeutic effect because it serves as an anchor to deliver cytotoxins directly to cancer cells. Brentuximab vedotin and T-DM1 are the standard of care in, respectively, patients with CD30-positive Hodgkin lymphoma or anaplastic large-cell lymphoma and patients with HER2-overexpressing metastatic breast cancer. More than 80 ADCs are or have been in clinical development.

The only molecular imaging study performed with a radiolabeled ADC involved brentuximab vedotin (36). In mice bearing xenograft tumors with various levels of CD30 expression, tumor-to-blood ratios of 15.05 and 0.78 were observed for 89Zr-brentuximab vedotin and 124I-brentuximab vedotin, respectively, 144 h after administration. These data suggested that 89Zr was a more suitable radionuclide for this ADC.

Radiolabeling of ADCs themselves is considered to increase the risk of instability of the molecule. Therefore, radiolabeling of the naked antibody that is part of an ADC for PET imaging is a safe alternative. Naked antibody uptake is assumed to reflect ADC uptake and thus may predict whether a patient will respond to ADC therapy. This approach was used in 3 preclinical trials in mice and 1 study in both mice and nonhuman primates (37–40). Organ biodistribution

and tracer tumor uptake were assessed. In 1 study, 3 doses of 89Zr-labeled naked antibody were evaluated as part of an ADC targeting mesothelin in mice bearing human pancreatic tumor xenografts. Tumor uptake decreased with increasing doses of the naked mAb (38), indicating dose-dependent and saturable tracer distribution at doses of 25 and 100 μg in mice. Biodistribution and tumor uptake were also investigated with an 89Zr-labeled antimesothelin naked antibody in patients subsequently treated with a mesothelin-directed ADC (41). The results showed uptake of the radiolabeled naked antibody in pancreatic and ovarian tumors. Strickland et al. administered a carcinoembryonic cell adhesion molecule 6–directed ADC to monkeys and assessed biodistribution with a 64Cu-labeled anti–carcinoembryonic cell adhesion molecule 6 naked mAb (38). The highest tracer uptake was seen in the bone marrow. Neutropenia and anemia occurred in all animals treated with this ADC, suggesting that tissuespecific toxicity can be predicted by antibody tracer uptake.

In 2 clinical studies, radiolabeled trastuzumab is being evaluated as a biomarker for predicting the response to T-DM1 treatment for HER2-positive metastatic breast cancer. The ZEPHIR trial is designed to prospectively investigate the role of pretreatment 89Zr-trastuzumab PET combined with early response assessment using 18F-FDG PET in the selection of patients who have metastatic HER2-positive tumors and are unlikely to benefit from T-DM1 treatment (18).

An analysis of the data from the first 56 patients showed that negative 89Zr-trastuzumab PET findings and the absence of a response in early 18F-FDG PET resulted in a negative predictive value of 100% for a response according to RECIST 1.1 criteria. Substantial inter- and intrapatient heterogeneity of tracer uptake was observed. Sixteen of 56 HER2-positive patients (29%) had negative 89Zr-trastuzumab PET results, and intrapatient heterogeneity was detected in 46% of patients. The same approach is being used to evaluate whether 64Cu-labeled trastuzumab can predict a response to T-DM1 therapy (ClinicalTrials.gov identifier NCT02226276).

Antibodies can also function as targeted delivery vehicles for radionuclides as part of radioimmunotherapy to selectively kill tumor cells (Fig. 1F). 90Y-ibritumomab tiuxetan has been approved for the treatment of B-cell non-Hodgkin lymphoma. An example of radioimmunotherapy that is being investigated in mice is a 177Lu-labeled anti–CD37 antibody targeting B lymphocytes (42).

Translation of molecular antibody imaging to clinical practice

There are several challenges in translating (pre)clinical antibody imaging studies using theranostics to standardized and, ultimately, daily routine patient care. Knowledge from preclinical models often cannot be extrapolated to humans unconditionally because most antibodies are specific for human targets. In addition, until now, most clinical trials with mAbs or antibody-related therapeutics have been performed in relatively small groups of patients,

with antibodies

precluding firm conclusions regarding clinical relevance. Performing larger studies will require harmonization and standardization of radiolabeling and imaging procedures across centers as well as proper access to the required radionuclide. Larger studies using 89Zr are feasible, as the transport of this nuclide or 89Zr-labeled drugs can be well organized because of the relatively long physical half-life. The availability of 64Cu is more limited by its relatively fast decay.

When multicenter studies are performed, evidence that the final radiolabeled drug products and manufacturing processes are comparable should be provided for all steps in the manufacturing process that are conducted at more than 1 center. Fortunately, accessing templates for routine documentation, such as Investigational Medicinal Product Dossiers, is becoming easier for mAbs or antibody-related tracers (43,44).

We identified 46 medical centers—24 in the United States, 18 in Europe, 3 in Asia, and 1 in Australia—that recently participated in clinical trials with antibodies or antibody-based PET theranostics. 89Zr is by far the most frequently used positron-emitting nuclide for antibody labeling. It is encouraging that of the 24 antibodies or antibody-related therapeutics that have been labeled with several PET radionuclides and investigated as theranostics in patients, 11 were investigated in the multicenter setting.

Finally, the integration of antibody PET imaging in clinical practice is costly. For instance, mAb labeling and a series of PET scans in 1 patient cost several thousand U.S. dollars. However, when proven valuable for making clinical decisions based on whole-body information obtained with molecular antibody imaging, a theranostic approach may prevent expensive treatment of patients who will not benefit from therapy because of a lack of target expression or drug uptake and may therefore lead to fewer side effects and better outcomes.

Conclusion

Theranostics with antibodies and antibody-related therapeutics can provide meaningful in vivo insights about the biodistribution and tumor uptake of radiolabeled drugs. This approach is currently being investigated extensively across numerous centers. Properly powered studies are required to prove that theranostics can play an important role in drug development and become a valuable tool in the selection of patients for antibody-based therapies.

Disclosure

Elisabeth G.E. de Vries has a consultation/advisory role with Medivation, Merck, and Synthon. Mathilde Jalving has an advisory role with Merck. Research funding was provided by Amgen, Genentech/Roche, Chugai, Servier, Novartis, Synthon, AstraZeneca, and Radius Health.

Elisabeth G.E. de Vries received support from an IMPACT grant and RUG 2016-10034 from the Dutch Cancer Society, IMI grant TRISTAN, and ERC advanced grant OnQview. No other potential conflict of interest relevant to this article was reported.

Acknowledgments

We thank Anouk Funke and Jan Pruim for their assistance in figure design.

References

1. Loibl S, Gianni L. HER2-positive breast cancer. Lancet. December 6, 2016 [Epub ahead of print].

2. Ribas A, Puzanov I, Dummer R, et al. Pembrolizumab versus investigator-choice chemotherapy for ipilimumab-refractory melanoma (KEYNOTE-002): a randomized, controlled, phase 2 trial. Lancet Oncol. 2015;16:908–918.

3. El-Osta H, Shahid K, Mills GM, Peddi P. Immune checkpoint inhibitors: the new frontier in non-small-cell lung cancer treatment. Onco Targets Ther. 2016;9: 5101–5116.

4. Eggermont AM, Chiarion-Sileni V, Grob JJ, et al. Prolonged survival in stage III melanoma with ipilimumab adjuvant therapy. N Engl J Med. 2016;375: 1845–1855.

5. Perez EA, Romond EH, Suman VJ, et al. Trastuzumab plus adjuvant chemotherapy for human epidermal growth factor receptor 2-positive breast cancer: planned joint analysis of overall survival from NSABP B-31 and NCCTG N9831. J Clin Oncol. 2014;32:3744–3752.

6. Williams SP. Tissue distribution studies of protein therapeutics using molecular probes: molecular imaging. AAPS J. 2012;14:389–399.

7. van Dongen GA, Visser GW, Lub-de Hooge MN, de Vries EG, Perk LR. Immuno-PET: a navigator in monoclonal antibody development and applications. Oncologist. 2007;12:1379–1389.

8. van Schaijk FG, Broekema M, Oosterwijk E, et al. Residualizing iodine markedly improved tumor targeting using bispecific antibody-based pretargeting. J Nucl Med. 2005;46:1016–1022.

9. Verel I, Visser GW, Boellaard R, Stigter-van Walsum M, Snow GB, van Dongen GA. 89Zr immuno-PET: comprehensive procedures for the production of 89Zr-labeled monoclonal antibodies. J Nucl Med. 2003;44:1271–1281.

10. Ciprotti M, Tebbutt NC, Lee FT, et al. Phase 1 imaging and pharmacodynamic trial of CS-1008 in patients with metastatic colorectal cancer. J Clin Oncol. 2015;33:2609–2616.

11. de Korte MA, de Vries EG, Lub-de Hooge MN, et al. 111Indium-trastuzumab visualises myocardial human epidermal growth factor receptor 2 expression shortly after anthracycline treatment but not during heart failure: a clue to uncover the mechanisms of trastuzumab-related cardiotoxicity. Eur J Cancer. 2007;43:2046–2051.

12. Lamberts LE, Williams SP, Terwisscha van Scheltinga AGT, et al. Antibody positron emission tomography imaging in anticancer drug development. J Clin Oncol. 2015;33:1491–1504.

13. Dijkers EC, Oude Munnink TH, Kosterink JG, et al. Biodistribution of 89Zr-trastuzumab and PET imaging of HER2-positive lesions in patients with metastatic breast cancer. Clin Pharmacol Ther. 2010;87:586–592.

Theranostics with antibodies

14. Menke-van der Houven van Oordt CW, Gootjes EC, Huisman MC, et al. 89Zr-cetuximab PET imaging in patients with advanced colorectal cancer. Oncotarget. 2015;6:30384–30393.

15. Tibben JG, Boerman OC, Massuger LF, Schijf CP, Claessens RA, Corstens FH. Pharmacokinetics, biodistribution and biological effects of intravenously administered bispecific monoclonal antibody OC/TR F(ab')2 in ovarian carcinoma patients. Int J Cancer. 1996;66:477–483.

16. Perik PJ, Lub-De Hooge MN, Gietema JA, et al. Indium-111–labeled trastuzumab scintigraphy in patients with human epidermal growth factor receptor 2–positive metastatic breast cancer. J Clin Oncol. 2006;24:2276–2282.

17. Gaykema SBM, de Jong JR, Perik PJ, et al. 111In-trastuzumab scintigraphy in HER2-positive metastatic breast cancer patients remains feasible during trastuzumab treatment. Mol Imaging. 2014;13:1–6.

18. Gebhart G, Lamberts LE, Wimana Z, et al. Molecular imaging as a tool to investigate heterogeneity of advanced HER2-positive breast cancer and to predict patient outcome under trastuzumab emtansine (T-DM1): the ZEPHIR trial. Ann Oncol. 2016;27:619–624.

19. Gaykema SBM, Schroder CP, Vitfell-Rasmussen J, et al. 89Zr-trastuzumab and 89Zr-bevacizumab PET to evaluate the effect of the HSP90 inhibitor NVP-AUY922 in metastatic breast cancer patients. Clin Cancer Res. 2014;20:3945–3954.

20. Even AJ, Hamming-Vrieze O, van Elmpt W, et al. Quantitative assessment of zirconium-89 labeled cetuximab using PET/CT imaging in patients with advanced head and neck cancer: a theragnostic approach. Oncotarget. 2017;8: 3870–3880.

21. Oosting SF, Brouwers AH, van Es SC, et al. 89Zr-bevacizumab PET visualizes heterogeneous tracer accumulation in tumor lesions of renal cell carcinoma patients and differential effects of antiangiogenic treatment. J Nucl Med. 2015;56:63–69.

22. Bahce I, Huisman MC, Verwer EE, et al. Pilot study of 89Zr-bevacizumab positron emission tomography in patients with advanced non-small cell lung cancer. EJNMMI Res. 2014;4:35

23. Jansen MH, Veldhuijzen van Zanten SEM, van Vuurden DG, et al. Molecular drug imaging: 89Zrbevacizumab PET in children with diffuse intrinsic pontine glioma. J Nucl Med. 2017;58:711–716.

24. Oosting SF, van Asselt SJ, Brouwers AH, et al. 89Zr-bevacizumab PET visualizes disease manifestations in patients with von Hippel–Lindau disease. J Nucl Med. 2016;57:1244–1250.

25. Divgi CR, Uzzo RG, Gatsonis C, et al. Positron emission tomography/computed tomography identification of clear cell renal cell carcinoma: results from the REDECT trial. J Clin Oncol. 2013;31:187–194.

26. Heskamp S, Hobo W, Molkenboer-Kuenen JD, et al. Noninvasive imaging of tumor PD-L1 expression using radiolabeled anti-PD-L1 antibodies. Cancer Res. 2015;75:2928–2936.

27. Chatterjee S, Lesniak WG, Gabrielson M, et al. A humanized antibody for imaging immune checkpoint ligand PD-L1 expression in tumors. Oncotarget. 2016;7:10215–10227.

28. Lesniak WG, Chatterjee S, Gabrielson M, et al. PD-L1 detection in tumors using [64Cu]atezolizumab with PET. Bioconjug Chem. 2016;27:2103–2110.

29. Hettich M, Braun F, Bartholoma MD, Schirmbeck R, Niedermann G. Highresolution PET imaging with therapeutic antibody-based PD-1/PD-L1 checkpoint tracers. Theranostics. 2016;6:1629–1640.

Theranostics with antibodies

30. Josefsson A, Nedrow JR, Park S, et al. Imaging, biodistribution, and dosimetry of radionuclide-labeled PD-L1 antibody in an immunocompetent mouse model of breast cancer. Cancer Res. 2016;76:472–479.

31. Cole E, Kim J, Donnelly D, et al. Radiosynthesis and preclinical PET evaluation in healthy non-human primates of 89Zr-nivolumab [abstract]. J Nucl Med. 2016;57(suppl 2):1189.

32. England CG, Ehlerding EB, Hernandez R, et al. Preclinical pharmacokinetics and biodistribution studies of 89Zr-labeled pembrolizumab. J Nucl Med. 2017;58:162–168.

33. Brischwein K, Parr L, Pflanz S, et al. Strictly target cell-dependent activation of T cells by bispecific singlechain antibody constructs of the BiTE class. J Immunother. 2007;30:798–807

34. Warnders FJ, Waaijer SJ, Pool M, et al. Biodistribution and PET imaging of labeled bispecific T cell–engaging antibody targeting EpCAM. J Nucl Med. 2016;57:812–817.

35. Waaijer SJH, Warnders FJ, Lub-de Hooge MN, et al. Preclinical evaluation of the radiolabeled bispecific T-cell engager 89Zr-AMG 211 targeting CEA-positive tumors [abstract]. Mol Cancer Ther. 2015;14(12 suppl 2):A85.

36. Moss A, Gudan J, Albertson T, Whiting N, Law C. Preclinical microPET/CT imaging of 89Zr-Df-SGN-35 in mice bearing xenografted CD30 expressing and non-expressing tumors [abstract]. Cancer Res. 2014;74(19 suppl):104.

37. ter Weele EJ, Terwisscha van Scheltinga AGT, Kosterink JGK, et al. Imaging the distribution of an antibody-drug conjugate constituent targeting mesothelin with 89Zr and IRDye 800CW in mice bearing human pancreatic tumor xenografts. Oncotarget. 2015;6:42081–42090.

38. Strickland LA, Ross J, Williams S, et al. Preclinical evaluation of carcinoembryonic cell adhesion molecule (CEACAM) 6 as potential therapy target for pancreatic adenocarcinoma. J Pathol. 2009;218:380–390.

39. Ilovich O, Natarajan A, Hori S, et al. Development and validation of an immuno-PET tracer as a companion diagnostic agent for antibody-drug conjugate therapy to target the CA6 epitope. Radiology. 2015;276:191–198.

40. Rylova SN, Del Pozzo L, Klingeberg C, et al. Immuno-PET imaging of CD30-positive lymphoma using 89Zrdesferrioxamine-labeled CD30-specific AC-10 antibody. J Nucl Med. 2016;57:96–102.

41. Lamberts LE, Menke-van der Houven van Oordt CW, ter Weele EJ, et al. Immuno-PET with anti-mesothelin antibody in patients with pancreatic and ovarian cancer before anti-mesothelin antibody-drug conjugate treatment. Clin Cancer Res. 2016;22:1642–1652.

42. Repetto-Llamazares AH, Larsen RH, Giusti AM, et al. 177Lu-DOTA-HH1, a novel anti-CD37 radioimmunoconjugate: a study of toxicity in nude mice. PLoS One. 2014;9:e103070.

43. IND regulatory & manufacturing resources. National Cancer Institute website. https://imaging.cancer. gov/programs_resources/IND_regulatory_manufacturing.htm. Updated October 28, 2016. Accessed May 15, 2017.

44. Todde S, Windhorst AD, Behe M, et al. EANM guideline for the preparation of an Investigational Medicinal Product Dossier (IMPD). Eur J Nucl Med Mol Imaging. 2014;41:2175–2185.

89Zr-PET imaging to predict tumor uptake of 177Lu-NNV003 anti-CD37 radioimmunotherapy in mouse models of B cell lymphoma

Danique Giesen 1, Marjolijn N Lub-de Hooge 2, 3, Marcel Nijland 4, Helen Heyerdahl 5 , Jostein Dahle 5, Elisabeth G E de Vries 1, Martin Pool 1, 6

1 Department of Medical Oncology, University Medical Center Groningen, P.O. Box 30.001, 9700 RB, Groningen, The Netherlands. d.giesen@umcg.nl; 2 Department of Clinical Pharmacy and Pharmacology, 3 Department of Nuclear Medicine and Molecular Imaging, and 4 Department of Hematology, University Medical Center Groningen, Groningen, The Netherlands; 5 Nordic Nanovector ASA, Oslo, Norway; 6 Department of Clinical Pharmacy and Toxicology, Leiden University Medical Center, Leiden, The Netherlands. Sci Rep. 2022;12(1):6286.

Abstract

[177Lu]Lu-DOTA-NNV003, a radioimmunoconjugate targeting CD37, is developed as novel radioimmunotherapy (RIT) treatment for patients with B cell non-Hodgkin’s lymphoma (NHL).

Since patients are at risk for developing hematological toxicities due to CD37 expression on normal B cells, we aimed to develop 89Zr-labeled NNV003 for positron emission tomography (PET) imaging as a surrogate tool to predict [177Lu]Lu-DOTA-NNV003 RIT whole-body distribution and tumor uptake. NNV003 antibody was first radiolabeled with 89Zr. [89Zr]Zr-N-sucDf-NNV003 tumor uptake was evaluated by PET imaging of mice bearing human CD37-expressing REC1 B cell NHL or RAMOS Burkitt’s lymphoma xenograft tumors followed by ex vivo analysis. Finally, CD37-targeting of [89Zr]Zr-N-sucDf-NNV003 and [177Lu]Lu-DOTA-NNV003 RIT were compared. [89Zr]Zr-N-sucDf-NNV003 accumulated in REC1 tumors over time, which was not observed for non-specific, 111In-labeled IgG control molecule. In RAMOS tumor-bearing mice, [89Zr]ZrN-sucDf-NNV003 tumor uptake was higher than [111In]In-DTPA-IgG at all tested tracer protein doses (10 μg, 25 μg and 100 μg; P < 0.01), further confirming [89Zr]Zr-N-sucDf-NNV003 tumor uptake is CD37-mediated. [89Zr]Zr-N-sucDf-NNV003 and [177Lu]Lu-DOTA-NNV003 RIT showed similar ex vivo biodistribution and tumor uptake in the RAMOS tumor model. In conclusion, [89Zr]Zr-N-sucDf-NNV003 PET imaging can serve to accurately predict CD37-targeting of [177Lu] Lu-DOTA-NNV003. To enable clinical implementation, we established a good manufacturing practice (GMP)-compliant production process for [89Zr]Zr-N-sucDf-NNV003.

Introduction

Novel treatment options for patients with B cell non-Hodgkin’s lymphoma (NHL) are warranted, especially for those with poor prognosis, since large subgroups of these patients become refractory after initial CD20-based immunotherapy. Therapies directed against other targets expressed by B cells, for example leukocyte antigen CD37, may provide a therapeutic alternative. CD37-directed radioimmunotherapy (RIT) with the fully murine antibody lutetium-177 (177Lu; T1/2: 6.65 d)-lilotomab satetraxetan, was recently evaluated in a phase 1/2a study in relapsed/ refractory indolent NHL patients and showed encouraging results (1–3). Instead of a murine antibody, next-generation [177Lu]Lu-DOTA-NNV003 consists of a chimeric mouse-human antiCD37 antibody (NNV003) conjugated with p-SCN-Bn-DOTA that chelates the β -emitting radionuclide 177Lu (4).

CD37 is a highly glycosylated transmembrane protein selectively expressed by normal B cells and the majority of B cell lymphomas and is specifically of interest for RIT, since CD37 receptor-antibody complexes are highly internalized in tumor cells (5,6). In combination with 177Lu’s favorable physical properties for RIT, [177Lu]Lu-DOTA-NNV003 has a potentially improved toxicity profile. Compared to historically used radionuclides for RIT in B cell malignancies

such as yttrium-90 (90Y; T1/2: 2.66 d), 177Lu has a relatively short β -range (7). Also, 177Lu becomes intracellularly trapped in lysosomes through residualization, whereas iodine-131 (131I; T1/2: 8.02 d) diffuses passively out of cells after catabolization (8). This may result in enhanced tumor irradiation, while sparing surrounding healthy tissues.

RIT in B cell malignancies is generally effective in only a subset of patients, while often resulting in on-target, off-tumor toxicities due to target expression on normal B cells (9). Insight in CD37 expression may help to select patients who are more likely to respond or are at risk for developing CD37-induced hematological toxicities. Therefore, RIT could benefit from molecular imaging as a non-invasive approach to provide whole-body information on target presence and RIT distribution. Dosimetry for organs-at-risk in 177Lu radionuclide therapy is often calculated using single photon emission computed tomography (SPECT)/computed tomography (CT) imaging. Still, this procedure has limited sensitivity and quantitative evaluation of targetmediated uptake can be challenging (10,11).

We aimed to develop a surrogate image tool for the assessment of [177Lu]Lu-DOTA-NNV003 RIT whole-body distribution and tumor uptake, which could aid its clinical development and

FIGURE 1: Schematic overview of [89Zr]Zr-N-sucDf-NNV003 anti-CD37 PET imaging in mouse models of B cell lymphoma and its potential implementations. We aimed to develop [89Zr]Zr-N-sucDf-NNV003 as a surrogate image tool for the assessment of [177Lu]LuDOTA-NNV003 RIT whole-body distribution and tumor uptake. [89Zr]Zr-N-sucDf-NNV003 pre-therapy PET imaging in patients with B cell NHL may help to identify those more likely to respond. Furthermore, as a surrogate imaging tool, [89Zr]Zr-N-sucDf-NNV003 may aid in optimizing [177Lu]Lu-DOTA-NNV003 RIT doseregimens and post-therapy response evaluation.

89Zr-PET imaging to predict tumor uptake of 177Lu-NNV003 radioimmunotherapy

use (Fig. 1). The positron emission tomography (PET) radioisotope zirconium-89 (89Zr) is particularly suited for antibody imaging, as it yields high sensitivity and accurate quantification, while its physical half-life of 3.27 d matches the time that antibodies need for tumor accumulation. Studies in patients and mice showed the utility of 89Zr-PET imaging to predict the distribution of therapeutic radionuclides such as 177Lu and 90Y coupled to antibodies (12–14).

To assess whether [89Zr]Zr-N-sucDf-NNV003 PET imaging can serve to predict [177Lu]Lu-DOTANNV003 RIT biodistribution, we evaluated their whole-body distribution and tumor-targeting properties in mice bearing human B cell lymphomas. Furthermore, a good manufacturing practice (GMP)-compliant production process was established for [89Zr]Zr-N-sucDf-NNV003 to enable administration to patients.

Results

Development and quality control of 89Zr-labeled NNV003 For in vivo studies, NNV003 was conjugated to tetrafluorphenol-N-succinyldesferal (TFP-NsucDf) followed by radiolabeling with 89Zr. NNV003 was incubated with increasing molar ratios of TFP-N-sucDf, which resulted in approximately 60% conjugation efficiency (Supplementary Fig. S1A). NNV003-N-sucDf intermediate product was obtained with a mean yield of 54% (Supplementary Fig. S1B). To enable 89Zr-chelation by TFP-N-sucDf, protective Fe(III) must be removed from the hydroxamate groups. At least 40% Fe(III) was removed from NNV003-NsucDf by incubation with EDTA (Supplementary Fig. S1C). The NNV003:TFP-N-sucDf ratio did not impair immunoreactivity of NNV003-N-sucDf to CD37-expressing RAMOS cells, as the mean immunoreactive fractions (IRF) were similar for all tested ratios and all higher than 0.8 (Supplementary Fig. S1D). NNV003-N-sucDf intermediate product was obtained with ≥ 95% purity and no aggregation or fragmentation was observed (Supplementary Fig. S1E). NNV003 conjugated to ~ 1.2 TFP-N-sucDf chelators per antibody consistently bound 500 MBq 89Zr per mg (radiochemical purity; RCP ≥ 95%) with preserved immunoreactivity and is therefore considered most optimal for in vivo studies (Supplementary Fig. S1F).

[89Zr]Zr-N-sucDf-NNV003 PET imaging and biodistribution in REC1 tumor-bearing mice

[89Zr]Zr-N-sucDf-NNV003 whole-body distribution and tumor-targeting were evaluated in BALB/c nude mice bearing CD37-expressing human REC1 B cell NHL xenograft tumors. PET imaging revealed [89Zr]Zr-N-sucDf-NNV003 tumor uptake increased between day 1 and day 5 pi, whereas blood pool activity decreased (Fig. 2A). PET quantification showed the highest tumor uptake with mean standardized uptake value (SUVmean) of 2.1 (± 0.9) at both 3 and 5 days pi and highest tumor-to-blood ratio of 1.7 (± 0.7) at day 5 pi (Fig. 2B).

Ex vivo biodistribution in healthy tissues was similar for [89Zr]Zr-N-sucDf-NNV003 and non-

FIGURE 2: [89Zr]Zr-N-sucDf-NNV003 PET imaging and ex vivo biodistribution in REC1 xenografted mice. (A) Representative coronal PET images of 10 μg [89Zr]Zr-N-sucDf-NNV003 biodistribution in REC1 tumorbearing mice at 1, 3 and 5 days (d) pi. [89Zr]Zr-N-sucDf-NNV003 uptake is presented as standardized uptake value (SUV). The dashed circle indicates REC1 tumor. (B) Quantification of [89Zr]Zr-N-sucDf-NNV003 uptake in REC1 tumor and blood pool activity at 1, 3 and 5 days pi. [89Zr]Zr-N-sucDf-NNV003 uptake is presented as mean standardized uptake value (SUVmean). TBR indicates tumor-to-blood ratio at day 5 pi. (C) Ex vivo biodistribution results of 10 μg [89Zr]Zr-N-sucDf-NNV003 versus [111In]In-DTPA-IgG control in REC1 tumorbearing mice at 5 days pi. Tracer uptake per organ is presented as percentage of injected radioactivity dose per gram tissue (%ID/g). (D) Left: Ex vivo REC1 tumor uptake of 10 μg [89Zr]Zr-N-sucDf-NNV003 versus [111In]In-DTPA-IgG control at day 5 pi. Tumor uptake of tracer is presented as %ID/g. Right: Ex vivo tumor-toblood ratio of 10 μg [89Zr]Zr-N-sucDf-NNV003 versus [111In]In-DTPA-IgG control in REC1 tumor-bearing mice at day 5 pi. Data in (B–D) is shown as mean ± standard deviation (SD). **P < 0.01, *P < 0.05, ns not significant.

specific, indium-111 (111In; T1/2: 2.80 d)-labeled IgG control molecule, however [89Zr]Zr-N-sucDfNNV003 demonstrated lower activity in the blood pool, kidneys and colon, whereas uptake in bone was higher compared to [111In]In-DTPA-IgG at day 5 pi (Fig. 2C). Bone uptake is common for 89Zr-labeled molecules, as they are metabolized in vivo and unbound 89Zr has a high affinity for hydroxyapatite (15). Levels similar to the [89Zr]Zr-N-sucDf-NNV003 uptake in bone were found in mouse models for other 89Zr-labeled antibodies (16,17). For this reason, potential instability of [89Zr]Zr-N-sucDf-NNV003 linker chelation is considered to be limited. Ex vivo tumor

89Zr-PET imaging to predict tumor uptake of 177Lu-NNV003 radioimmunotherapy

uptake and tumor-to-blood ratio were higher for [89Zr]Zr-N-sucDf-NNV003 compared to [111In]In-DTPA-IgG (10.9 vs. 3.9%ID/g; P < 0.05 and 2.4 vs. 0.5; P < 0.01) (Fig. 2D), indicating [89Zr]ZrN-sucDf-NNV003 tumor uptake is CD37-mediated. Furthermore, [111In]In-DTPA-IgG tumor-toblood ratio was lower than 1, meaning tumor uptake is not target-mediated. [111In]In-DTPA-IgG can therefore be considered a suitable control molecule.

Dose- and CD37-dependent uptake in RAMOS tumor-bearing mice

To study CD37-mediated tumor uptake, we evaluated [89Zr]Zr-N-sucDf-NNV003 biodistribution at three total protein dose levels in human RAMOS Burkitt’s lymphoma tumor-bearing mice.

A radiolabeled antibody dose of 10 μg [89Zr]Zr-N-sucDf-NNV003 or [111In]In-DTPA-IgG was supplemented with 0, 15 or 90 μg of unlabeled NNV003 or IgG to obtain total protein doses of 10 μg, 25 μg and 100 μg for each tracer. PET quantification showed [89Zr]Zr-N-sucDf-NNV003 accumulation in CD37-expressing RAMOS tumors for all tested protein doses, with the highest tumor-to-blood ratios found for day 5 pi (Fig. 3A). Comparable tumor uptake was observed for 10, 25 and 100 μg [89Zr]Zr-N-sucDf-NNV003, with SUVmean of 1.8 (± 0.3), 1.7 (± 0.1) and 1.6 (± 0.2) respectively at day 5 pi (Fig. 3B). The highest tumor-to-blood ratio (1.7 ± 0.3) was observed for 10 μg [89Zr]Zr-N-sucDf-NNV003 at day 5 pi.

No relevant differences in ex vivo biodistribution of [89Zr]Zr-N-sucDf-NNV003 and [111In]In-DTPAIgG were observed, except for uptake in bone and tumor (Fig. 3C). [89Zr]Zr-N-sucDf-NNV003 bone uptake demonstrated some variation between 10 μg, 25 μg and 100 μg dose groups, potentially due to slight differences in radiochemical purity, which varied from 95 to 98%. Also, [89Zr]Zr-N-sucDf-NNV003 distribution to healthy tissues was not affected by the addition of unlabeled antibody dose. Ex vivo [89Zr]Zr-N-sucDf-NNV003 tumor uptake was higher compared to [111In]In-DTPA-IgG for all protein dose groups, confirming that [89Zr]Zr-N-sucDf-NNV003 tumor uptake is CD37-mediated (Fig. 3D). Furthermore, tumor uptake was comparable for 10, 25 and 100 μg [89Zr]Zr-N-sucDf-NNV003, with 10.6 (± 2.7), 12.4 (± 3.4) and 10.1 (± 2.1) %ID/g respectively, indicating CD37 is not saturated in tumor cells by addition of unlabeled antibody at these protein doses. The highest [89Zr]Zr-N-sucDf-NNV003 tumor-to-blood ratio of 1.9 (± 0.3) and 1.9 (± 0.5) was observed for the 10 and 25 μg total protein doses, compared to 1.3 (± 0.2) found for 100 μg [89Zr]Zr-N-sucDf-NNV003, suggesting CD37 in RAMOS tumors may be saturated by [89Zr]Zr-N-sucDf-NNV003 at higher protein doses (Fig. 3E). [89Zr]Zr-N-sucDf-NNV003 tumor uptake was the highest in the 25 μg dose group compared to [111In]In-DTPA-IgG (12.4 vs. 3.3%ID/g; P < 0.01).

[89Zr]Zr-N-sucDf-NNV003 tumor uptake was comparable for REC1 and RAMOS tumors, with both tumor types expressing similar levels of CD37 as measured by flow cytometry and immunohistochemistry (Fig. 4).

FIGURE 3: [89Zr]Zr-N-sucDf-NNV003 PET imaging and ex vivo biodistribution at increasing protein dose in RAMOS xenografted mice.

(A) Representative coronal PET images of 10, 25 and 100 μg [89Zr]Zr-N-sucDf-NNV003 biodistribution in RAMOS tumor-bearing mice at 1, 3 and 5 days (d) pi. [89Zr]Zr-N-sucDf-NNV003 uptake is presented as SUV. RAMOS tumor is indicated by a white, dashed circle. (B) Quantification of [89Zr]Zr-N-sucDf-NNV003 uptake in RAMOS tumor and blood pool activity for 10, 25 and 100 μg protein doses at 1, 3 and 5 days pi. [89Zr]Zr-NsucDf-NNV003 uptake is presented as SUVmean. TBR indicates tumor-to-blood ratio at day 5 pi. (C) Ex vivo biodistribution results of [89Zr]Zr-N-sucDf-NNV003 versus [111In]In-DTPA-IgG control dose-escalation in RAMOS tumor-bearing mice at 5 days pi. Tracer uptake per organ is presented as %ID/g. (D) Ex vivo RAMOS uptake of [89Zr]Zr-N-sucDf-NNV003 versus [111In]In-DTPA-IgG control for 10 μg, 25 μg and 100 μg protein doses at 5 days pi. Tumor uptake is presented as %ID/g. (E) Ex vivo tumor-to-blood ratios of [89Zr]Zr-N-sucDf-NNV003 versus [111In]In-DTPA-IgG control in RAMOS tumor bearing mice for 10 μg, 25 μg and 100 μg protein doses at 5 days pi. Data in (B–E) is shown as mean ± SD. **P < 0.01, *P < 0.05, ns not significant.

89Zr-PET imaging to predict tumor uptake of 177Lu-NNV003 radioimmunotherapy

FIGURE 4: CD37 expression in RAMOS and REC1 tumors.

(A) In vitro CD37 expression in REC1 and RAMOS tumors determined by flow cytometry. CD37 expression is presented as mean fluorescent intensity (MFI). HCC827 lung adenocarcinoma cells were used as negative control. Data is shown as mean ± SD. (B) Hematoxylin and eosin staining and CD37 immunohistochemistry on formalin-fixed, paraffin-embedded REC1 and RAMOS tumor tissue sections. Representative tumors are shown.

Lastly, we compared tumor uptake determined by PET quantification with ex vivo tumor uptake to confirm if [89Zr]Zr-N-sucDf-NNV003 PET imaging accurately visualizes tumor uptake. Ex vivo [89Zr]Zr-N-sucDf-NNV003 tumor uptake expressed as SUVmean correlated with PET-derived tumor SUVmean, but not SUVmax, which can be explained by the fact that SUVmax represents the highest voxel and may not reflect whole-tumor uptake (Fig. 5).

FIGURE 5: Correlation between in vivo (PET) and ex vivo (gammacounter) [89Zr]Zr-N-sucDf-NNV003 tumor uptake.

(A) In vivo [89Zr]Zr-N-sucDf-NNV003 tumor uptake expressed as SUVmean and SUV max versus ex vivo tumor uptake. Ex vivo SUV was calculated by correcting %ID/g for injected dose and body weight. Data is shown as mean ± SD. (B) Correlation between in vivo SUV mean and (C) SUV max versus ex vivo tumor uptake. Data is shown as mean with fitted regression curve ± 95% confidence interval (CI).

Comparison of [89Zr]Zr-N-sucDf-NNV003 and [177Lu]Lu-DOTA-NNV003 RIT biodistribution in the RAMOS tumor model

Next, we evaluated [89Zr]Zr-N-sucDf-NNV003 as a surrogate for [177Lu]Lu-DOTA-NNV003 RIT whole-body distribution in RAMOS tumor-bearing mice. Ex vivo tissue analysis showed decreasing [177Lu]Lu-DOTA-NNV003 activity in the blood pool between 1 h to 3 days pi, which coincided with decreasing uptake in well-perfused organs such as heart, lung and kidney (Fig. 6A). [177Lu]Lu-DOTA-NNV003 uptake in RAMOS tumors increased over time with the highest uptake of 13.6 ± 8.3%ID/g found at 3 days pi. Blood pool activity of [89Zr]Zr-N-sucDf-NNV003 at 5 days pi was comparable to [177Lu]Lu-DOTA-NNV003 at 3 days pi (5.7 ± 1.7%ID/g versus 7.6 ± 5.0%ID/g), validating a relevant comparison of both tracers’ ex vivo biodistribution results.

Comparison of ex vivo biodistribution results revealed highest uptake of [89Zr]Zr-N-sucDfNNV003 and [177Lu]Lu-DOTA-NNV003 in tumor compared to normal tissues (Fig. 6B). Also, similar uptake was found for [89Zr]Zr-N-sucDf-NNV003 and [177Lu]Lu-DOTA-NNV003 in RAMOS tumors (10.6 vs. 13.6%ID/g; P = 0.48), indicating tumor-targeting is not affected by chelator or radioisotope (TFP-N-sucDf for 89Zr versus p-SCN-Bn-DOTA for 177Lu). Furthermore, [177Lu]LuDOTA-NNV003 and [89Zr]Zr-N-sucDf-NNV003 showed similar biodistribution in healthy tissues, except for bone and spleen. Bone uptake was higher for [89Zr]Zr-N-sucDf-NNV003, which is common for 89Zr-labeled antibodies, as discussed previously (15). In patients, tracer uptake in red bone marrow can be clearly separated from bone uptake on PET. Therefore, toxicity from myelosuppression can be potentially recognized by high [89Zr]Zr-N-sucDf-NNV003 uptake in red bone marrow.

Higher spleen uptake was observed for [177Lu]Lu-DOTA-NNV003 compared to [89Zr]Zr-N-sucDfNNV003 (5.6 vs. 2.1%ID/g; P < 0.05). Murine CD37 is expected to be expressed on B cells present in the spleen, however NNV003 is not cross-reactive with murine CD37. Furthermore, [89Zr]ZrN-sucDf-NNV003 spleen uptake was similar to [111In]In-DTPA-IgG, indicating this uptake is not CD37-mediated. In mice, the spleen plays an important role in antibody pharmacokinetics due to its high blood flow and loose capillaries, but also due to the expression of Fcγ receptors. Therefore, differences in tracer spleen uptake are easily induced by slight differences in the immune status of mice (18).

Clinical-grade [89Zr]Zr-N-sucDf-NNV003 for patient studies

To enable PET imaging in patients, we developed and characterized the production of clinicalgrade [89Zr]Zr-N-sucDf-NNV003 (Supplementary Fig. S2). Quality control for three individual batches of NNV003-N-sucDf intermediate product and [89Zr]Zr-N-sucDf-NNV003 final product was according to specifications, indicating a robust manufacturing process (Supplementary Table S1). NNV003-N-sucDf intermediate product demonstrated stability up to 6 months

89Zr-PET imaging to predict tumor uptake of 177Lu-NNV003 radioimmunotherapy

FIGURE 6: [177Lu]Lu-DOTA-NNV003 biodistribution in RAMOS xenografted mice and comparison with [89Zr] Zr-N-sucDf-NNV003.

(A) Ex vivo biodistribution results of [177Lu]Lu-DOTA-NNV003 in RAMOS-tumor bearing mice at 1 h (h), 6 h, 1 day (d) and 3 days pi. Tracer uptake per organ is presented as %ID/g. Left: [177Lu]Lu-DOTA-NNV003 uptake in healthy tissues. Right: [177Lu]Lu-DOTA-NNV003 uptake in tumor. (B) [89Zr]Zr-N-sucDf-NNV003 (500 MBq/ mg) and [177Lu]Lu-DOTA-NNV003 (50–90 MBq/mg) ex vivo biodistribution results at 5 days pi and 3 days pi respectively. Uptake is presented as %ID/g. Data is shown as mean ± SD. **P < 0.01, *P < 0.05, ns not significant.

(Supplementary Table S2). Therefore, NNV003-N-sucDf shelf-life is currently set at 6 months, and may be extended if future stability time points remain within specifications. Furthermore, [89Zr]Zr-N-sucDf-NNV003 final product was stable up to 96 h when stored at 2–8 °C and up to 4 h when prepared in the syringe (Supplementary Table S3).

Discussion

Our study demonstrates the potential of using [89Zr]Zr-N-sucDf-NNV003 PET imaging to predict whole-body distribution and tumor uptake of [177Lu]Lu-DOTA-NNV003 RIT. We showed that [89Zr]Zr-N-sucDf-NNV003 accumulated in REC1 and RAMOS tumor tissues in a CD37-dependent manner, resulting in high tumor-to-blood ratios. Furthermore, [89Zr]Zr-N-sucDf-NNV003 biodistribution and tumor-targeting were similar to [177Lu]Lu-DOTA-NNV003 RIT.

177Lu-lilotomab satetraxetan, the murine version of [177Lu]Lu-DOTA-NNV003, is currently in phase II for patients with anti-CD20 refractory follicular lymphoma and has previously been studied in phase I/IIa clinical trials in patients with relapsed, CD37-positive, indolent and aggressive

NHL (ClinicalTrials.gov Identifier: NCT01796171, NCT02658968) (3). These studies showed high RIT uptake in tumors, but also in red bone marrow, liver, spleen, and kidneys (9,10). This may be explained by CD37 expression on mature, normal B cells in these tissues. Predosing with unlabeled lilotomab significantly reduced the absorbed radiation dose in healthy CD37expressing tissues due to 177Lu-lilotomab satetraxetan (9). Also, 177Lu-lilotomab satetraxetan dosimetry, biodistribution, and tumor targeting were improved by lilotomab predosing compared with rituximab predosing or no predosing (3). Informing clinicians on whether tumors are effectively targeted by [177Lu]Lu-DOTA-NNV003 and gaining insight into the amount of cold antibody dose required to saturate CD37 expression in healthy tissues are essential to optimize future [177Lu]Lu-DOTA-NNV003 RIT dose-regimens. We showed that [89Zr]Zr-N-sucDfNNV003 PET imaging can serve as a surrogate for [177Lu]Lu-DOTA-NNV003 RIT whole-body distribution and represents a potentially attractive tool to assess distribution to tumors and healthy tissues.

The combined approach of RIT and diagnostics such as molecular imaging may support precise cancer therapy in both palliative and curative settings (19). SPECT/CT imaging was routinely used for assessing biodistribution and dosimetry of 90Y- and 131I-based RIT antibodies in the early 2000s (20,21), but low quantities of γ-photons emitted by 177Lu complicate quantification. As a therapeutic agent, even low dose pre-treatment imaging of 177Lu may result in local toxicity, while this risk is limited for the low energy β+-rays of PET radioisotopes. In this respect, gallium-68 (68Ga)/177Lu is a commonly used theranostic pair for studying receptor expression or drug distribution. However, 68Ga, given its relatively short physical half-life of 67.6 min, provides no insight in internalizing properties of an antibody. As internalization is an essential factor for both efficacy and toxicity of 177Lu-based RIT agents, longer-lived PET radioisotopes such as 89Zr may better reflect 177Lu RIT in vivo behavior.

In a recent study, response at the tumor lesion level after treatment with 177Lu-lilotomab satetraxetan was evaluated by FDG PET/CT and did not correlate with tumor-absorbed dose (11). They hypothesized that the combination regimen of radiolabeled and cold antibodies might preclude such a correlation. In our study, we were able to quantitatively visualize tumor uptake in the presence of unlabeled antibody using [89Zr]Zr-N-sucDf-NNV003 PET imaging. In patients with relapsed B cell NHL, a pre-therapy scan with 89Zr-ibritumomab tiuxetan was used to predict radiation dosimetry during 90Y-ibritumomab tiuxetan therapy (14). Importantly, 89Zr-ibritumomab tiuxetan whole-body distribution was not affected by simultaneous 90Y-ibritumomab tiuxetan therapy. These findings emphasize the potential of an [89Zr]Zr-NsucDf-NNV003 pre-therapy scan to predict CD37-targeting by [177Lu]Lu-DOTA-NNV003 RIT, thereby allowing for patient stratification. Furthermore, a post-therapy [89Zr]Zr-N-sucDfNNV003 scan may inform on therapy response by evaluating CD37-mediated tumor uptake

89Zr-PET imaging to predict tumor uptake of 177Lu-NNV003 radioimmunotherapy

after [177Lu]Lu-DOTA-NNV003 RIT.

Next-generation anti-CD37 RIT antibody [177Lu]Lu-DOTA-NNV003, based on the chimeric mousehuman antibody NNV003, was hypothesized to be less immunogenic than the fully murine 177Lu-lilotomab satetraxetran. After a single dose of 177Lu-lilotomab satetraxetan, development of human-anti-mouse-antibodies (HAMAs) was reported for seven out of 74 subjects (3). Preclinically, [177Lu]Lu-DOTA-NNV003 in silico immunogenicity prediction tools revealed a lower immunogenicity potential compared to 177Lu-lilotumab satetraxetan (4). These results warrant further evaluation of [177Lu]Lu-DOTA-NNV003 RIT in patients with CD37-expressing B cell malignancies, and [89Zr]Zr-N-sucDf-NNV003 PET imaging could assist its clinical development and use.

Conclusion

We showed that [89Zr]Zr-N-sucDf-NNV003 PET imaging can accurately predict whole-body distribution and tumor uptake of [177Lu]Lu-DOTA-NNV003 therapy in B cell lymphoma xenograft models. To enable clinical implementation of this theranostic strategy, we established a GMP-compliant production process for [89Zr]Zr-N-sucDf-NNV003. [89Zr]Zr-N-sucDf-NNV003 pre-therapy PET imaging in patients with B cell NHL may help to identify those more likely to respond. Furthermore, as a surrogate imaging tool, [89Zr]Zr-N-sucDf-NNV003 may aid in optimizing [177Lu]Lu-DOTA-NNV003 RIT dose-regimens and post-therapy response evaluation.

Materials and methods

Cell lines and flow cytometry experiments

Human CD37-expressing cell lines RAMOS (Burkitt's lymphoma) and REC1 (Mantle cell lymphoma) were obtained from the American Type Culture Collection. RAMOS and REC1 cell lines were tested and authenticated in July and October 2019 respectively using short tandem repeat profiling. Cells were cultured in Roswell Park Memorial Institute (RPMI) medium, supplemented with 10% fetal calf serum (FCS) and incubated at 37 °C in a humidified atmosphere with 5% CO2.

CD37 expression by RAMOS and REC1 cells was determined by flow cytometry. Cells were harvested in 2% FCS in phosphate-buffered saline (PBS) and kept on ice prior to use. NNV003 and non-specific human IgG control molecule (Nanogam®, Sanquin) were diluted with 2% FCS in PBS to 20 μg/mL and incubated with 2×105 cells/mL for 1 h at 4 °C. Bound NNV003 and control antibodies were detected using a phycoerythrin-conjugated goat anti-human IgG secondary antibody (SouthernBiotech; 2040-09) diluted 1:50 with 2% FCS in PBS and analyzed on a BD Accuri C6 flow cytometer (BD Biosciences). Data analysis was performed using FlowJo v10 (Tree Star) and surface receptor expression was expressed as

mean fluorescent intensity (MFI).

Radiolabeling of NNV003 and IgG control for animal studies NNV003 antibody (IgG1, mouse variable regions, κ, and human constant region, κ; Nordic Nanovector) was conjugated to TFP-N-sucDf (ABX GmbH) and subsequently radiolabeled with 89Zr as described previously (22). To date, several 89Zr-labeled antibodies are produced according to this methodology and were evaluated in animals and patients without any sign of toxicity (16,17,23,24). In short, NNV003 was incubated with a twofold molar excess of TFP-NsucDf at pH 9.0–9.5. After incubation for 1 h at room temperature (RT), pH was set to 4.0–4.5. Ethylenediaminetetraacetic acid (EDTA; Hospital Pharmacy UMCG) 25 mg/mL was added and incubated for 30 min at 35 °C to transchelate Fe(III) from the TFP-N-sucDf hydroxamate groups. NNV003-N-sucDf was subsequently purified using a Vivaspin-2 concentrator (Sartorius GmbH), aliquoted and stored at −80 °C until use. On the day of tracer injection, NNV003-NsucDf was radiolabeled using GMP-grade 89Zr oxalate (Perkin Elmer). RCP of [89Zr]Zr-N-sucDfNNV003 was determined by trichloroacetic acid precipitation test (17). Furthermore, NNV003 antibody was conjugated to p-SCN-Bn-DOTA (Macrocyclics) and subsequently radiolabeled using 177Lu chloride (Perkin Elmer) as described previously (4,5).

Non-specific IgG control molecule was conjugated with a 50-fold molar excess of p-SCN-BnDTPA (Macrocyclics) as described previously (25). Radiolabeling of IgG-DTPA was performed using 111In chloride (Mallinckrodt) by incubation during 1–2 h in ammonium acetate pH 5.5. Radiochemical purity of [111In]In-DTPA-IgG was assessed by instant thin-layer chromatography using 0.1 M citrate buffer pH 6.0 as eluent.

[89Zr]Zr-N-sucDf-NNV003 quality control [89Zr]Zr-N-sucDf-NNV003 purity and concentration were determined by size-exclusion highperformance liquid chromatography (SE-HPLC). A Waters SE-HPLC system was equipped with a dual-wavelength absorbance detector, in-line radioactivity detector and TSK-Gel SW column G3000SWXL 5 μm, 7.8 mm (Joint Analytical Systems GmbH). PBS (9.0 mM sodium phosphate, 1.3 mM potassium phosphate, 140 mM sodium chloride, pH 7.2; Hospital Pharmacy UMCG) was used as mobile phase at a flow of 0.7 mL/min.

NNV003-N-sucDf IRF was determined on human CD37-expressing Burkitt's lymphoma RAMOS cells. Cells were harvested in PBS with 0.5% bovine serum albumin (BSA), diluted to 75 × 106 cells and 0.2 mL added per tube to a total of 5 tubes. CD37-specific binding sites were blocked in 2 tubes by incubation with 20 μg NNV003-N-sucDf for 15 min at RT. Subsequently, 8 ng [89Zr]Zr-N-sucDf-NNV003 (~ 9000 counts per minute) was added to each tube and incubated for 1 h at RT. Tubes were counted in a calibrated well-type gamma counter (LKB instruments),

89Zr-PET imaging to predict tumor uptake of 177Lu-NNV003 radioimmunotherapy

subsequently spun down and washed with 0.5% BSA in PBS for three times, after which tubes were counted again. IRF was expressed as the average percentage of CD37-bound [89Zr]ZrN-sucDf-NNV003 as a fraction of the percentage of total activity added in non-blocked tubes corrected for non-specific binding in the blocked tubes. Acceptance criteria were set at ≤ 5% non-specific binding in blocked tubes and NNV003-N-sucDf IRF at ≥ 0.8.

Animal studies

All experiments were performed in accordance with relevant guidelines and regulations. All methods were reported in accordance with recommendations in the ARRIVE guidelines. Animal studies involving [89Zr]Zr-N-sucDf-NNV003 were approved by the Institutional Animal Care and Use Committee of the University Medical Center Groningen. Male BALB/c OlaHsd-Foxn1nu mice (Envigo) 8–10 weeks of age were inoculated with 10×106 either REC1 or RAMOS cells. Murine NK cells were depleted to enhance tumor take-rate. This was achieved by administering the mice anti-asialo GM1 treatment on 1 day before and 4, 11, 18 and 25 days post tumor inoculation. When tumors measured a volume of at least 200 mm3 (~ 14 days post inoculation for REC1 tumors and ~ 19 days for RAMOS tumors), mice received intravenous injections of 10 μg (~ 5 MBq) [89Zr]Zr-N-sucDf-NNV003 supplemented with either 0, 15 or 90 μg of unlabeled NNV003 and coinjected with an equal total protein dose (~ 1 MBq) of [111In]In-DTPA-IgG control (n = 5–6 mice per group). By co-injection of [89Zr]Zr-N-sucDf-NNV003 and [111In]In-DTPA-IgG, tumor uptake and biodistribution results can be compared within the same animal, thereby providing valid results on target-specific uptake. Also, the number of animals required for these studies can be reduced using this strategy. Mice underwent microPET scanning at 1, 3 and 5 days post injection (pi), followed by ex vivo biodistribution.

MicroPET scans were performed using a Focus 220 rodent scanner (CTI Siemens). Scans were reconstructed using a 2-dimensional ordered-subset expectation maximization reconstruction algorithm with Fourier rebinning, 4 iterations, and 16 subsets. Data sets were corrected for decay, random coincidences, scatter, and attenuation. For in vivo quantification, regions of interest were drawn for tumor based upon ex vivo weight, assuming 1 g/cm3 tissue density, and heart using AMIDE medical image data examiner software v1.0.4. Tracer uptake was quantified as SUVmean and SUVmax, calculated from the mean or maximum activity in the region of interest and divided by the injected dose per gram body weight. For ex vivo biodistribution studies, relevant organs were collected, weighed and counted using a calibrated well-type gamma counter. Standards of injected tracer were included to correlate measured counts to the percentage of injected tracer activity. After correction for decay, ex vivo tissue uptake was expressed as the percentage of injected radioactivity dose per gram tissue (%ID/g) and standardized uptake value (SUV) by correcting for injected dose and mouse body weight.

Animal experiments involving [177Lu]Lu-DOTA-NNV003 were approved by the Norwegian Animal Research Authority. Biodistribution of [177Lu]Lu-DOTA-NNV003 was studied in the RAMOS tumor model. Female Hsd:Athymic Nude-Foxn1nu mice (Envigo) 7–11 weeks of age were subcutaneously injected with 100 μL RAMOS cell suspension from a donor mouse xenograft to enhance tumor take-rate. Mice received intravenous injections of 4–10 μg (0.5–0.9 MBq) [177Lu]Lu-DOTA-NNV003 (IRF 74.4–81.7%), followed by tissue collection and ex vivo biodistribution analysis at 1 h, 6 h, 1 day and 3 days pi (n = 4 mice per group).

Ex vivo tissue preparation and immunohistochemistry

For ex vivo tissue analysis, formalin-fixed paraffin-embedded (FFPE) tumor tissue blocks were prepared. FFPE blocks were sliced into 4 μm tumor tissue sections, fixated on microscope slides and dried overnight at 60 °C. For CD37 immunohistochemistry, tumor tissue sections were deparaffinized in xylene and rehydrated. Heat-induced antigen retrieval was performed in 10 mM citrate (pH 6.0) for 15 min at 95–100 °C. Endogenous peroxidase was blocked by 10-min incubation with 10% hydrogen peroxide in PBS. Slides were incubated with rabbit anti-human CD37 antibody (Proteintech; 21044-1) or rabbit IgG antibody control (Abcam; ab172730) diluted to 0.8 μg/mL in 1% BSA in PBS for 1 h at RT. Thereafter, slides were incubated with Dako EnVision horseradish peroxidase system (Agilent Technologies) for 30 min at RT, followed by 10-min incubation with diaminobenzidine chromogen. Hematoxylin counterstaining was applied routinely. For histological analysis of tumors, hematoxylin/eosin staining was performed on subsequent tissue sections. Digital scans of slides were acquired by a Hamamatsu NanoZoomer 2.0-HT multi-slide scanner and analyzed with NanoZoomer Digital Pathology viewer software.

Production and stability testing of clinical grade [89Zr]Zr-N-sucDf-NNV003

NNV003 was conjugated to TFP-N-sucDf at a 1:2 molar ratio and subsequently radiolabeled with GMP-grade 89Zr. Quality control was performed on both NNV003-N-sucDf intermediate product and [89Zr]Zr-N-sucDf-NNV003 final product. This included analysis on appearance, yield, purity, concentration, pH, radiochemical purity, residual solvents, sterility, endotoxin content and IRF. Analytical procedures were validated to demonstrate suitability for use in quality control testing of NNV003-N-sucDf and [89Zr]Zr-N-sucDf-NNV003. The production processes for NNV003-N-sucDf and [89Zr]Zr-N-sucDf-NNV003 were validated according to GMP guidelines by the production of three consecutive validation batches.

NNV003-N-sucDf intermediate product was stored in sterile vials (BioPure) at −80 °C. Stability of NNV003-N-sucDf was analyzed at 0, 1, 3 and 6 months after production. [89Zr]Zr-N-sucDfNNV003 final product was stored in sterile vials (BioPure) at 2–8 °C and stability was analyzed at 0 and 96 h. Stability of [89Zr]Zr-N-sucDf-NNV003 in the syringe at RT was analyzed at 0 and 4

89Zr-PET imaging to predict tumor uptake of 177Lu-NNV003 radioimmunotherapy

h. Stability tests consisted of quality control according to release specifications.

Statistical analysis

Data were analyzed for statistical significance in GraphPad Prism v7.0 using the Mann–Whitney U test for non-parametric data followed by Bonferroni post-test correction for comparison of more than two groups. Ex vivo biodistribution of [89Zr]Zr-N-sucDf-NNV003 and [177Lu]Lu-DOTANNV003 were compared with Welch’s t-test for unequal variances. Correlation was assessed by Spearman’s rank-order correlation test. In vitro experiments were repeated at least 3 times.

P values < 0.05 were considered significant.

Acknowledgements

We are thankful to L. Pot-de Jong for her assistance in the validation of a GMP-compliant production process for [89Zr]Zr-N-sucDf-NNV003.

Author contributions

D.G., M.N.L.-d.H., M.N., H.H., J.D., E.G.E.d.V. and M.P. participated in the design and/or interpretation of the reported experiments or results. D.G. and M.P. participated in the acquisition and/ or analysis of data. H.H. and J.D. participated in [177Lu]Lu-DOTA-NNV003 animal studies and provided data on ex vivo biodistribution. D.G., M.N.L-d.H., H.H., J.D., E.G.E.d.V and M.P. provided administrative, technical or supervisory support. All authors participated in drafting and/or revising the manuscript. All authors read and approved the final manuscript.

Competing interests

Nordic Nanovector ASA provided a research grant to E. G. E. de Vries, which was made available to her institution (UMCG). J. Dahle and H. Heyerdahl are current employees of Nordic Nanovector ASA, which developed and owns the intellectual property rights pertaining to NNV003. All other authors declare no competing interests.

References

1. Repetto-Llamazares, A. H. V. et al. Combination of 177Lu-lilotomab with rituximab significantly improves the therapeutic outcome in preclinical models of non-Hodgkin’s lymphoma. Eur. J. Haematol. 101, 522–531 (2018).

2. Malenge, M. M. et al. 177Lu-lilotomab satetraxetan has the potential to counteract resistance to rituximab in non-Hodgkin lymphoma. J. Nucl. Med. 61, 1468–1475 (2020).

3. Kolstad, A. et al. Phase 1/2a study of 177Lu-lilotomab satetraxetan in relapsed/refractory indolent nonHodgkin lymphoma. Blood Adv. 4, 4091–4101 (2020).

4. Maaland, A. F. et al. Targeting B-cell malignancies with the bèta-emitting anti-CD37 radioimmunoconjugate 177Lu-NNV003. Eur. J. Nucl. Med. Mol. Imaging. 46, 2311–2321 (2019).

89Zr-PET imaging to predict tumor uptake of 177Lu-NNV003 radioimmunotherapy

5. Dahle, J. et al. Evaluating antigen targeting and anti-tumor activity of a new anti-CD37 radioimmunoconjugate against non-Hodgkin’s lymphoma. Anticancer Res. 33, 85–96 (2013).

6. de Winde, C. M., Elfrink, S. & van Spriel, A. B. Novel insights into membrane targeting of B cell lymphoma. Trends Cancer. 3, 442–453 (2017).

7. Emmett, L. et al. Lutetium-177 PSMA radionuclide therapy for men with prostate cancer: A review of the current literature and discussion of practical aspects of therapy. J. Med. Radiat. Sci. 64, 52–60 (2017).

8. Repetto-Llamazares, A., Abbas, N., Bruland, Ø. S., Dahle, J. & Larsen, R. H. Advantage of lutetium-177 versus radioiodine immunoconjugate in targeted radionuclide therapy of B-cell tumors. Anticancer Res. 34, 3263–3269 (2014).

9. Stokke, C. et al. Pre-dosing with lilotomab prior to therapy with 177Lu-lilotomab satetraxetan significantly increases the ratio of tumor to red marrow absorbed dose in non-Hodgkin lymphoma patients. Eur. J. Nucl. Med. Mol. Imaging. 45, 1233–1241 (2018).

10. Blakkisrud, J. et al. Red marrow-absorbed dose for non-Hodgkin lymphoma patients treated with 177Lulilotomab satetraxetan, a novel anti-CD37 antibody-radionuclide conjugate. J. Nucl. Med. 58, 55–61 (2017).

11. Løndalen, A. et al. FDG PET/CT parameters and correlations with tumor-absorbed doses in a phase 1 trial of 177Lu-lilotomab satetraxetan for treatment of relapsed non-Hodgkin lymphoma. Eur. J. Nucl. Med. Mol. Imaging. 48, 1902–1914 (2021).

12. Verel, I. et al. Long-lived positron emitters zirconium-89 and iodine-124 for scouting of therapeutic radioimmunoconjugates with PET. Cancer Biother. Radiopharm. 18, 655–661 (2003).

13. Perk, L. R. et al. 89Zr as a PET surrogate radioisotope for scouting biodistribution of the therapeutic radiometals 90Y and 177Lu in tumor-bearing nude mice after coupling to the internalizing antibody cetuximab. J. Nucl. Med. 46, 1898–1906 (2005).

14. Rizvi, S. N. F. et al. Biodistribution, radiation dosimetry and scouting of 90Y-ibritumomab tiuxetan therapy in patients with relapsed B-cell non-Hodgkin’s lymphoma using 89Zr-ibritumomab tiuxetan and PET. Eur. J. Nucl. Med. Mol. Imaging. 39, 512–520 (2012).

15. Abou, D. S., Ku, T. & Smith-Jones, P. M. In vivo biodistribution and accumulation of 89Zr in mice. Nucl. Med. Biol. 38, 675–681 (2011).

16. Dijkers, E. C. et al. Development and characterization of clinical-grade 89Zr-trastuzumab for HER2/neu immunoPET imaging. J. Nucl. Med. 50, 947–981 (2009).

17. Nagengast, W. B. et al. In vivo VEGF imaging with radiolabeled bevacizumab in a human ovarian tumor xenograft. J. Nucl. Med. 48, 1313–1319 (2007).

18. Cataldi, M., Vigliotti, C., Mosca, T., Cammarota, M. R. & Capone, D. Emerging role of the spleen in the pharmacokinetics of monoclonal antibodies, nanoparticles and exosomes. Int. J. Mol. Sci. 18, 1249–1263 (2017).

19. Herrmann, K. et al. Radiotheranostics: A roadmap for future development. Lancet Oncol. 21, e146-156 (2020).

20. Wiseman, G. A. et al. Radiation dosimetry results for Zevalin radioimmunotherapy of rituximab-

refractory non-Hodgkin lymphoma. Cancer 94(SUPPL), 1349–1357 (2002).

21. Sgouros, G. et al. Patient-specific, 3-dimensional dosimetry in non-Hodgkin’s lymphoma patients treated with 131I-anti-B1 antibody: Assessment of tumor dose-response. J. Nucl. Med. 44, 260–268 (2003).

22. Verel, I. et al. 89Zr immuno-PET: Comprehensive procedures for the production of 89Zr-labeled monoclonal antibodies. J. Nucl. Med. 44, 1271–1281 (2003).

23. Dijkers, E. C. et al. Biodistribution of 89Zr-trastuzumab and PET imaging of HER2-positive lesisons in patients with metastatic breast cancer. Clin. Pharmacol. Ther. 87, 586–592 (2010).

24. Gaykema, S. B. M. et al. 89Zr-bevacizumab PET imaging in primary breast cancer. J. Nucl. Med. 54, 1014–1018 (2013).

25. Ruegg, C. L. et al. Improved in vivo stability and tumor targeting of bismuth-labeled antibody. Cancer Res. 50, 4221–4226 (1990).

Supplementary information

FIGURE S1: Quality control of [89Zr]Zr-N-sucDf-NNV003 production for in vivo studies. (A) Efficiency of NNV003 conjugation to TFP-N-sucDf. The NNV003 to TFP-N-sucDf (mAb:Df) ratio is presented on the x-axis, the effective amount of TFP-N-sucDf chelator conjugated per NNV003 antibody is presented on the y-axis. (B) NNV003 yield after conjugation to increasing molar ratios of TFP-N-sucDf. The mAb:Df ratio is presented on the x-axis, percentage yield after conjugation is presented on the y-axis. (C) Efficiency of Fe(III) transchelation from TFP-N-sucDf hydroxamate groups to EDTA. The mAb:Df ratio is presented on the x-axis, percentage of total Fe(III) transchelated to EDTA is presented on the y-axis. (D) [89Zr]Zr-N-sucDfNNV003 immunoreactivity to CD37 after conjugation and radiolabeling. The mAb:Df ratio is presented on the x-axis and NNV003-N-sucDf immune reactive fraction (IRF) is expressed on the y-axis. (E) Representative HPLC chromatograms showing purity and aggregates after conjugation of NNV003 to increasing molar excess of TFP-N-sucDf. (F) Radiochemical purity of [89Zr]Zr-N-sucDf-NNV003 at increasing specific activity.

Amount of 89Zr in MBq added to 1 mg of NNV003 is presented on the x-axis, effective amount of 89Zr labeled to NNV003 as percentage of total added radioactivity is presented on the y-axis. Data in A-D and F is shown as mean ± standard deviation.

89Zr-PET imaging to predict tumor uptake of 177Lu-NNV003 radioimmunotherapy

FIGURE S2: Production process for clinical-grade [89Zr]Zr-N-sucDf-NNV003. Steps-wise process for the production of good manufacturing practice (GMP)-compliant [89Zr]Zr-N-sucDfNNV003. Quality control (QC) was performed on NNV003-N-sucDf intermediate product (QC 1-11) and [89Zr]Zr-N-sucDf-NNV003 final product (QC 12-20).

89Zr-PET imaging to predict tumor uptake of 177Lu-NNV003 radioimmunotherapy

TABLE S1: Quality control of clinical-grade NNV003-N-sucDf and [89Zr]Zr-N-sucDf-NNV003.

Test Specification

Batch 1 Batch 2 Batch 3 NNV003 N sucDf intermediate product

Appearance Colorless to light yellow Colorless to light yellow Colorless to light yellow Colorless to light yellow Conjugation ratio 0.5 2.0 1.47 1.41 1.39

Filter integrity < 20% < 20% < 20% < 20% Yield > 50% 82.9% 92.5% 89.9%

pH 4.0 7.0 5.34 5.23 5.30

Purity ≤ 3% aggregates (280 nm) < 3% < 3% < 3 %

Concentration 9.0 11.0 mg/mL 9.7 mg/mL 9.6 mg/mL 9.8 mg/mL

Endotoxins < 2.5 EU/ml < 2.5 EU/mL < 2.5 EU/mL < 2.5 EU/mL

Sterility Sterile Sterile Sterile Sterile Residual solvents (acetonitrile) < 410 ppm < 410 ppm < 410 ppm < 410 ppm

IRF 50 100% 86.8% 67.8% 64.6% 89Zr[Zr] N sucDf NNV003 final product

Appearance Colorless to light yellow Colorless to light yellow Colorless to light yellow Colorless to light yellow

RCP ≥ 95% 99.1% 99.3% 99.4%

Filter integrity < 20% < 20% < 20% < 20%

Yield For 1 patient > 40.7 MBq For 2 patients > 77.7 MBq

49.77 MBq 64.03 MBq 80.28 MBq

pH 5.0 8.0 5.07 5.75 5.76

Purity ≤ 3% aggregates (280 nm) < 3% < 3% < 3 %

Concentration For information only 0.15 mg/mL 0.10 mg/mL 0.07 mg/mL

Endotoxins < 2.5 EU/ml < 2.5 EU/mL < 2.5 EU/mL < 2.5 EU/mL

Sterility Sterile Sterile Sterile Sterile

Quality control results for three individual batches of NNV003-N-sucDf intermediate product and [89Zr]ZrN-sucDf-NNV003 final product (37 MBq to ~1 mg). IRF was determined after labeling with 89Zr. EU: endotoxin units, IRF: immune reactive fraction, RCP: radiochemical purity, ppm: parts per million.

TABLE S2: Stability of clinical-grade NNV003-N-sucDf.

Test Specification t = 0 1 month 3 months 6 months

Appearance Colorless to light yellow Colorless to light yellow Colorless to light yellow Colorless to light yellow Colorless to light yellow pH 4.0 7.0 5.34 5.36 5.33 5.31

Purity £ 3% aggregates (280 nm) £ 3% £ 3% £ 3% £ 3%

Concentration 9.0 11.0 mg/mL 9.7 mg/mL 9.4 mg/mL 9.5 mg/mL 9.3 mg/mL Endotoxins < 2.5 EU/ml < 2.5 EU/mL nd nd nd

Sterility Sterile Sterile nd

nd

nd Residual solvents (acetonitrile) < 410 ppm < 410 ppm nd

RCP ≥ 95% 99.1% 99.2% 99.3% 99.6%

Quality control results for stability of NNV003-N-sucDf intermediate product (batch 1) at 1, 3 and 6 months. EU: endotoxin units, IRF: immune reactive fraction, nd: not determined, RCP: radiochemical purity, ppm: parts per million.

57 3 Immunoreactivity 50 100% 86.8% 69.3% 79.4% 82.2%

Table S3: Stability of clinical-grade [89Zr]Zr-N-sucDf-NNV003.

Test Specification t = 0 96 h (2 8 °C) 4 h syringe (RT)

Appearance Colorless to light yellow Colorless to light yellow Colorless to light yellow Colorless to light yellow pH 5.0 8.0 5.76 5.89 5.83

Purity £ 3% aggregates (280 nm) £ 3% £ 3% £ 3%

Concentration For information only 0.07 mg/mL 0.07 mg/mL 0.06 mg/mL

Endotoxins < 2.5 EU/ml < 2.5 EU/mL nd nd

Sterility Sterile Sterile Sterile Sterile RCP ≥ 95% 99.4% 97 9% 97.8%

Quality control results for stability of [89Zr]Zr-N-sucDf-NNV003 final product (produced from batch 3 of NNV003-N-sucDf intermediate product) for 96 h at 2-8 °C and for 4 h in the syringe at RT. EU: endotoxin units, nd: not determined, RCP: radiochemical purity, RT: room temperature, ppm: parts per million.

89Zr-pembrolizumab biodistribution is influenced by PD-1-mediated uptake in lymphoid organs

Elly L van der Veen 1, Danique Giesen 1, Linda Pot-de Jong 1, Annelies JorritsmaSmit 2, Elisabeth G E De Vries 1, Marjolijn N Lub-de Hooge 3, 4

1 Department of Medical Oncology, 2 Department of Clinical Pharmacy and Pharmacology, 3 Department of Clinical Pharmacy and Pharmacology, and 4 Department of Nuclear Medicine and Molecular Imaging, UMCG, Groningen, Groningen, Netherlands. m.n.de.hooge@umcg.nl.

J Immunother Cancer. 2020;8(2):e000938.

Abstract

Background: To better predict response to immune checkpoint therapy and toxicity in healthy tissues, insight in the in vivo behavior of immune checkpoint targeting monoclonal antibodies is essential. Therefore, we aimed to study in vivo pharmacokinetics and whole-body distribution of zirconium-89 (89Zr) labeled programmed cell death protein-1 (PD-1) targeting pembrolizumab with positron-emission tomography (PET) in humanized mice.

Methods: Humanized (huNOG) and non-humanized NOG mice were xenografted with human A375M melanoma cells. PET imaging was performed on day 7 post 89Zr-pembrolizumab (10 μg, 2.5 MBq) administration, followed by ex vivo biodistribution studies. Other huNOG mice bearing A375M tumors received a co-injection of excess (90 μg) unlabeled pembrolizumab or 89ZrIgG4 control (10 μg, 2.5 MBq). Tumor and spleen tissue were studied with autoradiography and immunohistochemically including PD-1.

Results: PET imaging and biodistribution studies showed high 89Zr-pembrolizumab uptake in tissues containing human immune cells, including spleen, lymph nodes and bone marrow. Tumor uptake of 89Zr-pembrolizumab was lower than uptake in lymphoid tissues, but higher than uptake in other organs. High uptake in lymphoid tissues could be reduced by excess unlabeled pembrolizumab. Tracer activity in blood pool was increased by addition of unlabeled pembrolizumab, but tumor uptake was not affected. Autoradiography supported PET findings and immunohistochemical staining on spleen and lymph node tissue showed PD-1 positive cells, whereas tumor tissue was PD-1 negative.

Conclusion: 89Zr-pembrolizumab whole-body biodistribution showed high PD-1-mediated uptake in lymphoid tissues, such as spleen, lymph nodes and bone marrow, and modest tumor uptake. Our data may enable evaluation of 89Zr-pembrolizumab whole-body distribution in patients.

Background

Immune checkpoint inhibitors targeting the programmed cell death protein-1 (PD-1)/ programmed death ligand-1 (PD-L1) pathway are showing impressive antitumor effects. However, not all patients respond and serious immune-related toxicity has been reported (1). This has raised interest in better understanding the behavior of these drugs in the human body. PD-L1 and PD-1 are expressed by a broad range of immune cells, including T-cells, B-cells, natural killer (NK) cells, monocytes and dendritic cells. PD-L1 can be highly expressed by tumor cells, whereas PD-1 expression is most prominent in T-cells and lower in other immune cells (2). Biodistribution of PD-1 and PD-L1 targeting drugs will likely be influenced by the dynamic

expression patterns of these targets.

Molecular imaging has proven to be an useful tool for studying drug biodistribution (3,4). In table 1, we summarized preclinical imaging studies that investigated biodistribution of radiolabeled molecules targeting PD-1 and PD-L1 (5–28). Most studies that we reviewed focused on tracer distribution in the tumor and its microenvironment, without considering PD-1 and PD-L1 expression in healthy immune tissues. Studies that do report on tracer uptake in lymphoid tissues are scarce and results are often limited to the spleen. Furthermore, most tracers targeting human PD-1/PD-L1 are not cross-reactive with murine proteins and relevant mouse models reconstituted with (parts of) a human immune system are rarely used. A limited number of studies used NOD scid gamma (NSG) mice engrafted with human peripheral blood mononuclear cells (hNSG model) (23–25,27). The hNSG model has a high level of functional T-cells, however, it is also characterized by aberrant distribution of immune cells to murine immune tissues and other cell lineages remain underdeveloped (29). Humanized mice that are engrafted with human CD34+ hematopoietic stem cells (HSCs) establish an immunecompetent model with a broader set of developed human immune cells present and might therefore be a better surrogate for the human immune environment.

To gain more insight in the in vivo behavior of a human PD-1 targeting monoclonal antibody (mAb), not cross-reactive with murine PD-1, we aimed to study the biodistribution of zirconium-89 (89Zr) radiolabeled pembrolizumab in melanoma-bearing humanized NOG mice (huNOG) engrafted with HSCs using positron-emission tomography (PET) imaging. To enable consecutive clinical translation of this approach, we developed and validated a good manufacturing practices (GMP) compliant production process for 89Zr-pembrolizumab. Finally, we put our data in perspective by summarizing results from current in vivo preclinical studies with PD-1 and PD-L1 targeting radiolabeled molecules.

Methods

Cell lines

The human melanoma cell line A375M was purchased from the American Type Culture Collection. Cell lines were confirmed to be negative for microbial contamination and were authenticated on August 6, 2018, by BaseClear using short tandem repeat profiling. A375M cells were routinely cultured in Roswell Park Memorial Institute 1640 medium (Invitrogen) containing 10% fetal calf serum (Bodinco BV), under humidified conditions at 37 °C with 5% CO2 Cells were passaged 1:10, twice a week. For in vivo experiments, cells in the exponential growth phase were used.

89Zr-pembrolizumab biodistribution in humanized mouse models

Development of 89Zr-pembrolizumab and 89Zr-IgG4

First, the buffer of pembrolizumab (25 mg/mL, Merck) was exchanged for NaCl 0.9% (Braun) using a Vivaspin-2 concentrator (30 kDa) with a polyethersulfon filter (Sartorius). Next, pembrolizumab was conjugated with the tetrafluorphenol-N-succinyldesferal-Fe(III) ester (TFP-N-sucDf; ABX) as described earlier, in a 1:2 TFP-N-sucDf:mAb ratio (30). Conjugated product was purified from unbound chelator using Vivaspin-2 concentrators and stored at −80 °C. On the day of tracer injection, N-sucDf-pembrolizumab was radiolabeled with 89Zr, delivered as 89Zr-oxalate dissolved in oxalic acid (PerkinElmer), as described previously (30). For in vivo studies, pembrolizumab was radiolabeled at a specific activity of 250 MBq/mg. IgG4 control molecule (Sigma-Aldrich) was conjugated with TFP-N-sucDf at a 1:3 molar ratio, followed by radiolabeling with 89Zr at similar specific activity of 250 MBq/mg.

Quality control of 89Zr-pembrolizumab Size exclusion high-performance liquid chromatography (SE-HPLC) was used to determine the final number of TFP-N-sucDf ligands per antibody (chelation ratio). SE-HPLC analysis was also performed to assess potential aggregation and fragmentation for both N-sucDfpembrolizumab and 89Zr-pembrolizumab. An HPLC system (Waters) equipped with an isocratic pump (Waters), a dual wavelength absorbance detector (Waters), in-line radioactivity detector (Berthold) and a TSK-GEL G3000SWXL column (Tosoh Biosciences) was used with phosphate buffered saline (PBS, sodium chloride 140.0 mmol/L, sodium hydrogen phosphate 0.9 mmol/L, sodium dihydrogen phosphate 1.3 mmol/L; pH 7.4) as mobile phase (flow 0.7 mL/ min). Radiochemical purity of 89Zr-pembrolizumab was measured by trichloroacetic acid precipitation assay (31). Immunoreactivity of 89Zr-pembrolizumab was analyzed by a competition binding assay with unlabeled pembrolizumab. Nunc-immuno break apart 96-wells plates (Thermo Scientific) were coated overnight at 4 °C with 100 μL of 1 μg/mL PD-1 extracellular domain (R&D Systems) in PBS, set to pH 9.6 with Na2CO3 2M. Plates were washed with 0.1% Tween 80 in PBS and blocked for 1 hour at room temperature (RT) with 150 μL 1% human serum albumin (Albuman, Sanquin) in PBS. Multiple 1:1 mixtures of 89Zr-pembrolizumab with unlabeled pembrolizumab were prepared, using a fixed concentration of 89Zr-pembrolizumab (7000 ng/mL) and varying concentrations of unlabeled pembrolizumab (from 3.75 ng/mL to 12.5×106 ng/mL). Of each mixture, 100 μL was added to the 96-wells plate and incubated for 2 hours at RT. After washing twice with washing buffer, radioactivity in each well was counted using a gamma counter (Wizard2 2480–0019, SW 2.1, PerkinElmer). Counts were plotted against the concentration of competing unlabeled pembrolizumab. The half maximal inhibitory concentration (IC50) was calculated using GraphPad Prism 7 (GraphPad software). Immunoreactivity was expressed as the IC50 value divided by the 89Zr-pembrolizumab concentration to calculate the immune reactive fraction (IRF).

Animal studies

All animal studies were approved by the Institutional Animal Care and Use Committee of the University of Groningen. Studies were performed in humanized NOG mice (NOD.Cg-Prkdcscid Il2rgtm1Sug/JicTac, Taconic) and non-humanized NOG mice (Taconic) were used for control experiments. HuNOG mice are sublethally irradiated 3 weeks after birth and subsequently reconstituted with human CD34+ hematopoietic stem cells derived from fetal cord blood to express a functional human immune system including B-cells, T-cells, NK-cells, dendritic cells and monocytes. HuNOG and NOG mice were subcutaneously xenografted with 5×106 A375M human melanoma cells in 300 μL of a 1:1 mixture of PBS and Matrigel (BD Biosciences) on the right flank. Tumor growth was assessed by caliper measurements. When tumor volumes reached 100 to 200 mm3 (after 2 weeks), 2.5 MBq 89Zr-pembrolizumab (10 μg) was administered via retro-orbital injection. Mice were anesthetized using isoflurane/medical air inhalation (5% induction, 2.5% maintenance).

The first group of huNOG mice received 10 μg 89Zr-pembrolizumab (n=5). In addition, a second group of huNOG mice xenografted with the same tumor model received a co-injection of 10 μg 89Zr-pembrolizumab and 90 μg unlabeled pembrolizumab (n=4). To a third group of huNOG mice, 2.5 MBq 89Zr-IgG4 control (10 μg) was administered (n=4). Control NOG mice received 10 μg 89Zr-pembrolizumab (n=4).

PET imaging and ex vivo biodistribution

On day 7 post tracer injection (pi), PET scanning was performed. We selected this day based on optimal tumor-to-blood ratio and technical aspects, including feasible tracer specific activity and animal welfare. Mice were placed in a Focus 220 rodent scanner (CTI Siemens) on heating matrasses. Acquisition time was 60 min. A transmission scan of 515 s was performed using a 57Co point source to correct for tissue attenuation. After scanning, mice were sacrificed for ex vivo biodistribution. Bone marrow was collected from the femur bone by centrifugalbased separation. All other organs were dissected and counted in a gamma-counter (Wizard2 2480–0019, SW 2.1, PerkinElmer). Tracer uptake in each organ was expressed as percentage of the injected dose per gram tissue weight, calculated by the following formula: %ID/g = [activity in tissue (MBq)/total injected activity (MBq)]/tissue weight (g)×100. To compare ex vivo and in vivo uptake, ex vivo uptake was also calculated as mean radioactivity per gram tissue, adjusted for total body weight (SUVmean ex vivo), calculated with the following formula: SUVmean ex vivo = [activity in tissue (MBq)/total injected activity (MBq)]×mouse weight (g). Calculations are corrected for decay and background.

PET data was reconstructed and in vivo quantification was performed using PMOD software (V.4.0, PMOD technologies LCC). Three-dimensional regions of interest were drawn around

89Zr-pembrolizumab biodistribution in humanized mouse models

the tumor. For other organs and tissues, a size-fixed sphere was drawn in representative tissue parts. PET data was presented as mean standardized uptake value (SUVmean in vivo), calculated by the following formula: SUVmean (g/mL) = [activity concentration (Bq/mL)/applied dose (Bq)]×weight (kg)×1000.

Autoradiography

Tumor and spleen from ex vivo biodistribution studies were formalin-fixed and paraffin embedded (FFPE). FFPE tissue blocks where cut into slices of 4 μM. These slices were exposed to a phosphor imaging screen (PerkinElmer) for 72 hours and then scanned using a Cyclone phosphor imager (PerkinElmer).

Immunohistochemistry

Subsequent slices of the same tumor, spleen and mesenteric lymph node tissue were stained for H&E, CD3, CD8 and PD-1. FFPE tumor, spleen and lymph node tissue were cut into 4 μm slices using a microtome (Microm Hm 355 s, Thermo Scientific) and mounted on glass slides. Tissue sections were deparaffinized and rehydrated using xylene and ethanol. Heat-induced antigen retrieval was performed in citrate buffer (pH=6) at 100 °C for 15 min. Endogenous peroxidase was blocked by 30 min incubation with 0.3% H2O2 in PBS. For CD3 staining, slides were incubated with rabbit anti-human CD3-antibody (Spring bioscience; clone SP162) in a 1:100 dilution in PBS/1% bovine serum albumin (BSA) at RT for 15 min. For CD8 staining, slides were incubated with rabbit anti-human CD8-antibody (Abcam; clone SP16) in a 1:50 dilution in PBS/1% BSA at 4 °C overnight. For PD-1 staining, slides were incubated with rabbit anti-human PD-1-antibody (Abcam, clone EPR4877(2)) in a 1:500 dilution in PBS/1% BSA at RT for 30 min. Human tonsil or lymph nodes tissues sections served ad positive control and were incubated with either CD3, CD8 or PD-1 antibody. As a negative control human tonsil or lymph nodes sections were incubated with rabbit IgG monoclonal antibody (Abcam, clone EPR25A) or PBS/1% BSA.

For CD3, CD8 and PD-1 staining, incubation with secondary antibody (anti-rabbit EnVision+, Dako) was performed for 30 min, followed by application of diaminobenzidine chromogen for 10 min. Hematoxylin counterstaining was applied and tissue sections were dehydrated using ethanol and imbedded using mounting medium (Eukitt). H&E staining served to analyze tissue viability and morphology. Digital scans were acquired by a Nanozoomer 2.0-HT multi slide scanner (Hamamatsu).

89Zr-pembrolizumab manufacturing according to GMP

To enable clinical application, GMP-compliant 89Zr-pembrolizumab was developed. First, N-sucDf-pembrolizumab intermediate product was produced on a larger scale (60 mg batch, divided in 2.5 mg aliquots) and subsequently radiolabeled with 89Zr, followed by purification,

dilution and sterile filtration (supplemental figure S1). Release specifications were defined, as shown in supplemental table S1. All analytical methods for quality control (QC) were validated. According to protocol validation of both N-sucDf-pembrolizumab and 89Zr-pembrolizumab, manufacturing consisted of three independent validation runs, including complete release QC. Stability of N-sucDf-pembrolizumab stored at −80 °C was studied up to 6 months and stability of 89Zr-pembrolizumab was determined up to 168 hours at 2 °C to 8 °C stored in a sterile, type 1 glass injection vial. In addition, in use stability was demonstrated at RT in a polypropylene syringe for up to 4 hours (supplemental table S2).

Statistical analysis

Data are presented as median ± IQR. A Mann-Whitney U test, followed by a Bonferroni correction was performed to compare groups (GraphPad, Prism 7). p values ≤0.05 were considered significant. If not indicated otherwise, results were not statistically significant.

Results

89Zr-pembrolizumab development for in vivo studies

We optimized the conjugation processes of pembrolizumab with the TFP-N-sucDf chelator and its subsequent radiolabeling with 89Zr. For in vivo studies, N-sucDf-pembrolizumab was produced with >60% yield and average 1.7 chelators per antibody (supplemental figure S2, table S1). N-sucDf-pembrolizumab was subsequently radiolabeled with 89Zr at a specific activity of 250 MBq/mg, with radiochemical purity of >95% after purification. Both N-sucDfpembrolizumab and 89Zr-pembrolizumab were stable, as shown in supplemental table S1, S2 and figure S2. Immunoreactivity was not impaired by conjugation or radiolabeling.

89Zr-pembrolizumab imaging and biodistribution in humanized mice

PET imaging revealed 89Zr-pembrolizumab uptake in tumor, but also in healthy tissues, including liver, spleen and lymph nodes, of A375M tumor-bearing huNOG mice (figure 1A, B). Consistent with these results, ex vivo biodistribution at day 7 pi showed highest 89Zr-pembrolizumab uptake in spleen (SUVmean 30.5, IQR 15.8 to 67.7), mesenteric lymph nodes (SUVmean 20.4, IQR 8.0 to 25.2), bone marrow (SUVmean 14.5, IQR 6.1 to 32.8), thymus (SUVmean 1.3, IQR 1.1 to 2.1), liver (SUVmean, IQR 6.0, IQR 3.4 to 9.9) and tumor (SUVmean 5.1, IQR 3.3 to 8.9) (figure 1C, supplemental table S3).

Tumor uptake of 89Zr-pembrolizumab was variable and slightly higher than tumor uptake observed for 89Zr-IgG4 control, however not significant due to small groups of mice (SUVmean 5.1, IQR 3.3 to 8.9 vs SUVmean 3.5, IQR 2.7 to 4.4) (figure 1C). This may be explained by low PD-1 expression found in all tumors by immunohistochemical (IHC) analysis (figure 2). 89Zrpembrolizumab tumor-to-blood ratio also did not differ from 89Zr-IgG4 control (figure 1D).

89Zr-pembrolizumab biodistribution in humanized mouse models

FIGURE 1: In vivo PET imaging and ex vivo biodistribution of 89Zr-pembrolizumab in immunocompetent humanized NOG mice.

Mice were xenografted with A375M tumor cells and received tracer injection at day 0. For blocking studies huNOG mice received a 10-fold excess of unlabeled pembrolizumab (huNOG excess). As a control for nonspecific uptake huNOG mice were injected with 89Zr-IgG4. PET imaging performed on day 7 post injection (pi). (A) In vivo PET examples (maximum intensity projections) at day 7 pi showing uptake in tumor (T), axillary lymph nodes (LN), liver (L) and spleen (S). (B) In vivo uptake of 89Zr-pembrolizumab in spleen, lymph nodes (axillary), liver and tumor, at day 7 pi. Uptake is expressed as SUVmean. (C) Ex vivo biodistribution of 89Zr-pembrolizumab in humanized NOG mice. Uptake is expressed as mean radioactivity per gram tissue, adjusted for total body weight (SUVmean ex vivo), tumor-to-blood ratio (D) and tumor-to-muscle ratio (E) Data is expressed as median ± IQR; *p ≤ 0.05. BAT, brown adipose tissue; huNOG, humanized NOG mice; MLN, mesenteric lymph nodes; PET, positron emission tomography.

89Zr-pembrolizumab in huNOG mice showed higher uptake in lymphoid tissues compared with 89Zr-IgG4 control: spleen (SUVmean 13.9, IQR 7.1 to 21.4, NS, p = 0.254), mesenteric lymph nodes (SUVmean 2.3, IQR 1.4 to 4.4, NS, p = 0.114), salivary gland (SUVmean 2.1, IQR 1.2 to 2.9, NS, p = 0.635), bone marrow (SUVmean 8.8, IQR 7.6 to 10.0, NS, p = 1.714) and thymus (SUVmean 0.5, IQR 0.4 to 1.1, p = 0.1714), indicating that 89Zr-pembrolizumab uptake in these tissues is, at least partly, PD-1-mediated. 89Zr-pembrolizumab tissue-to-blood (T:B) and tissue-to-muscle (T:M) ratios in lymphoid organs confirmed high uptake in these tissues (figure 1D, E). Additionally, relatively high 89Zr-IgG4 uptake was found in spleen, bone marrow and liver compared with

FIGURE 2: IHC analysis of spleen, mesenteric lymph node and tumor tissue humanized NOG mice.

Formalin-fixed and paraffin embedded tissue blocks where cut into slices of 4 μM and stained for PD-1, CD3 and CD8 (40x). H&E staining served to analyze tissue viability and morphology (40x). Scalebar: 50 μm. IHC, immunohistochemical; PD-1, programmed cell death protein-1.

other organs, suggesting 89Zr-pembrolizumab uptake in these tissues is also due to Fcγ receptor (FcγR)-binding of the antibody’s Fc-tail. High 89Zr-IgG4 uptake was less evident in lymph nodes and thymus.

89Zr-pembrolizumab spleen uptake in huNOG mice was blocked by the addition of a 10-fold excess unlabeled pembrolizumab (SUVmean 30.5, IQR 15.8 to 67.7 versus SUVmean 5.1, IQR 4.3 to 7.0, p = 0.032) (figure 1B, C). Uptake in other lymphoid organs and liver was also reduced by addition of unlabeled mAb dose, whereas uptake in non-lymphoid tissues was unaffected (supplemental table S3). Tracer activity in blood pool was increased by addition of unlabeled mAb (SUVmean 0.1, IQR 0.0 to 1.8 to SUVmean 2.2, IQR 1.4 to 7.4), but uptake in tumor did not change.

Autoradiography confirmed PET imaging results on a macroscopic level, showing high uptake in spleens of huNOG mice compared with spleens of mice that received an additional unlabeled pembrolizumab dose (figure 3). Furthermore, comparable tumor uptake was found for different dose groups. IHC analysis on spleen and lymph node tissue of huNOG mice

89Zr-pembrolizumab biodistribution in humanized mouse models

revealed that PD-1, CD3 and CD8 positive cells were present. CD3 and CD8 cells were also present in tumor tissue of huNOG mice (figure 2), however, PD-1 staining of these tumors was negative.

89Zr-pembrolizumab biodistribution in NOG control mice clearly showed a different pattern than in huNOG mice, with high uptake in liver (SUVmean 16.9, IQR 5.1 to 26.2) and spleen (SUVmean 49.6, IQR 16.6 to 135.6), whereas 89Zr-pembrolizumab tumor uptake in NOG mice was similar to huNOG mice (SUVmean 9.3, IQR 4.5 to 15.7 vs SUVmean 5.1, IQR 3.3 to 8.9) (supplemental figure S3). High 89Zr-pembrolizumab spleen uptake in this model may be unexpected, since limited T-cells are present in NOG mice (supplemental figure S3). However, high spleen uptake in severely immunocompromised mice has been described previously and is potentially Fcγ receptormediated (23,24,32). Moreover, spleen weights in NOG mice were lower than in huNOG mice (NOG: 0.017 ±0.015 g; huNOG: 0.037 ±0.016 g, p = 0.036), which resulted in higher tracer uptake expressed as %ID per gram spleen tissue for NOG mice. A low spleen weight may result from high radiosensitivity of NOG splenocytes, which can lead to toxicity (33).

FIGURE 3: Autoradiography of spleen and tumor tissue of humanized NOG mice (huNOG).

Formalin-fixed and paraffin embedded tissue blocks where cut into slices of 4 μM. These slices were exposed to a phosphor imaging screen for 72 hours and were then scanned using a Cyclone phosphor imager.

Critical steps in 89Zr-pembrolizumab manufacturing

The production processes for N-sucDf-pembrolizumab intermediate product and 89Zrpembrolizumab for in vivo studies were modified to comply with GMP requirements. In the conjugation reaction, pH is increased from 4.5 to 8.5, performed in small titration steps, as described earlier by Verel et al (30). During this pH transition, precipitation occurred at 6.5 to 7.0, which was re-dissolved at pH >7.5. No precipitation was observed when pH was changed abruptly, for example, by buffer exchange, to pH 8.5 during conjugation and to pH 4.5 for removal of Fe(III). This indicates potential instability of pembrolizumab at pH 6.5 to 7.0. Formation of aggregates may be explained by the fact that pembrolizumab is an IgG4 type mAb, which forms non-classical disulfide bonds. In contrast, IgG1 type antibodies can only form classical disulfide bonds. There are many other determinants of antibody stability

besides disulfide bond formation, however, this phenomenon was not seen previously with the radiolabeling of IgG1 type antibodies (31,33,34).

Immunoreactivity was not affected when pembrolizumab showed precipitation during pH transition, demonstrated by comparable IRF for precipitated N-sucDf-pembrolizumab and for non-precipitated N-sucDf-pembrolizumab (supplemental figure S4). However, it is unknown whether the pembrolizumab structure is modified by the formation of precipitates. Therefore, the method for pH transition by buffer exchange was incorporated in the conjugation protocol for pembrolizumab. Production of clinical grade 89Zr-pembrolizumab was performed as previously described by Verel et al (30).

89Zr-pembrolizumab GMP validation

Three consecutive batches of conjugated and radiolabeled pembrolizumab were produced at clinical scale and complied with all release specifications (supplemental tables S1 and S2), indicating that our process for manufacturing clinical grade 89Zr-pembrolizumab is consistent and robust. 89Zr-pembrolizumab was obtained with a specific activity of 37 MBq/mg and mean IRF of 1.35±0.6 (n=3). Stability studies revealed that N-sucDf-pembrolizumab remained compliant to release specifications up to 6 months storage at −80 °C, therefore N-sucDfpembrolizumab shelf-life was set at 6 months. Stability studies are ongoing and shelf-life may be extended if future time points remain within specifications. Data obtained during process development and validation were used to compile the investigational medicinal product dossier (IMPD), which includes all information regarding quality control, production and validation of 89Zr-pembrolizumab. Based on this IMPD, 89Zr-pembrolizumab has been approved by competent authorities for use in clinical studies.

Discussion

This study reveals 89Zr-pembrolizumab whole-body distribution in tumor-bearing huNOG mice established with a broad set of developed immune cells. Tumor uptake of 89Zr-pembrolizumab was markedly lower than uptake in lymphoid tissues such as spleen, lymph nodes and bone marrow, but higher than uptake in other organs. Importantly, high uptake in lymphoid tissues could be reduced with a 10-fold excess of unlabeled pembrolizumab. This contrasts with 89Zr-pembrolizumab tumor uptake, which was not reduced by the addition of unlabeled pembrolizumab.

Our study nicely shows the in vivo behavior of 89Zr-pembrolizumab, which, apart from IgG pharmacokinetics determined by its molecular weight and Fc tail, is predominantly driven by its affinity for PD-1 (Kd:~30 pM). The PD-1 cell surface receptor is primarily expressed on activated T-cells and pro B-lymphocytes, which are abundantly present in our huNOG mouse model.

89Zr-pembrolizumab biodistribution in humanized mouse models

Lymphocytes are highly concentrated in organs that are key players of the immune system: lymph nodes, spleen, thymus, bone marrow as well as tonsils, adenoid and Peyer’s patches. From our PET imaging and ex vivo biodistribution data, we learned that 89Zr-pembrolizumab distributed mainly to lymphoid organs, where PD-1 expressing immune cells are present. 89Zr-pembrolizumab showed relatively low and variable tumor uptake, however, this uptake could be visualized with PET imaging 7 days pi and was higher than in non-lymphoid tissues. We hypothesized there may be PD-1-mediated 89Zr-pembrolizumab tumor uptake, but we also found tumor uptake for 89Zr-IgG4, suggesting part of the 89Zr-pembrolizumab tumor uptake is FcγR-mediated. In our mouse model, few PD-1 positive immune cells may have traveled to the tumor, thereby potentially limiting 89Zr-pembrolizumab tumor uptake. Interestingly, the addition of unlabeled pembrolizumab did not influence tumor uptake. This is likely caused by substantial increase of 89Zr-pembrolizumab in blood pool as a direct consequence of adding excess unlabeled pembrolizumab, warranting a continuous pembrolizumab supply to the tumor.

Ex vivo immunohistochemical analysis revealed CD3 and CD8 positive lymphocytes were present in tumor, but limited PD-1-expression was found. Immune checkpoint protein expression status in tumor-infiltrating lymphocytes is highly dynamic (35,36). This socalled ‘immune phenotype’ depends on several factors, including tumor type, location and mutational burden. Our results indicate that, whereas PD-1 expression may demonstrate large variation, 89Zr-pembrolizumab PET imaging is able to capture PD-1 dynamics in both tumor and healthy tissues.

Compared with earlier preclinical studies with radiolabeled pembrolizumab in the hNSG model, we found higher 89Zr-pembrolizumab uptake in spleen and other lymphoid tissues (23,24). This likely reflects the presence of multiple hematopoietic cell lineages, including B-cells, T-cells, NK-cells, dendritic cells and monocytes, and thus higher PD-1 expression, in our huNOG model compared with the hNSG model. Molecular imaging studies with radiolabeled antibodies generally show distribution to the spleen. It also known that Fc/FcγR-mediated immunobiology of the experimental mouse model plays a key role in the in vivo biodistribution and tumor targeting (33). In our mouse model, we also observed 89Zr-IgG4 uptake in lymphoid tissues, indicating 89Zr-pembrolizumab uptake in these organs may have an FcγR-mediated component. For most radiolabeled antibodies without an immune target, spleen uptake in patients is ~5 %ID/kg (37). This supports the idea that, independent of their target, antibodies often show distribution to the spleen. However, spleen uptake may be higher if PD-1 or PD-L1 is present.

Pembrolizumab has an IgG4κ backbone with a stabilizing SER228PRO sequence alteration in the Fc-region to prevent the formation of half molecules. The IgG4 backbone of pembrolizumab may slightly differ from the IgG4 control molecule that we used for our experiments, however, FcγR-binding affinity and kinetics of pembrolizumab appears to be very similar to IgG4 (38). We, therefore, consider the used IgG4 control molecule to provide a useful indication of the extent of FcγR-mediated uptake. In this respect, FcγR-mediated uptake may be present in the spleen but potentially also in liver and tumor, since these tissues demonstrate relatively high uptake of 89Zr-IgG4

PD-1 is predominantly expressed on activated T-cells while its ligand PD-L1 is expressed by a broader range of immune cells as well as tumor cells. It is therefore to be expected that biodistribution of antibody tracers targeting PD-L1 may deviate from the biodistribution results that we described here for 89Zr-pembrolizumab. In table 1, we presented an overview of preclinical imaging and biodistribution studies using anti-PD-1 and anti-PD-L1 tracers. Data turned out to be highly variable, mostly focused on tumor and not on the immune system, and therefore not just comparable. From our results, we increasingly realize that it is extremely important for interpretation of these type of data to know the characteristics of the antibody (origin, cross-reactivity, Fc-backbone, target, target-affinity and dose), the animal model (mouse strain, age, immune status and tumor cell line) and time points, variables we detailed in the table.

As for preclinical studies, data on the distribution of PD-1 and PD-L1 targeting antibodies to lymphoid organs in patients is still limited. A clinical imaging study in 13 patients demonstrated modest 89Zr-nivolumab spleen uptake of SUVmean 5.8±0.7, whereas uptake of this radiolabeled antibody targeting PD-1 in other lymphoid tissues was not addressed (39). 89Zr-atezolizumab (anti-PD-L1 antibody) imaging in 22 patients revealed spleen uptake with an SUVmean of 15. 89Zr-atezolizumab also distributed to other lymphoid tissues and sites of inflammation, whereas uptake in non-lymphoid organs was low. The high spleen uptake could at least partly be explained by presence of PD-L1 in endothelial littoral cells of the spleen (40). To perceive what can be expected for 89Zr-pembrolizumab PET imaging in patients, how results may be interpreted and potentially translated to predicting response, knowledge on which immune cells express PD-1 and where these cells are located in the human body is of utmost importance.

With our study, we validated the use of 89Zr-pembrolizumab PET imaging to evaluate PD-1mediated uptake in tumor and immune tissues in a setting that allowed for comparing tracer uptake and whole tumor tissue analysis. To enable evaluation of 89Zr-pembrolizumab biodistribution in humans, we developed clinical grade 89Zr-pembrolizumab. Clinical 89Zr-

89Zr-pembrolizumab biodistribution in humanized mouse models

pembrolizumab PET imaging in patients with melanoma and NSCLC before treatment with pembrolizumab is currently performed at our center (ClinicalTrials.gov Identifier NCT02760225), and may elucidate if tracer tumor uptake correlates to response and if uptake in healthy PD-1 expressing tissues correlates to toxicity.

Conclusion

We demonstrated the in vivo biodistribution of 89Zr-pembrolizumab in humanized mice, and found uptake in tumor with the highest uptake in the lymphoid system, reflecting the presence of PD-1. Insight in the in vivo behavior and biodistribution of immune checkpoint targeting monoclonal antibodies might aid in better understanding immune checkpoint inhibition therapy and could potentially help explaining variation in response as well as potential toxicity due to uptake in healthy (immune) tissues.

Contributors

ELvdV was involved in project design and conceptualization, was involved in tracer development and GMP validation, wrote the IMPD, performed animal studies, performed ex vivo analyses, data analysis and wrote the manuscript; DG was involved in study conceptualization, data analysis, performed ex vivo analyses and wrote the manuscript; LPdJ was involved in tracer development and GMP validation, performed animal studies, performed ex vivo analyses and edited the manuscript; AJS was involved in GMP validation, wrote the IMPD and edited the manuscript; EGEdV was involved in project design and conceptualization, supervised the study and edited the manuscript; MNLdH was involved in project design and conceptualization, supervised the study and edited the manuscript. All authors read and approved the final manuscript.

Funding

The research leading to these results received funding from the Innovative Medicines Initiatives 2 Joint Undertaking under grant agreement No 116106 (TRISTAN). This Joint Undertaking receives support from the European Union’s Horizon 2020 research and innovation program and EFPIA.

Competing interests

EGEdV reports grants from IMI TRISTAN (GA no.116106), during the conduct of the study; consulting and advisory role for NSABP, Daiichi Sankyo, Pfizer, Sanofi, Merck, Synthon Biopharmaceuticals; grants from Amgen, Genentech, Roche, Chugai Pharma, CytomX Therapeutics, Nordic Nanovector, G1 Therapeutics, AstraZeneca, Radius Health, Bayer, all made available to the institution, outside the submitted work.

89Zr-pembrolizumab biodistribution in humanized mouse models

References

1. Postow MA, Sidlow R, Hellmann MD. Immune-Related adverse events associated with immune checkpoint blockade. N Engl J Med 2018;378:158–68.

2. Nguyen LT, Ohashi PS. Clinical blockade of PD1 and LAG3–potential mechanisms of action. Nat Rev Immunol 2015;15:45–56.

3. van der Veen EL, Bensch F, Glaudemans AWJM, et al. Molecular imaging to enlighten cancer immunotherapies and underlying involved processes. Cancer Treat Rev 2018;70:232–44.

4. Lamberts LE, Williams SP, Terwisscha van Scheltinga AGT, et al. Antibody positron emission tomography imaging in anticancer drug development. J Clin Oncol 2015;33:1491–504.

5. Heskamp S, Hobo W, Molkenboer-Kuenen JDM, et al. Noninvasive imaging of tumor PD-L1 expression using radiolabeled anti-PD-L1 antibodies. Cancer Res 2015;75:2928–36.

6. Josefsson A, Nedrow JR, Park S, et al. Imaging, biodistribution, and dosimetry of radionuclide-labeled PD-L1 antibody in an immunocompetent mouse model of breast cancer. Cancer Res 2016;76:472–9.

7. Chatterjee S, Lesniak WG, Gabrielson M, et al. A humanized antibody for imaging immune checkpoint ligand PD-L1 expression in tumors. Oncotarget 2016;7:10215–27.

8. Lesniak WG, Chatterjee S, Gabrielson M, et al. PD-L1 Detection in Tumors Using [64Cu]-atezolizumab with PET. Bioconjug Chem 2016;27:2103–10.

9. Nedrow JR, Josefsson A, Park S, et al. Imaging of programmed cell death ligand 1: impact of protein concentration on distribution of anti-PD-L1 SPECT agents in an immunocompetent murine model of melanoma. J Nucl Med 2017;58:1560–6.

10. Kikuchi M, Clump DA, Srivastava RM, et al. Preclinical immunoPET/CT imaging using Zr-89-labeled antiPD-L1 monoclonal antibody for assessing radiation-induced PD-L1 upregulation in head and neck cancer and melanoma. Oncoimmunology 2017;6:e1329071.

11. Truillet C, Oh HLJ, Yeo SP, et al. Imaging PD-L1 expression with ImmunoPET. Bioconjug Chem 2018;29:96–103.

12. Heskamp S, Wierstra PJ, Molkenboer-Kuenen JDM, et al. Pd-L1 microSPECT/CT imaging for longitudinal monitoring of PD-L1 expression in syngeneic and humanized mouse models for cancer. Cancer Immunol Res 2019;7:150–61.

13. Chatterjee S, Lesniak WG, Miller MS, et al. Rapid PD-L1 detection in tumors with PET using a highly specific peptide. Biochem Biophys Res Commun 2017;483:258–63.

14. Gonzalez Trotter DE, Meng X, McQuade P, et al. In vivo imaging of the programmed death ligand 1 by 18F positron emission tomography. J Nucl Med 2017;25:1852–7.

15. Broos K, Keyaerts M, Lecocq Q, et al. Non-Invasive assessment of murine PD-L1 levels in syngeneic tumor models by nuclear imaging with nanobody tracers. Oncotarget 2017;8:41932–46.

16. Maute RL, Gordon SR, Mayer AT, et al. Engineering high-affinity PD-1 variants for optimized immunotherapy and immuno-PET imaging. Proc Natl Acad Sci U S A 2015;112:E6506–14.

17. Donnelly DJ, Smith RA, Morin P, et al. Synthesis and biologic evaluation of a novel 18F-labeled adnectin as a PET radioligand for imaging PD-L1 expression. J Nucl Med 2018;59:529–35

18. Mayer AT, Natarajan A, Gordon SR, et al. Practical immuno-PET radiotracer design considerations for human immune checkpoint imaging. J Nucl Med 2017;58:538–46.

19. Natarajan A, Patel CB, Ramakrishnan S, et al. A novel engineered small protein for positron emission tomography imaging of human programmed death ligand-1: validation in mouse models and human cancer tissues. Clin Cancer Res 2019;25:1774–85.

20. De Silva RA, Kumar D, Lisok A, et al. Peptide-based 68Ga-PET radiotracer for imaging PD-L1 expression in cancer. Mol Pharm 2018;15:3946–52.

21. Kumar D, Lisok A, Dahmane E, et al. Peptide-based PET quantifies target engagement of PD-L1 therapeutics. J Clin Invest 2019;129:616–30.

22. Natarajan A, Mayer AT, Xu L, et al. Novel radiotracer for immunoPET imaging of PD-1 checkpoint expression on tumor infiltrating lymphocytes. Bioconjug Chem 2015;26:2062–9.

23. Natarajan A, Mayer AT, Reeves RE, et al. Development of novel immunoPET tracers to image human PD-1 checkpoint expression on tumor-infiltrating lymphocytes in a humanized mouse model. Mol Imaging Biol 2017;19:903–14.

24. England CG, Ehlerding EB, Hernandez R, et al. Preclinical pharmacokinetics and biodistribution studies of 89Zr-labeled pembrolizumab. J Nucl Med 2017;58:162–8.

25. England CG, Jiang D, Ehlerding EB, et al. 89Zr-labeled nivolumab for imaging of T-cell infiltration in a humanized murine model of lung cancer. Eur J Nucl Med Mol Imaging 2018;45:110–20.

26. Cole EL, Kim J, Donnelly DJ, et al. Radiosynthesis and preclinical PET evaluation of 89Zr-nivolumab (BMS936558) in healthy non-human primates. Bioorg Med Chem 2017;25:5407–14

27. Natarajan A, Patel CB, Habte F, et al. Dosimetry prediction for clinical translation of 64Cu-pembrolizumab ImmunoPET targeting human PD-1 expression. Sci Rep 2018;8:633.

28. Hettich M, Braun F, Bartholomä MD, et al. High-Resolution PET imaging with therapeutic antibody-based PD-1/PD-L1 checkpoint tracers. Theranostics 2016;6:1629–40.

29. De La Rochere P, Guil-Luna S, Decaudin D, et al. Humanized mice for the study of immuno-oncology. Trends Immunol 2018;39:748–63.

30. Verel I, Visser GWM, Boellaard R, et al. 89Zr immuno-PET: comprehensive procedures for the production of 89Zr-labeled monoclonal antibodies. J Nucl Med 2003;44:1271–81.

31. Nagengast WB, de Vries EG, Hospers GA, et al. In vivo VEGF imaging with radiolabeled bevacizumab in a human ovarian tumor xenograft. J Nucl Med 2007;48:1313–9.

32. Liu H, May K. Disulfide bond structures of IgG molecules: structural variations, chemical modifications and possible impacts to stability and biological function. MAbs 2012;4:17–23

33. Sharma SK, Chow A, Monette S, et al. Fc-Mediated anomalous biodistribution of therapeutic antibodies in immunodeficient mouse models. Cancer Res 2018;78:1820–32.

34. Dijkers ECF, Kosterink JGW, Rademaker AP, et al. Development and characterization of clinical-grade 89Zr-trastuzumab for HER2/neu immunoPET imaging. J Nucl Med 2009;50:974–81.

35. Havel JJ, Chowell D, Chan TA. The evolving landscape of biomarkers for checkpoint inhibitor immunotherapy. Nat Rev Cancer 2019;19:133–50.

36. Simon S, Labarriere N. Pd-1 expression on tumor-specific T cells: friend or foe for immunotherapy? Oncoimmunology 2017;7:e1364828.

37. Bensch F, Smeenk MM, van Es SC, et al. Comparative biodistribution analysis across four different 89Zrmonoclonal antibody tracers-The first step towards an imaging warehouse. Theranostics 2018;8:4295–304.

38. Zhang T, Song X, Xu L, et al. The binding of an anti-PD-1 antibody to FcγRI has a profound impact on its biological functions. Cancer Immunol Immunother 2018;67:1079–90.

39. Niemeijer AN, Leung D, Huisman MC, et al. Whole body PD-1 and PD-L1 positron emission tomography in patients with non-small-cell lung cancer. Nat Commun 2018;9:4664.

40. Bensch F, van der Veen EL, Lub-de Hooge MN, et al. 89Zr-atezolizumab imaging as a non-invasive approach to assess clinical response to PD-L1 blockade in cancer. Nat Med 2018;24:1852–8.

89Zr-pembrolizumab biodistribution in humanized mouse models

Supplementary data

FIGURE S1: Flow chart of the manufacturing process of the conjugated N-sucDf pembrolizumab and the 89Zr-pembrolizumab formulation and filling process, including in process control (IPC) and release quality control (QC) steps.

FIGURE S2: Representative SE-HPLC chromatograms for quality control of N-sucDf pembrolizumab and 89Zr-pembrolizumab.

(A) SE-HPLC chromatograms for determination of the final number of TFP-N-sucDf ligands per antibody (chelation ratio). Left panel shows chromatograms before purification, right panel after purification. (B) SE-HPLC chromatograms for determination of stability. Left panels shows chromatograms for stability of N-sucDf-pembrolizumab and 89Zr-pembrolizumab at t = 0 and t = 6 months. Right panel shows chromatograms for stability of 89Zr-pembrolizumab at t = 0, t = 4 h and t = 72 h. Abbreviations: SE-HPLC: size exclusion high-performance liquid chromatography.

89Zr-pembrolizumab biodistribution in humanized mouse models

Mice were xenografted with A375M tumor cells and received tracer injection at day 0. PET imaging performed on day 7 pi. (A) In vivo PET example (maximum intensity projection) at day 7 pi showing uptake in tumor (T) uptake, lymph nodes (LN), liver (L) and spleen (S). (B) In vivo uptake of 89Zr-pembrolizumab in liver, spleen, lymph nodes (axillary), liver and tumor, at day 7 pi. Uptake is expressed as SUVmean (C) Ex vivo biodistribution of 89Zr-pembrolizumab in NOG mice. Uptake is expressed as mean radioactivity per gram tissue, adjusted for total body weight (SUVmean ex vivo). (D) Ex vivo biodistribution of 89Zr-pembrolizumab in NOG mice, expressed as tumor-to-blood ratio. (E) Ex vivo biodistribution of 89Zr-pembrolizumab in NOG mice, expressed as tumor-to-muscle ratio. Data expressed as median ± IQR; *p ≤ 0.05. Abbreviations: MLN: mesenteric lymph nodes; BAT: brown adipose tissue. (F) IHC analysis and autoradiography of spleen and tumor tissue of NOG mice. Formalin-fixed and paraffin embedded (FFPE) tissue blocks where cut into slices of 4 μM and stained for PD-1, CD3 and CDS (40x). Hematoxylin & eosin (H&E) staining served to analyze tissue viability and morphology (40x). For autoradiography slices were exposed to a phosphor imaging screen for 72 hours and were then scanned using a Cyclone phosphor imager. Scalebar: 50 μm.

FIGURE S3: In vivo PET imaging and ex vivo biodistribution of 89Zr-pembrolizumab in immunodeficient NOG mice.

89Zr-pembrolizumab biodistribution in humanized mouse models

FIGURE S4: Immunoreactivity assays of different batches of N-sucDf-pembrolizumab. N sucDf-pembrolizumab, which was precipitated during conjugation (orange) and N-sucDf pembrolizumab which was not precipitated during conjugation, by changing pH directly in one step (blue).

TABLE S1: GMP manufacturing of N-sucDf-pembrolizumab and 89Zr-pembrolizumab.

Specification Batch 1 Batch 2 Batch 3 Batch 1 6 M at 80°C

N sucDf pembrolizumab

Colorless Colorless Colorless Colorless

1.55

SucDfpembrolizumab

Not applicable

Not applicable

Not

protocol

Concentration

9.0 11.0 mg/mL 10.13 mg/mL 9.65 mg/mL 10.13 mg/mL 10.83 mg/mL

pH pH 4.0 6.0 4.7 4.9 5.0 4.66

Radiochemical purity (test labeling)

> 95% 99.3% 99.4% 99.3% 99.7%

Endotoxins ≤ 2.5 EU/mL < 2.5 EU/mL < 2.5 EU/ml < 2.5 EU/mL ND according to protocol

Sterility Sterile Sterile Sterile Sterile ND according to protocol

TABLE S1: Continued.

Residual solvents (ACN)

< 410 ppm < 100 ppm < 100 ppm < 100 ppm ND according to protocol

89Zr pembrolizumab

Appearance Colorless to light yellow Colorless Colorless Colorless Colorless

Radiochemical purity prepurification

≥ 70% 97.4% 97.8% 95.5% 97.6%

Radiochemical purity postpurification ≥ 95% 99.3% 99.4% 99.3% 99.7%

pH pH 5.0 8.0 5.65 5.18 5.66 6.1

Filter integrity ≤ 20% 14% 14% 12% 15% Impurities of 89Zr pembrolizumab < 10% < 10% < 10% < 10% < 10%

Concentration For information only 0.262 mg/mL 0.094 mg/mL 0.133 mg/mL 0.200 mg/mL

Bacterial endotoxins ≤ 2.5 EU/mL 0.450 EU/mL 0.412 EU/mL 0.532 EU/mL ND according to protocol

Sterility Sterile Sterile Sterile Sterile ND according to protocol

Immunoreactivity towards PD 1 > 70% 202% 88% 114% 231%

Abbreviations: ACN: acetonitrile; ND: not determined; ppm: parts per million

Batch 1, 2, and 3 fulfill release criteria. In addition, stability data are shown for N-sucDf-pembrolizumab stored at -80 °C for 6 months. All release specifications are still met.

TABLE S2: Stability data of 89Zr-pembrolizumab.

Test Specification

Batch 1

Original result 4 h RT syringe

vial

Original result

2

RT syringe 168 h 2

vial

Appearance Colorless to light yellow Colorless Colorless Colorless Colorless Colorless Colorless pH pH 4.0 6.0 5.65 6.16 6.33 5.40 5.75 6.14

Radiochemical purity (postpurification) > 95% 99.3% 98.6% 97.1% 99.6% 99.1% 98.2%

Impurities of 89Zr pembrolizumab < 10% < 10% < 10% < 10% < 10% < 10% < 10%

Stability data are shown for two batches 89Zr-pembrolizumab stored 4 h at room temperature (RT) or 168 h at 2-8 °C in the vial. All release specifications are still met.

89Zr-pembrolizumab imaging as a non-invasive approach to assess clinical response to PD-1 blockade in cancer

I C Kok 1, J S Hooiveld 1 *, P P van de Donk 1 *, D Giesen 1, E L van der Veen 2, M N Lub-de Hooge 2, A H Brouwers 3, T J N Hiltermann 4, A J van der Wekken 4, L B M Hijmering-Kappelle 4, W Timens 5, S G Elias 6, G A P Hospers 1, H J M Groen 4 , W Uyterlinde 7, B van der Hiel 8, J B Haanen 7, D J A de Groot 1, M Jalving 1 , E G E de Vries 9

* These authors contributed equally to this work

1 Department of Medical Oncology, 2 Department of Clinical Pharmacy and Pharmacology, 3 Medical Imaging Center, 4 Department of Pulmonary Medicine, and 5 Department of Pathology and Medical Biology, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands;

6 Department of Epidemiology, Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht University, Utrecht, The Netherlands; 7 Department of Medical Oncology, Netherlands Cancer Institute, Amsterdam, The Netherlands. 8 Department of Nuclear Medicine, Netherlands Cancer Institute, Amsterdam, The Netherlands; 9 Department of Medical Oncology, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands. e.g.e.de.vries@umcg.nl. Ann Oncol. 2022;33(1):80-88.

Abstract

Background: Programmed cell death protein 1 (PD-1) antibody treatment is standard of care for melanoma and non-small-cell lung cancer (NSCLC). Accurately predicting which patients will benefit is currently not possible. Tumor uptake and biodistribution of the PD-1 antibody might play a role. Therefore, we carried out a positron emission tomography (PET) imaging study with zirconium-89 (89Zr)-labeled pembrolizumab before PD-1 antibody treatment.

Patients and methods: Patients with advanced or metastatic melanoma or NSCLC received 37 MBq (1 mCi) 89Zr-pembrolizumab (~2.5 mg antibody) intravenously plus 2.5 or 7.5 mg unlabeled pembrolizumab. After that, up to three PET scans were carried out on days 2, 4, and 7. Next, PD-1 antibody treatment was initiated. 89Zr-pembrolizumab tumor uptake was calculated as maximum standardized uptake value (SUVmax) and expressed as geometric mean. Normal organ uptake was calculated as SUVmean and expressed as a mean. Tumor response was assessed according to (i)RECIST v1.1.

Results: Eighteen patients, 11 with melanoma and 7 with NSCLC, were included. The optimal dose was 5 mg pembrolizumab, and the optimal time point for PET scanning was day 7. The tumor SUV max did not differ between melanoma and NSCLC (4.9 and 6.5, P = 0.49). Tumor 89Zr-pembrolizumab uptake correlated with tumor response (Ptrend = 0.014) and progressionfree (P = 0.0025) and overall survival (P = 0.026). 89Zr-pembrolizumab uptake at 5 mg was highest in the spleen with a mean SUVmean of 5.8 (standard deviation ±1.8). There was also 89Zrpembrolizumab uptake in Waldeyer's ring, in normal lymph nodes, and at sites of inflammation.

Conclusion: 89Zr-pembrolizumab uptake in tumor lesions correlated with treatment response and patient survival. 89Zr-pembrolizumab also showed uptake in lymphoid tissues and at sites of inflammation.

Introduction

Immune checkpoint blockade with monoclonal antibodies targeting programmed cell death protein 1 (PD-1), such as pembrolizumab, is a standard of care for numerous tumor types, including melanoma and non-small-cell lung cancer (NSCLC) (1,2). However, only a subset of treated patients respond to this therapy, while all patients are at risk for treatment-induced side-effects (3). Immunohistochemical programmed death-ligand 1 (PD-L1) expression in the tumor is a predictive biomarker for patients with NSCLC (4). However, not all patients with high tumor PD-L1 expression respond to PD-1 blockade (5,6). Moreover, patients lacking immunohistochemical tumor PD-L1 expression can still experience treatment benefit from PD-1 blockade. The Food and Drug Administration also approved pembrolizumab for adults and

children with tumor mutational burden-high solid tumors (7). However, again, not all patients with such a high mutational burden will respond. For melanoma, no approved companion diagnostic for patient selection is available.

There are likely multiple factors involved in tumor response to PD-1 antibodies, including PD-1 expression by tumor-infiltrating T cells and the amount of PD-1 antibody reaching its target. Molecular imaging using a radiolabeled antibody and positron emission tomography (PET) can provide insight into these aspects by giving non-invasive whole-body information.

The impact of PD-1 differs between tumor types. CD8 and PD-1 RNA expression by tumorinfiltrating T cells appear to be better determinants of response to immune checkpoint inhibitors in patients with melanoma than in patients with NSCLC (8). Therefore, it is interesting to explore whether the uptake of PD-1 antibody varies between tumor types. Moreover, imaging with a PD-1 antibody will provide information about its biodistribution and uptake by the immune system, which is currently unknown.

Therefore, we carried out a study with zirconium-89 (89Zr)-pembrolizumab in patients with locally advanced or metastatic melanoma and NSCLC. The aim was to assess 89Zrpembrolizumab tumor uptake and whole-body biodistribution before treatment with a PD-1 antibody and explore its relationship with patient outcome.

Patients and methods

Patient population

Patients with histologically or cytologically documented locally advanced or metastatic melanoma or NSCLC, eligible for PD-1 antibody treatment, were included. Other inclusion criteria were age ≥18 years, Eastern Cooperative Oncology Group performance status of 0-1, ability to comply with protocol, life expectancy ≥12 weeks, and RECIST v1.1 measurable disease (9). The study was approved by the Medical Ethical Committee of the University Medical Center Groningen (UMCG) and registered with ClinicalTrials.gov (NCT02760225). All patients gave written informed consent. All procedures carried out in the study involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Study design

This two-center study was carried out in the UMCG and the Netherlands Cancer InstituteAntoni van Leeuwenhoek Hospital (NKI-AvL). The study contained two parts, parts A and B. In part A, the optimal tracer protein dose of 89Zr-pembrolizumab and the optimal time point for

89Zr-pembrolizumab imaging to assess clinical response

PET imaging were assessed. Patients received ~2.5 mg pembrolizumab labeled with 37 MBq (1 mCi) 89Zr-oxalate intravenously, combined with 2.5 mg or 7.5 mg unlabeled pembrolizumab, resulting in a total dose of 5 or 10 mg. These two doses were based on the pharmacokinetic results of a phase I study with pembrolizumab (10). 89Zr-pembrolizumab PET scans were carried out on days 2, 4, and 7 after tracer injection. The unlabeled antibody dose was considered sufficient when the mean standardized uptake value (SUVmean) in the blood pool on day 4 was comparable to other 89Zr-monoclonal antibodies with well-known kinetics over time (11,12). In part B, PET imaging was carried out at the optimal pembrolizumab dose and day of imaging as determined in part A.

When feasible, a tumor biopsy was carried out in part B shortly after the day 7 PET scan. Participation in the 89Zr-pembrolizumab PET imaging study was followed by PD-1 antibody treatment as per standard of care (pembrolizumab or nivolumab ± ipilimumab). Tumor response assessment was carried out every 12 weeks from the start of treatment, according to RECIST v1.1, and when applicable, iRECIST (9).

89Zr-pembrolizumab PET scanning

89Zr-pembrolizumab was produced as described previously (13). PET scans were combined with a low-dose computed tomography (CT) scan for attenuation correction and anatomic reference. In the UMCG, PET scans were carried out with a Biograph mCT 64-slice PET/CT camera or a Biograph mCT 40-slice PET/CT camera [both Siemens (Siemens Healthcare, Erlangen, Germany)] and in the NKI-AvL with a Philips (Philips Medical Systems, Best, The Netherlands) GEMINI TF Big Bore, 16-slice PET/CT camera. PET acquisitions on days 2 and 4 were carried out from head to upper thigh in up to six bed positions with 5 min/bed position and the legs in up to nine bed positions with 2 min/bed position. Day 7 post-injection head to upper thigh was scanned in up to six bed positions with 10 min/bed position and the legs in up to nine bed positions with 4 min/bed position. All PET images were reconstructed using the algorithm for multicenter 89Zr-monoclonal antibody PET scan trials (14). Image analysis was executed using the Accurate tool (IDL version 8.4; Harris Geospatial Solutions, Bloomfield, NJ) for volume-of-interest (VOI)-based background and lesion analysis (15). Firstly, when in the field of view, the tumor lesions were identified on a contrast-enhanced baseline CT of the chest and abdomen and contrast-enhanced CT or magnetic resonance imaging of the head. Secondly, spherical VOIs, encompassing each whole tumor lesion, were placed on the 89Zr-pembrolizumab PET/low-dose CT around all lesions identified on the diagnostic baseline CT. For normal organ biodistribution measurements, spherical VOIs with fixed sizes per organ were drawn. Bodyweight-corrected SUVs were calculated using the amount of injected activity, body weight, and the amount of radioactivity detected within a VOI. We report the SUV max for tumor lesions and the SUV mean for normal organ 89Zr-pembrolizumab uptake. 89Zr-

pembrolizumab uptake in non-malignant lymph nodes and Waldeyer's ring was compared qualitatively to the surrounding tissue uptake at day 7 post-injection.

Other study assessments

Available archival tumor tissues, obtained within 50 weeks before 89Zr-pembrolizumab administration, were collected. Fresh tumor biopsies obtained after the last PET scan were formalin-fixed and paraffin-embedded. Tumor tissue sections of 4 μM were stained with hematoxylin–eosin and immunohistochemically for PD-1, PD-L1, and CD8.

PD-1 expression was determined with the anti-PD-1 antibody ERP4877(2) (Abcam) as described previously (13), with a few modifications to increase signal-to-background ratio. PD-L1 expression was assessed with the anti-PD-L1 SP263 clone (Ventana Medical Systems) and CD8 with the mouse CD8 monoclonal antibody C4/144B (DAKO/Agilent), all according to the manufacturer's protocols. Human tonsil sections served as a positive control for PD-1 and PDL1 expression. For PD-L1, the percentage of tumor cells with PD-L1-positive membrane staining, at any intensity, was estimated and classified as 0%, <1%, 1%-49%, ≥50%. PD-1 was scored as present or absent in the tumor area. CD8 was scored as negative 0, sporadic 1+, average 2+, abundant 3+ within tumor cell areas, and separately in the stroma directly surrounding the tumor.

89Zr-pembrolizumab binding kinetics was determined by studying the accumulation and dissociation of PD-1-bound 89Zr-pembrolizumab in peripheral blood mononuclear cells (PBMCs) with a radioactive binding assay (Supplementary Methods). The PBMCs were pre-stimulated with phytohemagglutinin and interleukin-2 to increase PD-1 expression (Supplementary Figure S1). Internalization of pembrolizumab, chelator-conjugated (pembrolizumab-N-succinyl desferal; pembrolizumab-N-sucDf), pembrolizumab-N-sucDf, and the anti-PD-1-antibody nivolumab was assessed in PD-1-expressing PBMC flow cytometrically (Supplementary Methods).

Statistical analyses

We used standard descriptive statistics to report patient characteristics. PET analyses were primarily carried out for all tumor lesions and after excluding small lesions (defined as tumor lesions with a long axis <1 cm and malignant lymph nodes with a short axis <1 cm) to take partial volume effects into account (16). Previously irradiated lesions were excluded. To study 89Zr-pembrolizumab tumor uptake in relation to time post-injection, primary tumor type, metastatic site, immunohistochemistry (PD-1, PD-L1 CD8), and best tumor response at a patient level, we used linear mixed-effect models with the natural logarithm of SUVmax as dependent variable to account for its right-skewed distribution and used random intercepts to

89Zr-pembrolizumab imaging to assess clinical response

account for within-patient clustering of data (and, if applicable, within-tumor clustering). From these models, we report geometric mean SUVmax values and corresponding 95% confidence intervals (CIs) and Wald P values, after back transformation of the estimated mean ln(SUVmax), using Satterthwaite's degrees of freedom approximation and restricted maximum likelihood. Likelihood ratio P values were obtained under maximum likelihood. We used Akaike's information criterion to select the best fitting time post-injection versus tumor uptake curve from a linear, log-linear, or quadratic fit. We similarly assessed tumor-to-background uptake ratios and the biodistribution in normal tissues using SUVmean as the dependent variable (the latter without natural logarithmic transformation due to its approximate normal distribution). These time–activity curves were assessed in cohort A, adjusted for the tracer dose, and projected at the 5-mg dose level. All other analyses were carried out in cohorts A and B combined.

Finally, to explore the relationship between per-patient geometric mean pre-treatment 89Zrpembrolizumab tumor uptake and progression-free survival (PFS) and overall survival (OS), we used Cox regression with Firth's bias correction for small samples. For this, we analyzed perpatient geometric mean tumor SUVmax continuously, assuming linearity and binning patients in above and below median uptake groups.

All reported P values are two-sided, and we used R version 3.2.1 for macOS for analyses, particularly using packages lme4 (1.1-11), lmerTest (2.0-20), and coxphf (1.11).

Results

Eighteen patients were enrolled between October 2016 and January 2019, 6 patients in part A, 12 in part B, 11 with metastatic melanoma, and 7 with metastatic NSCLC. Patient characteristics are shown in Table 1. With data cut-off set at 27 January 2021, their median follow-up was 22 months (range: 4-50+ months). No tracer-related adverse events were observed following 89Zr-pembrolizumab administration.

Dose finding and imaging time point for 89Zr-pembrolizumab PET

In part A, the first two patients received 2.5 mg 89Zr-pembrolizumab plus 7.5 mg unlabeled pembrolizumab. This resulted in an SUVmean in the blood pool of 6.7 and 8.4 on day 2, 5.6 and 7.8 on day 4, and 4.0 and 5.3 on day 7 post-injection. The next four patients in cohort A received 2.5 mg 89Zr-pembrolizumab plus 2.5 mg unlabeled pembrolizumab, resulting in a median SUVmean in the blood pool of 8.7 (range: 8.1-9.6), 6.4 (5.4-7.0), and 4.5 (4.2-7.5) on days 2, 4, and 7 postinjection, respectively.

The six patients in part A had 35 tumor lesions, of which 5 were previously irradiated bone metastases. In the 30 non-irradiated lesions, the geometric mean SUVmax increased from day 2

TABLE 1: Patient characteristics.

Total number of patients 18

Median age, years (range) 55.3 (24.9 72.7)

Sex, n (%)

Male 10 (56)

Female 8 (44)

Primary tumor, n (%)

NSCLC 7 (39)

Melanoma 11 (61)

ECOG performance status, n (%)

0 11 (61)

1 7 (39)

Numer of lines of therapy, n (%)

0 15 (83)

1 3 (17)

EOCG, Eastern Cooperative Oncology Group; NSCLS, non small cell lung cancer.

to 7, from 5.1 (95% CI 3.0-8.6) to 6.7 (95% CI 3.9-11.3) (Figure 1A; adjusted for dose and projected at 5 mg). The two most frequent metastatic sites were lung (n = 16 lesions, five patients) and bone (n = 5 lesions, three patients). The increasing tumor-to-blood ratio from day 2 to 7 for all 30 lesions and lung and bone metastasis separately is shown in Figure 1B. Altogether, the 5-mg protein dose for 89Zr-pembrolizumab in part B was determined to be superior to the 10mg dose. The optimal time point, given the highest tumor-to-blood ratio, for PET imaging was day 7.

89Zr-pembrolizumab biodistribution

89Zr-pembrolizumab uptake measured over time, as carried out in cohort A, was low in the brain, lung, bone cortex, subcutaneous tissue, and the abdominal cavity (Figure 1C and D).

The 16 patients receiving 5 mg pembrolizumab showed high spleen and bone marrow uptake with a mean SUV mean of 5.8 ± 1.8 and 2.4 ± 0.9 on day 7, respectively. There was also 89Zrpembrolizumab uptake in the liver and kidneys (Supplementary Table S1). Uptake in the spleen increased from day 2 to 7, while uptake in the blood pool decreased, indicating specific 89Zrpembrolizumab uptake (Figure 1C and D). In 12 patients, there was clear uptake in Waldeyer's ring, and in 6 patients in normal lymph nodes (Supplementary Figure S2). Moreover, 89Zrpembrolizumab uptake was visually present at sites of inflammation, due to autoimmune disease, prior surgery, and infection (Supplementary Figure S3). No clear tracer uptake was seen in the tissues where four patients later experienced an immune-related adverse event to anti-PD-1 antibody therapy.

89Zr-pembrolizumab tumor uptake

Combining data from cohorts A and B, a total of 103 non-irradiated tumor lesions in 18 patients were analyzed. Tracer uptake in tumor lesions ranged from 0.08 to 34.5 SUVmax. Uptake in tumor

89Zr-pembrolizumab imaging to assess clinical response

FIGURE 1: 89Zr-pembrolizumab positron emission tomography (PET) biodistribution. (A) 89Zr-pembrolizumab tumor uptake [geometric mean maximum standardized uptake value (SUVmax), 95% confidence interval (CI)] of patients imaged within cohort A (n = 6 patients, 30 lesions) adjusted for pre-dose and projected to 5 mg total protein dose according to time post-injection. (B) The geometric mean tumor SUVmax-to-background SUVmean ratio of lung metastases (purple, n = 16; five patients), bone metastases (teal, n = 5; three patients), and tumor-to-blood ratio (red, n = 30; six patients), adjusted for dose and projected to 5 mg total protein dose according to time post-injection. Lines represent fitted regression lines accompanied by 95% CI bands. (C,D) Two graphs showing 89Zr-pembrolizumab tracer uptake [mean SUVmean (95% CI)] in healthy tissues at 2, 4, and 7 days after tracer injection of six patients. Two patients received 10 mg tracer and four patients 5 mg tracer. Tracer uptake (SUVmean) is adjusted for total protein dose and extrapolated to the 5-mg total protein dose level.

lesions varied between and sometimes also within patients (Figure 2). 89Zr-pembrolizumab uptake did not differ between patients with NSCLC and melanoma (geometric mean SUVmax of 6.5, 95% CI 3.4-12.5, n = 45 lesions versus 4.9, 95% CI 2.8-8.4, n = 58 lesions; P = 0.49; Figure 3B).

89Zr-pembrolizumab tumor uptake was strongly related to the lesion site (P = 0.000019), with

FIGURE 2: 89Zr-pem-brolizumab tumor uptake for the 103 lesions expressed as maximum standardized uptake value (SUVmax) on day 7 positron emission tomography (PET) scan, depicted per patient (n = 18). Patients are categorized per tumor type and ordered according to increasing geometric mean tumor SUV max per patient (horizontal black lines). Each tumor lesion is one circle, with the circle size corresponding to computed tomography (CT)-derived lesion size and color depicting lesion location. NSCLC, non-smallcell lung cancer; SD, standard deviation.

lymph node metastases showing the highest uptake with a geometric mean SUVmax of 6.7 (95% CI 4.4-10.4), and brain metastases the lowest with a geometric mean SUVmax of 1.9 (95% CI 1.1-3.3) (Figure 3C). Examples of 89Zr-pembrolizumab tumor uptake on PET are shown in Supplementary Figure S4.

Exclusion of small lesions (n = 30) yielded overall higher geometric mean SUVmax estimates but did not substantially change the relationship between uptake and tumor type or metastatic site described above (data not shown).

89Zr-pembrolizumab

PET uptake and response to therapy

Fourteen patients received pembrolizumab 2 mg/kg in a 3-weekly schedule and two patients received nivolumab 240 mg every 2 weeks. One patient with NSCLC received three cycles of pembrolizumab, followed by nivolumab. One patient with melanoma received ipilimumab 3 mg/kg and nivolumab 1 mg/kg every 3 weeks until cycle 4, followed by nivolumab monotherapy, 240 mg every 2 weeks. The three patients with brain metastases received radiotherapy shortly after the initiation of immunotherapy. Median treatment duration was 177 days (range: 21-717 days) overall, 185 days (range: 21-589 days) in patients with melanoma, and 119 days (range: 87-717 days) in patients with NSCLC. At the end of follow-up, all patients were off treatment.

89Zr-pembrolizumab imaging to assess clinical response

FIGURE 3: 89Zr-pembrolizumab tumor uptake on day 7.

(A) Maximum intensity projection of 89Zr-pembrolizumab positron emission tomography (PET) scan 7 days after tracer injection; red arrows indicate tumor lesions. (B) Violin plot of the distribution of tumor maximum standardized uptake value (SUVmax) day 7 according to primary tumor type with bottom and top

1% of SUV max values truncated (1st, 50th, and 99th SUV max percentile: 1.6, 6.8, 21.0 for non-small-cell lung cancer (NSCLC) and 0.1, 4.4, 28.4 for melanoma); black vertical lines are 95% confidence intervals (CIs) of geometric mean SUVmax, white dots within black lines and values below the violin plot are the actual geometric means; two-sided Wald P values for geometric mean SUVmax comparison between tumor types above the graph; NSCLC: 45 lesions in 7 patients, melanoma: 58 lesions in 11 patients. (C) Violin plot of the distribution of tumor SUV max according to lesion site for sites with at least 10 observations with bottom and top 1% of SUVmax values truncated (1st, 50th, and 99th SUV max percentile: 0.1, 1.5, 4.1 for the brain; 3.6, 4.9, 7.1 for soft tissue; 1.5, 4.8, 22.3 for lung; 2.8, 7.4, 30.8 for lymph node); otherwise representation as for panel B; brain: 10 lesions in 3 patients, soft tissue: 10 lesions in 1 patient, lung/pleural: 37 lesions in 4 patients, lymph node: 28 lesions in 10 patients. Likelihood ratio test for the overall effect of lesion site on geometric mean SUVmax, P = 0.000019.

Eight patients were alive, of whom three were without evidence of disease. Treatment response and reasons for termination of immunotherapy are shown in Table 2. The median follow-up time of the eight patients alive at data cut-off was 34 months (range: 22-50+ months).

The geometric mean SUVmax of the 103 tumor lesions was positively associated with tumor response (P for trend = 0.014; Figure 4A) and target lesion size change (Figure 4B). Patients with a high geometric mean 89Zr-pembrolizumab uptake (above the median of 5.8) showed a longer PFS and OS than patients with low uptake (P = 0.0025 and P = 0.026, respectively; Figure 4C and D). One of the two patients with high 89Zr-pembrolizumab uptake who died during follow-up experienced a partial response. However, this patient died 21 months after starting pembrolizumab treatment due to pembrolizumab-induced antiphospholipid syndrome, with intestinal ischemia and respiratory failure due to pneumonitis. When analyzed continuously, the hazard ratio for PFS for each unit decrease in geometric mean SUVmax per patient was 1.22 (95% CI 1.05-1.48; P = 0.0097), and 1.13 (95% CI 0.97-1.34; P = 0.13) for OS. The exclusion of small lesions did not substantially change the above-described relationship between uptake and patient outcome (data not shown).

Patients with melanoma, n (%)

Complete response 3 (27)

Partial response 3 (27)

Stable disease 2 (18)

Progressive disease 3 (27)

Patients with NSCLC, n (%)

Complete response 0 (0)

Partial response 3 (43)

Stable disease 1 (14)

Progressive disease 3 (43)

Reason for termination of treatment, n (%)

Progressive disease 9 (50)

Completed treatment 5 (28)

Immune related toxicity 4 (22)

NSCLC, non small cell lung cancer; PD 1, programmed cell death protein 1.

Other study assessments

TABLE 2: Achieved tumor response and reason for stopping PD-1 antibody.

Tumor samples of 13 patients were analyzed immunohistochemically, 10 were archival tissues, and 3 fresh tumor biopsies were taken after the last 89Zr-pembrolizumab PET scan. PD-1 expression was found in five samples. In these PD-1-positive samples, CD8 expression was also observed. Results are shown in Supplementary Table S2. Three of the PD-1-positive tumor tissues expressed PD-L1. PD-1 expression assessed immunohistochemically did not correlate with the geometric mean SUVmax 89Zr-pembrolizumab uptake of that patient (Supplementary Figure S5). Of the three fresh tumor biopsy samples, two contained enough tissue to assess PD-1 expression. Both lesions showed high uptake on the 89Zr-pembrolizumab PET scan (SUVmax

89Zr-pembrolizumab imaging to assess clinical response

FIGURE 4: 89Zr-pembrolizumab tumor uptake and clinical outcome measures. (A) 89Zr-pembrolizumab tumor uptake as geometric mean maximum standardized uptake value (SUVmax) on day 7 and best tumor response (n = 18 patients). Gray violin plot areas show the distribution of SUVmax at the tumor level per best response category, with bottom and top 1% values truncated [1st, 50th, and 99th

SUV max percentile: 0.1, 4.4, 14.5 for progressive disease (PD); 1.6, 4.2, 9.5 for stable disease (SD); 1.6, 9.9, 30.2 for partial response (PR); 4.2, 7.9, 15.4 for complete response (CR)]; points show geometric mean uptake per patient, with colors indicating tumor type [red, melanoma; dark blue, non-small-cell lung cancer (NSCLC)]; black vertical lines are 95% confidence intervals (CIs) of geometric mean SUVmax, and white dots within black lines and values below the violin plot are the actual geometric means; with two-sided Wald P values, supplemented with a two-sided likelihood ratio P for trend; PD = 27 lesions in six patients, SD = 29 in three patients, PR = 41 in six patients, CR = 6 in three patients. (B) Waterfall plots depicting percentage change in sum of longest diameters of the target lesions (SLD) from baseline [measured on computed tomography (CT)], with color scale indicating geometric mean SUVmax of the tumor lesions per patient; aindicates patient with PD, however no SLD change data are available. (C) Progression-free survival according to geometric mean tumor SUV max per patient (red depicts the group above and dark blue the group below the median geometric mean uptake of an SUVmax of 5.8). (D) Overall survival of the patients binned and represented as in panel C (red depicts the group above and dark blue the group below the median geometric mean uptake of an SUVmax of 5.8).

17.0 and 21.0), but were negative for PD-1 expression measured immunohistochemically. Small amounts of 89Zr-pembrolizumab accumulated in PD-1-expressing PBMCs; hence dissociation of PD-1-bound tracer occurred (Supplementary Figure S1). Pembrolizumab showed modest internalization in pre-stimulated, PD-1-expressing PBMCs with 13.6% (±10.7%) after 2-h incubation (Supplementary Figure S1). Pembrolizumab internalization was not affected by antibody conjugation. Internalization rates for pembrolizumab and nivolumab, which target distinct epitopes of PD-1, with different affinities, were comparable (17).

Discussion

In this study, we demonstrate that 89Zr-pembrolizumab PET imaging is a safe and non-invasive imaging modality for whole-body visualization of PD-1 and pembrolizumab biodistribution. Tumor 89Zr-pembrolizumab uptake correlated with tumor response, PFS, and OS. 89Zrpembrolizumab uptake was also seen in lymphoid tissues reflecting the presence of PD-1 in normal tissues and at sites of inflammation. These findings illustrate that all major sites for T cells are visualized with 89Zr-pembrolizumab PET imaging.

89Zr-labeled anti-PD-1 antibody nivolumab was studied in 13 patients with NSCLC (18). They report higher 89Zr-nivolumab uptake in tumor lesions responding to nivolumab treatment. However, they only report response for individual tumor lesions and not for the patient as a whole. Recently, another study using 89Zr-pembrolizumab was published performing PET scans in 12 patients with NSCLC before pembrolizumab monotherapy (19). This study implemented a different dosing strategy by administering a labeled dose of 2 mg without adding any unlabeled pembrolizumab. Fourteen days later, 2 mg labeled 89Zr-pembrolizumab was

administered on the same day as the first full therapeutic dose of 200 mg pembrolizumab. The 2-mg labeled dose alone suffers from fast clearance and early trapping in sink organs, whereas the 200-mg unlabeled pembrolizumab pre-dose resulted in low uptake in tumor lesions likely due to saturation. This study observed a trend between tracer uptake in tumor lesions and response to therapy. However, this was not statistically significant. We demonstrate that 89Zr-pembrolizumab uptake in tumor lesions correlates not only with response to therapy, but also with PFS and OS. Interestingly, for 89Zr-atezolizumab targeting PD-L1, we also observed a relationship between tumor 89Zr-atezolizumab uptake and response PFS, and OS (12).

Therefore, this is our second study that demonstrates that PET imaging using PD-1 and PD-L1 radiolabeled antibodies may predict response to therapy and survival. Both studies showing these correlations are of modest size. Therefore, whole-body tumor uptake of radiolabeled anti-PD-1 or anti-PD-L1 antibodies deserves to be studied in a larger patient population.

In melanoma, CD8 and PD-1 messenger RNA expression are better determinants of response than in NSCLC (8). In our relatively small number of patients with melanoma and NSCLC, we

89Zr-pembrolizumab imaging to assess clinical response

found similar 89Zr-pembrolizumab tumor uptake. We were unable to study the relationship between uptake and patient outcome within primary tumor subtypes due to the limited sample size. The antitumor efficacy of immune checkpoint inhibitors depends on multiple factors. Hence, a future study might also include immune checkpoint inhibitor tumor uptake as a key treatment response factor.

A tracer dose of 5 mg protein in total was found to be sufficient with adequate activity in the blood pool at day 4 post-injection, based on experience with other radiolabeled antibodies (11). In normal tissues, there was a clear 89Zr-pembrolizumab uptake in the spleen and bone marrow, and in most patients, uptake in Waldeyer's ring and part of the normal lymph nodes. Information about PD-1 antibody uptake in the lymphoid system was, until now, only available from preclinical studies. In humanized NOG mice engrafted with human CD34+ hematopoietic stem cells and xenografted with human A375M melanoma cells, the highest 89Zr-pembrolizumab uptake occurred in the spleen, mesenteric lymph nodes, bone marrow, thymus, and tumor (13). In healthy Cynomolgus monkeys, a similar biodistribution pattern was found with preferential uptake in lymph nodes, spleen, and tonsils (20). Tracer uptake in malignant lymph nodes was highest of all tumor localizations. Therefore, it might be interesting to evaluate whether 89Zr-pembrolizumab PET can distinguish benign from malignant lymph nodes. Our current study was not sufficiently large enough nor designed to address this question, but this may be relevant to assess in a future study.

Other PD-1 PET imaging studies report uptake in the spleen and bone marrow as a readout for the lymphoid system (18,19). The 89Zr-pembrolizumab study also observed uptake in nonmalignant lymph nodes. The SUVs found in the spleen and in the bone marrow were nearly identical and in line with the results of our study (18,19). In the 89Zr-atezolizumab PET imaging study, tracer uptake was higher in the spleen and bone marrow than 89Zr-pembrolizumab with an SUV mean of 14.9 and 3.1, respectively, at day 7 post-injection (12). This is likely because PD-L1 is next to lymphocytes, also expressed by macrophages, dendritic cells, and endothelial littoral cells in the spleen.

Interestingly, we also show 89Zr-pembrolizumab uptake at sites of chronic inflammation, a phenomenon we also found using 89Zr-atezolizumab (12). This is likely the result of the fact that PD-1 and PD-L1 are modulators of the immune response at sites of chronic inflammation (21,22).

89Zr-pembrolizumab and 89Zr-nivolumab seem to demonstrate similar uptake in tumor lesions, spleen, and bone marrow (18). This occurred despite major differences in their affinity for the PD-1 receptor. Pembrolizumab binds the CD loop of the PD-1 receptor with KD = 29 pM, while nivolumab targets the N-loop epitope with ~100-fold lower binding affinity. Moreover, after the

PD-1-bound antibody internalization, 89Zr is trapped in the cell. PET imaging at 7 days may, therefore, in part reflect residualized 89Zr. We found limited PD-1 expression in unstimulated PBMCs, which moderately increased after stimulation. Pembrolizumab and nivolumab showed comparable binding and internalization in these PD-1-expressing PBMCs. This further supports the result that PET imaging using these immune checkpoint inhibitors is not affected by their different affinities for PD-1.

In the current study, we observed no correlation between lymphocytic markers, such as PD-1 and CD8 expression, and 89Zr-pembrolizumab tumor uptake. Interestingly, in other PET imaging studies of immune checkpoint molecules, similar results are found (12,18,19). These discordant results between PET imaging and immunohistochemistry are likely due to the heterogeneity of PD-1 expression within and between tumor lesions. A biopsy specimen might not accurately reflect expression levels in all tumor lesions. The heterogeneity of tracer uptake in tumor lesions, as depicted in Figure 2, illustrates this problem. Several studies have found different expression levels of immune checkpoint molecules in different tumor lesions within the same patient (23-26). This further illustrates the possible additive value of whole-body PET scanning versus a tumor biopsy of a part of a single lesion. Limitations of this study are the number of patients included and the limited availability of tumor tissue for immunohistochemical analysis. A larger and more homogeneous study is required to validate these results. The collection of tumor tissue to correlate PET imaging findings with immunohistochemistry and autoradiography results will aid further validation of this approach.

In summary, this study shows that 89Zr-pembrolizumab tumor uptake with PET imaging correlates with response to PD-1 antibody treatment, including PFS and OS. These findings require validation in larger studies to prove their impact on patient selection for PD-1 blockade.

Funding

This work was supported by the Dutch Cancer Society Grant POINTING [grant number RUG 2016-10034] and the Innovative Medicines Initiatives 2 Joint Undertaking project TRISTAN [grant number GA no. 116106]. This Joint Undertaking receives support from the European Union's Horizon 2020 research and innovation program and EPFIA (no grant number).

Disclosure

MJ reports consultancy fees from AstraZeneca (paid to UMCG). JBH reports consultancy roles for Achilles Therapeutics, BioNTech, BMS, GSK, Immunocore, Instil Bio, Molecular Partners, MSD, Merck Serono, Neogene Therapeutics, Novartis, Pfizer, PokeAcel, Roche/Genentech, Sanofi, T-Knife, and Third Rock Ventures and research grants from Amgen, Asher-Bio, BMS, BioNTech, MSD, Novartis, and Neogene Therapeutics (paid to the Netherlands Cancer Institute). WT reports

89Zr-pembrolizumab imaging to assess clinical response

fees from Merck, Sharp, Dohme, and Bristol-Myers-Squibb (paid to UMCG). EGEdV reports an advisory role at Daiichi Sankyo, NSABP, and Sanofi and research funding from Amgen, AstraZeneca, Bayer, Chugai Pharma, Crescendo, CytomX Therapeutics, G1 Therapeutics, Genentech, Nordic Nanovector, Radius Health, Regeneron, Roche, Servier, and Synthon (paid to UMCG). GAPH reports consulting and advisory role at Amgen, Roche, MSD, BMS, Pfizer, Novartis, and Pierre Fabry and research funding from BMS and Seerave (paid to UMCG). TJNH reports consultancy fees (paid to UMCG) from BMS, MSD, Merck, Boehringer, AstraZeneca, and Roche. AJvdW reports an advisory role at Janssen, Takeda, and Boehringer-Ingelheim (paid to UMCG) and research funding from AstraZeneca, Boehringer-Ingelheim, Pfizer, Roche, and Takeda. MNL-dH reports research funding from Merck, Bayer, and Amgen (paid to UMCG). HJMG reports an advisory role at Eli Lilly, MSD, BMS, and Novartis. All other authors have declared no conflicts of interest.

References

1. Schmidt EV, Chisamore MJ, Chaney MF, et al. Assessment of clinical activity of PD-1 checkpoint inhibitor combination therapies reported in clinical trials. JAMA Netw Open. 2020;3:e1920833.

2. Boussiotis VA. Molecular and biochemical aspects of the PD-1 checkpoint pathway. N Engl J Med. 2016;375:1767-1778.

3. Xu C, Chen YP, Du XJ, et al. Comparative safety of immune checkpoint inhibitors in cancer: systematic review and network meta-analysis. Br Med J. 2018;363:k4226.

4. Dolled-Filhart M, Roach C, Toland G, et al. Development of a companion diagnostic for pembrolizumab in non-small cell lung cancer using immunohistochemistry for programmed death ligand-1. Arch Pathol Lab Med. 2016;140:1243-1249.

5. Aguiar PN, De Mello RA, Hall P, Tadokoro H, de Lima Lopes G. PD-L1 expression as a predictive biomarker in advanced non-small-cell lung cancer: updated survival data. Immunotherapy. 2017;9:499-506.

6. Rittmeyer A, Barlesi F, Waterkamp D, et al. Pembrolizumab versus docetaxel in patients with previously treated non-small-cell lung can- cer (OAK): a phase 3, open-label, multicentre randomised controlled trial. Lancet. 2017;389:255-265.

7. Marabelle A, Fakih M, Lopez J, et al. Association of tumour mutational burden with outcomes in patients with advanced solid tumours treated with pembrolizumab: prospective biomarker analysis of the multicohort, open-label, phase 2 KEYNOTE-158 study. Lancet Oncol. 2020;20:30445-302259.

8. Lee SJ, Ruppin E. Multiomics prediction of response rates to therapies to inhibit programmed cell death 1 and programmed cell death 1 ligand 1. JAMA Oncol. 2019;5:1614-1618.

9. Seymour L, Bogaerts J, Perrone A, et al. iRECIST: guidelines for response criteria for use in trials testing immunotherapeutics. Lancet Oncol. 2017;18:e143-e152.

10. Patnaik A, Kang SP, Rasco D, et al. Phase I study of pembrolizumab (MK-3475; anti-PD-1 monoclonal antibody) in patients with advanced solid tumors. Clin Cancer Res. 2015;21:4286-4293.

11. Bensch F, Smeenk MM, van Es SC, et al. Comparative biodistribution analysis across four different 89Zr-

89Zr-pembrolizumab imaging to assess clinical response

monoclonal antibody tracersdthe first step towards an imaging warehouse. Theranostics. 2018;8:42954304.

12. Bensch F, van der Veen EL, Lub-de Hooge MN, et al. 89Zr-atezolizumab imaging as a non-invasive approach to assess clinical response to PD-L1 blockade in cancer. Nat Med. 2018;24:1852-1858.

13. van der Veen EL, Giesen D, Pot-de Jong L, Jorritsma-Smit A, de Vries EGE, Lub-de Hooge MN. 89Zrpembrolizumab biodistribution is influenced by PD-1 mediated uptake in lymphoid organs. J Immunother Cancer. 2020;8:e000938.

14. Makris NE, Boellaard R, Visser EP, et al. Multicenter harmonization of 89Zr PET/CT performance. J Nucl Med. 2014;55:264-267.

15. Frings V, van Velden FH, Velasquez LM, et al. Repeatability of metabolically active tumor volume measurements with FDG PET/CT in advanced gastrointestinal malignancies: a multicenter study. Radiology. 2014;273:539-548.

16. Gallivanone F, Canevari C, Gianolli L, et al. A partial volume effect correction tailored for 18F-FDG-PET oncological studies. Biomed Res Int. 2013;2013:780458.

17. Fessas P, Lee H, Ikemizu S, Janowitz T. A molecular and preclinical comparison of the PD-1-targeted T-cell checkpoint inhibitors nivolu- mab and pembrolizumab. Semin Oncol. 2017;44:136-140.

18. Niemeijer AN, Leung D, Huisman MC, et al. Whole body PD-1 and PD- L1 positron emission tomography in patients with non-small cell lung cancer. Nat Commun. 2018;9:4664.

19. Niemeijer AN, Oprea Lager DE, Huisman MC, et al. First-in-human study of 89Zr-pembrolizumab PET/CT in patients with advanced stage non-small-cell lung cancer. J Nucl Med. 2021. https://doi.org/10.2967/ jnumed.121.261926.

20. Li W, Wang Y, Rubins D, et al. PET/CT Imaging of 89Zr-N-sucDf-pembrolizumab in healthy cynomolgus monkeys. Mol Imaging Biol. 2021;23:250-259.

21. Jubel JM, Barbati ZR, Burger C, Wirtz DC, Schildberg FA. The role of PD-1 in acute and chronic infection. Front Immunol. 2020;11:487.

22. Sharpe AH, Pauken KE. The diverse functions of the PD1 inhibitory pathway. Nat Rev Immunol. 2018;18:153167.

23. Madore J, Vilain RE, Menzies AM, et al. PD-L1 expression in melanoma shows marked heterogeneity within and between patients: implications for anti-PD-1/PD-L1 clinical trials. Pigment Cell Melanoma Res. 2015;28:245-253.

24. Madore J, Strbenac D, Vilain R, et al. PD-L1 negative status is associated with lower mutation burden, differential expression of immune-related genes, and worse survival in stage III melanoma. Clin Cancer Res. 2016;22:3915-3923.

25. Bassanelli M, Sioletic S, Martini M, et al. Heterogeneity of PD-L1 expression and relationship with biology of NSCLC. Anticancer Res. 2018;38:3789-3796.

26. Haragan A, Field JK, Davies MPA, Escriu C, Gruver A, Gosney JR. Heterogeneity of PD-L1 expression in non-small cell lung cancer: implications for specimen sampling in predicting treatment response. Lung Cancer. 2019;134:79-84.

Supplementary methods

Pembrolizumab internalization, accumulation and dissociation Peripheral blood mononuclear cells (PBMCs) were prepared from healthy blood donor buffy coats after informed consent (Sanquin) by Ficoll gradient centrifugation in LeucoSep-tubes (Greiner Bio-One). To increase PD-1 expression, PBMCs were stimulated with 6000 U/mL interleukin-2 (IL-2) and 10 μg/mL phytohemagglutinin (PHA) for 72 hours at 37 °C.

Internalization of pembrolizumab, pembrolizumab-N-sucDf, nivolumab or IgG4 control antibody (BioLegend; 403701) was studied as described previously (27). Primary antibodies were diluted in phosphate-buffered saline containing 2% fetal calf serum to 20 μg/mL and detected by PE-conjugated goat anti-human IgG secondary antibody (Southern Biotech; 2040-09). Samples were analyzed on a BD FACS Verse flow cytometer (BD Biosciences). Duplicate samples were measured for each treatment condition, corrected for background fluorescence and unspecific antibody binding. Data analysis was performed with FlowJo v10 (Tree Star). Surface receptor expression was expressed as mean fluorescent intensity and normalized to 100% PD-1-bound antibody at t = 0.

Next to internalization, we studied 89Zr-pembrolizumab binding to PD-1 in PBMCs. 89Zrpembrolizumab (0.5 μg-250 MBq/mg) was added to 450,000 stimulated PBMCs and incubated for 45 min on ice. For 89Zr-pembrolizumab accumulation, total PD-1-bound 89Zr-pembrolizumab was determined by measuring cell-associated activity in a calibrated well-type gamma counter, followed by incubation for 1 and 2 hours at 37 °C, while control samples were kept on ice. Cells were counted for remaining cell-associated activity, i.e., membrane-bound and internalized, in a calibrated well-type gamma counter. To study 89Zr-pembrolizumab dissociation, activity present in the supernatant after incubation for 1 and 2 hours at 37 °C was measured. Accumulation and dissociation were expressed as a percentage of total PD-1bound 89Zr-pembrolizumab at t = 0, corrected for 4 °C control samples.

Supplementary references

27. Kol A, Terwisscha van Scheltinga A, Pool M, Gerdes C, de Vries E, de Jong S. ADCC responses and blocking of EGFR-mediated signaling and cell growth by combining the anti-EGFR antibodies imgatuzumab and cetuximab in NSCLC cells. Oncotarget. 2017;8:45432-45446

Supplementary figures

SUPPLEMENTARY FIGURE 1: Pembrolizumab binding and internalization. (A) Schematic representation of PD-1-bound pembrolizumab internalization in PBMCs. For internalization experiments, PD-1-bound pembrolizumab to the cell membrane was detected by PE-conjugated antihuman IgG secondary antibody with flow cytometry or by 89Zr measurements in a gamma counter. (B) Accumulation and dissociation of 89Zr-pembrolizumab in stimulated PBMCs determined by measuring cell-associated activity and activity present in the cell medium with and without the presence of excess 89Zr-pembrolizumab. (C) Internalization of pembrolizumab and pembrolizumab-N-sucDf in pre-stimulated, PD-1-expressing PBMCs after incubation at 37 °C for 3 hours. (D) Internalization of pembrolizumab and nivolumab in PBMCs (n = 3). Results were normalized to 100% PD-1-bound antibody at t = 0. Data in B-D are presented as mean ± standard deviation.

89Zr-pembrolizumab imaging to assess clinical response

SUPPLEMENTARY FIGURE 2: 89Zrpembrolizumab PET-CT image examples at day 7 of uptake in tonsils and lymph nodes.

Red arrows indicate region of interest. All images are scaled 0-5 SUV. (A) Clear tracer uptake in tonsils of the waldeyer’s ring. (B) Uptake in a normal inguinal lymph node indicated by the red arrow.

Red arrows indicate the region of interest. All images are scaled 0-5 SUV. (A) Patient with Hashimoto’s thyroiditis. (B) Patient who underwent a craniotomy for a brain metastasis 59 days before the PET scan. (C) Patient with pleural effusion at the left side. Cytological analysis of the fluid showed a lymphocyte-rich exudate without malignant cells. (D) Patient with influenza-A infection at the time of the PET scan.

SUPPLEMENTARY FIGURE 3: 89Zr-pembrolizumab PET-CT image examples at day 7 of high uptake at sites of inflammation.

89Zr-pembrolizumab imaging to assess clinical response

SUPPLEMENTARY FIGURE 4: 89Zr-pembrolizumab PET-CT image examples at day 7 of tumor lesions in various patients.

Red arrows indicate tumor lesions. Images A and B are scaled 0-8 SUV, images C and D are scaled 0-5 SUV. (A) Tumor lesion left upper lobe, NSCLC. (B) Pulmonary metastasis, melanoma. (C) Brain metastasis, melanoma (SUVmax 4.20). (D) Abdominal wall metastasis, melanoma (SUVmax 5.02).

SUPPLEMENTARY FIGURE 5: 89Zr-pembrolizumab tumor uptake as geometric mean SUVmax on day 7 and immunohistochemical tumor stainings.

(A) Tracer uptake in tumor lesions (58 lesions) versus PD-1 staining (9 tumor biopsy samples). (B) Tracer uptake in tumor lesions (76 lesions) versus PD-L1 staining (13 biopsy samples). (C) Tracer uptake in tumor lesions (76 lesions) versus CD8 staining in the center of the tumor lesion (13 biopsy samples). (D) Tracer uptake in tumor lesions (76 lesions) versus CD8 staining in the stromal tissue surrounding the tumor (13 biopsy samples).

89Zr-pembrolizumab imaging to assess clinical response

SUPPLEMENTARY TABLE 1: 89Zr-pembrolizumab biodistribution mean SUVmean (± SD)

Supplementary Table 1. 89Zr pembrolizumab biodistribution mean SUVmean (± SD)

Organ Day 2 (n = 4) Day 4 (n = 4) Day 7 (n = 16)

Blood 8.7 (0.7) 6.3 (0.8) 5.0 (1.2) *1 Brain 0.3 (0.0) 0.2 (0.1) 0.2 (0.0)

Lung 1.2 (0.3) 0.6 (0.7) 1.0 (0.4) *2

Spleen 6.0 (0.3) 4.1 (2.7) 5.8 (1.8)

Liver 4.2 (0.3) 4.2 (0.3) 4.5 (1.2) Intestine 4.6 (1.7) 4.8 (3.2) 3.4 (1.3)

Kidney 5.0 (0.6) 4.5 (0.8) 4.3 (1.0)

Bone marrow 2.5 (0.4) 2.1 (0.3) 2.4 (0.9) *1

Bone cortex 0.8 (1.0) 0.7 (0.6) 1.2 (0.7)

Muscle 0.7 (0.2) 0.6 (0.4) 0.7 (0.2) Subcutis 0.2 (0.1) 0.2 (0.1) 0.1 (0.1)

*1 n = 15, *2 n = 14

SUPPLEMENTARY TABLE 2: Immunohistochemical staining tumor tissues

Supplementary Table 2. Immunohistochemical staining tumor tissues

Immunohistochemical staining Tumor biopsies, n (%)

PD L1 expression on tumor cells negative / 0% <1% 1 50% >50%

PD 1 expression on immune cells

Positive Negative 5 (56) 4 (44)

CD8 expression in tumor lesion negative / 0% sporadic / 1+ average / 2+ abundant / 3+

CD8 expression in tumor stroma negative / 0% sporadic / 1+ average / 2+ abundant / 3+

8 (62) 1 (8) 3 (23) 1 (8)

0 (0) 8 (62) 2 (15) 3 (23)

111 5 10 (77) 3 (23) 0 (0) 0 (0)

Probody therapeutic design of 89Zr-CX-072 promotes accumulation in PD-L1-expressing tumors compared to normal murine lymphoid tissue

Danique Giesen 1, Linda N Broer 1, Marjolijn N Lub-de Hooge 2, 3, Irina Popova 4, Bruce Howng 4, Margaret Nguyen 4, Olga Vasiljeva 5, Elisabeth G E de Vries 6, Martin Pool 1

1 Department of Medical Oncology, 2 Department of Clinical Pharmacy and Pharmacology, and 3 Department of Nuclear Medicine and Molecular Imaging, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands; 4 CytomX Therapeutics, Inc., South San Francisco, California; 5 CytomX Therapeutics, Inc., South San Francisco, California. ovasiljeva@cytomx.com; 6 Department of Medical Oncology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands. e.g.e.de.vries@umcg.nl.

Clin Cancer Res. 2020;26(15):3999-4009.

Abstract

Purpose: Probody therapeutic CX-072 is a protease-activatable antibody that is crossreactive with murine and human programmed death-ligand 1 (PD-L1). CX-072 can be activated in vivo by proteases present in the tumor microenvironment, thereby potentially reducing peripheral, anti–PD-L1-mediated toxicities. To study its targeting of PD-L1–expressing tissues, we radiolabeled CX-072 with the PET isotope zirconium-89 (89Zr).

Experimental Design: 89Zr-labeled CX-072, nonspecific Probody control molecule (PbCtrl) and CX-072 parental antibody (CX-075) were injected in BALB/c nude mice bearing human MDAMB-231 tumors or C57BL/6J mice bearing syngeneic MC38 tumors. Mice underwent serial PET imaging 1, 3, and 6 days after intravenous injection (pi), followed by ex vivo biodistribution. Intratumoral 89Zr-CX-072 distribution was studied by autoradiography on tumor tissue sections, which were subsequently stained for PD-L1 by IHC. Activated CX-072 species in tissue lysates were detected by Western capillary electrophoresis.

Results: PET imaging revealed 89Zr-CX-072 accumulation in MDA-MB-231 tumors with 2.1fold higher tumor-to-blood ratios at 6 days pi compared with 89Zr-PbCtrl. Tumor tissue autoradiography showed high 89Zr-CX-072 uptake in high PD-L1–expressing regions. Activated CX-072 species were detected in these tumors, with 5.3-fold lower levels found in the spleen. Furthermore, 89Zr-CX-072 uptake by lymphoid tissues of immune-competent mice bearing MC38 tumors was low compared with 89Zr-CX-075, which lacks the Probody design.

Conclusions: 89Zr-CX-072 accumulates specifically in PD-L1–expressing tumors with limited uptake in murine peripheral lymphoid tissues. Our data may enable clinical evaluation of 89ZrCX-072 whole-body distribution as a tool to support CX-072 drug development (NCT03013491).

Introduction

Immunotherapies targeting immune-regulatory checkpoints have acquired a clear role in clinical cancer care. These therapies improve survival of patients with advanced stages of several tumor types, although not all patients respond (1). Immune checkpoint inhibition can elicit a unique spectrum of immune-related adverse events (irAE) due to the role of these immune checkpoints in maintaining immunologic homeostasis, including toxicities of endocrine, hepatologic, dermatologic, cardiac, and gastro-enteric origin, that can be lifethreatening (2). Combining immune checkpoint inhibitors improves response rates and overall survival for specific cancers (3–7), but these combinations often show increases in rate and severity of side effects (8–10). Immune checkpoint inhibitors with reduced peripheral, immune-related toxicities are therefore of interest.

CX-072 is a Probody therapeutic that targets the programmed death-ligand 1 (PD-L1) immune checkpoint. It is currently studied in a phase I/II clinical trial (ClinicalTrials.gov identifier NCT03013491). CX-072 potentially limits irAEs, as it is activated preferentially in the tumor microenvironment (11–13). Tumor-associated proteases can remove the masking peptide that blocks the PD-L1–binding region, yielding activated antibody with approximately 100-fold increased target affinity compared with its intact, inactivated form (Fig. 1).

Due to its design, the CX-072 tissue distribution profile is expected to diverge from other PDL1–targeting antibodies. PET imaging is a powerful, noninvasive technique to determine in vivo antibody distribution when used with radiolabeling. It provides quantitative spatial and temporal information on tissue-targeting and target-expression. The PET isotope zirconium-89 (89Zr; t1/2 = 78.4 hours) is favorable for radiolabeling antibodies, as its physical half-life matches the time antibodies require for tumor accumulation, resulting in an optimal tumor-tobackground signal (14).

Several preclinical imaging studies have reported high uptake of radiolabeled PD-L1–targeting antibodies in murine lymphoid tissues, including spleen, lymph nodes, and thymus, but also in brown adipose tissue (BAT; refs. 15, 16). PET imaging with radiolabeled anti–PD-L1 antibody 89Zratezolizumab in patients measured high, heterogeneous uptake in tumor lesions as well as in spleen, nonmalignant lymph nodes and Waldeyer's ring (17).

FIGURE 1: CX-072 mechanism of action.

Schematic overview of the CX-072 mechanism of action in healthy and tumor tissues. Both light chains of the anti–PD-L1 parental antibody are modified at their N-termini by addition of a protease-cleavable linker peptide that tethers a mask to the antibody. In healthy tissue, this mask prevents binding of CX-072 to PDL1. The masking peptide is removed by proteases commonly upregulated in the tumor microenvironment, yielding fully activated, PD-L1–targeting antibody.

Therapeutic design of anti-PD-L1 Probody CX-072

Translational Relevance

Combining immune checkpoint inhibitors improves survival of patients with advanced stages of several tumor types, but can elicit severe immune-related adverse events (irAE). These irAEs may be caused by immune checkpoint blockade in healthy tissues, and immune checkpoint–inhibiting antibodies with tumor-restricted activity are therefore of interest. Recently, imaging of 89Zr-atezolizumab whole-body distribution in patients with cancer showed high uptake in healthy lymphoid tissues, including spleen, lymph nodes, and Waldeyer's ring. Our preclinical imaging study in mice reveals anti–PD-L1 Probody therapeutic CX-072 is preferentially activated in tumors, followed by PD-L1–mediated uptake, whereas accumulation in spleen and other PD-L1–expressing peripheral lymphoid tissues is limited. These findings demonstrate CX-072 may reduce anti–PD-L1-mediated toxicities in healthy tissues, thereby potentially expanding its use in combination therapies. We developed and characterized clinical grade 89Zr-CX-072, which is currently studied in patients as part of a phase I/II clinical trial (NCT03013491) to support CX-072 drug development.

We performed a PET imaging study in murine models with 89Zr-labeled CX-072 to reveal its whole-body distribution. Also, we compared 89Zr-CX-072 targeting of tumor and lymphoid tissues in both an immune-compromised and an immune-competent setting. To enable clinical PET imaging of 89Zr-CX-072 distribution to tumor and lymphoid tissues in patients, we characterized and developed a good manufacturing practice (GMP)–compliant tracer.

Materials and Methods

Radiolabeling of CX-072, PbCtrl, and CX-075 CX-072, nonspecific Probody control molecule (PbCtrl), and parental antibody CX-075 (CytomX Therapeutics) were allowed to react with an 1:2 molar excess of tetrafluorphenolN-succinyldesferal (TFP-N-sucDf; ABX GmbH) as described previously (18), with the following modification: pH was set at 4.0 to 4.5 using 1.0 mol/L ammonium acetate instead of 0.025 mol/L sulfuric acid to prevent aggregate formation. CX-072-N-sucDf, PbCtrl-N-sucDf, and CX-075-N-sucDf were purified using a Vivaspin-2 concentrator, aliquoted, and stored at −80 °C. Concentration and purity were determined by a Waters size exclusion high-performance liquid chromatography system equipped with a dual-wavelength absorbance detector (280 nm vs. 430 nm), in-line radioactivity detector, and TSK-Gel SW column G3000SWXL 5 μm, 7.8 mm (Joint Analytical Systems; mobile phase: PBS 9.0 mmol/L sodium phosphate, 1.3 mmol/L potassium phosphate, 140 mmol/L sodium chloride, pH 7.2; Hospital Pharmacy UMCG; flow: 0.7 mL/min).

CX-072-N-sucDf, PbCtrl-N-sucDf, and CX-075-N-sucDf were radiolabeled with clinical grade 89Zr (Perkin Elmer) as described previously (18). Radiochemical purity was assessed by a trichloroacetic acid precipitation assay (19). For all experiments, radiochemical purity of ≥95% was required.

Immunoreactivity

Immunoreactivity after conjugation to TFP-N-SucDf was assessed by indirect ELISA. Note that 96-well plates (Nunc Maxisorp) were coated with 1 μg/mL human extracellular PD-L1 domain (R&D Systems; 156-B7-100) diluted in PBS (Gibco; 0.7 mmol/L sodium phosphate, 1.5 mmol/L potassium phosphate, 154 mmol/L sodium chloride, pH 7.2) and incubated overnight at 4 °C. Wells were blocked for 2 hours at room temperature (RT) with 1% BSA (Sigma-Aldrich), 0.05% Tween 20 in PBS. After blocking, plates were incubated with either unconjugated CX-072, PbCtrl, or CX-075 or their respective N-sucDf-conjugates in a concentration ranging from 0.0914 to 600 nmol/L for 60 minutes at RT. Plates were subsequently washed with 0.05% Tween 20 in PBS and incubated with horseradish peroxidase–labeled anti-human IgG antibody (SigmaAldrich; A0293) for 60 minutes at RT. Detection was performed with single-component TMB peroxidase substrate (BioRad), and optical density read-out was performed at 450 nm using a micro plate-reader. Immunoreactivity was analyzed by nonlinear regression Log(agonist) versus response in Graphpad Prism v7.0. and was expressed as the effective concentration needed for 50% of receptor occupation (EC50).

Cell lines

PD-L1–expressing human triple-negative breast cancer cell line MDA-MB-231 was a kind gift from Dr. Janet Price, MD Anderson Cancer Center (Houston, TX). Cell lines were confirmed to be negative for microbial contamination and were authenticated in January 2018 by BaseClear using short tandem repeat profiling. The murine PD-L1–positive colon cancer cell line MC38 was obtained from Dr. Walter Storkus, University of Pittsburgh (Pittsburgh, PA). MDA-MB-231 and MC38 cells were cultured in DMEM (Gibco) containing 1.0 g/L glucose and 4.5 g/L glucose, respectively, supplemented with 10% FCS. Cells were incubated at 37 °C in a humidified atmosphere with 5% CO2 and were passaged no longer than 6 months. For all experiments, cells were collected in the exponential growth phase.

Flow cytometry

Flow cytometry experiments in MDA-MB-231 and MC38 cells were performed using CX-075 parental antibody, which is cross-reactive to human and murine PD-L1. Cells were trypsinized and harvested in PBS with 2% FCS and kept on ice prior to use. IgG4 isotype control antibody and CX-075 were diluted in PBS with 2% FCS to 20 μg/mL and incubated with 2×105 cells in 1 mL for 1 hour at 4 °C. Bound primary antibody was detected using a phycoerythrin-conjugated

Therapeutic design of anti-PD-L1 Probody CX-072

goat anti-human IgG secondary antibody (Southern Biotech; 2040-09) diluted 1:50 in 2% FCS in PBS and analyzed on a BD Accuri C6 flow cytometer (BD Biosciences). Data analysis was performed with FlowJo v10 (Tree Star), and surface receptor expression was expressed as mean fluorescent intensity.

For flow cytometry experiments in lymphoid tissues, spleen, lymph nodes, bone marrow, BAT, and thymus were collected from BALB/c nude and C57BL/6J mice. Single-cell suspensions were prepared using a cell strainer (Fisher Scientific). Red blood cell lysis buffer (BioLegend) was used for spleen and bone marrow samples. Single cells were analyzed for PD-L1 expression by flow cytometry, using Brilliant Violet 421 anti-mouse PD-L1 (CD274) antibody (BioLegend; 124315) and Brilliant Violet 421 Rat IgG2b isotype control antibody (BioLegend; 400639). Zombie aqua (BioLegend; 423101) was used for detection of viable cells. In splenocytes, PD-L1 expression on specific immune cell types was measured using fluorescein isothiocyanate–conjugated rat anti-mouse CD3 antibody (Thermo Scientific; 11-0032-82) diluted 1:200 in PBS with 2% FCS, allophycocyanin-eFluor780–conjugated rat anti-mouse CD335 antibody (Thermo Scientific; 47-3351-80) diluted 1:20, peridinin chlorophyll protein complex-cyanine5.5-conjugated rat anti-mouse CD11b antibody (Thermo Scientific; 45-0112-80) diluted 1:80, and allophycocyaninconjugated rat anti-mouse CD19 antibody (Thermo Scientific; 17-0193-80) diluted 1:160. Samples were analyzed on a BD FACS Verse (BD Biosciences). BAT was excluded from measurements, as insufficient amounts of immune cells were present in these samples.

89Zr-CX-075 internalization in MDA-MB-231 and MC38 cell lines 89Zr-CX-075 was used for internalization experiments because it acts as a surrogate for potential internalization of 89Zr-CX-072 after removal of its masking peptide by tumorassociated proteases. Internalization in MDA-MB-231 and MC38 cells was assessed by adding 50 ng 89Zr-CX-075 (200 MBq/mg) to 1×106 cells. For control of binding without internalization, cells were incubated for 1 hour on ice. Cells were subsequently washed with ice-cold 1% human serum albumin (Sanquin) in PBS. Total PD-L1–bound 89Zr-CX-075 was determined by measuring cell-associated activity in a calibrated well-type gamma counter (LKB instruments) followed by incubation for 1 and 2 hours at 37 °C, whereas controls were kept on ice. Cells were stripped of cell-surface–bound antibody using 0.05 mol/L glycine and 0.1 mol/L sodium chloride (pH 2.8) and subsequently counted for remaining cell-associated activity, ie, internalized PD-L1–bound 89Zr-CX-075, in a calibrated well-type gamma counter. Internalization was expressed as percentage of total PD-L1–bound 89Zr-CX-075, corrected for internalization in 4 °C controls.

Animal studies

All animal experiments were performed in accordance with the Dutch code of practice “Animal experiments in cancer research” and were approved by the institutional animal care and use

committee of the University of Groningen, the Netherlands.

PD-L1–expressing MDA-MB-231 human triple-negative breast cancer cells were s.c. engrafted in male BALB/cOlaHsd-Foxn1nu mice (Envigo). Tumors were allowed to grow 4 weeks, yielding tumor volumes of ±200 mm3. To study 89Zr-CX-072 target–specific tumor uptake and the potential for antigen saturation, mice received 10-μg 89Zr-CX-072, 89Zr-PbCtrl, or 89Zr-CX-075 (~5 MBq) supplemented with 0, 40, or 240 μg of unlabeled CX-072, PbCtrl, or CX-075, respectively, via penile vein injection (n = 5–6 per group). To evaluate biodistribution in a fully immunecompetent, syngeneic model, male C57BL/6JOlaHsd mice (Envigo) were implanted s.c. with MC38 murine colon adenocarcinoma cells. After 16 days of tumor growth to reach a volume of ±200 mm3, mice received 10 μg of 89Zr-CX-072, 89Zr-PbCtrl, or 89Zr-CX-075 via penile vein injection (n = 5–6 per group).

Mice underwent serial in vivo PET scans 1, 3, and 6 days post intravenous injection (pi) in a Focus 200 microPET scanner (CTI Siemens), followed by tissue collection for ex vivo biodistribution analysis. Tissues were weighed and counted in a calibrated well-type gamma counter. Tracer uptake per organ was quantified by percentage injected dose per gram tissue (%ID/g).

PET data were reconstructed using a two-dimensional ordered-subset expectation maximization reconstruction algorithm with Fourier rebinning, four iterations, and 16 subsets. Data sets were corrected for decay, random coincidences, scatter, and attenuation. PET images were analyzed using AMIDE medical image data examiner software 1.0.5, and regions of interest were drawn for tumor, spleen, and blood pool (i.e., heart). Tracer uptake was quantified as mean standardized uptake value (SUVmean).

Tracer integrity

Tracer integrity was studied ex vivo by SDS-PAGE in plasma obtained after sacrifice on 6 days pi. Total protein concentration was determined by the Bradford assay (20). Mini-Protean TGX precast protein gels 4%–15% (BioRad) were loaded with 40 μg of total plasma protein. As a positive control, freshly radiolabeled (intact) 89Zr-CX-072 was diluted to match activity levels of plasma samples. Gels were allowed to run for 30 to 45 minutes at 100 V. Detection was performed by exposing gels to a multipurpose phosphor plate (Perkin Elmer) overnight at −20 °C; exposures were captured using a Cyclone phosphor imager.

Assessment of activated and intact CX-072 in tumor lysates

Tumor homogenates were prepared in Pierce IP lysis buffer (Thermo Scientific) with added Halt protease inhibitor cocktail kit (Thermo Scientific) using a Barocycler (Pressure Biosciences). Protein lysates in IP lysis buffer with protease inhibitor/EDTA were analyzed by the Wes system

Therapeutic design of anti-PD-L1 Probody CX-072

(ProteinSimple). Activated, intact CX-072 was detected using mouse anti–CX-075 primary antibody (CytomX Therapeutics) and anti-mouse secondary antibody (ProteinSimple; 042-205). Concentrations of activated CX-072 were analyzed using Compass software (ProteinSimple).

Ex vivo tissue preparation, autoradiography, and PD-L1 IHC

For ex vivo tissue analysis, formalin-fixed, paraffin-embedded (FFPE) tumor tissue blocks were prepared. For autoradiography, FFPE blocks were sliced into 4-μm tumor tissue sections and exposed to a phosphor plate overnight at −20 °C. Exposures were captured using a phosphor imager (Cyclone).

For PD-L1 IHC, previously autoradiographed FFPE tumor tissue sections were deparaffinized in xylene and rehydrated. Heat-induced antigen retrieval was performed in 10 mmol/L citrate buffer (pH 6.0) for 15 minutes at 95 °C to 100 °C. Endogenous peroxidase was blocked by 10-minute incubation with dual endogenous enzyme-blocking reagent (Agilent Technologies). Slides were rinsed in 137 mmol/L sodium chloride, 20 mmol/L tris(2,3-dibromopropyl) phosphate, 0.1% Tween-20 (pH 7.6), and subsequently incubated with Dako serum free protein block (Agilent Technologies) for 30 minutes at RT. For human PD-L1 staining, slides were incubated with rabbit anti-human PD-L1 antibody (Abcam; ab205921) or rabbit IgG antibody control (Abcam; ab172730) diluted to 5 μg/mL in Dako serum-free protein block for 1 hour at RT. Thereafter, slides were incubated with Dako EnVision HRP system (Agilent Technologies) for 30 minutes at RT, followed by 10-minute incubation with diaminobenzidine chromogen. Hematoxylin counterstaining was applied routinely. For histologic analysis of tumors, hematoxylin/eosin staining was performed on tissue sections that were sliced subsequent to the sections used for autoradiography. Digital scans of slides were acquired by a Hamamatsu NanoZoomer 2.0-HT multi slide scanner and analyzed with NanoZoomer Digital Pathology viewer software.

Murine PD-L1 staining was performed as described previously (21). Heat-induced antigen retrieval was performed in 10 mmol/L citrate (pH 6.0) in a Lab Vision PT module (Thermo Scientific) at 95 °C. Endogenous peroxidase was blocked by 10-minute incubation at RT with 3% hydrogen peroxide in PBS, followed by 15-minute avidin/biotin-blocking (Vector Labs). Slides were preincubated with 10% Dako normal rabbit serum (Agilent Technologies) for 30 minutes at RT. Goat anti-mouse PD-L1 antibody (R&D systems; AF1019) or normal goat IgG control antibody (R&D systems; AB-108-C) diluted to 0.4 μg/mL in PBS with 1% BSA was incubated overnight at 4 °C, followed by 30-minute incubation at RT with rabbit anti-goat biotinylated secondary antibody (Agilent Technologies; E046601-2) diluted 1:400 in PBS with 1% BSA. For detection, slides were incubated with Vectastain Elite ABC HRP-kit (Vector Labs) for 30 minutes at RT.

Production of clinical grade 89Zr-CX-072 and stability testing

Methods for conjugation and radiolabeling were transferred to GMP environment. Analytical procedures were validated to demonstrate suitability for use in quality control testing of CX072-N-sucDf and 89Zr-CX-072. For validation of the manufacturing process, three batches of GMP-compliant CX-072-N-sucDf intermediate product were produced and radiolabeled with 89Zr. For quality control, these batches met specifications on conjugation ratio, activity yield, purity, concentration, pH, radiochemical purity, residual solvents, sterility, and endotoxin content. Preservation of immunoreactivity after conjugation was determined by ELISA as previously described for preclinical conjugated CX-072.

CX-072-N-sucDf intermediate product was stored in sterile vials (Biopure) at −80 °C. Stability of CX-072-N-sucDf batch 1 was analyzed at 0, 1, 3, 6, and 12 months after production. Quality control was performed after radiolabeling with 89Zr according to release specifications.

Statistical analysis

Data were analyzed for statistical significance in GraphPad Prism v7.0 using the Mann–Whitney U test for nonparametric data followed by Bonferroni post test correction for comparison of more than two groups. Experiments were performed at least three times. P values < 0.05 were considered significant.

Results

CX-072 characterization and binding

CX-072, a recombinant, protease-activatable antibody targeting PD-L1, is based on CX-075, a phage-derived, fully human IgG4 antibody that blocks interactions of PD-L1 with PD-1 and B7-1 molecules. In its intact form, both light chains of CX-072 are modified at their N-terminus by addition of a peptide prodomain, which serves to mask the PD-L1–binding region of the antibody. Proteolytic cleavage of the substrate within the prodomain yields the active, PDL1–binding form of CX-072.

Binding of CX-075 parental antibody, which lacks a masking peptide, was comparable for both human (Kapp = 0.25 nmol/L) and murine (Kapp = 0.30 nmol/L) PD-L1 (Supplementary Fig. S1A), whereas CX-072 binding to murine PD-L1 was approximately 15-fold weaker than for human PD-L1. CX-072 bound to murine PD-L1 at K app of 152 nmol/L and bound to human PD-L1 at K app of 9.9 nmol/L, as determined by ELISA (Supplementary Fig. S1B).

In vivo 89Zr-CX-072 tumor and spleen uptake over time

For in vivo PET imaging studies, CX-072, PbCtrl, and CX-075 were conjugated to an average of approximately 1.2 TFP-N-sucDf chelators per antibody and thereafter radiolabeled with

Therapeutic design of anti-PD-L1 Probody CX-072

500 MBq/mg 89Zr at ≥95% radiochemical purity. Immunoreactivity to PD-L1 was preserved for CX-072-N-sucDf and CX-075-N-sucDf, whereas PbCtrl-N-sucDf remained non-avid (Supplementary Fig. S1C and S1D).

To study human tumor targeting, we performed in vivo studies in human PD-L1–expressing MDA-MB-231 xenograft tumor-bearing BALB/c nude mice. In this model, we compared tumor targeting with targeting of the spleen, because this lymphoid organ is well-developed in immune-compromised mice and can be quantified with imaging. Serial PET scans on days 1, 3, and 6 pi showed tumor accumulation over time for 89Zr-CX-072 and 89Zr-CX-075, but not for 89Zr-PbCtrl (Fig. 2A). PET quantification revealed a 1.5-fold higher spleen uptake for 89Zr-CX-075 than for 89Zr-CX-072 at day 6 pi (P < 0.01). 89Zr-CX-075 spleen uptake was already higher than blood pool levels at day 1 pi and increased to a spleen-to-blood ratio of 2.6 ± 0.5 at day 6 pi (Fig. 2B), suggesting this uptake is PD-L1–mediated. Tracer activity in the blood pool decreased over time, resulting in increasing tumor-to-blood ratios for 89Zr-CX-072 and 89Zr-CX-075 from day 1 to 6 pi with highest tumor uptake observed at day 6 pi. Day 6 was selected for ex vivo biodistribution studies of subsequent mouse cohorts based on highest 89Zr-CX-072 tumor-toblood ratio of 0.9 ± 0.2, compared with 0.4 ± 0.0 found for 89Zr-PbCtrl (P < 0.01).

Ex vivo 89Zr-CX-072 biodistribution in tumor-bearing mice

Ex vivo tracer uptake quantitation per organ revealed similar biodistribution of 89Zr-CX-072 and 89Zr-PbCtrl in healthy, nontumor tissues (Fig. 3A). Compared with 89Zr-CX-072, blood pool levels and uptake in the heart were lower for 89Zr-CX-075, whereas its uptake in liver, pancreas, stomach, ilium, bone, skin, and spleen was higher.

Next, we assessed whether tracer uptake in tumor and spleen is PD-L1–mediated through a protein dose-escalation study. A radiolabeled antibody dose of 10-μg 89Zr-CX-072, 89Zr-PbCtrl, or 89Zr-CX-075 was supplemented with 0, 40, or 240 μg of unlabeled CX-072, PbCtrl, or CX-075 to obtain total protein doses of 10, 50, and 250 μg for each tracer. We found that 89Zr-CX-072 and 89Zr-PbCtrl biodistribution in healthy tissues was not affected by increased protein dose, but 89Zr-CX-075 uptake in these healthy organs decreased with increasing total protein dose (Supplementary Fig. S2). PET quantification showed blood pool levels of all three radiolabeled molecules were unaffected by increased protein dose (Supplementary Fig. S3).

Ex vivo tracer uptake quantitation showed a protein dose-dependent tumor uptake for 89ZrCX-072 and 89Zr-CX-075, which was not observed for 89Zr-PbCtrl (Fig. 3B). 89Zr-CX-072 tumor uptake decreased from 8.7 ± 1.0 %ID/g for the 10-μg total protein dose to 6.0 ± 1.3 %ID/g and 4.3 ± 0.7 %ID/g for the 50- and 250-μg dose groups, respectively, indicating competition of tracer with unlabeled CX-072 for PD-L1 receptor-binding. This shows 89Zr-CX-072 tumor uptake is

FIGURE 2: In vivo tumor and spleen accumulation over time in MDA-MB-231 tumor-bearing mice.

(A) Representative coronal PET images of 10-μg 89Zr-CX-072, 89Zr-PbCtrl, and 89Zr-CX-075 in MDA-MB-231 tumor-bearing mice at 1, 3, and 6 days pi. Tracer uptake is presented as standardized uptake value (SUV). H, heart; T, tumor; S, spleen. (B) PET imaging quantification of 89Zr-CX-072, 89Zr-PbCtrl, and 89Zr-CX-075 uptake in MDA-MB-231 tumor, blood pool (surrogated by heart), and spleen at 1, 3, and 6 days pi. Tracer uptake is presented as mean standardized uptake value (SUVmean). Data are shown as mean ± standard deviation (SD).

PD-L1–driven. Similarly, 89Zr-CX-075 tumor uptake was reduced by unlabeled antibody. 89ZrPbCtrl tumor uptake was independent of total protein dose, confirming its nonspecificity for PD-L1.

Although immune-compromised mice were used for this model, PD-L1–mediated spleen uptake was observed for 89Zr-CX-075. Spleen uptake of 89Zr-CX-075 decreased from 25.8 ± 4.1 %ID/g at the 10-μg total protein dose to 10.8 ± 2.8 %ID/g for the 50-μg dose group and 5.3

Therapeutic design of anti-PD-L1 Probody CX-072

± 2.6 %ID/g for the 250-μg dose group. 89Zr-CX-072 did not show protein dose-dependent spleen uptake, similar to 89Zr-PbCtrl, indicating this spleen uptake is not PD-L1–driven, but part of normal, nonspecific antibody distribution.

89Zr-CX-072 and 89Zr-CX-075 showed a comparable tumor uptake of 8.7 ± 1.0 %ID/g and 8.8 ± 2.9 %ID/g, respectively, for the 10-μg total protein dose, whereas only 3.8 ± 0.2 %ID/g was observed for 10-μg 89Zr-PbCtrl. This demonstrates that CX-072's design affects biodistribution to healthy organs, but not its tumor-targeting properties.

Tumor-specific activation of CX-072 species

MDA-MB-231 tumor and spleen lysates were analyzed for presence of activated CX-072 to confirm whether CX-072 is specifically activated by proteases in the tumor microenvironment. Flow cytometry revealed PD-L1 expression by MDA-MB-231 tumor cells and splenocytes of BALB/c nude mice (Fig. 3C). IHC on ex vivo tumor and spleen tissues also confirmed presence of PD-L1. MDA-MB-231 tumor lysates had 6.9 ng/mL activated CX-072 species at the 10-μg total protein dose, 21.2 ng/mL at the 50-μg total protein dose, and highest concentration of 81.7 ng/ mL was found for the 250-μg dose group (Fig. 3D). Activated CX-072 level detected in spleen lysates at the 250-μg total protein dose was 5.3-fold lower compared with tumor lysates (P < 0.05). This demonstrates that CX-072 is preferentially activated in tumor tissue and thereafter remains predominantly within the tumor microenvironment. Furthermore, circulating 89ZrCX-072 tracer remained intact in the blood pool at 6 days pi, as confirmed by SDS-PAGE (Fig. 3E).

89Zr-CX-072 targeting of tumor and lymphoid tissues in a syngeneic mouse model

To assess 89Zr-CX-072 targeting of relevant PD-L1–expressing tissues besides tumor, tracer distribution in BALB/c nude mice was compared with fully immune-competent C57BL/6J mice bearing murine PD-L1–expressing MC38 tumors. PET imaging at 6 days pi revealed high spleen uptake for 89Zr-CX-075 in both mouse models, which was not visible for 89Zr-CX-072 and 89Zr-PbCtrl (Fig. 4A). Importantly, 89Zr-CX-072 and 89Zr-CX-075 showed comparable uptake in syngeneic MC38 tumors at 6 days pi (Supplementary Fig. S4A), whereas 3.1-fold higher spleen uptake was observed for 89Zr-CX-075 compared with 89Zr-CX-072 in C57BL/6J mice (P < 0.01).

PD-L1 expression in spleen, mesenteric lymph nodes, axillary lymph nodes, BAT, and thymus was confirmed by IHC and flow cytometry (Figs. 4B and 5A). Although limited immune cells were found in BAT, IHC revealed PD-L1 expression by adipocytes. High 89ZrCX-075 uptake was found in these peripheral PD-L1–expressing tissues ex vivo (Fig 4C). In contrast, 89Zr-CX-072 uptake was low in these tissues and comparable with 89ZrPbCtrl, indicating that CX-072's design limits uptake in lymphoid tissues and BAT. Notably,

FIGURE 3: Ex vivo biodistribution in MDA-MB-231 tumor-bearing mice.

(A) Ex vivo biodistribution of 10-μg 89Zr-CX-072, 89Zr-PbCtrl, and 89Zr-CX-075 in MDA-MB-231 tumor-bearing mice at 6 days pi. Tracer uptake per organ is presented as %ID/g. Data are shown as mean ± SD. (B) Ex vivo tumor and spleen uptake of 89Zr-CX-072, 89Zr-PbCtrl, and 89Zr-CX-075 at increasing total protein dose. Tracer uptake is presented as %ID/g. Data are shown as mean ± SD. (C) PD-L1 expression in MDA-MB-231 cells and splenocytes detected with flow cytometry and in FFPE tumor and spleen tissue sections detected with IHC. IgG4 antibody was used as isotype control. (D) Activated CX-072 species detected ex vivo in MDA-MB-231 tumor tissue and spleen by Western capillary electrophoresis. Data are shown as mean ± SD. (E) Ex vivo tracer integrity of 89Zr-CX-072 in plasma samples 6 days pi, as determined by SDS-PAGE. Detection of signal was performed by autoradiography. Data from a representative experiment are shown. *, P < 0.05; **, P < 0.01; and ns, not significant.

89Zr-CX-075 spleen uptake was comparable in both tumor-bearing BALB/c nude and C57BL/6J mice. We found few T cells among BALB/c nude mouse splenocytes, whereas monocytes, NK cells, and B cells were abundant and showed high levels of PD-L1 expression (Fig. 5B).

Ex vivo 89Zr-CX-072 uptake was higher in MDA-MB-231 tumors compared with MC38 tumors (Supplementary Fig. S5). Flow cytometry confirmed MDA-MB-231 and MC38 cell lines both express PD-L1 in vitro, albeit at a higher level in MDA-MB-231 cells (Supplementary Fig. S5A and S5B). After 2 hours of incubation, 31.7 ±1.2% of PD-L1–bound 89Zr-CX-075 was internalized in MDA-MB-231 cells (Supplementary Fig. S5C). This finding supports the idea that 89Zr-CX-072 tracer activity

Therapeutic design of anti-PD-L1 Probody CX-072

FIGURE 4: Biodistribution in tumor-bearing BALB/c nude and C57BL/6J mice.

(A) Representative maximum intensity projections of 89Zr-CX-072, 89Zr-PbCtrl and 89Zr-CX-075 in MC38 tumor-bearing C57BL/6J mice and MDA-MB-231 tumor-bearing BALB/c nude mice imaged at 6 days pi. H, heart; T, tumor; S, spleen. (B) PD-L1 IHC in lymphoid tissues of tumor-bearing BALB/c nude and C57BL/6J mice. MLN, mesenteric lymph nodes; ALN, axillary lymph nodes. (C) Ex vivo uptake of 89Zr-CX-072, 89Zr-PbCtrl, and 89Zr-CX-075 in lymphoid tissues of BALB/c nude and C57BL/6J mice at 6 days pi. Tracer uptake per organ is presented as %ID/g. Data are shown as mean ± SD. *, P < 0.05; **, P < 0.01; and ns, not significant.

can residualize in the tumor after removal of its masking peptide. Tracer internalization and 89Zr residualization may augment the PET signal, resulting in better tumor visualization. Negligible internalization was found for PD-L1–bound 89Zr-CX-075 in MC38

FIGURE 5: PD-L1 expression by specific immune cell types in BALB/c nude and C57BL/6J mice. (A) PD-L1 expression detected with flow cytometry in single-cell suspensions of BALB/c nude and C57BL/6J lymphoid tissues. LN, lymph nodes; BM, bone marrow. (B) Within splenocytes, PD-L1 expression was measured on specific immune cell types using flow cytometry.

cells: 2.5 ±0.5% after 2 hours of incubation. Low levels of PD-L1 expression combined with limited internalization in MC38 tumors may explain why the observed 89Zr-CX-072 tumor uptake was not significantly different from 89Zr-PbCtrl (Supplementary Fig. S5D).

Similar to our observations in BALB/c nude tumor-bearing mice, ex vivo tracer uptake quantitation in C57BL/6J tumor-bearing mice revealed comparable biodistribution in healthy organs for 89Zr-CX-072 and 89Zr-PbCtrl (Supplementary Fig. S4B). 89Zr-CX-075 uptake was higher in liver, ilium, and brain, whereas blood pool levels were lower compared with 89Zr-CX-072. 89ZrCX-075 is potentially affected by target-mediated drug disposition, whereas 89Zr-CX-072 is not, presumably due to protection by its masking peptide. In addition, residual activity measured in tumor-bearing mice at 1, 3, and 6 days pi suggests a slightly faster metabolism for 89Zr-CX-075 compared with 89Zr-CX-072 (Supplementary Fig. S6).

89Zr-CX-072 tumor uptake on autoradiography correlates to PD-L1–expressing tissue Finally, we studied ex vivo macroscopic tracer distribution in FFPE tumor tissue slices using autoradiography to further confirm PD-L1–specific tumor targeting. Autoradiography revealed a heterogeneous distribution pattern for 89Zr-CX-072 and 89Zr-CX-075, but not for 89Zr-PbCtrl (Fig. 6). PD-L1 staining was observed in both viable and necrotic tumor tissue and correlated to regions showing high uptake of 89Zr-CX-072 on autoradiography. Tumor tissue regions showing low 89Zr-CX-072 uptake on autoradiography had corresponding low levels of PD-L1 expression. In contrast, 89Zr-PbCtrl distributed to nontumor areas, whereas PD-L1 expression was present in viable tumor tissue, indicating that the observed uptake is not PD-L1–specific. 89Zr-CX-075 also distributed to tumor tissue regions expressing high levels of PD-L1.

Therapeutic design of anti-PD-L1 Probody CX-072

FIGURE 6: Ex vivo macroscopic tumor tissue distribution.

Autoradiography images of 89Zr-CX-072, 89Zr-PbCtrl, and 89Zr-CX-075 in FFPE MDA-MB-231 tumor tissue sections, followed by PD-L1 IHC. Hematoxylin/eosin (H&E) staining was performed on an adjacent tissue section to demonstrate viability of tumor tissue. Representative data are shown.

Clinical grade 89Zr-CX-072 for patient studies

To enable PET imaging in patients, we optimized and validated a robust, GMP-compliant manufacturing process for clinical grade 89Zr-CX-072. Three validation batches of clinical grade CX-072-N-sucDf intermediate product were produced and subsequently radiolabeled with 89Zr, which met prior set quality specifications. CX-072-N-sucDf intermediate product demonstrated stability up to 12 months. Therefore, CX-072-N-sucDf shelf-life is currently set at 12 months and may be extended if the product remains within specifications at future time points. An investigational medicinal product dossier for 89Zr-CX-072 was compiled, submitted, and approved by the competent authority as part of clinical trial application. A clinical study evaluating 89Zr-CX-072 biodistribution in patients is currently ongoing at our center as part of the CX-072 phase I/II study.

Discussion

This study demonstrates that 89Zr-CX-072 is activated in tumor tissue, followed by PD-L1–mediated tumor uptake, whereas PD-L1–mediated accumulation in healthy, peripheral PD-L1–expressing tissues is limited. Importantly, we found comparable tumor uptake for 89Zr-CX-072 and its parental antibody 89Zr-CX-075 in both human xenograft and syngeneic tumor-bearing mice, indicating that CX-072's required activation by proteases does not hamper tumor uptake.

Low uptake of 89Zr-CX-072 was found in PD-L1–expressing lymphoid tissues and BAT, similar to 89Zr-labeled nonspecific Probody control molecule, which shows this uptake is not mediated by PD-L1. This finding is remarkably different from lymphoid tissue-targeting properties of other antibodies whose PD-L1–binding regions are not masked. PD-L1 is abundantly expressed by subsets of immune cells present in healthy lymphoid tissues of mice and humans (15, 22).

As a consequence, PD-L1–targeting antibodies generally show high uptake in the spleen, which potentially affects their biodistribution and pharmacokinetic profile. Several preclinical imaging studies using radiolabeled PD-L1–targeting antibodies in tumor-bearing mice have reported high tracer uptake in the spleen (15, 16, 21, 23–29). PET imaging of PD-L1 in patients with cancer clearly demonstrated high 89Zr-atezolizumab uptake in the spleen among other lymphoid tissues (17).

Most preclinical PD-L1–imaging studies have used radiolabeled antibodies specific for either murine or human PD-L1 (15, 16, 21, 25, 27, 28, 30). 89Zr-CX-072 human/murine cross-reactivity enabled us not only to evaluate its human tumor targeting, but also to acquire data on its uptake in murine lymphoid tissues. Peripheral PD-L1 expression in healthy organs, including lymphoid tissues, strongly affected biodistribution of 89Zr-CX-075 parental antibody, but not biodistribution of 89Zr-CX-072 in tumor-bearing BALB/c nude mice. However, lymphoid tissues in such immune-compromised mice are underdeveloped, leading to absence or anomalous frequencies of specific immune cell types. To overcome this drawback, we used syngeneic MC38 tumor-bearing immune-competent C57BL/6J mice to study CX-072's biodistribution in greater detail.

Interestingly, PET scans and ex vivo biodistribution revealed high uptake of 89Zr-CX-075 parental antibody in spleen and lymph nodes in both BALB/c nude and C57BL/6J mice. Few T cells were present in BALB/c nude mice, but B cells, NK cells, and macrophages were abundant and have been shown to be fully functional (31). We found that PD-L1 is highly expressed by monocytes of BALB/c nude mice, in line with previous research (32). B cells and NK cells also demonstrated PD-L1 expression, but only limited amounts of PD-L1 were present on T cells. Nevertheless, BALB/c nude mice demonstrated levels of PD-L1 expression comparable with immune-competent

Therapeutic design of anti-PD-L1 Probody CX-072

mice. This shows that even though BALB/c nude mice lack a fully functional immune system, they were well-suited for studying 89Zr-CX-072 targeting of lymphoid tissues, and potentially for other PD-L1–targeting antibodies.

89Zr-CX-072 tumor-specific uptake depends on high protease activity within the tumor microenvironment and limited protease levels in healthy organs, including PD-L1–expressing lymphoid tissues (11, 12). Proteases that can activate CX-072 are generally upregulated during tumorigenesis. They include urokinase-type plasminogen activator (uPA), matriptase (MT-SP1), and selected matrix metalloproteinases, which are all associated with many types and stages of cancer (33–37). As an indirect read-out of protease activity in the tumor microenvironment, we measured high concentrations of activated CX-072 in tumor lysates. This indicates that sufficient levels of proteases were present in tumor tissue to activate CX-072.

Our conclusions are limited by the fact that CX-072, in its intact form, binds with lower affinity to murine PD-L1 compared with human PD-L1. This finding suggests that the masking peptide has greater ability to block CX-072 binding to murine PD-L1. However, once 89Zr-CX-072 is activated, the tracer has similar binding affinity for both murine and human PD-L1. Uptake in PD-L1–expressing lymphoid tissues may be higher in the human setting than we found in mice if large amounts of inactivated 89Zr-CX-072 are taken up by these tissues. Clinical 89Zr-CX-072 PET imaging will reveal whether the tumor-specific activation we observed in mouse models is translatable to patients. To this end, we validated the production process of a GMP-compliant 89Zr-CX-072 tracer for clinical PET imaging as a tool to support CX-072 drug development.

In conclusion, we found that 89Zr-CX-072 accumulates specifically in PD-L1–expressing tumor xenografts with limited uptake in murine peripheral lymphoid tissues and BAT, thus supporting the hypothesis that CX-072 may reduce anti–PD-L1-mediated toxicities in these healthy tissues.

Disclosure of potential conflicts of interest

A research grant to E. G. E. de Vries was obtained from CytomX Therapeutics and made available to the institution. I. Popova, B. Howng, M. Nguyen and O. Vasiljeva are employees of CytomX Therapeutics, which developed and owns the intellectual property rights pertaining to CX-072.

Authors' contributions

Conception and design: D. Giesen, L.N. Broer, M.N. Lub-de Hooge, O. Vasiljeva, E.G.E. de Vries, M. Pool; Development of methodology: D. Giesen, L.N. Broer, M.N. Lub-de Hooge, B. Howng, O. Vasiljeva, M. Pool; Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): D. Giesen, L.N. Broer, B. Howng, M. Nguyen, O. Vasiljeva, E.G.E. de Vries, M.

Therapeutic design of anti-PD-L1 Probody CX-072

Pool; Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): D. Giesen, L.N. Broer, I. Popova, B. Howng, M. Nguyen, O. Vasiljeva, E.G.E. de Vries, M. Pool; Writing, review, and/or revision of the manuscript: D. Giesen, L.N. Broer, M.N. Lub-de Hooge, I. Popova, B. Howng, M. Nguyen, O. Vasiljeva, E.G.E. de Vries, M. Pool; Administrative, technical, or material support (ie, reporting or organizing data, constructing databases): D. Giesen, L.N. Broer, M. Nguyen, O. Vasiljeva, E.G.E. de Vries, M. Pool; Study supervision: M.N. Lub-de Hooge, O. Vasiljeva, E.G.E. de Vries, M. Pool.

Acknowledgments

We thank Linda Pot-de Jong (University Medical Center Groningen) for assistance in the validation and production processes of clinical grade 89Zr-CX-072; Shanti Davur, Eric Ureno, and Sridhar Viswanathan (CytomX Therapeutics, Inc.) for assistance with production and characterization of 89Zr-CX-072 and 89Zr-PbCtrl; and Gerwin Sandker (Radboud University Medical Center) for helping with IHC on murine PD-L1. Research support from CytomX Therapeutics, Inc. was made available to the institution. The study drug and control molecules were supplied by CytomX Therapeutics, Inc.

PROBODY is a U.S. registered trademark of CytomX Therapeutics, Inc. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

References

1. Balar AV, Galsky MD, Rosenberg JE, Powles T, Petrylak DP, Bellmunt J, et al. Atezolizumab as first-line treatment in cisplatin-ineligible patients with locally advanced andmetastatic urothelial carcinoma: A single-arm, multicentre, phase 2 trial. Lancet 2017;389:67–76.

2. Postow MA, Sidlow R, Hellmann MD. Immune-related adverse events associated with immune checkpoint blockade. N Engl J Med 2018;378:158–68.

3. Fuc a G, de Braud F, Di Nicola M. Immunotherapy-based combinations. Curr Opin Oncol 2018;30:345–51.

4. Wolchok JD, Chiarion-Sileni V, Gonzalez R, Rutkowski P, Grob JJ, Cowey CL, et al. Overall survival with combined nivolumab and ipilimumab in advanced melanoma. N Engl J Med 2017;377:1345–56.

5. Hellmann MD, Ciuleanu T-E, Pluzanski A, Lee JS, Otterson GA, Audigier-Valette C, et al. Nivolumab plus ipilimumab in lung cancer with a high tumor mutational burden. N Engl J Med 2018;378:2093–104.

6. Overman MJ, Lonardi S, Wong KYM, Lenz H-J, Gelsomino F, Aglietta M, et al. Durable clinical benefit with nivolumab plus ipilimumab in DNA mismatch repair-deficient/microsatellite instability-high metastatic colorectal cancer. J Clin Oncol 2018;36:773–9.

7. Motzer RJ, Tannir NM, McDermott DF, Ar en Frontera O, Melichar B, Choueiri TK, et al. Nivolumab plus ipilimumab versus sunitinib in advanced renal-cell carcinoma. N Engl J Med 2018;378:1277–90.

8. Larkin J, Chiarion-Sileni V, Gonzalez R, Grob JJ, Cowey CL, Lao CD, et al. Combined nivolumab and ipilimumab ormonotherapy in untreated melanoma. N Engl J Med 2015;373:23–34

9. Yang H. Safety and efficacy of durvalumab (MEDI4736) in various solid tumors. Drug Des Devel Ther 2018;12:2085–96.

10. Fumet J-D, Isambert N, Hervieu A, Zanetta S, Guion J-F, Hennequin A, et al. Phase Ib/II trial evaluating the safety, tolerability and immunological activity of durvalumab (MEDI4736) (anti-PD-L1) plus tremelimumab (anti-CTLA-4) combined with FOLFOX in patients with metastatic colorectal cancer. ESMO Open 2018;3:e000375.

11. Desnoyers LR, Vasiljeva O, Richardson JH, Yang A,Menendez EEM, Liang TW, et al. Tumor-specific activation of an EGFR-targeting probody enhances therapeutic index. Sci Transl Med 2013;5:1–10.

12. Wong KR, Menendez E, Craik CS, Kavanaugh WM, Vasiljeva O. In vivo imaging of protease activity by Probody therapeutic activation. Biochimie 2016;122:62–7.

13. Wong C, Mei L,Wong KR, Menendez EEM, Vasiljeva O, Richardson JH, et al. A PD-L1-targeted Probody provides antitumor efficacy while minimizing induction of systemic autoimmunity. Cancer Immunol Res 2016;4(1 suppl): abstract A081.

14. Williams S. Tissue distribution studies of protein therapeutics using molecular probes: molecular imaging. Am Assoc Pharm Sci 2012;14:389–99.

15. Hettich M, Braun F, Bartholomä MD, Schirmbeck R, Niedermann G. Highresolution PET imaging with therapeutic antibody-based PD-1/PD-L1 checkpoint tracers. Theranostics 2016;6:1629–40.

16. Josefsson A, Nedrow JR, Park S, Banerjee SR, Rittenbach A, Jammes F, et al. Imaging, biodistribution, and dosimetry of radionuclide-labeled PD-L1 antibody in an immunocompetent mouse model of breast cancer. Cancer Res 2016;76:472–9.

17. Bensch F, van der Veen EL, Lub-de Hooge MN, Jorritsma-Smit A, Boellaard R, Kok IC, et al. 89Zratezolizumab imaging as a non-invasive approach to assess clinical response to PD-L1 blockade in cancer. Nat Med 2018;24:1852–8.

18. Verel I, Visser GWM, Boellaard R, van Walsum-Stigter M, Snow GB, van Dongen GAMS. 89Zr immunoPET: comprehensive procedures for the production of 89Zr-labeled monoclonal antibodies. J Nucl Med 2003;44:1271–81.

19. Nagengast WB, de Vries EGE, Hospers GA, Mulder NH, de Jong JR, Hollema H, et al. In vivo VEGF imaging with radiolabeled bevacizumab in a human ovarian tumor xenograft. J Nucl Med 2007;48:1313–9.

20. Bradford MM. A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem 1976;72:248–54.

21. Heskamp S, Wierstra PJ, Molkenboer-Kuenen JD, Sandker GW, Thordardottir S, Cany J, et al. PD-L1 microSPECT/CT imaging for longitudinal monitoring of PD-L1 expression in syngeneic and humanized mouse models for cancer. Cancer Immunol Res 2018;7:150–61.

22. Sun C, Mezzadra R, Schumacher TN. Regulation and function of the PD-L1 checkpoint. Immunity 2018;48:434–52.

23. Truillet C, Oh HLJ, Yeo SP, Lee CY, Huynh LT, Wei J, et al. Imaging PD-L1 expression with immunoPET.

Therapeutic design of anti-PD-L1 Probody CX-072

Bioconjug Chem 2018;29:96–103.

24. Chatterjee S, LesniakWG, GabrielsonM, LisokA, Wharram B, Sysa-Shah P, et al. A humanized antibody for imaging immune checkpoint ligand PD-L1 expression in tumors. Oncotarget 2016;7:10215–27.

25. Li D, Cheng S, Zou S, Zhu D, Zhu T, Wang P, et al. Immuno-PET imaging of 89Zr-labeled anti-PD-L1 domain antibody. Mol Pharm 2018;15:1674–81.

26. Lesniak WG, Chatterjee S, Gabrielson M, Lisok A, Wharram B, Pomper MG, et al. PD-L1 detection in tumors using 64Cu-atezolizumab with PET. Bioconjug Chem 2016;27:2103–10.

27. Nedrow JR, Josefsson A, Park S, Ranka S, Roy S, Sgouros G. Imaging of programmed cell death ligand 1: impact of protein concentration on distribution of anti-PD-L1 SPECT agents in an immunocompetent murine model of melanoma. J Nucl Med 2017;58:1560–6.

28. Kikuchi M, Clump DA, Srivastava RM, Sun L, Zeng D, Diaz-Perez JA, et al. Preclinical immunoPET/CT imaging using Zr-89-labeled anti-PD-L1 monoclonal antibody for assessing radiation-induced PD-L1 upregulation in head and neck cancer and melanoma. Oncoimmunology 2017;6:1–13.

29. Jagoda EM, Vasalatiy O, Basuli F, Opina ACL, Williams MR, Wong K, et al. Immuno-PET imaging of the programmed cell death-1 ligand (PD-L1) using a zirconium-89 labeled therapeutic antibody, avelumab. Mol Imaging 2019;18:1–14.

30. Heskamp S, Hobo W, Molkenboer-Kuenen JDM, Olive D, Oyen WJG, Dolstra H, et al. Noninvasive imaging of tumor PD-L1 expression using radiolabeled anti-PD-L1 antibodies. Cancer Res 2015;75:2928–36.

31. Beliz ario JE. Immunodeficient mouse models: an overview. Open Immunol J 2009;2:79–85.

32. Hartley G, Regan D, Guth A, Dow S. Regulation of PD-L1 expression on murine tumor-associated monocytes and macrophages by locally produced TNF-a. Cancer Immunol Immunother 2017;66:523–35.

33. Ulisse S, Baldini E, Sorrenti S, D'Armiento M. The urokinase plasminogen activator system: a target for anti-cancer therapy. Curr Cancer Drug Targets 2009;9:32–71.

34. Vasiljeva O, Hostetter DR, Moore SJ, Winter MB. The multifaceted roles of tumor-associated proteases and harnessing their activity for prodrug activation. Biol Chem 2019;400:965–77.

35. Overall CM, Kleifeld O. Validatingmatrix metalloproteinases as drug targets and anti-targets for cancer therapy. Nat Rev 2006;6:227–39.

36. Kessenbrock K, Plaks V, Werb Z. Matrix metalloproteinases: regulators of the tumor microenvironment. Cell 2010;141:52–67.

37. LeBeau AM, Lee M, Murphy ST, Hann BC, Warren RS, Delos Santos R, et al. Imaging a functional tumorigenic biomarker in the transformed epithelium. Proc Natl Acad Sci U S A 2013;110:93–8.

Supplementary data

FIGURE S1: CX-072, PbCtrl and CX-075 binding properties. (A) Binding of CX-075 parental antibody to human and murine PD-L1 determined by ELISA. Bound CX-075 is presented as optical density (OD) measured at 450 nm. Data is shown as mean ± standard deviation (SD). (B) CX-072 and CX-075 binding affinity for human and murine PD-L1. Affinity for PD-L1 is expressed as the apparent binding constant (Kapp). (C) Binding of CX-072, PbCtrl and CX-075, after conjugation with TFPN-SucDf, to PD-L1 determined by ELISA. Bound CX-075 or conjugated antibody (mAb-N-sucDf) is presented as OD measured at 450 nm. Data is shown as mean ± SD. (D) CX-072 and CX-075 immunoreactivity to PDL1 before and after conjugation with TFP-N-sucDf. Immunoreactivity to PD-L1 is presented as the effective concentration needed for 50% receptor occupation (EC50).

Therapeutic design of anti-PD-L1 Probody CX-072

FIGURE S2: Ex vivo biodistribution with escalating protein dose of 89Zr-CX-072, 89Zr-PbCtrl and 89Zr-CX-075. Biodistribution of (A) 89Zr-CX-072, (B) 89Zr-PbCtrl and (C) 89Zr-CX-075 at increasing total protein dose of 10, 50 and 250 μg, respectively, in MDA-MB-231 tumor-bearing mice at 6 days pi. Tracer uptake per organ is presented as percentage of injected dose per gram tissue (%ID/g). Data is shown as mean ± SD.

FIGURE S3: In vivo quantification of 89Zr-CX-072, 89Zr-PbCtrl and 89Zr-CX-075 PET imaging. Quantification of (A) 89Zr-CX-072, (B) 89Zr-PbCtrl and (C) 89Zr-CX-075 uptake 6 days pi in MDA-MB-231 tumor and blood pool at increasing total protein dose. Tracer uptake is presented as mean standardized uptake value (SUVmean). Data is shown as mean ± SD.

FIGURE S4: PET imaging and biodistribution in syngeneic MC38 tumor-bearing mice.

(A) In vivo quantification of 89Zr-CX-072, 89Zr-PbCtrl and 89Zr-CX-075 uptake in MC38 tumor and blood pool at 6 days pi. Tracer uptake is presented as SUVmean (B) Ex vivo biodistribution of 89Zr-CX-072, 89Zr-PbCtrl and 89Zr-CX-075 in MC38 tumor-bearing mice at 6 days pi. Tracer uptake per organ is presented as %ID/g. Data is shown as mean ± SD.

FIGURE S5: In vitro PD-L1 expression versus ex vivo uptake in MC38 and MDA-MB-231 tumors. (A) PD-L1 expression detected with flow cytometry in MDA-MB-231 and MC38 cell lines. CX-075 was used for detection of PD-L1 positive cells, and IgG4 antibody was used as isotype control. Data from a representative experiment is shown. (B) PD-L1 expression in MDA-MB-231 and MC38 cell lines presented as mean fluorescent intensity (MFI). Data is shown as mean ± SD. (C) Internalization of 89Zr-CX-075-PD-L1 complexes in MDAMB-231 and MC38 cells. Internalization was determined after 1 and 2 hours incubation at 37 °C, while control samples were kept at 4 °C. Internalization is presented as percentage of total PD-L1-bound 89Zr-CX-075.

Data is shown as mean ± SD. (D) Uptake of 89Zr-CX-072, 89Zr-PbCtrl and 89Zr-CX-075 in MDA-MB-231 and MC38 tumors at 6 days pi. Tracer uptake per organ is presented as %ID/g.

FIGURE S6: In vivo kinetics of 89Zr-CX-072, 89Zr-PbCtrl and 89Zr-CX-072. Residual 89Zr activity in BALB/c nude and C57BL/6J tumor-bearing mice measured at 1, 3 and 6 days pi of 89ZrCX-072, 89Zr-CX-PbCtrl and 89Zr-CX-075. Results are presented as radioactive decay corrected percentage of injected dose (%ID). Data is shown as mean ± SD.

Therapeutic design of anti-PD-L1 Probody CX-072

Laura Kist de Ruijter 1, Jahlisa S Hooiveld-Noeken 1, Danique Giesen 1, Marjolijn N Lub-de Hooge 2, Iris C Kok 1, Adrienne H Brouwers 3, Sjoerd G Elias 4, Margaret T L Nguyen 5, Hong Lu 5, Jourik A Gietema 1, Mathilde Jalving 1, Derk J A de Groot 1, Olga Vasiljeva 5, Elisabeth G E de Vries 1

1 Department of Medical Oncology, 2 Department of Clinical Pharmacy and Pharmacology, and 3 Medical Imaging Center, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands; 4 Department of Epidemiology, Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht University, Utrecht, the Netherlands; 5 CytomX Therapeutics Inc., South San Francisco, California.

Clin Cancer Res. 2021;27(19):5325-5333.

First-in-Human study of the biodistribution and pharmacokinetics of 89Zr-CX-072, a novel immunoPET tracer based on an anti-PD-L1 Probody

Abstract

Purpose: CX-072, a PD-L1–targeting Probody therapeutic, is engineered to be activated by tumor proteases that remove a masking peptide. To study effects on biodistribution and pharmacokinetics, we performed 89Zr-CX-072 positron emission tomography (PET) imaging.

Experimental Design: Patients received ~1 mg, 37 MBq 89Zr-CX-072 plus 0, 4, or 9 mg unlabeled CX-072 and PET scans at days 2, 4, and 7. After that, treatment comprised 10 mg/kg CX-072 q2 weeks (n = 7) + 3 mg/kg ipilimumab q3w 4× (n = 1). Normal organ tracer uptake was expressed as standardized uptake value (SUV)mean and tumor uptake as SUVmax. PD-L1 expression was measured immunohistochemically in archival tumor tissue.

Results: Three of the eight patients included received 10-mg protein dose resulting in a blood pool mean SUVmean ± SD of 4.27 ± 0.45 on day 4, indicating sufficient available tracer. Tumor uptake was highest at day 7, with a geometric mean SUVmax 5.89 (n = 113) and present in all patients. The median follow-up was 12 weeks (4–76+). One patient experienced stable disease and two patients a partial response. PD-L1 tumor expression was 90% in one patient and ≤1% in the other patients. Mean SUVmean ± SD day 4 at 10 mg in the spleen was 8.56 ± 1.04, bone marrow 2.21 ± 0.46, and liver 4.97 ± 0.97. Four patients out of seven showed uptake in normal lymph nodes and Waldeyer's ring. The tracer was intact in the serum or plasma.

Conclusions: 89Zr-CX-072 showed tumor uptake, even in lesions with ≤1% PD-L1 expression, and modest uptake in normal lymphoid organs, with no unexpected uptake in other healthy tissues.

Introduction

Immune-checkpoint inhibitors yield impressive responses in patients with locally advanced and metastatic malignancies and can result in long-term survival in a subset of cancer patients. Currently, ipilimumab targeting cytotoxic T lymphocyte-associated antigen 4 (CTLA4) and several programmed cell death protein 1 (PD-1)– and PD-1 ligand 1 (PD-L1)–targeting medicines are registered for multiple tumor types. The antitumor efficacy of immunecheckpoint inhibitors is higher when combined, as seen for nivolumab plus ipilimumab (1–3). Immune-related toxicity is induced by all immune-checkpoint inhibitors. However, the percentage of patients with side effects increases from 20% for single immune-checkpoint inhibitors to 54% when ipilimumab and nivolumab are combined (4).

These immune-related toxicities have stimulated the development of medicines that have similar pharmacologic activity but fewer side effects. CX-072, a Probody therapeutic directed against PD-L1, is a prime example of this. This recombinant monoclonal antibody prodrug is

First-in-human PET imaging of 89Zr-CX-072 anti-PD-L1 Probody

designed to be conditionally activated by proteases in the tumor microenvironment. CX-072 contains a mask with a protease-cleavable substrate at the amino-terminus of the light chain. This mask is devised to block PD-L1 binding until it is released in the tumor microenvironment. The effect of such modifications on antibody biodistribution and kinetics is unknown.

Molecular imaging is increasingly seen as an approach to facilitate drug development (5). An antibody labeled to a positron emission tomography (PET) isotope, such as zirconium-89 (89Zr), which complements the long biological half-life of antibodies, allows the collection of data on whole-body drug distribution (6, 7). PD-L1 is expressed by tumor cells, as well as by macrophages, dendritic cells, and T cells. Clinical PET imaging with 89Zr-atezolizumab, a zirconium-labeled PD-L1 antibody, showed uptake in tumor lesions and the spleen, lymph nodes, Waldeyer's ring, and sites of inflammation (8). Whole-body PET imaging may also provide information as a biomarker. For example, higher 89Zr-atezolizumab tumor uptake correlated with better tumor response and overall survival in cancer patients treated with atezolizumab (8), and uptake on 89Zr-trastuzumab PET was predictive for effect of trastuzumab–emtansine therapy in patients with breast cancer (9).

In immune-competent mice, 89Zr-CX-072 PET showed specific drug accumulation in PDL1–expressing human tumors with minor uptake in lymphoid tissues (10). In patients with advanced solid tumors, long-term treatment with CX-072 administered as a single agent or in combination with ipilimumab has shown durable objective responses in patients with advanced solid tumors (11, 12).

We aimed to obtain insight into CX-072 biodistribution and pharmacokinetics (PK) and to learn whether CX-072's Probody therapeutic design influences them. Therefore, we studied 89Zr-CX-072 biodistribution in patients with locally advanced or metastatic malignancies.

Materials and Methods

Study design and participants

This is a single-center, prospective substudy of an international multicenter phase I–II treatment study. Eligible patients had measurable advanced or metastatic solid malignancies, archival tumor tissue available, age >18 years, Eastern Cooperative Oncology Group performance status (ECOG PS) of 0–1, an anticipated life expectancy of at least 3 months, and stable hematologic and chemistry laboratory values. The main study and the substudy were approved by the medical ethical committee of the University Medical Center Groningen (UMCG). PD-L1 expression in the tumor was not required for eligibility to the substudy. All patients gave written informed consent. The study is registered in ClinicalTrials.gov (NCT03013491).

Translational Relevance

The PD-L1–targeting antibody CX-072 is engineered to be conditionally activated in the tumor microenvironment by tumor-associated proteases and to remain predominantly masked in the circulation. This has raised interest in the effects of this design on biodistribution and pharmacokinetics. Therefore, we performed whole-body 89Zr-CX-072 PET imaging in patients. 89Zr-CX-072 showed evident tumor uptake, even in lesions with ≤1% PD-L1 expression in tumor biopsies, modest uptake in normal lymphoid organs, and no unexpected uptake in other healthy tissues. The highest normal tissue uptake was in the spleen, although less than the nonconditionally activated 89Zr-labeled PD-L1 antibody 89Zr-atezolizumab. 89Zr-CX-072 was found to be intact in the circulation, predominantly in its masked form. Our findings indicate that 89Zr-CX-072 PET provides insight into the whole-body distribution of a newly designed therapeutic drug candidate that can be achieved with a small number of patients.

Procedures

Patient screening was performed within 30 days before tracer injection and comprised a baseline computed tomography (CT), electrocardiogram (ECG), and laboratory assessment. After completing the imaging substudy part, patients received CX-072 at their assigned dose cohort 10 mg/kg every 2 weeks, with or without an initial combination with 4 cycles of ipilimumab given at 3 mg/kg at 3 weekly intervals. Treatment with CX-072 was initiated within 10 days after the last PET scan.

Tumor response was assessed every 8 weeks and after one year every 12 weeks. The objective response was defined as a complete or partial response on two consecutive tumor assessments at least 4 weeks apart, according to RECIST version 1.1 and irRECIST (13, 14).

PD-L1 expression in archival tissue biopsies, performed at the latest 27 months before the study start, was determined by PD-L1 immunohistochemistry (IHC) 22C3 pharmDx assay (Dako) and scored as the percentage of viable tumor cells showing partial or complete membrane staining of PD-L1.

Tracer development and PET imaging

Clinical-grade 89Zr-CX-072 was developed and produced in the UMCG, as described previously (10). For quality control, 89Zr-CX-072 met all release specifications on conjugation ratio, radioactive yield, protein purity, concentration, pH, radiochemical purity, residual solvents, sterility, and endotoxin content. Preservation of immunoreactivity after conjugation was verified by enzyme-linked immunosorbent assay (ELISA).

First-in-human PET imaging of 89Zr-CX-072 anti-PD-L1 Probody

Cohorts of 2 to 3 patients received 37 MBq of ~1 mg 89Zr-CX-072 plus 0 mg, 4 mg, or 9 mg unlabeled CX-072 (~1 mg, 5 mg, or 10 mg total tracer protein doses) until sufficient blood pool levels were reached with satisfactory visualization of tumor lesions. Sufficient unlabeled dose supplementation was determined by comparing 89Zr-CX-072 SUV mean in the blood pool at day 4 with other 89Zr antibody tracers with well-known kinetics over time (9, 15, 16), thereby taking into account deposition in sink organs known for a PD-L1 antibody (8).

89Zr-CX-072 safety was assessed through changes in laboratory test results, changes in vital signs, and summaries of adverse events (AE) before and after exposure to 89Zr-CX-072. AE data were recorded according to NCI Common Terminology Criteria for Adverse Events v4.0.

All PET scans were performed on days 2, 4, and 7 after tracer administration and combined with a low-dose CT scan for attenuation correction and anatomic reference, on a Siemens Biograph mCT 40 or 64-slice PET/CT camera, as described previously (8). PET images were reconstructed with the parameters advised for multicenter 89Zr-monoclonal antibody PET scan trials, according to EARL1 (17).

Biodistribution and tumor lesion quantification

Quantification of tracer uptake in normal tissues and tumor lesions was performed in the ACCURATE tool by manually placing a 3D-sphere as the volume of interest (VOI) on each of the three post tracer administration PET scans for each patient (18) (RRID: SCR_020955). Tumor lesions were identified before treatment by conventional imaging techniques according to RECIST1.1. Standardized uptake value (SUV) was calculated using net injected dose, body weight, and measured radioactivity within the VOI on the PET image, corrected for decay. The tracer uptake in healthy tissue was expressed as mean standardized uptake value (SUVmean; average uptake within a VOI), which is standard for calculating physiologic uptake in homogeneous tissues. SUV mean is reported with ± standard deviation. In tumor lesions, uptake was calculated as maximum SUV (SUVmax; maximum uptake within a VOI) to assess target-specific uptake.

Uptake of 89Zr-CX-072 in Waldeyer's ring, axillary, and inguinal lymph nodes was assessed visually. The uptake was compared to the background as well as liver and spleen.

Plasma PK and ADA assay

Blood samples for PK analysis were obtained at 30 minutes and on days 2, 4, and 7 after tracer administration.

Magnetic beads coated with protein A were used to enrich for immunoglobulin (including intact and cleaved CX-072) in K2EDTA plasma samples. Following protein denaturation,

reduction, and alkylation, the proteins were digested with trypsin, and two peptide fragments to CX-072 were monitored: 1 peptide from the CX-072 heavy chain present in both the intact and active forms of CX-072 (for quantitation of total CX-072) and 1 peptide from the CX072 prodomain present only in the intact form of CX-072 (for quantitation of intact CX-072). Corresponding stable isotope-labeled peptides were used as internal standards. The final extract was analyzed via high-performance liquid chromatography mass spectrometry with tandem mass spectrometry detection using positive ion electrospray. The assay has a quantifiable range of 0.657 to 328 nmol/L for intact and total CX-072.

Blood samples were obtained before tracer infusion to study the presence of endogenous antidrug antibodies (ADA) against CX-072. A validated method using solid-phase extraction with acid dissociation (SPEAD) sample pretreatment followed by a direct electrochemiluminescent assay was used to detect anti-CX-072 antibodies in serum. To maximize the detection of ADAs to CX-072, including antibodies directed against the target binding region, ADAs were extracted from serum using a 1:1 mixture of biotinylated CX-072 and biotinylated CX-075 (parental antibody of CX-072). Following SPEAD processing steps, anti-CX-072 antibodies were detected using a 1:1 mixture of sulfo-tagged CX-072 and sulfo-tagged CX-075. Sample testing was conducted in a tiered manner beginning with a screening assay, which identified presumptive positive or negative samples. This was followed by confirmatory assays. The screening sensitivity is 2.72 ng/mL.

Tracer integrity

The intactness of the 89Zr-CX-072 was studied in serum or plasma, depending on sample availability, collected days 2, 4, and 7 with sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) as described previously (10). To detect a ±4 kDa difference between masked and unmasked antibody, 89Zr-CX-072 was reduced to its heavy and light chains using β-mercaptoethanol. Intact 89Zr-CX-072 was used as control and diluted to match sample radioactivity. 89Zr-CX-072 was detected autoradiographically by exposing gels to a multipurpose phosphor plate (PerkinElmer) overnight at −20 °C. Exposures were captured using a Cyclone phosphor imager. Intact tracer and masked versus unmasked antibody were quantified using ImageJ (version 1.52p, RRID: SCR_003070).

Statistical analysis

Descriptive statistics (i.e., mean and SDs), median counts, and percentages were used to provide an overview of the patient population and 89Zr-CX-072 uptake values. Despite the early-phase nature of this study (without a priori–defined testable hypotheses and supporting sample size calculations), and the limited number of patients studied, we performed various exploratory statistical analyses making optimal use of the wealth of repeated measurements

First-in-human PET imaging of 89Zr-CX-072 anti-PD-L1 Probody

that are obtainable using PET imaging in patients with metastatic cancer. To describe the 89ZrCX-072 biodistribution according to administered dose and time post-tracer administration in terms of tumor uptake, we used linear mixed models. The natural logarithmic of the SUVmax was used as dependent variable (to account for its right-skewed distribution), taking withinpatient and within-tumor clustering into account using nested random intercepts. The results of these models are expressed as geometric mean uptake levels. To allow post-injection time–tumor–uptake patterns to be variable between dose groups, interaction terms between days post-injection and dose were added to the models. They were tested for statistical significance using the likelihood ratio test under maximum likelihood (such tests also served to test the contribution of other variables to tumor uptake). Besides analyzing days post-tracer administration as a categorical variable, it was also analyzed continuously, choosing the best-fitting form from a linear, a log-linear, or a quadratic curve using the Akaike's Information Criterion under maximum likelihood. The 89Zr-CX-072 biodistribution in normal tissues was assessed using similar linear mixed-effect models using SUVmean as the dependent variable, for which a natural logarithmic (or other) transformation was not necessary.

Again, as strictly exploratory, the relation between best treatment response per patient and tumor uptake was analyzed by similar linear mixed-effect models. The relation between geometric mean tumor SUVmax per patient at day 7 post-injection and overall survival was studied using Firth's penalized maximum likelihood small-sample bias reduction method for Cox regression and assuming linearity and proportionality. All reported statistical tests are two-sided, without correction for multiple testing. Main analyses were performed using R version 3.2.1 for macOS, particularly using packages lme4 (1.1-11, RRID: SCR_015654), lmerTest (2.0-20, RRID: SCR_015656), and coxphf (1.11).

Results

Eight patients were enrolled between March 2018 and November 2018. Patient characteristics are listed in Table 1. All patients completed the entire PET scan series, and none experienced tracer-related AEs. Two patients received a total protein dose of 1 mg, three patients 5 mg, and three patients 10 mg. Thereafter treatment consisted of CX-072 monotherapy, and one patient additionally received ipilimumab (Supplementary Table S2).

89Zr-CX-072 uptake in tumor lesions

In total, 118 tumor lesions were identified, of which 5 were not included in the following analyses as these were irradiated within 3–12 weeks before tracer injection. Tumor characteristics, best response, and the geometric mean of SUVmax per patient at day 7 post-injection are shown in Fig. 1.

Table 1: Patient and tumor characteristics.

Characteristics

n

Age, median (range), years 56 (40–68)

Sex, n (%)

Male 3 (37.5)

Female 5 (62.5)

Primary tumor, n (%)

MSI high colorectal carcinoma 3 (37.5)

Pancreatic carcinoma 1 (12.5)

Ovarian carcinoma 1 (12.5)

Cervical carcinoma 1 (12.5)

Anaplastic thyroid carcinoma 1 (12.5)

Germ cell tumor 1 (12.5)

ECOG PS, n (%)

0 4 (50)

1 4 (50)

Prior lines of therapy, n (%)

0 1 (12.5)

1 1 (12.5)

³ 2 6 (75)

Tumor lesion sites, n (%)

Total number lesions 118 (100)

Lung 36 (30.5)

Lymph node 36 (30.5)

Soft tissue 27 (21.6)

Liver 8 (6.8)

Bone 4 (3.4)

Pancreas 2 (1.7)

Spleen 2 (1.7)

Intestine 1 (0.8)

Kidney 1 (0.8)

Thyroid 1 (0.8)

IHC tumor PD L1, n (%) 7 (87.5)

£ 1% 90% 1 (12.5)

Tumor uptake was heterogeneous within and between patients. The highest tumor lesion uptake and the highest geometric mean of SUVmax per patient were seen at the 10-mg dose. All patients showed uptake in their tumor lesions, with an overall geometric mean SUVmax of 4.73 (95% CI: 2.81–7.94; averaged over day 2–7 post-injection). Tumor uptake at 10-mg dose increased between day 2–7 post-injection, but not at other dose levels (Fig. 2). Of the 113 nonirradiated lesions defined by conventional imaging techniques, 84 were visually detected by conventional imaging and by 89Zr-CX-072 PET. These included lymph nodes with a short axis of less than 1.5 cm but defined as malignant based on other clinical or CT features, and 5 skin lesions not visible on CT. At the same time, 28 lesions were not distinguishable from the background or did not show uptake by 89Zr-CX-072 PET, but were detectable by conventional

FIGURE 1: SUV max 89Zr-CX-072 for the 113 tumor lesions of the eight patients for the three different tracer protein doses indicated by the color bars at the top; 1 mg (n = 2), 5 mg (n = 3), and 10 mg (n = 3). The geometric mean SUVmax of all lesions within a patient is indicated by the black horizontal line. A single lesion is marked with either a dot or “×” in case of an unmeasurable lesion. The dots' size corresponds with the anatomic size of the lesion and the color with the location of the lesion. Best responses are mentioned on the x-axis above tumor type.

imaging techniques. One lesion was visible on 89Zr-CX-072 PET that was not visible on CT imaging at baseline, but appeared malignant on the CT 6 weeks later.

The median follow-up at data cutoff on November 7, 2020, is 12 weeks (range, 4–76+). One patient experienced stable disease (SD) and two a partial response (PR). The geometric mean SUV max in tumor lesions from patients with at least stable disease as best response was 6.84 (95% CI: 2.25–20.78; 3 patients with 52 lesions) and was 3.62 (95% CI: 1.57–8.36; 5 patients with 61 lesions) in patients with progressive disease (PD) (likelihood ratio P = 0.14; averaged over day 2–7 post-injection and adjusted for differences in administered dose between patients). The patient with anaplastic thyroid cancer who experienced a PR for 18 months had high 89ZrCX-072 tumor uptake with geometric mean SUVmax 11.49 (Figs. 1 and 2). Furthermore, a patient with microsatellite instable (MSI) high colorectal cancer who responded as well showed uptake in an irresectable sacral local recurrence with SUVmax 3.25 on day 7 (blood pool day 7 at 1 mg mean SUVmean 1.65), lacking expression of PD-L1 by IHC. This tumor became resectable and was surgically removed after 15 months of treatment. At data cutoff, this patient was

First-in-human PET imaging of 89Zr-CX-072 anti-PD-L1 Probody

FIGURE 2: Uptake of 89Zr-CX-072 in tumor lesions.

(A) Maximum intensity projection of 89Zr-CX-072 PET on day 7 post-injection in a patient with anaplastic thyroid cancer with 10-mg total protein dose. Arrows show uptake in a thyroid tumor mass, cutaneous tumor lesions on the chest, malignant lymph nodes in the neck and axilla, and a bone lesion in the pelvis. (B) Axial view of 89Zr-CX-072 PET-CT shows uptake in thyroid mass and malignant lymph nodes. (C) Axial view of 89ZrCX-072 PET-CT shows uptake in cutaneous lesions on the chest and malignant axillary lymph nodes. (D) Axial view of 89Zr-CX-072 PET-CT shows uptake in a pelvic bone lesion. (E) SUV max 89Zr-CX-072 tumor uptake per tracer protein dose. Tumor uptake is highest at 10 mg (albeit nonsignificantly: likelihood ratio test for effect of dose on uptake P = 0.25). Between days 2 and 7, the uptake increases in the 10-mg protein group, whereas the uptake stays stable in the other dose groups (likelihood ratio test for interaction for a different time-tumor–uptake pattern across dose groups P = 0.000044).

free of disease for 11 months. In an exploratory univariable survival analysis accounting for small sample bias (8 patients, 6 events), the hazard ratio for dying was estimated to be 1.32 (95% CI: 0.98–1.93) per unit decrease in per-patient geometric mean tumor SUVmax (likelihood ratio P = 0.070). PD-L1 tumor expression measured immunohistochemically was 90% in the patient with anaplastic thyroid tumor, in the other patients this was ≤1%.

Biodistribution and pharmacokinetics of 89Zr-CX-072

The mean SUV mean in the blood pool at the 10-mg protein dose on day 4 was 4.27 ± 0.45, indicating sufficient tracer levels available to reach tumor lesions. At this dose, the mean SUV mean at day 4 in the liver was 5.05 ± 0.75 and the kidney 2.71 ± 0.64. Figure 3 shows the SUV mean for days 2, 4, and 7 after tracer injection with 1-, 5-, and 10-mg protein dose across the

FIGURE 3: Uptake in 89Zr-CX-072 in normal tissues per tracer protein dose.

SUV mean days 2, 4, and 7 after tracer injection with 1-, 5-, and 10-mg protein dose in blood pool, brain, lung, liver, kidney, spleen, intestine, bone marrow, bone cortex, and muscle. SUVmean 89Zr-CX-072 in the spleen shows the lowest uptake at 10-mg protein dose (likelihood ratio test for effect of dose on uptake: P = 0.017). Uptake in the spleen stays stable over time (likelihood ratio test for a different time–spleen–uptake pattern across dose groups: P = 0.54). Fitted regression lines with 95% CI are based on linear mixed-effect models taking clustering within patients (and for spleen and liver uptake multiple regions) into account.

normal tissues.

Remarkable was the spleen SUVmean, which was the highest of all healthy organs. Corrected for decay at each time point, the SUVmean of the spleen was highest at 1 mg (n = 2), lower at 5 mg (n = 3), and the lowest at 10 mg (n = 3) (test for effect of dose on uptake: likelihood ratio P = 0.017). In the liver, the average SUVmean also lowered with increasing tracer protein doses. The spleen uptake was already high on day 2; however, in contrast to what was seen in the tumor lesions, it did not further increase the days thereafter. In both spleen and liver, uptake was stable over time. In other healthy parenchymal tissues, the PET signal decreased over the days after tracer injection.

The bone marrow mean SUV mean at 10 mg was 2.54 ± 0.49 and remained stable over time. We also studied 89Zr-CX-072 uptake in other lymphoid tissues. The 89Zr-CX-072 uptake in the Waldeyer's ring on day 7 was present in four patients. Uptake of 89Zr-CX-072 was present on day 7, in axillary lymph nodes in four out of seven patients, and in inguinal lymph nodes in five out of eight patients. Percentages of visual uptake in Waldeyer's ring and lymph nodes are shown in Fig. 4. Visual uptake did not differ between tracer protein doses and occurred at all

First-in-human PET imaging of 89Zr-CX-072 anti-PD-L1 Probody

doses (Supplementary Table S1).

89Zr-CX-072 uptake in sites of inflammation

89Zr-CX-072 uptake was also observed at sites of inflammation in two patients: one with polyarthritis and another with unilateral rhinosinusitis (Fig. 4).

Pharmacokinetic analysis

Intact and total CX-072 protein concentrations at evaluated time plots for subjects receiving a total dose of 10 mg are depicted in Fig. 5B. Notably, CX-072 concentration was below the limit of quantitation (LQ) in 32% of all PK samples and corresponded to those patients who received 1- and 5-mg doses. Following the 10-mg dose, intact and total CX-072 concentrations remained above the LQ for PK sampling duration. Measurement of 89Zr blood radioactivity was not performed. All subjects were ADA negative at baseline.

Tracer integrity and degree of unactivated CX-072

89Zr-CX-072 integrity in blood samples collected up to 7 days after injection in five patients showed intactness of 89Zr-bound CX-072. CX-072 was predominantly present in its masked, unactivated form, up to day 7 (Fig. 5B–E).

FIGURE 4: Uptake of 89Zr-CX-072 in lymphoid tissues and sites of inflammation.

Top, examples of visual moderate to clear 89Zr-CX-072 PET-CT uptake in lymphoid tissues on day 7 postinjection. (A) Axial view of the left axilla region with uptake in lymph nodes. (B) Axial view of the inguinal region with uptake in inguinal lymph nodes. (C) Axial view of Waldeyer's ring region with high uptake in tonsils. (D) Percentages of patients with visual uptake on day 7 of different lymphoid tissue stations: tonsils (n = 8), axillary lymph nodes (n = 7), and inguinal lymph nodes (n = 8). Right, examples of 89Zr-CX-072 PETCT of sites of inflammation. (E) Patient with polyarthritis and uptake in left shoulder and (F) knee joints. (G) Patient with unilateral rhinosinusitis.

First-in-human PET imaging of 89Zr-CX-072 anti-PD-L1 Probody

FIGURE 5: Pharmacokinetic analysis, tracer integrity, and assessment of CX-072 activation.

(A) Schematic structure of CX-072 heavy and light chain variants. (B) Intact and total CX-072 concentration versus time for subjects who received the 10-mg tracer protein dose (n = 3). (C) Representative example of ex vivo serum or plasma SDS-PAGE analysis, 89Zr-CX-072 remained intact up to 7 days. (D) Quantification of intact tracer in samples obtained from n = 5 patients at 30 minutes and 2, 4, and 7 days after tracer administration. (E) After reducing 89Zr-CX-072 to its heavy and light chains, SDS-PAGE showed 89Zr-CX-072 is predominantly present in its masked form up to 7 days. Intact 89Zr-CX-072 was used as control.

Discussion

This is the first-in-human study with 89Zr-labeled Probody therapeutic PET imaging to study CX-072 drug biodistribution. We showed clear 89Zr-CX-072 tumor uptake in all patients, even in lesions of patients lacking IHC PD-L1 tumor expression. There was modest uptake in normal lymphoid organs without unexpected uptake in other healthy tissues.

The rationale behind the Probody therapeutic design is that the molecule is conditionally activated in the tumor. This should reduce the distribution of its activated form in normal tissues compared with current, unmodified, PD-L1 antibodies, while drug uptake in tumor lesions is not affected. In immune-competent mice, 89Zr-CX-072 PET did indeed show drug

accumulation in PD-L1 expressing human tumors similar to the parental antibody and minor uptake in lymphoid tissues (10). Our current study with 89Zr-CX-072 PET allowed further analysis of these properties in humans. We have shown that 89Zr-CX-072 was intact in the blood for up to 7 days and predominantly in its masked form in circulation, consistent with Probody therapeutic design and in agreement with findings from the PK analyses from the treatment studies (19). CX-072 shows a linear PK in doses that are comparable to the used tracer proteins doses in this imaging study (19), so the biodistribution of the tracer is representative for CX-072 in therapeutic dose.

All patients showed 89Zr-CX-072 uptake into tumor lesions. At 10 mg, uptake in tumor lesions increased between days 2 and 7. In vitro studies have priorly shown 89Zr-CX-072 bound to PDL1 on tumor cells can internalize, followed by intracellular residualization of radioactivity (10). In the current imaging study, at all doses, the tumor-to-blood ratio increased over time and was highest in the 10-mg cohort. Normal healthy organs without target show nonspecific 89ZrCX-072 uptake on the PET scan, which lowers with decreasing blood pool activity. Therefore, increasing 89Zr-CX-072 uptake in tumor lesions over time suggests that this tumor uptake is specific and that 89Zr-CX-072 is retained in the tumor by its interaction with PD-L1.

Tumor uptake was heterogeneous within and between patients. Also, 89Zr-CX-072 tumor uptake was seen in patients whose tumors had low PD-L1 expression by IHC on archival tissue. The same was the case in the 89Zr-atezolizumab PET imaging study, in which a biopsy was performed immediately after the last PET scan (8). The 89Zr-atezolizumab tumor uptake into that specific biopsied lesion was also seen when PD-L1 staining was negative.

Tumor tracer uptake showed target-mediated, specific kinetics. Therefore, this discrepancy may be explained by heterogeneous PD-L1 expression within one tumor lesion, where tissue analysis is limited when taking only a small biopsy (20). Also, heterogeneity for PD-L1 expression between archival tissue material and current metastases may play a role. Moreover, PD-L1 expression by tumor cells is dynamic, causing temporal differences (20). Therefore, molecular imaging with a tracer that internalizes, including 89Zr-CX-072, may provide an accumulated image of target expression over time. In contrast, IHC shows target expression at a given time point and could miss this dynamic expression. Finally, as molecular imaging shows target expression in the total lesion and in the whole body, it overcomes biases on sampling and heterogeneity between and within tumor lesions.

89Zr-CX-072 tumor uptake was highest for the 10-mg dose, although the high PD-L1 expression in one patient may have played a role in this. Exploratory analyses suggest a positive association between uptake and patient outcome, corroborating our previous findings in the

First-in-human PET imaging of 89Zr-CX-072 anti-PD-L1 Probody

89Zr-atezolizumab PET imaging study. However, the number of patients is too low to draw firm conclusions. Highest SUVmax occurred in a patient with anaplastic thyroid cancer with tumor PD-L1 expression of 90%, who experienced a partial response. Recent studies of patients with anaplastic thyroid cancer suggest that molecular-based personalized therapies, including PD-L1 antibody treatment, show improvements in survival (21, 22).

At the 10-mg 89Zr-CX-072 dose, we had the opportunity to compare the results of PET imaging with 10 mg of the 89Zr-labeled PD-L1 antibody atezolizumab (8). At that dose, 89Zr-CX-072 uptake in the bone marrow and lymph nodes was indeed lower than 89Zr-atezolizumab. Remarkably, after initial significant 89Zr-CX-072 uptake in the spleen on day 2, no further increase occurred through day 7. This 89Zr-CX-072 spleen uptake was similar to 89Zr-atezolizumab at 1 hour after injection. However, 89Zr-atezolizumab spleen uptake increased subsequently. PD-L1 is expressed in the spleen by immune cells but also by littoral cells that line the splenic sinuses of the red pulp. High perfusion of the spleen is the primary filter for the blood (23, 24). Therefore, the rapid high uptake of 89Zr-CX-072, already on day 2 post-injection, may well be explained by specific binding of 89Zr-CX-072 to PD-L1 expressing littoral cells.

The affinity of CX-072 to PD-L1 (KD = 9.9 nmol/L) is lower compared with atezolizumab (KD = ~0.30 nmol/L) when masked and similar when activated (KD = 0.25 nmol/L; ref. 10). So, even in its masked form, CX-072 still has measurable, although low, affinity for PD-L1, which could result in PD-L1 binding without protease cleavage of the linker and mask.

Interestingly, spleen uptake decreased when protein dose supplementation of CX-072 as part of the tracer increased. This phenomenon was not studied with 89Zr-atezolizumab PET imaging, as only a 10-mg dose was applied, but it was also reported for the labeled PD-L1 antibody 89Zr-durvalumab (25). Therefore, the rapid high uptake, already on day 2 post-injection, likely by binding of 89Zr-CX-072 to PD-L1 expressing littoral cells, is rapidly partly saturated by higher protein doses. The same saturation did not occur in other healthy tissues.

For other lymphoid tissues, for example, bone marrow 89Zr-CX-072 quantitative uptake was approximately half as much compared with the 89Zr-atezolizumab PET imaging study. Uptake of 89Zr-CX-072 in nonmalignant lymph nodes and of 89Zr-atezolizumab was scored the same way. Visual uptake of lymph nodes in the 89Zr-CX-072 study was ~20% lower for axillary and inguinal stations compared with 89Zr-atezolizumab PET imaging study. Uptake in other healthy tissues was comparable with uptake seen with other antibodies, namely, low uptake in the brain, lung, cortical bone, muscle, a subcutaneous tissue, and higher uptake in liver, kidney, and intestine, reflecting antibody metabolism and elimination (9, 15, 16).

Although procedures, equipment, and reconstruction protocol were uniform, results comparing 89Zr-CX-072 with 89Zr-atezolizumab should be interpreted cautiously, as the current study was not a head-to-head comparison of the two antibodies. However, a comparison of the kinetics of both tracers shows 89Zr-CX-072 has similar kinetics of nonspecific uptake in normal tissues, similar kinetics of specific uptake in tumor lesions, and less specific uptake in organs with the target outside the tumor microenvironment.

This is the first report of imaging in humans using a 89Zr-labeled Probody therapeutic that is conditionally activated in the tumor microenvironment. The findings of accumulation in the tumor, and modest uptake in normal lymphoid organs, support tumor-associated protease cleavage of the mask with subsequent target engagement in the tumor and decreased target engagement outside the tumor. Taking together, our findings suggest that already significant insight into the whole-body distribution of a new specialized designed medicine can be reached with a small number of patient subjects.

Authors' Disclosures

M.N. Lub-de Hooge reports conflict of interest related to CytomX-UMCG. A.H. Brouwers reports grants from CytomX Therapeutics, Inc. and nonfinancial support from CytomX Therapeutics, Inc. during the conduct of the study. M.T.L. Nguyen is a full-time employee of CytomX Therapeutics, Inc. H. Lu is a full-time employee of CytomX Therapeutics, Inc. J.A. Gietema reports grants from Roche, AbbVie, and Siemens during the conduct of the study. D.J.A. de Groot reports grants from CytomX Therapeutics, Inc. during the conduct of the study. O. Vasiljeva is a full-time employee of CytomX Therapeutics, Inc. E.G.E. de Vries reports grants from CytomX Therapeutics, Inc. during the conduct of the study, as well as other support from Daiichi Sankyo, NSABP, and Crescendo Biologics, and grants from Amgen, Bayer, Crescendo Biologics, G1 Therapeutics, Genentech, Regeneron, Roche, Servier, and Synthon outside the submitted work. No disclosures were reported by the other authors.

Authors' Contributions

L. Kist de Ruijter: Writing–original draft, writing–review and editing, acquisition of data (acquired and managed patients, etc.), analysis and interpretation of data. J.S. Hooiveld-Noeken: Writing–review and editing, acquisition of data (acquired and managed patients, etc.). D. Giesen: Writing–review and editing, acquisition of data. M.N. Lub-de Hooge: Conceptualization, writing–review and editing, acquisition of data. I.C. Kok: Writing–review and editing, acquisition of data (acquired and managed patients, etc.). A.H. Brouwers: Methodology, writing–review and editing, image interpretation. S.G. Elias: Data curation, formal analysis, visualization, writing–review and editing, analysis and interpretation of data (statistical analysis, biostatistics, computational analysis). M.T.L. Nguyen: Methodology, writing–review and editing.

First-in-human PET imaging of 89Zr-CX-072 anti-PD-L1 Probody

H. Lu: Data curation, methodology, writing–review and editing. J.A. Gietema: Writing–review and editing, acquisition of data (acquired and managed patients, etc.). M. Jalving: Writing–review and editing, acquisition of data (acquired and managed patients, etc.). D.J.A. de Groot: Conceptualization, formal analysis, writing–review and editing, acquisition of data (acquired and managed patients, etc.). O. Vasiljeva: Conceptualization, data curation, writing–review and editing. E.G.E. de Vries: Conceptualization, data curation, writing–original draft, writing–review and editing.

Acknowledgments

We thank Dr. R. Boellaard for support PET analyses. S. Davur, E. Ureno, and S. Viswanathan (CytomX Therapeutics, Inc.) assisted with producing and characterizing clinical-grade 89ZrCX-072. We thank the Clinical Development team of CytomX Therapeutics, Inc. (especially V. Huels, M. Will, and S. Yalamanchili) for the support of substudy design and execution. A research grant of CytomX Therapeutics, Inc. supported this study to the UMCG, Groningen, the Netherlands. The study drug CX-072 and control molecules were supplied by CytomX Therapeutics, Inc. (PROBODY is a U.S. registered trademark of CytomX Therapeutics, Inc.).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

References

1. Hodi FS, Chesney J, Pavlick AC, Robert C, Grossmann KF, McDermott DF, et al. Combined nivolumab and ipilimumab versus ipilimumab alone in patients with advanced melanoma: 2-year overall survival outcomes in a multicentre, randomised, controlled, phase 2 trial. Lancet Oncol 2016;17:1558–68.

2. Motzer RJ, Rini BI, McDermott DF, Frontera OA, Hammers HJ, Carducci MA, et al. Nivolumab plus ipilimumab versus sunitinib in firstline treatment for advanced renal cell carcinoma: extended follow-up of efficacy and safety results from a randomised, controlled, phase 3 trial. Lancet Oncol 2019;20:1370–85.

3. Hellmann MD, Paz-Arez L, Bernabe Caro R, Zurawski B, Kim SW, Costa EC, et al. Nivolumab plus ipilimumab in advanced non–small-cell lung cancer. N Engl J Med 2019;381:2020–31.

4. Martins F, Sofiya L, Sykiotis GP, Lamine F, Maillard M, Fraga M, et al. Adverse effects of immune-checkpoint inhibitors: epidemiology, management and surveillance. Nat Rev Clin Oncol 2019;16:563–80.

5. de Vries EGE, de Ruijter LK, Lub-de Hooge MN, Dierckx RA, Elias SG, Oosting SF. Integrating molecular nuclear imaging in clinical research to improve anticancer therapy. Nat Rev Clin Oncol 2019;16:241–55.

6. van de Donk PP, Ruijter LK, Lub-de Hooge MN, Brouwers AH, van der Wekken, Oosting SF, et al. Molecular imaging biomarkers for immune checkpoint inhibitor therapy. Theranostics 2020;10:1708–18.

7. Waaijer SJH, Kok IC, Eisses B, Schr€oder CP, Jalving M, Brouwers AH, et al. Molecular imaging in cancer drug development. J Nucl Med 2018;59:726–32.

8. Bensch F, van der Veen EL, Lub-de Hooge MN, Jorritsma-Smit A, Boellaard R, Kok IC, et al. 89Zratezolizumab imaging as a noninvasive approach to assess clinical response to PD-L1 blockade in cancer. Nat Med 2018:24;1852–8.

9. Gebhart G, Lamberts LE, Wimana Z, Garcia C, Emonts P, Ameye L, et al. Molecular imaging as a tool to investigate heterogeneity of advanced HER2-positive breast cancer and to predict patient outcome under trastuzumab emtansine (T-DM1): the ZEPHIR trial. Ann Oncol 2016;27:619–24

10. Giesen D, Broer LN, Lub-de Hooge MN, Popova I, Howng B, Nguyen M, et al. Probody therapeutic design of 89Zr-CX-072 promotes accumulation in PD-L1-expressing tumors compared to normal murine lymphoid tissue. Clin Cancer Res 2020;26:3999–4009.

11. Naing A, Thistlethwaite F, de Vries EGE, Eskens FALM, Uboha N, Ott PA, et al. CX-072 (Pacmilimab), a probody PD-L1 inhibitor in advanced or recurrent solid tumors (PROCLAIM-CX-072): an open-label dose-finding and first-inhuman study. J Immunother Cancer 2021; in press

12. Sanborn RE, Hamid O, de Vries EGE, Ott PA, Garcia-Corbacho J, Boni V, et al. CX-072 (Pacmilimab), a Probody PD-L1 inhibitor, in combination with ipilimumab in subjects with advanced solid tumors (PROCLAIMCX-072): a first-in-human, dose-finding study. J Immunother Cancer 2021; in press.

13. Eisenhauer EA, Therasse P, Bogaerts J, Schwartz LH, Sargent D, Ford R, et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur J Cancer 2009;45:228–47.

14. Bohnsack O, Hoos A, Ludajic K. Adaptation of the immune related response criteria: irRECIST. Ann Oncol 2014;25:iv361–iv372.

15. Lamberts TE, Menke-van der Houven van Oordt CW, ter Weele EJ, Bensch F, Smeenk MM, et al. ImmunoPET with anti-mesothelin antibody in patients with pancreatic and ovarian cancer before anti-mesothelin antibody-drug conjugate treatment. Clin Cancer Res 2016;22:1642–52.

16. Bensch F, Brouwers AH, Lub-deHoogeMN, de Jong JR, van derVegt B, Sleijfer S, et al. 89Zr-trastuzumab PET supports clinical decision making in breast cancer patients, when HER2 status cannot be determined by standard work up. Eur J Nucl Med Mol Imaging 2018;45;2300–6.

17. Makris NE, Boellaard R, Visser EP, de Jong JR, Vanderlinden B,Wierts R, et al. Multicenter harmonization of 89Zr PET/CT performance. J Nucl Med 2014;55;264–7.

18. Boellaard R. Quantitative oncology molecular analysis suite: ACCURATE. J Nucl Med 2018;59: (Suppl; Abstr 1753).

19. Stroh M, Green M, Miljard BL, Apgar JF, Burke JM, Garner W, et al. Modelinformed drug development of the masked anti-PD-L1 antibody CX-072. Clin Pharmacol Ther 2021;109:383–93.

20. Doroshow DB, Bhalla S, Beasley MB, Sholl LM, Kerr KM, Gnjatic S, et al. PD-L1 as a biomarker of response to immune-checkpoint inhibitors. Nat Rev Clin Oncol 2021;12. doi: 10.1038/s41571–021–00473–5. Epub ahead of print.

21. Maniakas A, Dadu R, Busaidy NL, Wang JR, Ferrarotto R, Lu C, et al. Evaluation of overall survival in patients with anaplastic thyroid carcinoma, 2000–2019. JAMA Oncol 2020;6:1397–404

22. Capdevila J, Wirth LJ, Ernst T, Aix SP, Lin CC, Ramlau R, et al. PD-1 blockade in anaplastic thyroid carcinoma. J Clin Oncol 2020;38:2620–7.

First-in-human PET imaging of 89Zr-CX-072 anti-PD-L1 Probody

23. Bronte V, Pittet M. The spleen in local and systemic regulation of immunity. Immunity 2013;39:806–18.

24. Cataldi M, Vigliotti C, Mosca T, Cammarota MR, Capone D. Emerging role of the spleen in the pharmacokinetics ofmonoclonal antibodies, nanoparticles and exosomes. Int J Mol Sci 2017;18:1249.

25. Verhoeff S, van de Donk PP, Aarntzen EHJG, Miedema IHC, Oosting SF, Voortman J, et al. 89Zr-durvalumab PD-L1 PET in recurrent or metastatic (R/M) squamous cell carcinoma of the head and neck. J Clin Oncol 2020:38 (Suppl;Abstr 3573).

Supplementary data

SUPPLEMENTARY FIGURE S1: Maximum intension projection (MIP) images of the 89Zr-CX-072 PET day 7 of all patients, scaled SUV 0-5. Color bars indicate the different tracer protein dose cohorts.

First-in-human PET imaging of 89Zr-CX-072 anti-PD-L1 Probody

SUPPLEMENTARY TABLE S1: Visual uptake in healthy lymphoid tissue per patient.

Subject number (protein dose)

1 (1 mg)

Tonsils

Axillary lymph nodes Inguinal lymph nodes

2 (1 mg) + + +

3 (5 mg)

4 (5 mg) + +

5 (5 mg) + + +

6 (10 mg) + +

7 (10 mg) + + +

8 (10 mg) NE

Legend: (+) Visual uptake compared to surrounding background tissue present. ( ) No visual uptake compared to surrounding background tissue present. NE: not evaluable, due to malignant lymph nodes in this location.

SUPPLEMENTARY TABLE S2: Overview of treatment and response per patient.

Subject Indication Treatment CX-072

1

2

Best overall response

Pancreatic carcinoma 10 mg/kg + ipilimumab 3 mg/kg PD

MSI high colorectal 10 mg/kg monotherapy PR carcinoma

3 Ovarian carcinoma 10 mg/kg monotherapy PD

4 Germ cell tumor 10 mg/kg monotherapy PD

5

MSI high colorectal 10 mg/kg monotherapy PD carcinoma

6

MSI high colorectal 10 mg/kg monotherapy PD carcinoma

7 Cervical carcinoma 10 mg/kg monotherapy SD

8 Anaplastic thyroid 10 mg/kg monotherapy PR carcinoma Abbreviations: MSI, microsatellite instability; PD, progressive disease; PR, partial response; SD, stable disease.

Preclinical PET imaging of bispecific antibody ERY974 targeting CD3 and glypican 3 reveals that tumor uptake correlates to T cell infiltrate

Stijn J H Waaijer 1, Danique Giesen 1, Takahiro Ishiguro 2, Yuji Sano 2, Naofumi Sugaya 2, Carolina P Schröder 1, Elisabeth G E de Vries 1 , Marjolijn N Lub-de Hooge 3, 4

1 Department of Medical Oncology, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands; 2 Research Division, Chugai Pharmaceuticals Co Ltd, Chuo-ku, Tokyo, Japan; 3 Department of Clinical Pharmacy and Pharmacology, and 4 Department of Nuclear Medicine and Molecular Imaging, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands. m.n.de.hooge@umcg.nl.

J Immunother Cancer. 2020;8(1):e000548.

Abstract

Background: Bispecific antibodies redirecting T cells to the tumor obtain increasing interest as potential cancer immunotherapy. ERY974, a full-length bispecific antibody targeting CD3ε on T cells and glypican 3 (GPC3) on tumors, has been in clinical development However, information on the influence of T cells on biodistribution of bispecific antibodies, like ERY974, is scarce. Here, we report the biodistribution and tumor targeting of zirconium-89 (89Zr) labeled ERY974 in mouse models using immuno-positron emission tomography (PET) imaging.

Methods: To study both the role of GPC3 and CD3 on the biodistribution of [89Zr]Zr-N-suc-DfERY974, 89Zr-labeled control antibodies targeting CD3 and non-mammalian protein keyhole limpet hemocyanin (KLH) or KLH only were used. GPC3 dependent tumor targeting of [89Zr] Zr-N-suc-Df-ERY974 was tested in xenograft models with different levels of GPC3 expression. In addition, CD3 influence on biodistribution of [89Zr]Zr-N-suc-Df-ERY974 was evaluated by comparing biodistribution between tumor-bearing immunodeficient mice and mice reconstituted with human immune cells using microPET imaging and ex vivo biodistribution. Ex vivo autoradiography was used to study deep tissue distribution.

Results: In tumor-bearing immunodeficient mice, [89Zr]Zr-N-suc-Df-ERY974 tumor uptake was GPC3 dependent and specific over [89Zr]Zr-N-suc-Df-KLH/CD3 and [89Zr]Zr-N-suc-Df-KLH/ KLH. In mice engrafted with human immune cells, [89Zr]Zr-N-suc-Df-ERY974 specific tumor uptake was higher than in immunodeficient mice. Ex vivo autoradiography demonstrated a preferential distribution of [89Zr]Zr-N-suc-Df-ERY974 to T cell rich tumor tissue. Next to tumor, highest specific [89Zr]Zr-N-suc-Df-ERY974 uptake was observed in spleen and lymph nodes.

Conclusion: [89Zr]Zr-N-suc-Df-ERY974 can potentially be used to study ERY974 biodistribution in patients to support drug development.

Background

Cancer treatment regimens increasingly contain monoclonal antibodies (mAb) with established mechanisms of action. These mAbs include drugs designed to block immune checkpoints, which are approved for multiple indications including melanoma, non-smallcell lung cancer and renal cell carcinoma (1). Unfortunately, not all patients with these tumor types treated benefit. Moreover, immunotherapy is still not effective in many tumor types. Therefore, novel approaches are exploited. This includes challenging approaches such as T cell redirecting bispecific antibodies, which target both T cells and tumor cells (2). By engaging T cells via CD3 and an antigen on tumor cells, T cells get activated to kill cancer cells (3, 4). The activity of the anti-CD19/CD3 bispecific T cell engager blinatumomab in adult and pediatric

patients with relapsed and refractory precursor B-cell acute lymphoblastic leukemia and with minimal residual disease resulted in the first approval of such an approach by the US Food and Drug administration and European Medicines Agency (5). For solid tumors, currently, several T cell redirecting bispecific antibodies are studied in clinical trials (4).

ERY974 is a T cell redirecting bispecific antibody that targets human CD3 on T cells (Kd 207 nM) and glypican 3 (GPC3; Kd 1.5 nM) on tumor cells (6). GPC3 is an oncofetal cell surface protein that is overexpressed in several tumor types while its expression is suppressed in healthy tissue (7). ERY974 is a fully humanized IgG4 antibody. It has preserved neonatal Fc receptor binding properties to allow extended circulating half-life by reducing lysosomal degradation, but lacks binding to Fcγ receptors (FcγR) to prevent GPC3-indepenent cytokine release by engaging FcγR and CD3 (6). Preclinically, ERY974 inhibits growth of several solid tumor types in a mouse model with reconstituted human immune cells (6). ERY974 is in clinical development (8).

Molecular imaging could accelerate drug development by gaining insight in biodistribution and target engagement (9). Recent interesting preclinical data showed that distribution of a radiolabeled bispecific antibody targeting CD3ε and human epidermal growth factor receptor 2 (HER2) is largely dependent on the affinity of the CD3ε arm of the antibody (10). It is difficult to predict the drug distribution of an engineered drug such as ERY974 that has two targets with different affinities. Studying ERY974’s biodistribution might be informative for optimal treatment of patients. Prior to a clinical study, a preclinical study would allow studying additional experimental conditions by using multiple tumors and mouse models combined with different control antibodies. Therefore, to improve the understanding of ERY974’s behavior we aimed to characterize the impact of T cells on the biodistribution of ERY974 in a mouse model. We coupled ERY974 to chelator N-succinyl desferal (N-suc-Df), followed by radiolabeling with the positron emission tomography (PET) isotope zirconium-89 (89Zr) to enable non-invasive molecular imaging of ERY974 to study its behavior in a tumor-bearing mouse model. To study both the role of GPC3 and CD3 on the biodistribution of [89Zr]Zr-N-sucDf-ERY974, 89Zr-labeled control antibodies targeting CD3 and non-mammalian protein keyhole limpet hemocyanin (KLH) or KLH only were used. CD3 influence was evaluated by comparing biodistribution between tumor-bearing immunodeficient mice and mice reconstituted with human immune cells. In addition, GPC3-dependent tumor targeting was tested in xenograft models with different levels of GPC3 expression. Ex vivo autoradiography was used to study deep tissue distribution.

Methods

Bispecific antibody constructs and cell lines

Bispecific antibody constructs ERY974 (IgG4), KLH/CD3 (IgG1), KLH/KLH (IgG1) and bivalent GPC3

Biodistribution of CD3/glypican 3-targeting bispecific antibody ERY974

antibody were provided by Chugai Pharmaceutical. The dissociation constants for the binding of ERY974 to human GPC3 and human CD3ε were 1.5±0.4 nM and 207±7, respectively. ERY974 was formulated in 150 mM arginine, 20 mM histidine, 171 mM L-aspartic acid and 0.52 mg/mL poloxamer 188, pH 6.0. KLH/CD3 binds KLH and human CD3ε with similar affinity than ERY974, whereas KLH/KLH binds KLH only. The molecular weight of the antibodies is approximately 146 kDa and all were engineered to abolish FcγR binding (6).

The human hepatocellular carcinoma cell line HepG2 (GPC3+), human ovarian clear cell carcinoma cell line TOV-21G (GPC3+) and human hepatocellular carcinoma cell line SKHEP-1 (GPC3 ) were used. All cell lines were obtained from American Type Culture Collection and confirmed to be negative for microbial contamination. Cell lines were authenticated by BaseClear using short tandem repeat profiling. This was repeated once a cell line has been passaged for more than 6 months after previous short tandem profiling. HepG2 cells were cultured in Dulbecco’s Modified Eagle’s Medium (DMEM; Invitrogen), TOV-21G in a 1:1 mixture of Ham’s F12/DMEM, SK-HEP-1 in DMEM with high glucose (Invitrogen) supplemented with 10% fetal bovine serum (Bodinco BV). All cells were cultured under humidified conditions at 37 °C with 5% CO2.

Animal experiments

We used female NOD.Cg-Prkdcscid Il2rgtm1Sug/JicTac (NOG; median body weight 22 g with IQR of 21–23 g) mice (Taconic) or female humanized NOG (huNOG; 22 g with IQR 20–23) mice which were engrafted with human CD34+ hematopoietic stem cells (Taconic, ref. 11). Human leukocyte reconstitution was checked by flow cytometry at 16 weeks postengraftment and CD3+ engraftment was similar between experimental groups (supplementary figure S1). Mice were housed per five mice in specific pathogen-free cages, with cage enrichment, on a 12hour day/night cycle, and ad libitum access to food and water. Mice were allowed to acclimate for at least 1 week on arrival. At approximately 25 weeks of age 10×106 HepG2, TOV-21G or SKHEP-1 cells in 1:1 ratio of medium and Matrigel (BD Biosciences; 0.3 mL) were subcutaneously injected for xenograft development. Tumor growth was assessed by caliper measurements and body weight was monitored twice weekly. Retro-orbital tracer injection (for description of tracer manufacturing see supplementary additional methods) was performed when tumors reached a size of 200 mm3. This was reached for TOV-21G in 14 days and for HepG2 and SKHEP-1 in 24 days. Anesthesia during microPET scanning was performed with isoflurane/oxygen inhalation (5% induction, 2.5% maintenance). Details regarding number of animals, microPET scans and time of biodistribution are included in the figure legends.

MicroPET scanning and ex vivo biodistribution

All microPET scans were performed in a Focus 200 rodent scanner (CTI Siemens). Mice were kept

warm on heating mats. A transmission scan of 515 s was obtained using a 57Co point source for tissue attenuation. The reconstruction of microPET scans was performed as previously described (12). After reconstruction, images were interpolated with trilinear interpolation using PMOD software (v. 3.7, PMOD Technologies). Coronal microPET images or maximal intensity projection images were used for display. Volumes of interest (VOI) of the whole tumor were drawn based on biodistribution tumor weight. For the heart, a 92 mm3 VOI in the coronal plane was drawn. VOIs were subsequently quantified. Data are expressed as the mean standardized uptake value (SUVmean).

For all ex vivo biodistribution studies, tumor, whole blood and organs of interest were collected and weighed. Samples together with tracer standards were counted in a calibrated well-type g-counter (LKB Instruments). Uptake is expressed as the percentage injected dose per gram of tissue (%ID/g).

To determine binding to peripheral blood mononuclear cells, whole blood of huNOG mice was separated using SepMate-15 tubes (STEMCELL Technologies) with Ficoll-Paque PLUS (GE Healthcare). Buffy coat fraction was washed twice using phosphate buffered saline (PBS; 140 mM/L NaCl, 9 mM/L Na2HPO4, 1.3 mM/L NaH2PO4, pH 7.4, UMCG) with 2% fetal calf serum. Radioactivity was counted of whole blood and PBMCs.

Ex vivo autoradiography and immunohistochemistry

Tumors, spleen or mesenteric lymph nodes were fixed in formalin overnight, followed by paraffin embedding. Four μm sections were subsequently exposed overnight to a phosphor screen (PerkinElmer) in an X-ray cassette. Signal was detected with a Cyclone Storage Phosphor System (PerkinElmer). Slides used for ex vivo autoradiography were deparaffinized then stained with H&E and digitalized with NanoZoomer and NDP software (Hamamatsu). Subsequent slides were stained for GPC3 (tumor only) and CD3ε (supplementary additional methods).

For ex vivo tissue, autoradiography quantification of tumor sections, regions of interest (ROIs) were drawn for tumor cells and stromal regions based on H&E. ROIs were exported to ImageJ (National Institutes of Health, USA), rescaled for ex vivo autoradiography and ROIs were measured.

For tumor lysate and plasma analysis, samples were heated for 10 min at 70 °C and 40 μg protein of tumor lysates or mouse plasma from three mice, tracer alone as positive control were loaded on mini-PROTEAN TGX Precast Gels (Bio-Rad). Gels were exposed overnight to phosphor imaging screens (PerkinElmer) in X-ray cassettes and analyzed using a Cyclone

Biodistribution of CD3/glypican 3-targeting bispecific antibody ERY974

Storage Phosphor System (PerkinElmer).

Statistical analysis

Statistical analyses were performed using GraphPad Prim v. 7.02. Unless otherwise stated, data are presented as median ± IQR. Mann-Whitney U test was performed to test differences between two groups. Bonferroni correction was applied when more than two groups were compared. P values ≤ 0.05 were considered significant.

Results

In vitro characterization of [89Zr]Zr-N-suc-Df-ERY974 and control tracers

We successfully radiolabeled ERY974 and control antibodies with 89Zr with a molar activity (Am) of 72.8 MBq/nmol at the end of synthesis. Radiochemical purity exceeded 95% after labeling and high molecular weight species were below 5%. In vitro, the intermediate N-sucDf-ERY974 binds GPC3 and CD3ε similarly to unconjugated ERY974 in an ELISA-based binding assay, indicating preserved binding to both targets (supplementary figure S2A and B). The T cell activation potency of N-suc-Df-ERY974 was not affected, being similar to unmodified ERY974 (supplementary figure S2C). This in vitro data demonstrate that N-suc-Df-ERY974 can be used as a surrogate for ERY974 to study its biodistribution. In GPC3-positive HepG2 tumor cells, 12.9%±3.2% [89Zr]Zr-N-suc-Df-ERY974 internalized after 4 hours (supplementary figure S2D). Control tracers KLH/CD3 and KLH/KLH did not bind to GPC3 and only [89Zr]Zr-N-suc-DfKLH/CD3 bound CD3ε (supplementary figure S2E and F).

Tumor xenograft accumulation of [89Zr]Zr-N-suc-Df-ERY974 in time in immunodeficient NOG mice

The optimal time point for microPET imaging with [89Zr]Zr-N-suc-Df-ERY974 providing the highest tumor-to-blood ratio in HepG2 xenograft bearing immunodeficient NOG mice was 168 hours after injection. MicroPET images revealed clear tumor uptake of [89Zr]Zr-N-suc-DfERY974 already at 24 hours after intravenous injection (figure 1A). After tracer injection over time, the blood levels decreased and tumor levels increased up to 120 hours, resulting in a maximal tumor-to-blood ratio of 2.2±0.3 at 168 hours (figure 1B, C).

Biodistribution and tumor uptake of [89Zr]Zr-N-suc-Df-ERY974 in immunodeficient NOG mice bearing tumor xenografts with different GPC3 expression

In both GPC3 positive tumor models, HepG2 (GPC3 high) and TOV-21G (GPC3 low; supplementary figure S3A), [89Zr]Zr-N-suc-Df-ERY974 tumor uptake was visualized with microPET, whereas GPC3-negative xenograft SK-HEP-1 revealed lower uptake (figure 2A, B; supplementary figure S3A). Ex vivo biodistribution at 168 hours after tracer administration confirmed GPC3-dependent uptake of [89Zr]Zr-N-suc-Df-ERY974, with higher uptake in TOV-

FIGURE 1: [89Zr]Zr-N-suc-Df-ERY974 distribution in time. (A) Maximal intensity projection images of [89Zr]Zr-N-suc-Df-ERY974 in HepG2 tumor (white dotted circle) bearing NOG mice at 24 hours, 72 hours, 120 hours and 168 hours post 10 μg [89Zr]Zr-N-suc-Df-ERY974 injection. (B) Quantification of tumor and blood pool (n=6). data shown as median SUVmean and IQR. (C) Tumor-to-blood ratio based on SUVmean. Data shown as median tumor-to-blood ratio and IQR. Cr, cranial; Ca, caudal; H, heart; L, liver; SUVmean, standardised uptake value; T, tumor.

21G xenografts and lower uptake in SK-HEP-1 compared with HepG2 xenografts (figure 2C). [89Zr]Zr-N-suc-Df-ERY974 tumor-to-blood ratio of was highest in HepG2 tumors (figure 2D). Ex vivo autoradiography of tumor tissue at 168 hours after tracer administration, showed [89Zr]ZrN-suc-Df-ERY974 presence mainly confined to GPC3 expressing tumor tissue areas of HepG2 and TOV-21G xenografts (supplementary figure S3A). In contrast, GPC3 negative SK-HEP-1 xenografts showed predominantly 89Zr localization in non-tumor tissue (supplementary figure S3A). In all xenograft models, 89Zr-uptake reflected intact but also fragments of [89Zr]Zr-N-sucDf-ERY974 in tumor lysates and blood plasma (supplementary figure S3B).

[89Zr]Zr-N-suc-Df-ERY974 biodistribution demonstrated highest physiological uptake in spleen followed by liver, lung and kidney at 168 hours after tracer administration (figure 2C). [89Zr]ZrN-suc-Df-ERY974 blood levels in TOV-21G xenograft bearing mice were higher and liver uptake was lower, compared with HepG2 xenograft bearing mice (figure 2C).

Biodistribution of CD3/glypican 3-targeting bispecific antibody ERY974

FIGURE 2: [89Zr]Zr-N-suc-Df-ERY974 distribution in immunodeficient NOG mice bearing different tumor xenografts.

(A) Coronal microPET images (upper panel) and maximal intensity projection (MIP) images (lower panel) of NOG mice bearing HepG2, TOV-21G or SK-HEP-1 xenografts (white circle) 168 hours post 10 μg [89Zr]ZrN-suc-Df-ERY974 injection. (B) Quantification of HepG2 (n=6), TOV-21G (n=6) or SK-HEP-1 (n=6) uptake of [89Zr]Zr-N-suc-Df-ERY974 (upper graph) and corresponding blood pool uptake (lower graph) at 72 hours and 168 hours post-tracer injection. Data shown as median SUVmean and IQR. (C) Ex vivo biodistribution of [89Zr]Zr-N-suc-Df-ERY974 168 hours post-tracer administration. Data are expressed as median with IQR.

*P≤0.05; **P≤0.01; ***P≤0.001 (Mann-Whitney U). (D) Tumor-to-blood ratio based on ex vivo biodistribution of C. Data are expressed as median with IQR. **P≤0.01; ***P≤0.001 (Mann-Whitney U). Cr, cranial; Ca, caudal; H, heart; L, liver; PET, positron emission tomography; SUVmean, standardised uptake value; T, tumor; 89Zr, zirconium-89.

Spleen and bone marrow uptake of [89Zr]Zr-N-suc-Df-ERY974 in immunodeficient NOG mice [89Zr]Zr-N-suc-Df-ERY974 uptake in the spleen was ±18 %ID/g in all tumor-bearing NOG mice and thereby the highest uptake observed compared with any organ (figure 2C). As shown earlier

Biodistribution of CD3/glypican 3-targeting bispecific antibody ERY974

for severely immunodeficient NOD.Cg-Prkdcscid Il2rgtm1Wjl (NSG) mice (13), spleen uptake in our experiment was also influenced by FcγR-modification of the mAb, with lower spleen uptake for FcγR-silenced mAbs (supplementary figure S4A and B). In addition, with high radioactive dose spleens of our mice shrunk in time and aplasia was observed 168 hours after injection with median weight of 10 mg IQR 8–12 at higher radioactive dose versus 24 mg IQR 20–33 at a lower radioactive dose (supplementary figure S4C-G13). Spleen uptake is also affected by mouse strain, as our less immunodeficient nude (BALB/cOlaHsd-Foxn1nu) mice demonstrated lower relative spleen uptake of [89Zr]Zr-N-suc-Df-ERY974 than NOG mice (supplementary figure S4H), similarly described for an 89Zr-labeled antibody targeting delta-like protein 3 (14). Interestingly, absolute spleen uptake of [89Zr]Zr-N-suc-Df-ERY974 was higher in nude mice (supplementary figure S4I and J). Relative bone marrow uptake of [89Zr]Zr-N-suc-Df-ERY974 was also lower in nude mice than NOG mice, while cortical bone uptake was similar (supplementary figure S4K). In summary, a relative high spleen uptake of 89Zr-mAb in NOG mice is related to host and mAb characteristics and mediated by radiation dose, as demonstrated earlier in NSG mice (13).

Influence of ERY974 protein dose on [89Zr]Zr-N-suc-Df-ERY974 tumor xenograft uptake and biodistribution in immunodeficient NOG mice

To test the effect of protein dose on [89Zr]Zr-N-suc-Df-ERY974 tumor targeting and biodistribution, 10 μg [89Zr]Zr-N-suc-Df-ERY974 was supplemented with a dose of unlabeled ERY974. A 200-fold excess of unlabeled protein (2 mg) was unable to reduce HepG2 tumor uptake of [89Zr]Zr-N-suc-Df-ERY974 and even increased uptake in HepG2 xenografts at 168 hours after injection (supplementary figure S5A). However, this dose did lower [89Zr]Zr-N-sucDf-ERY974 liver uptake while [89Zr]Zr-N-suc-Df-ERY974 blood levels remained unchanged (supplementary figure S5B and D). HepG2 xenograft uptake of [89Zr]Zr-N-suc-Df-ERY974 could also not be blocked with 100-fold excess of unlabeled bivalent GPC3 mAb, but resulted in increased blood levels leading to a lower tumor-to-blood ratio (supplementary figure S5B and C). In TOV21G xenografts, which express less GPC3, data suggests that 2000 μg total protein dose ERY974 reduced, although non-significant, tumor uptake of [89Zr]Zr-N-suc-Df-ERY974 with 62.1% (supplementary figure S5A; P=0.07).

Specificity of HepG2 xenograft uptake of [89Zr]Zr-N-suc-Df-ERY974 in immunodeficient NOG mice

To demonstrate that tumor uptake is GPC3 dependent, control tracers [89Zr]Zr-N-suc-Df-KLH/ CD3 and [89Zr]Zr-N-suc-Df-KLH/KLH were administered to HepG2 bearing immunodeficient NOG mice. At 168 hours after injection, tumor uptake on microPET images of [89Zr]Zr-N-sucDf-ERY974 was better visible than of [89Zr]Zr-N-suc-Df-KLH/CD3 or [89Zr]Zr-N-suc-Df-KLH/KLH (figure 3A). Scan quantification of tumor uptake showed less tumor uptake for [89Zr]Zr-Nsuc-Df-KLH/CD3 and [89Zr]Zr-N-suc-Df-KLH/KLH, with a median SUVmean of 0.75 with an IQR of

0.59–0.91 and 1.08 (IQR 0.71–1.82) vs 2.91 (IQR 2.74–3.03) for [89Zr]Zr-N-suc-Df-KLH/CD3, [89Zr]ZrN-suc-Df-KLH/KLH and [89Zr]Zr-N-suc-Df-ERY974, respectively (figure 3B). Furthermore, ex vivo biodistribution results at 168 hours after injection supported the findings with the microPET images. This also revealed additional differences in tracer distribution such as lower blood levels for [89Zr]Zr-N-suc-Df-KLH/CD3 and higher for [89Zr]Zr-N-suc-Df-KLH/KLH compared with

FIGURE 3: Distribution of [89Zr]Zr-N-suc-Df-ERY974 and control tracers in HepG2 xenograft bearing immunodeficient NOG mice.

(A) coronal microPET images of HepG2 (white dotted circle) bearing NOG mice injected with 10 μg [89Zr]Zr-N-suc-Df-ERY974, [89Zr]Zr-N-suc-Df-KLH/CD3 or [89Zr]Zr-N-suc-Df-KLH/KLH 168 hours post tracer injection. (B) Quantification of [89Zr]Zr-N-suc-Df-ERY974 (n=6), [89Zr]Zr-N-suc-Df-KLH/CD3 (n=5) or [89Zr] Zr-N-suc-Df-KLH/KLH (n=6) uptake in HepG2 tumor and blood pool. Data shown as median SUVmean and IQR. *P≤0.05; **P≤0.01 (Mann-Whitney U). (C) Ex vivo biodistribution of [89Zr]Zr-N-suc-Df-ERY974, [89Zr]Zr-Nsuc-Df-KLH/CD3 and [89Zr]Zr-N-suc-Df-KLH/KLH 168 hours post-tracer administration. Data are expressed as median with IQR. *P≤0.05; **P≤0.01; ***P≤0.001 (Mann-Whitney U). Cr, cranial; Ca, caudal; KLH, keyhole limpet hemocyanin; L, liver; PET, positron emission tomography; SUVmean, standardised uptake value; T, tumor; 89Zr, zirconium-89.

Biodistribution of CD3/glypican 3-targeting bispecific antibody ERY974

[89Zr]Zr-N-suc-Df-N-suc-Df-ERY974 (figure 3C). Similarly, liver uptake was higher for [89Zr]ZrN-suc-Df-KLH/KLH and lower for [89Zr]Zr-N-suc-Df-KLH/CD3 compared with [89Zr]Zr-N-suc-DfERY974 (figure 3C). [89Zr]Zr-N-suc-Df-ERY974, [89Zr]Zr-N-suc-Df-KLH/CD3 and [89Zr]Zr-N-sucDf-KLH/KLH all showed relatively high distribution to spleens of immunodeficient NOG mice (figure 3C).

Biodistribution and tumor xenograft uptake of [89Zr]Zr-N-suc-Df-ERY974 in NOG mice engrafted with human immune cells

To study the biodistribution and tumor-targeting properties of [89Zr]Zr-N-suc-Df-ERY974 with additional availability of human CD3 on T cells, HepG2 tumor-bearing huNOG mice reconstituted with human immune cells were studied. Compared with HepG2 xenograft bearing NOG mice, [89Zr]Zr-N-suc-Df-ERY974 showed increased tumor uptake in HepG2 bearing huNOG mice with median SUV mean of 7.3 (IQR 4.3–9.3) at 168 hours after injection (figure 4A, B; supplementary figure S6). Compared with [89Zr]Zr-N-suc-Df-ERY974, control tracers [89Zr]Zr-N-suc-Df-KLH/CD3 and KLH/KLH administered to huNOG mice were taken up lower by the tumor with a median SUV mean of 0.6 (IQR 0.3–0.7) and 1.6 (IQR 1.5–1.9), respectively (figure 4B).

Apart from tumor uptake, highest organ uptake for [89Zr]Zr-N-suc-Df-ER974 was observed for spleen, followed by mesenteric lymph nodes (MLN) and liver (figure 4C). Both CD3 targeting molecules [89Zr]Zr-N-suc-Df-ERY974 and [89Zr]Zr-N-suc-Df-KLH/CD3 revealed twofold to threefold higher uptake in lymphoid organs such as spleen and MLN than [89Zr]Zr-N-suc-DfKLH/KLH (figure 4C). Median spleen weight of huNOG mice, as determined by biodistribution at 7 days after tracer administration, was 26 mg (IQR 22–32) and not different between groups. Interestingly, [89Zr]Zr-N-suc-Df-KLH/CD3 showed higher blood levels in huNOG mice than in NOG mice at 168 hours after tracer administration (supplementary figure S6B). Furthermore, higher binding of [89Zr]Zr-N-suc-Df-ERY974 and [89Zr]Zr-N-suc-Df-KLH/CD3 than [89Zr]ZrN-suc-Df-KLH/KLH to peripheral blood mononuclear cells of huNOG mice was observed (supplementary figure S7). Collectively, [89Zr]Zr-N-suc-Df-ERY974 demonstrated increased tumor uptake between NOG and huNOG mice and CD3 specific uptake in lymphoid tissues of huNOG mice.

Ex vivo autoradiography of tumor and lymphoid tissues of huNOG mice

We performed ex vivo autoradiography and evaluated T cell infiltration to determine microscopic colocalization of tracers in tumors and lymphoid tissues of mice engrafted with human immune cells. In HepG2 tumors, [89Zr]Zr-N-suc-Df-ERY974 gave higher radioactive signal in stromal regions with high CD3+ T cells infiltrate than tumor nests (figure 5A–C). Total radioactivity signal measured by ex vivo autoradiography correlated well with %ID/g of the individual tumor (R2=0.7178; figure 5C). For lymphoid organs spleen and mesenteric

FIGURE 4: Distribution of [89Zr]Zr-N-suc-Df-ERY974 and control tracers in HepG2 xenograft bearing humanized NOG mice.

(A) coronal microPET images of HepG2 (white dotted circle) bearing humanized NOG mice reconstituted with human immune cells injected with 10 μg [89Zr]Zr-N-suc-Df-ERY974, [89Zr]Zr-N-suc-Df-KLH/CD3 or [89Zr]Zr-N-suc-Df-KLH/KLH 72 hours and 168 hours post-tracer injection. (B) Quantification of [89Zr]Zr-Nsuc-Df-ERY974 (n=5), [89Zr]Zr-N-suc-Df-KLH/CD3 (n=4) or [89Zr]Zr-N-suc-Df-KLH/KLH (n=6) uptake in HepG2 tumor and blood pool. Data shown as median SUVmean and IQR. (C) Ex vivo biodistribution of [89Zr]Zr-N-sucDf-ERY974, [89Zr]Zr-N-suc-Df-KLH/CD3 and [89Zr]Zr-N-suc-Df-KLH/KLH 168 hours post-tracer administration. Data are expressed as median with IQR. *P≤0.05 (Mann-Whitney U). Cr, cranial; Ca, caudal; KLH, keyhole limpet hemocyanin; L, liver; MLN, mesenteric lymph node; PET, positron emission tomography; SUVmean, standardised uptake value; T, tumor; 89Zr, zirconium-89.

lymph nodes, both CD3-targeting tracers located to regions with human CD3+ cells (figure 5E, F). [89Zr]Zr-N-suc-Df-KLH/KLH distributed independent of CD3+ cells (figure 5G).

The T cell infiltration of HepG2 xenografts observed in huNOG mice injected with [89Zr]ZrN-suc-Df-ERY974 was not observed following injection of [89Zr]Zr-N-suc-Df-KLH/CD3 and [89Zr]Zr-N-suc-Df-KLH/KLH, in both tumor and stromal regions (supplementary figure S8A).

Quantification of CD3+ T cells confirmed less T cell infiltration with 41 (IQR 21–303) CD3+ cells/mm2 for [89Zr]Zr-N-suc-Df-KLH/CD3, 50 (IQR 22–85) CD3+ cells/mm2 for [89Zr]ZrN-suc-Df-KLH/KLH and 707 (IQR 670–1133) CD3+ cells/mm2 for [89Zr]Zr-N-suc-Df-ERY974

Biodistribution of CD3/glypican 3-targeting bispecific antibody ERY974

FIGURE 5: Intratumoral distribution of [89Zr]Zr-N-suc-Df-ERY974 in HepG2 bearing huNOG mice. (A) autoradiography (first panel) of [89Zr]Zr-N-suc-Df-ERY974 in HepG2 bearing huNOG mice with subsequent slides stained for human CD3 (second panel), (H&E; third panel) and human GPC3 (fourth panel). →

→ Scale bar length represents 5 mm for whole tissue and 500 μm for magnified tissue. (B) ROI drawing of tumor cells and stromal regions based on H&E staining (upper panel) and the overlay on autoradiography (lower panel). Scale bar length represents 2.5 mm. (C) Quantification of tumor and stromal regions on autoradiography. Circle and square with an inner dot represent the quantification of the tumor and stroma shown in (B). Checked circle and square represents the quantification of tumor and stroma depicted in (A). (D) Mean total tumor value on autoradiography correlated with the %ID/g of the same tumor. (E) Autoradiography (top panel), CD3 immunohistochemistry (middle panel) and H&E staining (bottom panel) of spleen and mesenteric lymph node (MLN) of huNOG mice injected with [89Zr]Zr-N-suc-Df-ERY974. Autoradiography and H&E were performed on the same slide. scale bar length represents 2.5 mm. (F) Autoradiography (top panel), CD3 immunohistochemistry (middle panel) H&E staining (bottom panel) of spleen and MLN of huNOG mice injected with [89Zr]Zr-N-suc-Df-KLH/CD3. Scale bar length represents 2.5 mm. (G) Autoradiography (top panel), CD3 immunohistochemistry (middle panel) and H&E staining (bottom panel) of spleen and MLN of huNOG mice injected with [89Zr]Zr-N-suc-Df-KLH/KLH. Scale bar length represents 2.5 mm. GPC3, glypican 3; huNOG, humanised NOG; KLH, keyhole limpet hemocyanin; 89Zr, zirconium-89.

(supplementary figure S8B). This difference might be explained by the pharmacological effect of 10 μg of [89Zr]Zr-N-suc-Df-ERY974 that leads to immune cell infiltration as demonstrated earlier (6), which effect is absent for [89Zr]Zr-N-suc-Df-KLH/CD3 and [89Zr]Zr-N-suc-Df-KLH/KLH. Tumor weight at 7 days after [89Zr]Zr-N-suc-Df-ERY974 administration was lower compared with [89Zr]Zr-N-suc-Df-KLH/CD3 or [89Zr]Zr-N-suc-Df-KLH/KLH, with median weight of 238 mg (IQR 194–367) vs 676 mg (IQR 479–807), respectively (P<0.05; data not shown). We therefore studied the distribution of [89Zr]Zr-N-suc-Df-KLH/CD3 and [89Zr]Zr-N-suc-Df-KLH/KLH in a HepG2 bearing huNOG mice co-injected with 10 μg of N-suc-Df-ERY974. N-suc-Df-ERY974 coinjection resulted in strong CD3+ T cell infiltration in HepG2 tumors of huNOG administered with [89Zr]Zr-N-suc-Df-KLH/CD3 or [89Zr]Zr-N-suc-Df-KLH/KLH, in similar range as [89Zr]Zr-N-sucDf-ERY974 (supplementary figure S9). ERY974 coinjection did not result in major differences in physiological organ uptake of [89Zr]Zr-N-suc-Df-KLH/CD3 or [89Zr]Zr-N-suc-Df-KLH/KLH, but did increase tumor uptake from 3.4 (IQR 2.0–3.6) to 10.0 (IQR 9.9–16.3) and from 6.1 (IQR 4.0–7.9) to 16.6 %ID/g (IQR 10.4–23.8) for [89Zr]Zr-N-suc-Df-KLH/CD3 and [89Zr]Zr-N-suc-Df-KLH/KLH, respectively (supplementary figure S6).

Discussion

This is the first study using molecular imaging to study the influence of T cells on the distribution and tumor uptake of the therapeutic T cell redirecting bispecific antibody ERY974 targeting CD3ε and GPC3. [89Zr]Zr-N-suc-Df-ERY974 tumor uptake was higher in human immune cell engrafted mice than in immunodeficient mice, and localized in the T cell infiltrated stromal regions. Next to tumor uptake, second highest [89Zr]Zr-N-suc-Df-ERY974 uptake was found in

CD3+ lymphoid tissues such as spleen and mesenteric lymph nodes, followed by liver.

The role of T cells on the biodistribution of T cell redirecting antibodies is poorly understood since its biodistribution is mainly studied in immunocompromised mice, which prohibits to study the influence of the CD3 arm of the bispecific antibody (12, 15). The current study shows that in the presence of T cells, specific physiological uptake is observed in CD3+ T cell containing organs such as spleen and lymph nodes. In contrast, for antibodies targeting tumor cell and growth factors highest physiological uptake is usually observed in liver (16). However, a recent clinical molecular imaging study with the immune checkpoint inhibitor atezolizumab demonstrated that lymphoid tissues can also be an important compartment for drug biodistribution (17). For example, [89Zr]Zr-N-suc-Df-atezolizumab, targeting the immune checkpoint programmed death-ligand 1, showed highest physiological uptake in spleen of patients with cancer. Furthermore, in the majority of the patients molecular PET imaging with [89Zr]Zr-N-suc-Df-atezolizumab was able to detect healthy lymph nodes.

The role of the CD3ε directed arm of a bispecific T cell redirecting antibody will be impacted by its affinity. An interesting study evaluated the impact of a CD3ε binding arm on the biodistribution of T cell redirecting antibodies in a human CD3ε transgenic mouse model using a HER2/CD3 bispecific antibody with different affinities for CD3ε (10). A low affinity for CD3ε (CD3εL), which was considered 50 nM, did not redirect a non-tumor targeting bispecific antibody to lymphoid organs. In contrast, subnanomolar (0.5 nM) and picomolar (0.05 nM) affinity to CD3ε showed extensive distribution to lymphoid tissues such as spleen and lymph nodes, which could be blocked by a 100-fold excess of unlabeled CD3 single arm antibody. For the tumor targeting of HER2/CD3ε antibody, the antibodies with high affinity for CD3ε lowered tumor targeting compared with HER2/CD3εL. Tumor uptake of HER2/CD3εL could only be reduced by an excess of unlabeled HER2 single arm antibody and not by CD3 single arm antibody, suggesting limited effect of CD3ε on tumor uptake. [89Zr]Zr-N-suc-Df-ERY974 and [89Zr]Zr-N-suc-Df-KLH/ CD3, with a submicromolar (207 nM) affinity for CD3ε, have a lower affinity compared with the CD3εL. Despite the lower affinity for CD3ε, specific uptake in spleen and lymph nodes was shown for [89Zr]Zr-N-suc-Df-ERY974 and [89Zr]Zr-N-suc-Df-KLH/CD3 in huNOG mice. Moreover, tumor uptake of [89Zr]Zr-N-suc-Df-ERY974 was threefold higher compared with uptake in the tumors of mice lacking T cells. On administration of a pharmacological active dose of [89Zr]Zr-N-suc-Df-ERY974 (6) tumors got heavily infiltrated with T cells, both in intratumoral and in stromal regions. Autoradiography illustrated that [89Zr]Zr-N-suc-Df-ERY974 was predominantly localized to T cell infiltrated stromal regions. The T cell infiltration of the tumor might be due to local proliferation of resident T cells on activation by the bispecific antibody as demonstrated in earlier studies (6, 18, 19) or by bispecific antibody mediated migration. In our study, [89Zr]Zr-N-suc-Df-ERY974 specifically bound to peripheral blood cells. On encountering

Biodistribution of CD3/glypican 3-targeting bispecific antibody ERY974

the tumor, [89Zr]Zr-N-suc-Df-ERY974-bound T cells might get into the tumor due to GPC3 expression, resulting in increased tumorous T cell infiltration and ERY974 uptake. An immunoPET study in a CD3 transgenic mouse model bearing a mucin 16 (MUC16) positive xenograft with an 89Zr-labeled full-length bispecific T cell redirecting antibody REGN4018 targeting MUC16 and CD3ε also demonstrated tumor and secondary lymphoid organ targeting (20). However, affinities for both targets were not disclosed and tumor uptake of 89Zr-labeled REGN4018 was not compared with control tracers or tumor-bearing wild type mice.

As T cell redirecting bispecific antibodies have not shown antitumor effects in patients with solid tumors yet, it is key to better understand its behavior to support drug development. Clinical information regarding the biodistribution of T cell redirecting antibodies could provide additional insight in tumor targeting properties and off-target distribution. However, clinical data are strikingly limited. The distribution of an 89Zr-labeled 54 kDa bispecific T cell engager AMG211 targeting CD3ε (affinity 310 nM) and carcinoembryonic antigen (affinity 5.5 nM) was studied in a feasibility study in nine patients with advanced gastrointestinal adenocarcinomas (21). [89Zr]Zr-N-suc-Df-AMG211 clearly accumulated in spleen and bone marrow, both CD3rich tissues. Uptake of [89Zr]Zr-N-suc-Df-AMG211 in tumor lesions was highly heterogeneous within and between patients. For [89Zr]Zr-N-suc-Df-ERY974, evaluation in a clinical trial would be helpful to further understand ERY974’s behavior and tumor targeting properties. Whole body non-invasive imaging of [89Zr]Zr-N-suc-Df-ERY974 might help to inform potential targetrelated drug impact in vivo including off-tumor/on-target or off-target uptake. Information regarding distribution of [89Zr]Zr-N-suc-Df-ERY974 at baseline and during ERY974 treatment, could further aid in the drug development of ERY974. Biopsies would allow to correlate PET signal with multiple parameters such as tumorous T cell infiltration and GPC3 protein expression. Furthermore, biopsies combined with tissue autoradiography could provide information regarding 89Zr distribution in both stroma and tumor areas, as shown in this preclinical study. With the potency of T cell redirecting bispecific antibodies, radiolabeling a low protein dose would be a challenge in a clinical study. In the end, results from such a clinical study could support rational future trial design.

Although the engraftment of human immune cells allowed us to study the influence of T cells on the biodistribution of [89Zr]Zr-N-suc-Df-ERY974, the mouse model used in this study does not fully recapitulate the human environment. The reconstituted immune cells that were present in the mice in this study did home to lymphoid tissues and were able to infiltrate xenografts, leading to CD3 specific uptake of [89Zr]Zr-N-suc-Df-ERY974 and [89Zr]Zr-N-suc-DfKLH/CD3. Uptake of [89Zr]Zr-N-suc-Df-ERY974 found in lymphoid organs of these mice needs to be interpreted and potentially translated with caution. We observed CD3 independent uptake of [89Zr]Zr-N-suc-Df-ERY974 in spleens due to the genetic background of the mice used in this

study. This effect was mediated by radioactivity-related spleen aplasia and mAb affinity for FcγR. The non-specific spleen uptake was observed earlier in SCID mice (22), NOG mice (23) and NSG mice (13). In a more comprehensive study, the role of radioactivity, mouse strain and FcγR affinity of the mAb confirmed the underlying mechanism (14). Besides the lack of a fully intact and functional human immune system, lack of cross-reactivity toward GPC3 did not allow us to study the non-tumor targeting of ERY974 to organs with physiological GPC3 expression and is therefore a limitation of this study. However, GPC3 expression in healthy tissues in adult mammals is rarely observed and is, therefore, less likely to influence biodistribution of [89Zr]Zr-N-suc-Df-ERY974.

In our study, we did not observe a reduction in HepG2 xenograft uptake of [89Zr]Zr-N-sucDf-ERY974 in the presence of an excess of unlabeled ERY974. This might be due to several reasons. First, high GPC3 expression of HepG2 in vitro and ex vivo was observed, and might be too high to be blocked by excess of unlabeled ERY974 in vivo. In the TOV-21G xenograft, which expresses lower levels of GPC3, an excess of unlabeled ERY974 was able to reduce tumor uptake of [89Zr]Zr-N-suc-Df-ERY974. Second, the monovalent character of ERY974 might be partial responsible for the lack of reduction in HepG2 xenograft uptake on an excess of unlabeled ERY974. Adding an excess of bivalent GPC3 mAb namely reduced tumor-to-blood ratio. Furthermore circulating GPC3 protein shedded from the HepG2 tumor xenograft might redirect [89Zr]Zr-N-suc-Df-ERY974 to the liver and prevent tumor uptake. This phenomenon has been described for the epidermal growth factor receptor (EGFR) and an EGFR targeting mAb imgatuzumab. Shed EGFR in the circulation derived from the xenograft A431, that expresses high levels of EGFR, was able to redirect the 89Zr-labeled imgatuzumab to the liver at a low tracer protein dose (24). In that study, increasing the tracer protein dose led to a reduced liver and increased tumor uptake. Although we do not have data on the levels of circulating GPC3 protein in our study, in the presence of an excess of unlabeled ERY974 liver uptake of [89Zr]Zr-N-suc-Df-ERY974 was reduced. The cell line HepG2 has been shown to gradually increase secretion of GPC3 protein up to 48 hours of in vitro culture (25). In addition, concentration of GPC3 protein in serum has been described in patients with hepatocellular carcinoma ranging from 150 to 3000 ng/mL (26).

These data provide a rationale to study the biodistribution and tumor targeting properties of [89Zr]Zr-N-suc-Df-ERY974 in a clinical trial to support ERY974 drug development.

Acknowledgments

We thank Linda Pot for assistance with immunohistochemical stainings and Iris Kluft for assistance with the manufacturing of the conjugated antibodies. Furthermore, we thank Athos Gianella-Borradori, Shohei Kishishita, Kenji Hashioto, Norihisa Ohishi for their input in the design

Biodistribution of CD3/glypican 3-targeting bispecific antibody ERY974

of this study.

Contributors

Study concept and design: all authors. Acquisition of data: SJHW and DG. Analysis and interpretation of data: all authors. Study supervision: CPS, ML-dH and EDV. Writing, review and/ or revision of the manuscript: all authors.

Funding

A research grant to EDV was obtained from Chugai and made available to the institution.

Competing interests

TI, YS and NS are employees of Chugai Pharmaceutical and have ownership interest (including stocks and patents) in Chugai Pharmaceutical. A research grant to EDV was obtained from Chugai and made available to the institution.

References

1. Wei SC, Duffy CR, Allison JP. Fundamental mechanisms of immune checkpoint blockade therapy. Cancer Discov 2018;8:1069–86.

2. Carter PJ, Lazar GA. Next generation antibody drugs: pursuit of the 'high-hanging fruit'. Nat Rev Drug Discov 2018;17:197–223.

3. Krishnamurthy A, Jimeno A. Bispecific antibodies for cancer therapy: a review. Pharmacol Ther 2018;185:122–34.

4. Suurs FV, Lub-de Hooge MN, de Vries EGE, et al. A review of bispecific antibodies and antibody constructs in oncology and clinical challenges. Pharmacol Ther 2019;201:103–19.

5. Kantarjian H, Stein A, Gökbuget N, et al. Blinatumomab versus chemotherapy for advanced acute lymphoblastic leukemia. N Engl J Med 2017;376:836–47.

6. Ishiguro T, Sano Y, Komatsu S-I, et al. An anti-glypican 3/CD3 bispecific T cell-redirecting antibody for treatment of solid tumors. Sci Transl Med 2017;9:eaal4291.

7. Moek KL, Fehrmann RSN, van der Vegt B, et al. Glypican 3 overexpression across a broad spectrum of tumor types discovered with functional genomic mRNA profiling of a large cancer database. Am J Pathol 2018;188:1973–81.

8. Ogita Y, Weiss D, Sugaya N, et al. A phase 1 dose escalation (DE) and cohort expansion (CE) study of ERY974, an anti-Glypican 3 (GPC3)/CD3 bispecific antibody, in patients with advanced solid tumors. JCO 2018;36:2599.

9. de Vries EGE, Kist de Ruijter L, Lub-de Hooge MN, et al. Integrating molecular nuclear imaging in clinical research to improve anticancer therapy. Nat Rev Clin Oncol 2019;16:241–55.

10. Mandikian D, Takahashi N, Lo AA, et al. Relative target affinities of T-cell-dependent bispecific antibodies determine biodistribution in a solid tumor mouse model. Mol Cancer Ther 2018;17:776–85.

Biodistribution of CD3/glypican 3-targeting bispecific antibody ERY974

11. Yahata T, Ando K, Nakamura Y, et al. Functional human T lymphocyte development from cord blood CD34+ cells in nonobese diabetic/Shi-scid, IL-2 receptor gamma null mice. J Immunol 2002;169:204–9.

12. Warnders FJ, Waaijer SJH, Pool M, et al. Biodistribution and PET imaging of labeled bispecific T cellengaging antibody targeting EpCAM. J Nucl Med 2016;57:812–7.

13. Sharma SK, Chow A, Monette S, et al. Fc-mediated anomalous biodistribution of therapeutic antibodies in immunodeficient mouse models. Cancer Res 2018;78:1820–32.

14. Sharma SK, Pourat J, Abdel-Atti D, et al. Noninvasive interrogation of DLL3 expression in metastatic small cell lung cancer. Cancer Res 2017;77:3931–41.

15. Waaijer SJH, Warnders FJ, Stienen S, et al. Molecular imaging of radiolabeled bispecific T-cell engager 89Zr-AMG211 targeting CEA-positive tumors. Clin Cancer Res 2018;24:4988–96.

16. Bensch F, Smeenk MM, van Es SC, et al. Comparative biodistribution analysis across four different 89Zrmonoclonal antibody tracers-The first step towards an imaging warehouse. Theranostics 2018;8:4295–304.

17. Bensch F, van der Veen EL, Lub-de Hooge MN, et al. 89Zr-atezolizumab imaging as a non-invasive approach to assess clinical response to PD-L1 blockade in cancer. Nat Med 2018;24:1852–8.

18. Bacac M, Fauti T, Sam J, et al. A novel carcinoembryonic antigen T-cell bispecific antibody (CEA TCB) for the treatment of solid tumors. Clin Cancer Res 2016;22:3286–97.

19. Li J, Ybarra R, Mak J, et al. IFNγ-induced chemokines are required for CXCR3-mediated T-cell recruitment and antitumor efficacy of Anti-HER2/CD3 bispecific antibody. Clin Cancer Res 2018;24:6447–58.

20. Crawford A, Haber L, Kelly MP, et al. A mucin 16 bispecific T cell-engaging antibody for the treatment of ovarian cancer. Sci Transl Med 2019;11:eaau7534.

21. Moek KL, Waaijer SJH, Kok IC, et al. 89Zr-labeled bispecific T-cell engager AMG 211 PET shows AMG 211 accumulation in CD3-rich tissues and clear, heterogeneous tumor uptake. Clin Cancer Res 2019;25:3517–27.

22. Burvenich IJG, Parakh S, Lee F-T, et al. Molecular imaging of T cell co-regulator factor B7-H3 with 89ZrDS-5573a. Theranostics 2018;8:4199–209.

23. Marquez BV, Ikotun OF, Zheleznyak A, et al. Evaluation of 89Zr-pertuzumab in breast cancer xenografts. Mol Pharm 2014;11:3988–95.

24. Pool M, Kol A, Lub-de Hooge MN, et al. Extracellular domain shedding influences specific tumor uptake and organ distribution of the EGFR PET tracer 89Zr-imgatuzumab. Oncotarget 2016;7:68111–21.

25. Nakatsura T, Yoshitake Y, Senju S, et al. Glypican-3, overexpressed specifically in human hepatocellular carcinoma, is a novel tumor marker. Biochem Biophys Res Commun 2003;306:16–25

26. Capurro M, Wanless IR, Sherman M, et al. Glypican-3: a novel serum and histochemical marker for hepatocellular carcinoma. Gastroenterology 2003;125:89–97.

Supplementary additional methods

Conjugation, radiolabeling and quality control ERY974, KLH/CD3, KLH/KLH and IgG4 were conjugated with tertrafluorphenol-N-succinyl desferal-Fe (N-suc-Df; ABX) as described before (1, 2). In short, antibodies were purified using Vivaspin-2 30,000 MWCO PES centrifugal concentrators (Sartorius) in 0.9% NaCl (Braun). After pH adjustment to 9.0 using 0.1 M Na2CO3, a 4-fold excess of N-suc-Df was added for 30 minutes. Subsequently Fe3+ was removed using EDTA and the solution was purified using PD-10 column (GE Healthcare) and 0.9% NaCl as eluent. Quality of conjugated antibody was assessed using size exclusion high-performance liquid chromatography as described before (1), using a TSKgel G3000SWXL column (Tosoh). Radiolabeling of antibodies with [89Zr]Zr-oxalate (PerkinElmer) was performed as described before (2). After 1 hour incubation, radiochemical purity was above 95% for all experiments and purification was not performed. Molar activity for all experiments was 72.8 MBq/nmol, unless stated otherwise.

Binding to GPC3 and CD3ε was tested using an ELISA based method. Recombinant human GPC3 (10088-H08H; Sino Biologicals Inc.) or CD3ε (10977-H08H; Sino Biologicals Inc.) were diluted in 0.05M Na2CO3 to a concentration of 0.1 μg/mL. Nunc-Immuno 96 well MicroWell MaxiSorp plates (Thermo Fisher Scientific) were coated with 100 μL recombinant protein at 4 °C overnight. Wells were washed with 0.05% Tween20 in phosphate buffered saline (PBS; 140 mM/L NaCl, 9 mM/L Na2HPO4, 1.3 mM/L NaH2PO4, pH 7.4, UMCG). Next, wells were blocked with 0.5% bovine serum albumin (BSA), 0.05% Tween20 in PBS for 2 hours at room temperature (RT). After blocking, wells were incubated with a concentration series (0.02 nM – 137.4 nM) of mAb diluted in 0.5% BSA/0.05% Tween 20/PBS for 1 hour at RT. Subsequently, wells were washed three times with 0.05% Tween20/PBS followed by 1 hour incubation at RT of rabbit anti-human IgA, IgG, IgM, Kappa, Lambda HRP (1:8000; Agilent DAKO). Again, wells were washed three times with 0.05% Tween 20/PBS followed by addition of 100 μL substrate SureBlue Reserve TMB microwell substrate (KPL Inc.). Reaction was stopped with 1 M hydrochloric acid (UMCG) and absorbance at 450 nm was determined with a microplate reader (Bio-Rad).

T cell activation potency was determined using a co-culture of HepG2 cells with Jurkat cells that express a luciferase reporter driven by a Nuclear Factor of Activated T cells response element (Jurkat-NFAT; Promega). In a 96-well plate, 12,500 HepG2 cells and 75,000 Jurkat-NFAT effector cells were incubated overnight at 37 °C with a concentration of ERY974 or N-suc-DfERY974 ranging from 0.05 pM to 137.4 nM. After incubation, 75 μL Bio-Glo reagent (Promega) was added and bioluminescence was determined with a Synergy plate reader (Biotek).

Internalization of [89Zr]Zr-N-suc-Df-ERY974

To determine internalization of [89Zr]Zr-N-suc-Df-ERY974, 106 HepG2 cells were incubated with

50 ng [89Zr]Zr-N-suc-Df-ERY974 in 1 mL medium on ice for 1 hour. After initial binding, unbound [89Zr]Zr-N-suc-Df-ERY974 was washed three times using 1% human serum albumin in PBS. Next, cells were incubated at 4 °C or 37 °C for 1, 2, or 4 hours. After incubation, cell membranes were stripped with 1 mL stripping buffer (0.05 M glycine, 0.1 M NaCl, pH 2.8) at 4 °C. Radioactivity of the cell pellet (internalization) was expressed as percentage of radioactivity initially bound to cells.

Immunohistochemistry

Formalin-fixed paraffin-embedded 4 μm tissue slides were stained with immunohistochemistry using 2 μg/mL rabbit monoclonal GPC3 antibody (SP86; Abcam) or isotype control (EPR25A; Abcam), followed by rabbit EnVision HRP (Agilent). Human placenta and HepG2 tumor of [89Zr] Zr-N-suc-Df-ERY974 injected huNOG mice were used as positive control tissue (Additional file 1 Fig. S10A). For CD3, tissues were stained using 0.15 μg rabbit monoclonal CD3 antibody (SP162; Abcam) or isotype control (EPR25A; Abcam), followed by rabbit EnVision HRP (Agilent). Human liver and HepG2 tumors of [89Zr]Zr-N-suc-Df-ERY974 injected mice were used as positive control tissue (Additional file 1 Fig. S10B). CD3+ cells were quantified using positive cell detection using QuPath (3).

Flow cytometry

HepG2, TOV-21G and SK-HEP-1 cells were harvested and suspended in 20 μg/mL of ERY974 or human IgG4 in 0.5% fetal bovine serum (FBS)/2 mM EDTA/PBS. Cells were incubated for 1 hour at 4 °C, subsequently washed twice with 0.5% FBS/2 mM EDTA/PBS and incubated with PE-labeled goat anti-human IgG (1:50; Thermo Fisher Scientific) at for 1 hour 4 °C. After two more washes with 0.5% FBS/2 mM EDTA/PBS, cells were measured using a BD Accuri C6 flow cytometer (BD Biosciences).

Supplementary references

1. Warnders FJ, Waaijer SJ, Pool M, Lub-de Hooge MN, Friedrich M, Terwisscha van Scheltinga AG, et al. Biodistribution and PET imaging of labeled bispecific T cell-engaging antibody targeting EpCAM. J Nucl Med. 2016;57(5):812-7.

2. Verel I, Visser GW, Boellaard R, Stigter-van Walsum M, Snow GB, van Dongen GA. 89Zr immuno-PET: comprehensive procedures for the production of 89Zr-labeled monoclonal antibodies. J Nucl Med. 2003;44(8):1271-81.

3. Bankhead P, Loughrey MB, Fernandez JA, Dombrowski Y, McArt DG, Dunne PD, et al. QuPath: open source software for digital pathology image analysis. Sci Rep. 2017;7(1):16878.

Biodistribution of CD3/glypican 3-targeting bispecific antibody ERY974

Supplementary figures

SUPPLEMENTARY FIGURE S1: Human CD3+ engraftment in huNOG mice. Percentage of human CD3+ of human CD45+ cells in the experimental groups involving huNOG mice.

SUPPLEMENTARY FIGURE S2: In vitro characteristics of N-suc-Df-conjugated tracers.

(A) Representative binding curve of N-suc-Df-ERY974 and ERY974 binding to human GPC3 protein. (B) Representative binding curve of N-suc-Df-ERY974 and ERY974 binding to human CD3ε protein. (C) Potency of ERY974 and N-suc-Df-ERY974 to activate reporter T cells upon co-culture with HepG2 cells. (D) Internalization up to 4 h of [89Zr]Zr-N-suc-Df-ERY974 in HepG2 cells at 4 and 37 °C (n = 3). →

Biodistribution of CD3/glypican 3-targeting bispecific antibody ERY974

→ (E) Representative binding curve of N-suc-Df-KLH/CD3 and N-suc-Df-KLH/KLH to human GPC3 protein. (F) Representative binding curve of N-suc-Df-KLH/CD3 and N-suc-Df-KLH/KLH to human CD3ε protein.

SUPPLEMENTARY FIGURE S3: Tumor characteristics of HepG2, TOV-21G and SK-HEP-1.

(A) Hematoxylin and eosin (H&E), autoradiography and glypican-3 (GPC3) staining of HepG2, TOV-21G and SK-HEP-1 xenografts. Scale bar length represents 5 mm for HepG2, 1 mm for TOV-21G and 2.5 mm for SKHEP-1, and 250 μm for the zoomed slides. Autoradiography and H&E were performed on the same slide. For each cell line, flow cytometry was performed using ERY974 as primary antibody (black), including IgG4 as control (red; right panel). (B) SDS-PAGE autoradiography of different individual HepG2 (left), TOV-21G (middle) and SK-HEP-1 (right) lysates and corresponding plasma samples. '+' represents activity matched [89Zr]Zr-N-suc-Df-ERY974 tracer from injected solution. kDa = kilodalton.

(A) Spleen uptake at 168 h after administration of 10 μg of [89Zr]Zr-N-suc-Df-ERY974 (n = 6), [89Zr]Zr-Nsuc-Df-KLH/CD3 (n = 5), [89Zr]Zr-N-suc-Df-KLH/KLH (n = 6) and [89Zr]Zr-N-suc-Df-IgG4 (n = 5) expressed as median % injected dose per gram (%ID/g) with interquartile range. (B) Spleen weight of [89Zr]Zr-N-suc-DfERY974 (n = 6), [89Zr]Zr-N-suc-Df-KLH/CD3 (n = 5), [89Zr]Zr-N-suc-Df-KLH/KLH (n = 6) and [89Zr]Zr-N-suc-DfIgG4 (n = 5) expressed as median weight in mg with interquartile range. (C) Spleen weight of NOG mice injected with 10 μg of [89Zr]Zr-N-suc-Df-ERY974 labeled with 5 MBq (Am: 14.6 MBq/nmol) at 72 h (n = 2), 120 h (n = 2) and 168 h (n = 12) after administration expressed as median weight with interquartile range (IQR). A m = molar activity. (D) Spleen uptake of NOG mice injected with 10 μg of [89Zr]Zr-N-suc-Df-ERY974 labeled with 1 MBq (Am: 14.6 MBq/nmol; n = 6) or 5 MBq (Am: 72.8 MBq/nmol; n = 12) at 168 h expressed as median % injected dose per gram with IQR. (E) Spleen weight of NOG mice injected with 10 μg of [89Zr]Zr-N-suc-DfERY974 labeled with 1 MBq (Am: 14.6 MBq/nmol; n = 6) or 5 MBq (Am: 72.8 MBq/nmol; n = 12) at 168 h expressed as median weight with IQR. (F) Radioactivity dose of the spleen of NOG mice injected with 10 μg of [89Zr]ZrN-suc-Df-ERY974 labeled with 1 MBq (Am: 14.6 MBq/nmol n = 6) or 5 MBq (Am: 72.8 MBq/nmol; n = 12) at 168 h expressed as median dose with IQR. (G) Hematoxylin and eosin (H&E; 400x) staining of a NOG mice spleen →

SUPPLEMENTARY FIGURE S4: Influence of FcγR binding and radioactive dose on biodistribution of different tracers in mice.

Biodistribution of CD3/glypican 3-targeting bispecific antibody ERY974

→ injected with 1 MBq (Am: 14.6 MBq/nmol) or 5 MBq (Am: 72.8 MBq/nmol) of [89Zr]Zr-N-suc-Df-ERY974 at 168 h after tracer administration. Scale bar length represents 250 μm. (H) Uptake of [89Zr]Zr-N-suc-Df-ERY974 in spleen, bone, liver and blood in NOG (n = 6) and BALB/c nude (n = 6) at 168 h after tracer administration expressed as median % injected dose per gram of tissue (%ID/g) with interquartile range (IQR). (I) Uptake of [89Zr]Zr-N-suc-Df-ERY974 in spleen in NOG (n = 6) and BALB/c nude (n = 6) at 168 h after tracer administration expressed as median %ID/g with IQR. (J) Spleen weight of NOG (n = 6) and BALB/c nude (n = 6) mice at 168 h after tracer administration expressed as median weight with IQR. (K) Pooled data of [89Zr]Zr-N-suc-DfERY974 uptake in spleen, femur, cortical femur, femur bone marrow of NOG (n = 18) and BALB/c nude (n = 6) mice at 168 h after administration expressed as median %ID/g with IQR.

SUPPLEMENTARY FIGURE S5: Dose escalation of [89Zr]Zr-N-suc-Df-ERY974 in immunodeficient NOG mice bearing different tumor xenografts.

(A) Ex vivo biodistribution of [89Zr]Zr-N-suc-Df-ERY974 in HepG2 at 168 h post injection with 10 μg in (n = 12), 2000 μg (n = 6), or 1000 μg GPC3 bivalent (n = 3), and in TOV-21G with 10 μg (n = 6) or 2000 μg (n = 2). Doses higher than 10 μg were supplemented with non-labeled ERY974 or GPC3 bivalent antibody. Data is expressed as median %ID/g with interquartile range (IQR). **P≤ 0.01 (Mann-Whitney U). (B) Uptake of [89Zr]Zr-N-suc-Df-ERY974 dose groups in blood expressed as median %ID/g with IQR. *P ≤0.05 (Mann-Whitney U). (C) Tumor-to-blood ratio of [89Zr]Zr-N-suc-Df-ERY974 dose groups expressed as median with IQR. *P≤ 0.05; **P≤0.01 (Mann-Whitney U). (D) Uptake of [89Zr]Zr-N-suc-Df-ERY974 dose groups in liver expressed as median %ID/g with IQR. *P≤0.05 (Mann-Whitney U).

(A) Biodistribution of 10 μg [89Zr]Zr-N-suc-Df-ERY974 in NOG (n = 12) and huNOG (n = 5) mice expressed as median percentage injected dose per gram of tissue (%ID/g) with interquartile range (IQR). (B) Biodistribution of 10 μg [89Zr]Zr-N-suc-Df-KLH/CD3 in NOG (n = 5), huNOG (n = 4), or huNOG mice co-injected with 10 μg ERY974 (n = 3) expressed as median % ID/g with IQR. (C) Biodistribution of 10 μg [89Zr]Zr-N-suc-DfKLH/KLH in NOG (n = 6), huNOG (n = 6), or huNOG mice co-injected with 10 μg ERY974 (n = 3) expressed as median % ID/g with IQR.

SUPPLEMENTARY FIGURE S6: Ex vivo biodistribution of different tracers in different mice models at 168 h after tracer administration.

Biodistribution of CD3/glypican 3-targeting bispecific antibody ERY974

SUPPLEMENTARY FIGURE S7: Binding to peripheral blood mononuclear sites of huNOG mice injected with [89Zr]Zr-N-suc-Df-ERY974, [89Zr]Zr-N-suc-Df-KLH/CD3 or [89Zr]Zr-N-suc-Df-KLH/KLH.

Percentage of bound tracer to peripheral blood mononuclear cells (PBMCs) isolated from blood from huNOG mice injected with [89Zr]Zr-N-suc-Df-ERY974 (n = 3), [89Zr]Zr-N-suc-Df-KLH/CD3 (n = 4) or [89Zr]Zr-N-sucDf-KLH/KLH (n = 4).

SUPPLEMENTARY FIGURE S8: CD3 immunohistochemistry in HepG2 tumors of huNOG mice injected with [89Zr]Zr-N-suc-DfERY974, [89Zr]Zr-N-suc-DfKLH/CD3 or [89Zr]Zr-N-sucDf-KLH/KLH.

(A) Intratumoral (top panel; scale bar length represents 100 μm) and stromal (bottom panel; scale bar length represents 100 μm) CD3+ T cells in HepG2 tumors (middle panel; scale bar length represents 5 mm) of huNOG mice injected with [89Zr]Zr-N-sucDf-ERY974, [89Zr]Zr-N-sucDf-KLH/CD3 or [89Zr]Zr-Nsuc-Df-KLH/KLH.

(B) Quantification of T cell infiltrations expressed as CD3+ cells/mm2. Lines represent median with interquartile range. *P<0.05.

SUPPLEMENTARY FIGURE S9: CD3 immunohistochemistry in HepG2 tumors of huNOG mice co-injected with ERY974.

(A) Intratumoral CD3+ T cells in HepG2 tumors of huNOG mice injected with [89Zr]Zr-N-suc-DfKLH/CD3 or [89Zr]Zr-N-suc-DfKLH/KLH co-injected with ERY974.

Scale bar length represents 100 μm. (B) Quantification of CD3+ T cells expressed as CD3+ cells/ mm2

SUPPLEMENTARY FIGURE S10: Immunohistochemical staining validation.

(A) Glypican 3 (GPC3) or isotype control staining on human placenta tissue or huNOG HepG2 tumors. Scale bar length represents 100 μm for placenta and 2.5 mm for HepG2 tumor. (B) CD3 or isotype control staining on human liver or huNOG HepG2 tumors. Scale bar length represents 50 μm for liver and 500 μm for HepG2 tumor.

Whole-body CD8+ T-cell visualization before and during cancer immunotherapy

Laura Kist de Ruijter 1, Danique Giesen 1 *, Pim P. van de Donk 1 *, Jahlisa S. Hooiveld-Noeken 1 *, Sjoerd G. Elias 5, Marjolijn N. Lub-de Hooge 2, Sjoukje F. Oosting 1, Mathilde Jalving 1, Wim Timens 3, Adrienne H. Brouwers 4, Thomas C. Kwee 4, Jourik A. Gietema 1, Rudolf S.N. Fehrmann 1, Bernard M. Fine 6, Sandra M. Sanabria Bohorquez 6, Mahesh Yadav 6, Hartmut Koeppen 6, Jing Jing 6 , Sebastian Guelman 6, Mark T. Lin 6, Michael J. Mamounas 6, Jeffrey Eastham 6 , Patrick K. Kimes 6, Simon P. Williams 6, Alexander Ungewickell 6, Derk J.A. de Groot 1 & Elisabeth G.E. de Vries 1

* These authors contributed equally to this work

1 Department of Medical Oncology, 2 Department of Clinical Pharmacy and Pharmacology, 3 Department of Pathology and Medical Biology, and 4 Medical Imaging Center, University Medical Center Groningen, University of Groningen, the Netherlands; 5 Department of Epidemiology, Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht University, the Netherlands; 6 Genentech Inc., South San Francisco, CA, USA. Submitted

Abstract

Immune checkpoint inhibitors (ICI) that reinvigorate CD8+ T-cell mediated immunity have revolutionized cancer therapy, yet the systemic distribution of CD8+ T-cells as a potential predictive or pharmacodynamic biomarker of ICI response remains poorly characterized. Therefore we assessed the safety, imaging dose and time points, pharmacokinetics, and immunogenicity of the zirconium-89 labelled, CD8-specific one-armed antibody PET tracer 89ZED88082A in patients with solid tumours before and ~30 days after starting ICI therapy. The 39 patients experienced no tracer-related side effects. PET imaging with 10 mg anti-CD8 antibody revealed 89ZED88082A-uptake in normal lymphoid tissues, and tumour lesions across the body varying within and between patients 2 days after tracer injection (n = 38, median patient maximum standard uptake value (SUVmax) 5.2, IQI 4.0 – 7.0). Higher SUVmax was associated with mismatch-repair deficiency (8.0 vs 5.0, P = 0.01) and longer overall survival (P = 0.03). Uptake was higher in lesions with stromal/inflamed rather than desert immunophenotype (SUVmax 7.1 vs 4.3, P = 0.018), and tissue radioactivity was localized to areas with immunohistochemicallyconfirmed CD8 expression. Re-imaging patients on treatment (n = 22) showed no change in their average (geometric mean) tumour tracer uptake compared to baseline, but at the level of individual lesions markedly diverse changes were noted independent of tumour response. The imaging data suggest enormous heterogeneity in CD8+ T-cell distribution and pharmacodynamics within and between patients. In conclusion, 89ZED88082 is a novel tool to characterize the complex dynamics of CD8+ T-cells in the context of ICIs, and may inform future development of combination strategies and novel immunotherapeutics.

T-cell-enhancing immune checkpoint inhibitors (ICI) have gained their place in cancer treatment with impressive, durable anti-tumour efficacy in a remarkable variety of tumour types (1-3). However, response rates vary, and only a subset of patients benefits. A combination with another ICI or other medicines can improve response rates but can also increase the risk of adverse events (1). This highlights the clinical need for tools to optimize treatment strategies for individual patients. Several biomarkers have been identified to select patients for ICIs (3). These include programmed death-ligand 1 (PD-L1) expression, tumour mutational burden, deficiency of mismatch repair proteins (dMMR), and a T-cell-inflamed gene expression profile (4-6). However, no single biomarker or combination of biomarkers accurately predicts response to ICI.

CD8+ T-cells play an essential role in tumour cell destruction by the immune system. Their presence in the tumour is associated with responses to ICIs across several tumour types (610). An ICI treatment-emergent increase in CD8+ T-cell density in tumour biopsies has also been associated with tumour response. Most data are available for patients with advanced

melanoma with biopsies obtained at different time-points following the start of ICI. For example, increased CD8+ cell density in 25 paired tumour biopsies collected after 20-120 days pembrolizumab treatment was associated with response (11). Others reported a CD8+ T-cell expansion in 13 biopsies 2 weeks after anti-programmed cell-death (PD-1) antibody therapy initiation, but this was not the case in a study analyzing ten mostly late on-treatment biopsies after 0.7 – 26 months (9,10). Sampling bias may influence these differences and considerable heterogeneity can exist within or between different lesions within one patient (12,13).

Due to these inherent limitations for invasive tumour biopsies, remarkably little is known about the systemic kinetics and heterogeneity of CD8+ T-cell distribution among tumour types and individual tumour lesions in patients. To address this issue, we developed the zirconium-89 (89Zr) labelled one-armed antibody 89ZED88082A targeting CD8α, as 89Zr labelled antibodies or antibody fragments allow non-invasive whole-body visualization of a target with positron emission tomography (PET) (14-16). Firstly, 89ZED88082A-uptake with PET was shown in human CD8-expressing tumours xenografted in mice (17). We then performed 89ZED88082A PETscanning in patients with solid tumours before and ~30 days after starting ICI treatment with PD-L1 antibody, or PD-1 antibody with or without CTLA-4 antibody. The primary objectives of the study were to characterize the safety, imaging dose and time points, pharmacokinetics, and immunogenicity of 89ZED00802A in patients with solid tumours. Additional objectives included the potential to image whole-body CD8+ T-cells, correlations of CD8 PET imaging data with tumour based assessments, as well as correlations with clinical outcomes to ICI treatment.

Results

Trial population and safety

Between February 2019 and November 2020, 39 patients were enrolled. One patient with tracer extravasation was excluded from PET analyses (Extended Data Table 1). Twenty-two of the 29 consecutive patients included for repeated imaging did undergo this, with a median of 30 days following initiation of ICI treatment (IQI 28 – 36 days). Seven were not scanned during ICI therapy, due to withdrawal before (n = 1) and during treatment (n = 4), due to disease progression, patient anxiety (n = 1), and COVID-19 restrictions (n = 1).

No 89ZED88082A-related side effects occurred. Adverse events due to ICI were consistent with reports from previous studies (Extended Data Table 2).

In part A, two anti-CD8 tracer protein doses (89ZED88082A + unlabelled, DFO-conjugated onearmed antibody CED88004S) were evaluated: 4 mg (n = 3) or 10 mg (n = 6) with serial PET scans

0 (1 h), 2, 4, and 7 (± 1) days after administration, followed by a biopsy of a tumour lesion. The 10 mg dose allowed for sufficient blood pool tracer availability (average day 2 mean standard

Whole-body CD8+ T-cell visualization before and during cancer immunotherapy

uptake value (SUVmean) 2.9 (± 1.0), day 4 SUVmean 1.9 (± 0.3). Compared to 4 mg, the 10 mg dose showed less and stable splenic uptake, indicating abatement of splenic tracer sink effect (Extended Data Fig. 1a). The 10 mg protein dose visualized tumour lesions and lymphoid tissues (Fig. 1, online Supplementary Video), with highest uptake days 2 and 4 (Extended Data Fig. 2). In vitro, human peripheral blood mononuclear cells did not internalize the tracer (Extended Data Fig. 3), consistent with PET imaging data showing no further increase in tissue signal between days 2 - 7. Therefore, in part B, the 10 mg protein dose with PET-scanning on day 2 was considered optimal.

FIGURE 1: Normal tissue biodistribution of 89ZED88082A. (a) Representative 89ZED88082A PET scan maximum intensity projection day 2. A whole-body visualization is available as online supplementary video. (b-e) Axial views same scan fused with low-dose CT. Arrows indicating uptake (b) Waldeyer's ring, cervical lymph nodes, (c) spleen, bone marrow, (d) renal cortex, small intestine. (e) Inguinal lymph nodes. (f-g) Pretreatment uptake with 95% confidence bands across tissues adjusted for protein dose, projected at 10 mg dose, (n = 9) days 0 (1 h), 2, 4 and 7 (±1 day), (f) mean SUVmean, and (g) mean SUV max for lymph nodes and tonsils, not visible on day 0.

Uptake in tumour lesions at baseline

Baseline 89ZED88082A-uptake in all non-irradiated lesions (n = 266 in 38 patients) showed an overall geometric mean SUVmax of 5.6 (geometric coefficient of variation (GCV) 0.72) on day 2. Lesions were detected in all major organs. Median geometric mean SUVmax per patient was 5.2 (IQI 4.0 – 7.4). Heterogeneity in tumour uptake was observed between and within patients (intraclass correlation coefficient (ICC) 0.46, Fig. 2a, 2bii, Extended Data Fig. 4). In 10 patients, four with dMMR tumours, 16 lesions (6 dMMR) showed a pronounced tumour-rim uptake (Fig. 2b, Extended Data Fig. 4f-h). Among the 13 evaluable lesions out of these 16, only three had CT evidence of central necrosis.

89ZED88082A-uptake was related to the lesion's organ location and highest in malignant lymph nodes (Fig. 2c). Malignant lymph nodes also exhibited 62% higher SUVmax than normal lymph nodes (95%CI 45 – 80%, P ≤ 0.001). We took two approaches to verify whether potential differences in CD8 tracer uptake did reflect CD8-related tumour characteristics. Firstly, we showed that 89ZED88082A-tumour uptake was higher in the nine patients with dMMR than the 25 with pMMR tumours (Fig. 2d). Secondly, we studied with CD8 immunohistochemistry (IHC) the tumours of 24 patients with 22 pre- and 12 on-treatment samples. This showed 4 inflamed, 15 stromal, and 15 desert phenotypes (Extended Data Fig. 5a, Supplementary Information Fig. S1). The SUVmax was higher in inflamed or stromal phenotype lesions than desert phenotype lesions before and during treatment (Fig. 2e). Lesions with a CD8 desert phenotype had a geometric mean SUVmax of 4.3 (95%CI 3.1 – 6.0), while lesions with a stromal or inflamed phenotype had a geometric mean SUVmax of 7.1 (95%CI 5.4 – 9.4) (P = 0.018). Localized CD8+ T-cell density by IHC correlated with the autoradiography signal magnitude in tumour tissues (τ = 0.45, P = 0.015) (Fig. 3a, Extended Data Fig. 5.b-d).

As of October 13th, 2021, median patient follow-up was 5.6 months; 35 of 38 patients were evaluable for best overall response, four patients experienced a complete response (CR), eight a partial response (PR), four stable disease (SD), and 19 progressive disease (PD). Baseline tracer tumour uptake showed a positive trend with best overall RECIST response (Ptrend = 0.064, Extended Data Fig. 6a), and uptake was 40% (95%CI 0 – 94%) higher in patients with SD/PR/CR as best overall response during ICI (P = 0.047, Extended Data Fig. 6b). Patients with an above-median baseline 89ZED88082A-uptake geometric mean SUVmax (i.e., >5.2) showed a trend towards superior progression-free survival (PFS) (median 1.5, 95%CI 1.3 – not reached; versus 3.9, 95%CI 2.6 – not reached, P = 0.058) and had superior overall survival (OS) than patients with an uptake below the median (median 6.5, 95%CI 3.3 – not reached, versus 13.8 95%CI 11.3 – not reached, P = 0.030) (Fig. 4). Analyzed continuously, baseline 89ZED88082Auptake geometric mean SUVmax (per standard deviation decrease) showed for PFS an HR of 1.60 (95%CI 1.03 –2.78; P = 0.034) and for OS of 1.59 (95%CI 1.04 – 2.72; P = 0.031).

Whole-body CD8+ T-cell visualization before and during cancer immunotherapy

FIGURE 2: 89ZED88082A uptake in non-irradiated tumour lesions.

(a) Pretreatment uptake in 266 lesions day 2 post tracer injection, ordered by increasing geometric mean SUV max per patient, visualizing lesion size and site, and aorta background uptake. (b) Axial views PET/CT scans, arrows indicate lesions (i) High, heterogeneous uptake in dMMR duodenal tumour. →

→ (ii) Uptake in a triple-negative right breast cancer lesion, moderate uptake in pleural, and no to minor uptake in lung lesions. (iii) Minor uptake in perivesical dMMR urothelial cell cancer lesion pretreatment increased with rim pattern during treatment (iv). (c) Violin plot SUVmax in lesions (n = 212) per site (lymph nodes n = 99, liver n = 35, bone n = 17, lung n = 42, skin n = 19) (d) Violin plot of SUVmax in patients with pMMR (n = 25) and dMMR tumours (n = 9) (e) Violin plot of SUVmax in lesions with desert (n = 15) and non-desert (n = 19) immune phenotype before and during treatment in 24 patients. (c-e) Violin plots with bottom and top 1% of SUVmax values truncated (c, d, not for e); coloured dots are the geometric means per patient (d) or lesion (e); black vertical lines are geometric mean SUVmax 95% CI; white dots within black lines and values below the violin plot the actual geometric means.

Uptake in tumour lesions during treatment

During treatment, the average 89ZED88082A-uptake in non-irradiated lesions in all patients (lesion n = 111) was lower compared to baseline (-4.6% change in geometric mean SUVmax per week of treatment, 95%CI -6.5 to -2.6%), a change that depended on best overall response with a greater decrease in patients with SD, PR, or CR (Pinteraction = 0.018) (Extended Data Fig. 6c). Of the eight patients who showed PR or CR on treatment, five already met criteria for PR at the time of the PET scan at 30 days. When taking into account tumour volume change and resulting tracer uptake underestimation due to partial volume effects in responding lesions, the estimated average tracer uptake change was -2.7% (95%CI -4.4 to -1.1%) per week treatment, which no longer depended on best overall response (Pinteraction = 0.71) (Extended Data Fig. 6d). No patient in the repeat imaging cohort experienced pseudoprogression.

Within patients, lesions demonstrated diverse changes in 89ZED88082A-uptake, with some decreasing and others increasing compared to baseline. Moreover, responding lesions displayed a variety of dynamics in 89ZED88082A-uptake change between the two PET series (Fig. 4c).

For 10 patients, paired tumour tissues of the same lesion with corresponding tumour volumes of interest (VOIs) on PET were available (Supplementary Information Fig. S1). Five of them reflected concordant treatment-emergent changes by IHC and imaging (Fig. 3b). In one patient, a lymph node metastasis with a SUVmax of 8.28 and stromal CD8 T-cell infiltration at baseline showed only normal lymph node tissue in the second biopsy, with SUVmax of 5.63 on the on-treatment PET.

Normal tissue biodistribution and pharmacokinetics of 89ZED88082A/CED88004S 89ZED88082A showed a specific uptake per organ (Fig. 1). The highest 89ZED88082A-uptake occurred in the spleen and was apparent within an hour of injection. From day 2 onwards, there was a clear 89ZED88082A-uptake in normal lymphoid tissues, including the bone marrow,

Whole-body CD8+ T-cell visualization before and during cancer immunotherapy

(a) Autoradiography image of 89ZED88082A-uptake in a dMMR colorectal cancer liver metastasis and accompanying CD8 IHC staining. Areas 1, 3, 5 with moderate to high CD8 expression; 2, 4 without CD8 expression. (b) Overview of SUV max and CD8 IHC expression pattern [density score] in lesions with corresponding paired biopsies before and during treatment in 10 patients. On the x-axis, primary tumour type and location of biopsy are shown. The symbol above the bar indicates the radiographic response of the lesion at 6 weeks. →

FIGURE 3: 89ZED88082A in tumour tissues related to CD8 by immunohistochemistry.

→ TNBC; triple negative breast cancer. NEC; neuroendocrine carcinoma. OCCC; ovarian clear cell carcinoma. HCC; hepatocellular carcinoma. SqCC; squamous cell carcinoma. UP; unknown primary. UCC dMMR; urothelial cell carcinoma deficient mismatch repair.

Waldeyer's ring, lymph nodes, and the small intestine (Extended Data Fig. 1). Sites with previous lymph node dissection lacked uptake. Furthermore, tracer uptake was present in the renal cortex and liver. Partial volume effects and spillover signal precluded the quantification of small tumour lesions contained within the renal cortex and the spleen. Tracer uptake was also observed at sites of inflammation (Extended Data Fig. 7). In two patients, 89ZED88082Auptake was lower in vertebrae irradiated < 12 months earlier than in non-irradiated vertebrae (Extended Data Fig. 7g-h). During treatment, the average tracer SUVmean in blood pool at 4 weeks was 13.3% lower compared to pretreatment. Equally, uptake in spleen and lymphoid tissues was limitedly decreased, the latter not correlated to best overall response (Extended Data Fig. 1b).

Several patients developed immune-related adverse events (irAE) after ICI initiation (Extended Data Table 2). One patient with Hashimoto's thyroiditis on stable thyroid replacement therapy experienced a flare-up requiring more replacement. Her elevated baseline thyroid SUVmean 3.32 increased during treatment to 8.07 (Extended Data Fig. 7e-f). In other patients experiencing irAE ≥ grade 3 within the time frame of PET scans or thereafter, no higher 89ZED88082A uptake at baseline or during treatment occurred in organs of interest. This included two patients who developed diarrhoea 4 and 14 days after the on-treatment CD8 PET. They were evaluated two days after start of diarrhoea with colonoscopy and a colonic biopsy, which showed minor inflammation in both patients. They were later treated with steroids because of clinical suspicion of ICI-induced colitis.

In part A, serum 89ZED88082A/CED88004S protein levels were comparable within the same dose groups (Extended Data Fig. 8a-b). The estimated serum half-life of 89ZED88082A/CED88004S was 1.19 ± 0.33 days. Tracer pharmacokinetics were not influenced by ICI (Extended Data Fig. 8c). 89ZED88082A was intact in serum, while only low molecular weight components, including free 89Zr, were detectable in urine (Extended Data Fig. 8d). 89ZED88082A administration did not affect T-cell, B-cell, and NK-cell blood counts (Extended Data Fig. 3a).

No patient had endogenous antibody-drug antibodies (ADAs) before tracer injection (n = 31), 19% developed ADAs 28 - 50 days after the first (n = 5 out of 26), and 8% 18 - 38 days after the second tracer injection (n = 1 out of 12). One out of the 22 patients imaged twice (pre- and ontreatment) developed ADAs after the first tracer injection. There was no apparent ADA effect on 89ZED88082A/CED88004S serum levels and imaging results.

Whole-body CD8+ T-cell visualization before and during cancer immunotherapy

FIGURE 4: 89ZED88082A-uptake related to tumour response.

(a) Progression-free survival, and (b) overall survival, according to baseline geometric mean SUVmax below and above median. (c) Changes during repeated imaging in tumour uptake and anatomic size, expressed as estimated changes per week treatment to account for variation between patients in the timing of the PET-scan/CT response evaluation. Patients (n = 19) are represented by two bars (blue, pink) and grouped per best overall treatment response. Blue bars: change in sum target lesions according to RECIST between pretreatment and first response evaluation. Pink bars: average SUVmax change. Dots are individual lesions (n = 111); individual lesion datapoints for size (blue) and uptake (red) are connected by grey lines. Blue dots: lesion blueness = RECIST diameter pretreatment, dot location = change in size versus baseline. Red dots: lesion redness = SUV max pretreatment, dot location = SUVmax change.

Discussion

A systemic characterization of the tumour microenvironment is critical for understanding an effective anticancer immune response following immunotherapies. This is a first-in-

human study with the CD8-targeting antibody 89ZED88082A characterizing the CD8+ T-cell biodistribution by PET imaging in patients with cancer at baseline and during ICI treatment. We demonstrated that the tracer is safe. Tracer uptake in tumour lesions correlated with CD8 IHC and autoradiography signal in those lesions. 89ZED88082A signal was conspicuous early on in the blood pool and kidneys as clearance organs, and in the spleen with extensive CD8 expression on the red pulp reticuloendothelial cells (18). However, progressive uptake was evident only in CD8-rich tissues with such as lymph nodes, further supporting the tracer’s CD8 specificity.

Overall, high 89ZED88082A tumour uptake at baseline was associated with a better OS, concordant with findings from CD8 IHC in tissues from clinical ICI trials (19,20). There was a major spatial heterogeneity within and between patients in 89ZED88082A-uptake by their lesions. We took two approaches to verify whether potential differences in CD8 tracer uptake did reflect CD8-related tumour characteristics. First, we showed higher tracer uptake in dMMR than in pMMR tumours imaged before treatment, reflecting the higher CD8+ T-cell infiltrate reported in dMMR tumours (21-25). Second, we showed that tumour lesions biopsied and known by IHC to have a high T-cell infiltrate (either ‘stromal’ or ‘inflamed’ phenotype), showed higher CD8 tracer uptake than the group with a low-T-cell ‘desert’ phenotype. The 89ZED88082A-uptake in a rim pattern in several tumours before and during treatment likely mirrors CD8+ T-cell tumour infiltration referred to as the invasive margin (11,24,26).

To improve insight into ICIs, their biodistribution has been studied with 89Zr-labelled anti-PD-1 and anti-PD-L1 antibodies (15,16,28,29). In patients receiving atezolizumab, pretreatment 89Zratezolizumab tumour uptake predicted tumour response, PFS, and OS, while PD-L1 expression assessed by IHC did not (15). Similar observations were made for 89Zr-pembrolizumab imaging (16). This demonstrates that T-cells in tumour lesions as key mediators of immunotherapy can be evaluated by whole-body PET imaging. CD8 imaging was recently described in a small phase 1 study involving CD8 PET imaging at a single time point either before, during, or after ICI or targeted therapy in 15 patients using different protein doses of the minibody 89Zr-DfIAB22M2C27. The 89Zr-minibody was safe and accumulated in CD8+ rich tissues and tumour lesions of 10 patients, supporting the CD8 PET approach.

We are the first to report ICI-induced dynamics of CD8+ T-cells in tumour lesions measured by repeat whole-body PET imaging in patients. We cannot exclude that we also visualized CD8+ NK-cells, but they are relatively rare and not likely to be confounding. Although we observed increasing signal in individual cases preceding a response, as also shown in some biopsy studies (9-11,30), overall SUVmax changes on 89ZED88082A PET at 30 days after initiation of ICI did not correlate with best overall response when adjusted for volume changes. Intriguingly,

Whole-body CD8+ T-cell visualization before and during cancer immunotherapy

we identified an enormous interlesional heterogeneity in tracer uptake on PET at 30 days in patients who responded. These findings indicate a remarkable spatio-temporal variability in systemic T-cell dynamics as an antitumour immune response unfolds. Interestingly, similar results have been seen in a well-controlled mouse model using in situ fluorescent imaging of tumour cells and immune cells. Thus, a large variety in immunophenotype evolution was visualized even within individual mice of one model of the same seeded tumour cell line (31). Moreover, in a human tumour fragment platform assay, PD-1 blockade resulted in different immune activation profiles among small tumour fragments derived from individual patient tumour lesions. Together, our results underscore the importance of timing and characterization of all tumour lesions in comprehensively evaluating the tumour-immune status and therapyinduced pharmacodynamic effects.

Some tumour types display faster response kinetics to ICIs than others (32,33). At 30 days, we captured a snapshot of patients and their lesions at different stages of their immune response, or lack thereof. In our repeat PET imaging cohort, 63% of responders experienced tumour regression meeting criteria for PR within the 5-week timeframe between PET assessments, indicating that earlier imaging time points are warranted to capture CD8+ T-cell dynamics that may be preceding the antitumour activity resulting in lesion shrinkage in these patients. Since various tumour types were included in our study, the numbers of individual tumour types enrolled were too small to define patient subset-specific CD8+ T-cell kinetics. To fully understand and assess antitumour immunity induced by ICIs beyond what is feasible with localized tumour biopsies, it is essential to image T-cell dynamics across lesions by wholebody evaluation over time. Because 89Zr has a relatively long half-life of 78.4 h, repeated PET imaging with 89Zr-tracers ideally requires an interval of 2 weeks to avoid residual radioactivity and allow full clearance of the antibody. New small molecule tracers targeting CD8 and labelled with fluorine-18 may more readily allow sequential imaging time points, increasing the chance of capturing a more complete time-course, to elucidate spatio-temporal changes in CD8+ T-cells following initiation of immunotherapy (34). For future studies, we envision also an earlier second imaging time point, namely within two weeks after starting ICI therapy, to capture pharmacodynamic changes prior to substantial tumour shrinkage.

Several issues challenged the interpretation of CD8 imaging changes following treatment. The uptake pattern changed rather than the magnitude of uptake in some tumour lesions, probably reflecting enhanced infiltration in a larger tumour volume. We expressed specific tumour uptake as SUVmax, commonly used to measure specific uptake. However, this may not properly reflect heterogeneous uptake or a change in distribution pattern.

In addition, we detected CD8+ T-cells in areas of non-malignant inflammation, supporting

the tracer's ability to visualize inflammatory processes in any setting. Within the 30 days’ time frame of ICI treatment assessed by CD8 PET imaging, we observed 89ZED88082A PET changes during ICI treatment in a patient with Hashimoto’s thyroiditis, a disease with high lymphocyte involvement (35). Therefore, CD8 PET may identify potential irAEs if patients are scanned in the relevant timeframe. However, it should be noted that not all irAEs are driven by CD8+ T-cells, and instead may involve multifactorial aetiologies including B-cell, complement, or auto-antibody driven mechanisms (36). Thus, the potential relevance of CD8 PET in the characterization, identification, and monitoring of irAEs will require further study and is currently limited to a single anecdote.

The tracer showed an organ-specific biodistribution in normal tissues without in vitro signs of cellular tracer internalization by immune cells. Uptake in the spleen was conspicuous within the first hour postinjection, likely due to high perfusion and facile access of the tracer to high CD8 levels by littoral cells lining the red pulp sinusoids (15,18). The higher spleen 89ZED88082A SUV mean at 4 mg than at 10 mg likely reflects partial CD8 saturation at the 10 mg dose, due to containing more unlabelled anti-CD8 antibody CED88004S.

High bone marrow uptake was observed early after injection, followed by a gradual decline. The initial high signal in this densely vascularized space is likely related to perfusion, while imaging at later time points likely reflects target-mediated 89ZED88082A-binding to CD8+ T-cells, which would be expected based on its role as a primary and secondary lymphoid organ and memory CD8+ T-cell localization (37,38). Moreover, we saw tracer uptake in the small intestine, likely showing CD8+ T-cells in the gut-associated lymphoid tissue, such as the Peyer's patches within the gut mucosa (39,40). High tracer uptake in these tissues matched sites of CD8 protein expression reported in the Protein Atlas (41), although these comparisons cannot be exact due to the relatively young and healthy sources of tissues in the atlas, and the relative complexity of delivering antibody tracer to the CD8 target in living subjects. Tracer signals in liver, renal cortex, urine, and large bowel probably reflected tracer clearance and metabolism rather than target-mediated binding. The tracer was intact in the blood pool. The renal cortex showed a persistent high radioactive signal irrespective of decreasing blood pool levels. This is presumably due to renal tracer clearance followed by resorption and catabolism with residualization of intracellular charged metal chelate catabolites such as lysine-DFO-89Zrbinding proteins. This is a known phenomenon for small molecules and antibody fragments (42,43).

Serial, whole-body characterization of CD8+ T-cells has several potential applications in clinical research. One application is to more fully characterize the pre-treatment CD8+ T-cell tumour infiltration, which may function as a predictive biomarker for subsequent response

Whole-body CD8+ T-cell visualization before and during cancer immunotherapy

to a particular immunotherapy (e.g., ICI). Furthermore, serial CD8 PET imaging has the potential to characterize treatment-emergent pharmacodynamic changes following novel immunotherapies or combinations of agents, and may therefore prove useful in guiding their clinical development. 89ZED88082A-PET may also be helpful to guide tumour biopsies to improve the chance of obtaining a tumour sample with high CD8+ T-cell infiltration. Ultimately, CD8 PET has the potential to become a clinical decision support tool to individualize immunotherapeutic approaches in patients. Describing and accepting the huge spatial and temporal heterogeneity of CD8+ T-cells is critical towards a more individualized treatment approach in the future. However, the generation of much larger CD8 PET imaging data sets and correlation with clinical outcomes will be needed to assess whether CD8 PET can guide treatment decisions.

In conclusion, 89ZED88082A PET specifically visualizes CD8 in vivo, offering the opportunity to assess whole-body CD8+ T-cell distribution, not obtainable with a single-lesion biopsy. We demonstrated that CD8+ T-cell presence in tumour lesions imaged before ICI could be predictive for OS, highlighting the potential of CD8 imaging as a predictive biomarker to personalize treatment for patients. The dynamics of intra-tumoural CD8 expression during ICI exposure is more complex and nuanced than previously reported and differs between patients and lesions in the same patient. To properly evaluate tumour-immune status, timing and evaluation across lesions are crucial. Our results provide a strong rationale to characterize the tumour-immune microenvironment using novel imaging technologies.

References

1. Hodi, F.S. et al. Nivolumab plus ipilimumab or nivolumab alone versus ipilimumab alone in advanced melanoma (CheckMate 067): 4-year outcomes of a multicentre, randomised, phase 3 trial. Lancet Oncol. 19, 1480–1492 (2018).

2. Vaddepally, R.K. et al. Review of indications of FDA-approved immune checkpoint inhibitors per NCCN guidelines with the level of evidence. Cancers 12, 738 (2020).

3. Chang, E. et al. Systematic review of PD-1/PD-L1 inhibitors in oncology: from personalized medicine to public health. Oncologist 26, e1786–e1799 (2021).

4. Havel, J.J., Chowell, D. & Chan, T.A. The evolving landscape of biomarkers for checkpoint inhibitor immunotherapy. Nat. Rev. Cancer 19, 133–150 (2019).

5. Herbst, R.S. et al. Predictive correlates of response to the anti-PD-L1 antibody MPDL3280A in cancer patients. Nature 515, 563–567 (2014).

6. Lee, J.S. & Ruppin, E. Multiomics prediction of response rates to therapies to inhibit programmed cell death 1 and programmed cell death 1 ligand 1. JAMA Oncol. 5, 1614–1648 (2019).

7. Wong, P.F. et al. Multiplex quantitative analysis of tumor-infiltrating lymphocytes and immunotherapy outcome in metastatic melanoma. Clin. Cancer Res. 25, 2442–2449 (2019).

Whole-body CD8+ T-cell visualization before and during cancer immunotherapy

8. Ribas, A. et al. Intratumoral immune cell infiltrates, FoxP3, and indoleamine 2,3-dioxygenase in patients with melanoma undergoing CTLA4 blockade. Clin. Cancer Res. 15, 390–399 (2009).

9. Edwards, J. et al. CD103+ tumor-resident CD8+ T cells are associated with improved survival in immunotherapy-naïve melanoma patients and expand significantly during anti-PD-1 treatment. Clin. Cancer Res. 24, 3036–3045 (2018).

10. Chen, P.L. et al. Analysis of immune signatures in longitudinal tumor samples yields insight into biomarkers of response and mechanisms of resistance to immune checkpoint blockade. Cancer Discov. 6, 827–837 (2016).

11. Tumeh, P.C. et al. PD-1 blockade induces responses by inhibiting adaptive immune resistance. Nature 515, 568–571 (2014).

12. Litchfield, K. et al. Representative sequencing: unbiased sampling of solid tumor tissue. Cell Rep. 31, 107550 (2020).

13. iménez-Sánchez, A. et al. Heterogeneous tumor-immune microenvironments among differentially growing metastases in an ovarian cancer patient. Cell 170, 927–938.e20 (2017).

14. de Vries, E.G.E. et al. Integrating molecular nuclear imaging in clinical research to improve anticancer therapy. Nat. Rev. Clin. Oncol. 16, 241–255 (2019).

15. Bensch, F. et al. 89Zr-atezolizumab imaging as a non-invasive approach to assess clinical response to PD-L1 blockade in cancer. Nat. Med. 24, 1852–1858 (2018).

16. Kok, I.C. et al. 89Zr-pembrolizumab imaging as a non-invasive approach to assess clinical response to PD-1 blockade in cancer. Ann. Oncol. Epub ahead of print (2021).

17. Gill, H. et al. The production, quality control, and characterization of ZED8, a CD8-specific 89Zr-labeled immuno-PET clinical imaging agent. AAPS J. 22, 22 (2020).

18. Ogembo, J. G. et al. SIRPα/CD172a and FHOD1 are unique markers of littoral cells, a recently evolved major cell population of red pulp of human spleen. J. Immunol. 9, 4496–4505 (2012).

19. Li, F. et al. The association between CD8+ tumor-infiltrating lymphocytes and the clinical outcome of cancer immunotherapy: A systematic review and meta-analysis. EClinicalMedicine 41, 101134 (2021).

20. Lee, J.S. & Ruppin, E. Multiomics prediction of response rates to therapies to inhibit programmed cell death 1 and programmed cell death 1 ligand 1. JAMA Oncol. 5, 1614–1618 (2019)

21. Le, D.T. et al. PD-1 blockade in tumors with mismatch-repair deficiency. New Engl. J. Med. 372, 2509–2520 (2015).

22. Prall, F. et al. Prognostic role of CD8+ tumor-infiltrating lymphocytes in stage III colorectal cancer with and without microsatellite instability. Hum. Pathol. 35, 808–816 (2004).

23. Millen, R. et al. CD8+ tumor-infiltrating lymphocytes within the primary tumor of patients with synchronous de novo metastatic colorectal carcinoma do not track with survival. Clin. Transl. Immunol. 9, e1155 (2020).

24. Yoon, H.H. et al. Intertumoral heterogeneity of CD3+ and CD8+ T-cell densities in the microenvironment of DNA mismatch-repair–deficient colon cancers: implications for prognosis. Clin. Cancer Res. 25,125–133 (2019).

25. Narayanan, S. et al. Tumor infiltrating lymphocytes and macrophages improve survival in microsatellite unstable colorectal cancer. Sci Rep. 9, 13455 (2019).

26. Gallon, J. & Bruni, D. Approaches to treat immune hot, altered and cold tumours with combination immunotherapies. Nat. Rev. Drug Discov. 18, 197–218 (2019).

27. Farwell, M.D. et al. CD8-targeted PET imaging of tumor infiltrating T cells in patients with cancer: A phase I first-in-human study of 89Zr-Df-IAB22M2C, a radiolabeled anti-CD8 minibody. J. Nucl. Med. 63: 720–726 (2022).

28. Niemeijer, A.N. et al. Whole body PD-1 and PD-L1 positron emission tomography in patients with nonsmall-cell lung cancer. Nat. Commun. 9, 4664 (2018).

29. van de Donk, P.P. et al. Molecular imaging biomarkers for immune checkpoint inhibitor therapy. Theranostics 10, 1708–1718 (2020).

30. Ribas, A. et al. PD-1 Blockade expands intratumoral memory T cells. Cancer Immunol. Res. 3, 194–203 (2016).

31. Ortiz-Muñoz, G. et al. Surveillance of in situ tumor arrays reveals early environmental control of cancer immunity. bioRxiv 2021.05.27.445482.

32. Borcoman, E. et al. Novel patterns of response under immunotherapy. Ann. Oncol. 30, 385–396 (2019).

33. Hamid, O. et al. Five-year survival outcomes for patients with advanced melanoma treated with pembrolizumab in KEYNOTE-001. Ann. Oncol. 30, 582–588 (2019).

34. Rosenberg, A. et al. Development of a fully automated method for radiosynthesis of fluorine-18 labeled CD8 PCC radiotracers. J. Nucl. Med. 62, 1201 (2021).

35. Liblau, R.S. et al. Autoreactive CD8 T cells in organ-specific autoimmunity: Emerging targets for therapeutic intervention. Immunity, 17, 1–6 (2012).

36. Postow, M.A. et al. Immune-related adverse events associated with immune checkpoint blockade. N Engl J Med. 378, 158–168 (2018).

37. Bonomo, A. et al. A T cell view of the bone marrow. Front. Immunol. 7, 184 (2016).

38. Shin, S.S. et al. lmmunoarchitecture of normal human bone marrow: A study of frozen and fixed tissue sections. Hum. Pathol. 23, 686–694 (1992).

39. Sathaliyawala, T. et al. Distribution and compartmentalization of human circulating and tissueResident memory T cell subsets. Immunity 38, 187–197 (2013).

40. Heel, K.A. et al. Review: Peyer’s patches. J. Gastroenterol. Hepatol. 12, 122–136 (1997).

41. Uhlén, M. et al. Tissue-based map of the human proteome. Science 347, 1260419 (2015). Website accessed: https://www.proteinatlas.org/ENSG00000153563-CD8A/tissue.

42. Behr, T.M. et al. Reduction of the renal uptake of radiolabeled monoclonal antibody fragments by cationic amino acids and their derivatives. Cancer Res. 55, 3825–3834 (1995).

43. Akizawa, H. et al. Renal uptake and metabolism of radiopharmaceuticals derived from peptides and proteins. Adv. Drug Deliv. Rev. 60, 1319–1328 (2008).

Methods Study design

This single-centre imaging study comprised parts A and B. Eligible patients for part A or B1 had a histologically confirmed locally advanced or metastatic cancer, whom, in the investigator's opinion, based on available clinical data, may benefit from anti-PD-L1 antibody treatment. They were required to have disease progression during or following first-line standard-ofcare therapy. In part B2, patients with melanoma eligible for standard-of-care anti-PD-1 antibody with or without ipilimumab, could participate. Eligible patients had measurable disease according to RECIST1.1, and were amenable to a tumour biopsy. All patients were ≥ 18 years of age and had an Eastern Cooperative Oncology Group performance status of 0–1, life expectancy ≥ 12 weeks, and adequate hematologic and end-organ function. Patients with concomitant or historical conditions or medication use that could compromise their safety with 89ZED88082A/CED88004S or atezolizumab treatment, or interpretation of study results, were excluded.

The study was performed with a companion treatment study with atezolizumab for part A and B1 at the UMCG, the Netherlands (NCT02478099). All patients provided written informed consent for the imaging study and, if applicable, for the companion treatment study. The studies were performed in compliance with all relevant ethical regulations and approved by the Medical Ethical Committee of the University Medical Center Groningen and the Central Committee on Research Involving Human Subjects. The study was registered in ClinicalTrials. gov (NCT04029181). The primary objectives were feasibility, specifically safety, imaging dose and time points, pharmacokinetics, and immunogenicity. Secondary outcomes included the potential to image whole-body CD8+ T-cells, and secondarily ICI effects.

Patients received 37 MBq 89-zirconium labelled CED88004S (89ZED88082A) together with unlabelled DFO-conjugated one-armed antibody CED88004S intravenously as two consecutive bolus injections. In the dose-finding part A, patients received tracer injection before treatment with atezolizumab, consisting of a fixed dose of 37 MBq (1.2-1.5 mg) 89ZED88082A with additional unlabelled CED88004S until a total protein dose of 4 mg (n = 3) or 10 mg (n = 6) was reached. The unlabelled antibody tracer dose was varied to allow for adequate tracer blood pool availability, comparable with other studies with adequate visualization of tumour lesions (44). The first two patients at each new dose level during dose-finding would stay overnight in the hospital for safety monitoring. After tracer injection, PET scans were performed 1 h, and days 2, 4, and 7 followed by a safely accessible tumour lesion biopsy identified before the PET scan. In part B, patients received tracer injection and PET scans before and early during cycle 2 of ICI, with protein dose and PET scan schedule based on results from part A. After the baseline PET scans and the tumour biopsy, patients from parts A and B1 received 1,200

Whole-body CD8+ T-cell visualization before and during cancer immunotherapy

mg atezolizumab intravenously every 3 weeks. Patients with melanoma received standard of care immunotherapy at the treating physician's discretion. After part A was closed, part B was opened. Cohort assignment was in the order of enrolment.

89ZED88082A tracer and PET procedures

Unlabelled, DFO-conjugated one-armed antibody CED88004S was provided by Genentech Inc. It was radiolabelled with 89Zr-oxalate (89ZED88082A) at the UMCG according to good manufacturing practice guidelines, as described earlier17. Analytical methods for release testing were validated before clinical manufacturing. Based on stability testing 89ZED88082A shelf-life was defined as 96 h at 2-8 °C in the vial and an additional 4 h at room temperature in the syringe. Release specifications were met for appearance, protein, and radiochemical purity, protein and radioactive concentration, pH, endotoxin content, and immunoreactive fraction. Sterility was determined post-release. For additional information, see Supplementary Information Table S1.

PET scans were acquired in combination with low-dose computerized tomography (CT) for attenuation correction and anatomic localization, with a Biograph mCT 64-slice, Biograph mCT 40-slice, or Biograph Vision (128-slice) PET/CT camera (all Siemens). PET scan acquisition consisted of total body mode (skull – feet) up to 15 bed positions depending on the patient's length (Biograph mCTs) or total 4 passes (Vision). Repeated PET scans in cohort B were always performed on the same machine as the baseline PET scan. According to harmonization procedures, PET reconstruction was compatible with the EARL1 PET/CT accreditation and European Association of Nuclear Medicine guidelines (45). PET images were analyzed using the Accurate tool (46) as described earlier. Spherical VOIs were drawn around tumour lesions ≥ 1 cm and in organs of interest to assess the biodistribution of the tracer (Supplementary Information Fig. S1). Tumour lesions ≥ 1 cm in diameter were identified at baseline on diagnostic CT, MRI, or via clinical evaluation for (sub)cutaneous lesions, and VOIs were delineated manually for PET images analysis on tracer uptake. Tracer uptake in non-malignant lymph nodes was qualitatively assessed and quantified on the PET scan images in the cervical, axillary, and inguinal regions. Tracer uptake in Waldeyer's ring was omitted after previous tonsillectomy and/or adenoidectomy, and no visual uptake on PET. All PET scans were visually evaluated for unexpected tracer uptake.

SUV was calculated using bodyweight, net injected radioactivity dose, and radioactivity within a VOI. Tumour, lymph node, and Waldeyer's ring 89ZED88082A-uptake were analyzed as maximum SUV (SUVmax), in other organs as SUVmean. All SUVs reported are at 10 mg on day 2 postinjection unless specified otherwise.

Tumour tissue analyses

Tumour biopsies were performed within 10 days after tracer injection and within 4 days after the last PET scan. Whole tissue blocks of formalin-fixed, paraffin-embedded biopsies were analyzed with autoradiography. After that, 4 μm sections were hematoxylin and eosin stained, and CD8 was stained IHC with the mouse CD8 monoclonal antibody C4/144B (DAKO/Agilent) according to manufacturer's protocols. If no baseline biopsy was available, archival tumour tissue was studied. Tissue sections that did not contain tumour, were excluded from IHC/PET analyses.

CD8 expression was determined by a pathologist (HK) blinded for treatment outcome, and CD8+ T-cell infiltration was described as desert, stromal or inflamed phenotype (47,48). For stromal or inflamed tumour tissues, CD8+ T-cell density was assessed as 1 (minor), 2 (intermediate) or 3 (high) as a subjective estimate of the average density considering the entire tumour area to address intra-tumoural heterogeneity. Representative examples in Extended Data Fig. 4a.

Whole FFPE tumour tissue blocks were exposed 6-8 days to a multipurpose or multisensitive phosphor storage plate (PerkinElmer). Exposures were captured using a Cyclone phosphor imager. To correlate 89ZED88082A-uptake with the spatial patterning and intensity of CD8 expression, autoradiography images were scaled and registered to IHC images using manually selected control points and an affine transformation for 16 tumour slides. IHCevaluated CD8 expression was expressed as the percentage of CD8+ positive pixels across the manually defined region of interest (ROI) specific to tumour including tumour associated stroma per slide (excluding normal stroma and background tissue), thus CD8 IHC positive pixels / all pixels of the tumour tissue area. Tissue uptake of 89ZED88082A was measured as the digital autoradiograph signal for the ROI corrected by background subtraction on a per slide basis. Decay correction was applied to adjust for differences in the timing of sample scanning after injection. Slide-level analyses served to evaluate the ability of the tracer to distinguish specimens of relatively high and low CD8 expression (pixel based). For each slide, average IHC percent positivity and autoradiographic tracer intensity were computed globally as well as locally using overlapping square tiles of varying sizes (100x100px, 400x400px, 1000x1000px, ..., 8000x8000px). Only tiles with ≥ 25% overlap with tumour ROI were included. Image scaling, registration, and summarization were executed using MATLAB (Mathworks, Natick, MA, USA). A decay correction was applied to autoradiography tracer intensities to adjust for differences in the timing of sample collection after injection.

For all patients, tumour MMR proteins were determined. Tumours were considered dMMR if at least one of the following criteria was applicable (49): tumour showed loss of one or more MMR proteins MLH1, MSH2, MSH6, or PMS2, assessed by IHC; DNA analysis showed high microsatellite

Whole-body CD8+ T-cell visualization before and during cancer immunotherapy

instability; patients known with a germline mutation in MMR genes in the context of hereditary non-polyposis colorectal cancer syndrome. If unavailable at study entry, microsatellite status was assessed using IHC for MMR proteins on (archival) tumour tissue. If the result was equivocal, DNA analysis for microsatellite instability was performed.

Laboratory analyses

In part A, blood samples for pharmacokinetics were collected before injection and at 30 min, 3 h, 1 or 2 days, 4 days, and 7 days postinjection. In part B, blood samples were collected before injection, 30 minutes post-injection and at day of PET-scan, for pretreatment as wells as on-treatment PET series. Tracer levels were analyzed with a validated ELISA of serum 89ZED8802A/CED88004S, and with serum 89Zr-radioactivity measurements. Clinical samples, assay calibrators, and controls were captured on a microtiter plate using a rabbit monoclonal antibody to CED88004S. For detection, a biotin-conjugated anti-human IgG followed by a streptavidin-horseradish peroxidase incubation and a colourimetric reaction were used. The calibration curve range of this method is 149 to 2,500 ng ml-1. Half-life of 89ZED8802A/CED88004S was estimated by standard noncompartmental analysis using Phoenix WinNonlin (Certara Inc., version 6.4) and is presented as average ± standard deviation.

Serum samples, drawn before the first and second tracer injection and 30 days after the last injection, were analyzed for ADAs formation using a validated bridging ELISA assay with a relative sensitivity of 22 ng ml-1. ADA-positive subjects were defined as those who screened negative for ADAs at baseline and had ADAs following 89ZED88082A/CED88004S administration (positive in the ADA confirmatory assay).

In patients, blood was collected in sodium heparin tubes before and 2–7 days after the first tracer injection for peripheral blood lymphocyte analyses. PBMCs were isolated by Ficoll gradient centrifugation in LeucoSep-tubes (Greiner Bio-One) and resuspended in freeze medium using CTL-Cryo™ ABC Media Kit (CTL Europe GmbH). Cryovials were stored in liquid nitrogen until analysis. T-, B- and NK-cell enumeration was determined by flow cytometry.

89ZED88082A stability was studied in serum and urine collected at days 0, 4, and 7, with sodium dodecyl sulfate-polyacrylamide gel electrophoresis as described previously (50). Intact 89ZED88082A and radioactive degradation products were detected autoradiographically by exposing gels to a multipurpose phosphor plate (PerkinElmer) overnight at -20 °C. Exposures were captured using a Cyclone phosphor imager. Images were analyzed using ImageJ (version 1.52p).

For tracer CD8-receptor mediated binding and internalization analysis, PBMCs were prepared

from healthy blood donor buffy coats (Sanquin) with appropriate informed consent, by centrifugation in LeucoSep-tubes (Greiner Bio-One). Unstimulated PBMCs were diluted to 1 million cells per ml in phosphate-buffered saline containing 2% foetal calf serum (FACS buffer). CED88004S was diluted in FACS buffer to 20 μg ml-1 and incubated with the PBMCs for 1 or 2 h at 37 °C. CED88004S binding to CD8 and subsequent cellular internalization in PBMCs was determined by flow cytometry as described previously (51). Membrane-bound CED88004S was detected using an anti-human allophycocyanin-IgG F(ab')2 fragment within the total PBMC population (Extended Data Fig. 7, blue) or CD3-positive cell population (Extended Data Fig. 7, red). Samples were analyzed on a BD FACS Verse flow cytometer (BD Biosciences, Supplementary Information Fig. S2). Samples were measured in duplicate, corrected for background fluorescence and non-specific antibody binding. Data analysis was performed with FlowJo v10 (Tree Star). The presence of surface receptors was expressed as mean fluorescent intensity.

Clinical outcomes and CT analysis

Safety was assessed according to the common terminology criteria for adverse events (CTCAE) of the National Cancer Institute, v4.0. Tracer-related adverse events were collected from the first tracer injection until 30 days after the last tracer injection. For analyses of tracer uptake and immune-related toxicity due to immunotherapy, PET scans were evaluated for organs of interest in all patients who experienced immune-related adverse events grade ≥3.

Before starting therapy, all patients had a contrast-enhanced diagnostic CT-chest-abdomen and evaluation of the brain by CT or MRI. According to RECIST1.1 or iRECIST if applicable (52,53), response evaluation was performed every 6 weeks during atezolizumab treatment or 12 weeks in patients with melanoma (cohort B2). The sum of lesion diameter (SLD) is defined as the sum of the maximal diameter of target lesions, with short axis in the case of lymph nodes. Best overall response was defined as the most favourable response confirmed by a consecutive assessment. PFS and OS were determined from the first treatment dose until disease progression per RECIST1.1 or iRECIST if applicable, or death from any cause, for PFS whichever occurred first. For PFS, data from subjects without disease progression and death were censored at the date of the last tumour assessment, or, if no tumour assessments were made after the baseline visit, at the date of first treatment plus 1 day. To interpret tumour-rim uptake, tumour necrosis was defined as 10-30 Hounsfield Units in portal venous phase on CT.

Statistical analysis

We used standard descriptive statistics to describe the distribution of various characteristics, including 89ZED88082A-uptake.

Whole-body CD8+ T-cell visualization before and during cancer immunotherapy

As a general approach, the relation between 89ZED88082A-uptake in tumour lesions and in normal tissues with various determinants (time since tracer injection, protein dose level, tumour lesion organ location, MMR status, immune phenotype, best overall RECIST response, and ICI treatment status) was assessed using linear mixed models to account for repeated measurements within patients using random intercepts and, if applicable, within tumour lesions using additional random intercepts nested within patients. For tumour lesions and normal lymph nodes and tonsils, we used SUVmax as the 89ZED88082A-uptake measure, which was log-transformed in the analyses to account for its right-skewed distribution, and results were subsequently back-transformed to obtain estimates of geometric means and percent differences. The 89ZED88082A-uptake in other normal tissues was expressed as SUVmean, which was analyzed without transformation, yielding estimates of means and mean differences. To obtain 89ZED88082A-uptake estimates, we fitted the linear mixed models under restricted maximum likelihood and used Satterthwaite degrees of freedom to obtain 95% confidence intervals and Wald P-values. In addition, we obtained likelihood-ratio P-values from models fitted under maximum likelihood. A trend-test for the relation between best overall response and tumour 89ZED88082A-uptake was obtained by analyzing best overall response categories as a numerical variable (with PD, SD, PR, and CR expressed as 0, 1, 2, and 3, respectively).

More specifically, using data from study part A, post-injection time-uptake curves were fitted using post-injection imaging time-point both categorically as well as continuously, selecting the best curve-fit for the latter from a linear, a log-linear or a quadratic fit using the Akaike's Information Criterion (under maximum likelihood). Protein dose level varied in part A and was included in these models both as a main effect as well as using an interaction term with postinjection time. As the shape of the time-uptake curves did not substantially depend on protein dose level, the main results of these analyses included protein dose level as a main effect only, and the resulting estimates from these models were projected at the 10 mg protein dose level. This was the protein dose taken forward towards part B of the study. All other analyses were performed only in patients receiving a 10 mg protein dose level.

Regarding change in 89ZED88082A-uptake during ICI therapy, we defined the on-treatment measurement as the actual time between start of ICI therapy and the on-treatment 89ZED88082A PET assessment to account for variation between patients in the timing of the PET-scan (the pretreatment assessment was assumed to represent the situation at the start of ICI and therefore the time between pretreatment assessment and start ICI was set at zero days for this analysis). The results of these analyses are expressed as changes in 89ZED88082A-uptake per week of ICI therapy, also summarized as expected values at 30 days of ICI therapy which was the median timepoint across patients. To assess whether the change in 89ZED88082A-uptake depended on ICI treatment response, we used interaction terms

between treatment status and best overall response, binning patients into a PD and a non-PD group due to the limited number of patients prohibiting more detail.

For change in tumour uptake during ICI therapy, we attempted to account for possible shrinkage of individual tumour lesions during treatment leading to an underestimation of actual 89ZED88082A-uptake due to partial volume effects in a data-driven way. For this, we first assessed the relation between CT-measured tumour lesion volume (based on two orthogonal measurements assuming an oblate spheroid shape) and geometric mean 89ZED88082A-uptake in 238 lesions from 34 patients (all with 10 mg protein dose) only using the treatment-naive measurements and using a 5-knot restricted cubic spline. This showed that lesions below 2cc in volume exhibited a marked decrease in the measured geometric mean 89ZED88082A-uptake with decreasing volume (as expected), whereas for lesions between 2 and 65cc (the 95th percentile), there was no relation between volume and 89ZED88082A-uptake.

Using this observed relation between volume and geometric mean 89ZED88082A-uptake in the pretreatment data, we next expressed the observed 89ZED88082A-uptake of individual lesions as the absolute difference compared to the expected geometric mean uptake of lesions of identical volume based on the restricted cubic spline curve, both for the pretreatment as well as on-treatment measurements, and then adding to this difference between observed and expected 89ZED88082A-uptake the expected geometric mean pretreatment uptake of lesions of 5cc (arbitrarily chosen within the volume range without an observed relation with pretreatment 89ZED88082A-uptake). To account for the time-period between the on-treatment 89ZED88082A-PET scan and the first CT scan for response evaluation, we linearly interpolated the change in volume between baseline and the first on-treatment response CT-scan to obtain an expected lesion volume at the timing of the 89ZED88082A-PET. The resulting tumour volume-adjusted 89ZED88082A-uptake values can be interpreted as the absolute difference in 89ZED88082A-uptake compared to treatment-naive lesions of the same size, projected for all lesions towards a lesion volume of 5cc (i.e., resulting in an estimation of the amount of increased or decreased 89ZED88082A-uptake compared to an average lesion of 5cc). Finally, the resulting volume-adjusted 89ZED88082A-uptake variable was analyzed for its relationship with change in uptake during treatment and treatment response status in the same way as the actual measured 89ZED88082A-uptake. An important underlying assumption of the above approach is that the empirically observed relation between volume and 89ZED88082A-uptake in the pretreatment data accurately captures the true partial volume effect phenomenon. We specifically choose this approach beyond merely adjusting the analyses for estimated tumour volume directly, because of the potential mixing of on-treatment effects between volume and 89ZED88082A-uptake.

To investigate the relation between pretreatment 89ZED88082A-uptake and PFS and OS,

Whole-body CD8+ T-cell visualization before and during cancer immunotherapy

89ZED88082A-uptake was first expressed as the geometric mean SUVmax per patient and then analyzed both categorically (based on a median-split across patients) as well as continuously (expressed per population standard deviation—the entire per-patient geometric mean SUVmax distribution encompasses approximately six times this population standard deviation). Due to the small dataset, we specifically refrained from exploring potentially more optimal cutoff levels than the pre-defined median split to avoid overoptimistic results. Similarly, for the continuous analyses we assumed (log)linearity and refrained from exploring other functional forms. We used Kaplan-Meier curves and log-rank tests and obtained hazard ratios using Cox regression models with Firth's penalization to account for small sample bias. Above statistical analyses were performed using R version 4.1.1 for macOS, particularly using the lmer function for linear mixed models (lme4 1.1-27.1, lmerTest 3.1-3), coxphf for Cox models (coxphf 1.13.1), and rcspline.eval for restricted cubic splines (rms 6.2-0). All reported P values are based on twosided statistical tests without correction for multiple testing.

Slide-level correlation between autoradiography and IHC was assessed by Kendall’s rankbased correlation. Sub-slide (tile) level analyses were also performed to evaluate the ability of the tracer to identify localized regions of CD8 positivity within individual biopsies. For tile-level analyses, autoradiography images were scaled and aligned to CD8 IHC images using manually selected control points and an affine transformation. Local average autoradiography and IHC measurements for each slide were computed in overlapping tiles of varying sizes. Association between autoradiography and IHC was assessed using Kendall’s rank-based correlation within samples and after pooling across samples. Within the sample, tile-level correlations were calculated at each tile size only for samples with ≥6 tiles as the variance of estimated correlations is high at smaller sample sizes.

References methods

44. Bensch, F. et al. Comparative biodistribution analysis across four different 89Zr-monoclonal antibody tracers - The first step towards an imaging warehouse. Theranostics 8, 4295–4304 (2018).

45. Makris, N.E. et al. Multicenter harmonization of 89Zr PET/CT performance. J. Nucl. Med. 55, 264–267 (2014).

46. Boellaard, R. Quantitative oncology molecular analysis suite: ACCURATE. J. Nucl. Med. 59: Suppl. 1, 1753 (2018).

47. Hegde, P.S. et al. The where, the when, and the how of immune monitoring for cancer immunotherapies in the era of checkpoint inhibition. Clin Cancer Res. 22, 1865–1874 (2016).

48. Mariathasan, S. et al. TGFβ attenuates tumour response to PD-L1 blockade by contributing to exclusion of T cells. Nature 554, 544-548 (2018).

49. Weissman, S.M. et al. Genetic counseling considerations in the evaluation of families for Lynch syndrome - a review. J. Genet. Couns. 20, 5–19 (2011).

50. Giesen, D. et al. Probody therapeutic design of 89Zr-CX-072 promotes accumulation in PD-L1–expressing

Whole-body CD8+ T-cell visualization before and during cancer immunotherapy

tumors compared to normal murine lymphoid tissue. Clin. Cancer Res. 26, 3999–4009 (2020).

51. Kol, A. et al. ADCC responses and blocking of EGFR-mediated signaling and cell growth by combining the anti-EGFR antibodies imgatuzumab and cetuximab in NSCLC cells. Oncotarget 8, 45432–45446 (2017).

52. Eisenhauer, E.A. et al. New response evaluation criteria in solid tumours: revised RECIST guideline (version 1.1). Eur. J. Cancer. 45, 228–247 (2009).

53. Seymour, L. et al. iRECIST: guidelines for response criteria for use in trials testing immunotherapeutics. Lancet Oncol. 20, e242 (2019).

Acknowledgements

We thank R. Boellaard for assistance and support of PET analyses and L. Pot for clinical tracer production coordination. Financial supports came from the Dutch Cancer Society grant RUG 10034 Pointing; PPP-allowance of the Dutch Ministry of Economic Affairs and Climate Policy; and a research grant from Genentech made available to the institution.

Author contributions

Study design: E.G.E.V., A.U., B.M.F., S.P.W. Clinical patient coordination: L.K.R., J.S.H.N., P.P.D. Clinical investigators and supervision: S.F.O., M.J., J.A.G, D.J.A.G., E.G.E.V., M.L.H. Laboratory and tissue analyses: D.G., W.T. H.K., J.J., M.Y., P.K.K., J.R.E. Statistical analyses: S.G.E. Imaging data generation and interpretation: L.K.R., J.S.H.N., P.P.D., A.H.B., T.C.K., S.F.O., R.S.N.F., D.J.A.G., M.L.H., E.G.E.V., S.M.S.B., S.G., A.U., S.P.W., M.J.M., M.T.L. Manuscript: L.K.R. and E.G.E.V. wrote the manuscript with input, edits, and approval of all authors.

Competing interests

E.G.E.V. reports funding paid to the institution for clinical trials or contracted research from Amgen, AstraZeneca, Bayer, CytomX, Crescendo Biologics, G1 Therapeutics, GE Healthcare, Genentech, Regeneron, Roche, Servier, Synthon; fees paid to the institution for membership of the advisory board from Daiichi Sankyo, NSABP and Crescendo Biologics. S.F.O. reports research grants from Novartis, Celldex Therapeutics paid to the institution. W.T. reports fees paid to the institution for membership advisory boards from Merck Sharp Dohme, and Bristol-MyersSquibb. M.J. reports fees paid to the institution for membership of the advisory board from BMS, Merck, Novartis, Sanofi, AstaZeneca. J.A.G. reports research grants from Roche, AbbVie, Siemens, paid to the institution. S.M.S.B., M.Y., H.K., J.J., S.G., M.T.L., M.J.M., J.R.E., P.K.K., S.P.W. and A.U. are employees of Genentech Inc., member of the Roche group; M Y., S.G., M.J.M., S.P.W. and A.U. are also stockholders of Genentech, Inc/Roche. B.M.F. was employee of Genentech, Inc. and stockholder of Roche at time of work described; currently employee and stockholder of Gilead Sciences, Inc. The other authors declare no competing interests.

Extended data

EXTENDED DATA TABLE 1: Characteristics at study entry of all evaluable patients.

Total n = 38

Median age, years (range) 62 (32 80)

Gender

Female Male 20 (53%) 18 (47%)

Tumour types

dMMR colorectal 5 (13%), UCC 2 (5%), duodenal 1, pancreatic 1

Cervical carcinoma

Cutaneous SCC TNBC Cholangiocarcinoma Melanoma

Anorectal SCC

Vulvar SCC

NEC [cervical, gastric oesophageal] Oesophageal SCC

NSCLC

Hepatocellular carcinoma

Ovarian clear cell carcinoma

SCC of unknown primary

Tumour stage at study entry

9 (24%) 5 (13%) 4 (11%) 3 (8%) 3 (8%) 3 (8%) 2 (5%) 2 (5%) 2 (5%) 1 1 1 1 1

Loco regional irresectable Metastatic 3 (8%) 35 (92%)

ECOG performance status

0 1 19 (50%) 19 (50%)

Previous lines of systemic treatment in neo adjuvant or adjuvant setting

0 1 ≥ 2

Chapter 9 216 0 1 ≥ 2

Previous lines of systemic treatment in the locally advanced or metastatic setting

32 (84%) 4 (11%) 2 (5%)

29 (76%) 4 (11%) 5 (13%)

dMMR, mismatch repair deficient. ECOG, Eastern Cooperative Oncology Group. NEC, neuroendocrine carcinoma. NSCLC, non small cell lung cancer. SCC, squamous cell carcinoma. TNBC, triple negative breast cancer. UCC, urothelial cell carcinoma

Whole-body CD8+ T-cell visualization before and during cancer immunotherapy

EXTENDED DATA TABLE 2: Summary of immune-related serious adverse events (irSAEs) grade ≥3 or leading to treatment discontinuation.

Total number of patients 89ZED88082A PET study

n = 34 n = 1 n = 2

Total number of patients treated with ICI n = 39 n = 37 Treatment regimen Atezolizumab Nivolumab Nivolumab and ipilimumab

Total number of patients with irSAE grade ≥3 or leading to treatment discontinuation n = 11

Treatment related grade ≥3 AE Number of patients with events (%) Grade Treatment regimen

Time of occurrence (days)

Guillain Barré 1 (2.6) 4 atezolizumab 15

Cholangitis 1 (2.6) 3 atezolizumab 95

Pericarditis 1 (2.6) 3 atezolizumab 84

Flu like symptomsa 1 (2.6) 3 atezolizumab 15

Tubulointerstitial nephritis 1 (2.6) 3 atezolizumab 84

Infusion related reactionb 3 (7.9) 3 atezolizumab (n = 2) nivolumab + ipilimumab (n = 1) 21, 22 and 49

Colitis 3 (7.9) 3 atezolizumab 36, 38, and 335

Increased troponin T 2 (5.2) 1 nivolumab (n = 1) nivolumab + ipilimumab (n = 1) 22 and 73

Polymyalgia rheumatica 1 (2.6) 3 atezolizumab 161

Pneumonitis 1c (2.6) 5 atezolizumab 67

All adverse events were scored according to CTCAE v4.0. Only treatment related serious adverse events are listed here. Time of occurrence after initiation of ICI therapy is provided in days.

a Characterized by fever, headache and/or myalgia, which did not occur during infusion, and infection related causes were excluded by diagnostics. b Characterized by fever or chills during infusion.

217 9 c This patient experienced a grade 3 colitis after 2 cycles atezolizumab, followed by a pneumonitis after 24 days atezolizumab postponement, and died due to respiratory insufficiency.

EXTENDED DATA FIGURE 1: Biodistribution of 89ZED88082A in normal tissues. (a) Biodistribution per protein dose cohort pretreatment. Graphs show the average SUVmean with 95% confidence bands of 89ZED88082A in the blood pool and normal tissues at days 0 (1 h), 2, 4, and 7. →

→ Colours reflect the dose cohorts with 4 mg in blue (n = 3) and 10 mg in red (n = 6). Note the different scales of the y-axis. (b) Table showing average changes in tracer uptake values between pre- and ontreatment PET scans, projected at 4 weeks. Pinteraction is shown for the correlation between change and best overall response (PD vs. SD/PR/CR). ND = not determined.

(a) Whole-body maximum intensity projection (MIP) images of a patient with cervical cancer show biodistribution of 89ZED88082A over time at 10 mg tracer dose. Orange arrows indicated lung metastases; blue arrows indicate uptake in hilar lymph nodes. (b) 89ZED88082A tumour uptake pretreatment projected at 10 mg tracer dose as geometric mean SUVmax (line and dots) with 95% confidence bands for all nine patients in part A (lesions n = 70).

Whole-body CD8+ T-cell visualization before and during cancer immunotherapy

EXTENDED DATA FIGURE 2: 89ZED88082A biodistribution in time.

EXTENDED DATA FIGURE 3: 89ZED88082A/CED88004S effects in PBMCs. (a) Table shows no difference in mean ± standard deviation for T-cell, B-cell, and NK-cell counts in blood samples from patients before (day 0) and post-tracer injection (day 2), at baseline before the start of ICI. (b) CED88004S internalization experiments in PBMCs of healthy donors. CED88004S binding to CD8 and subsequent internalization was determined by flow cytometry in unstimulated PBMCs from healthy blood donor buffy coat. Membrane-bound CED88004S was detected using an anti-human allophycocyanin-IgG F(ab')2 fragment. CD8 membrane levels before incubation (T = 0 h) were set at 100%. Blue: Internalization in total PBMC population, red: Internalization in CD3 positive cells. CD8-bound CED88004S on the cell surface decreased during incubation at 37 °C (solid line), while membrane levels remained stable (dashed line).

EXTENDED DATA FIGURE 4: PET image examples of 89ZED88082A uptake in tumour lesions.

(a) Bone metastasis with high tracer uptake (SUVmax 18.9) in a patient with melanoma. (b) Uptake in a brain metastasis (SUVmax 1.6) of a patient with melanoma with corresponding MRI, whereas healthy brain showed low uptake with SUVmean 0.1. (c) High uptake in multiple cervical lymph node metastases in a patient with cutaneous squamous cell carcinoma. (d) Multiple liver metastases in a patient with ovarian clear cell carcinoma without 89ZED88082A uptake. (e) Uptake in a liver metastasis in a patient with squamous cell oesophageal cancer. (f-h) Several metastases with high rim uptake: (f) Liver metastases in a patient with dMMR colorectal cancer. (g) Bone lesion in a patient with squamous cell vulvar cancer. (h) Lung metastasis in a patient with cervical cancer.

Whole-body CD8+ T-cell visualization before and during cancer immunotherapy

EXTENDED DATA FIGURE 5: Tumour tissue IHC analyses and correlation with tracer signal. (a) Representative examples of IHC CD8 expression phenotypes: (i) Liver biopsy of a cholangiocarcinoma metastasis with a desert phenotype. (ii) A biopsy of a perivesical tumour mass of dMMR urothelial cell cancer with stromal CD8 expression phenotype [density 2]. (iii) A liver biopsy of dMMR colon carcinoma shows an inflamed phenotype [density 3]. (b) Correlation of mean CD8 staining pixel positivity and autoradiography signal across 16 samples with weighted quantile regression fit. Point sizes and regression weights are proportional to the size of each sample biopsy. (c) Using tile-based analysis, the correlation across and within samples of mean CD8 staining pixel positivity and autoradiography signal at subsample level. Crosssample correlations and corresponding 95% confidence intervals are displayed with horizontal and vertical lines at each tile size. Within-sample correlations are presented at each tile size for each sample as circles. At each tile size, only samples with ≥6 tiles are shown. Tiles containing less than <25% tissue were excluded. Only cross-sample correlations are shown at tiles sizes higher than 5000px2 as no single sample had >5 tiles. (d) Tiles of varying sizes are shown for a single representative clear cell ovarian cancer sample. →

→ (e) Violin plot of tumour-to-muscle ratio with desert (n = 15) and non-desert (n = 19) immune phenotype before and during treatment in 24 patients. Bottom and top 1% truncated, coloured dots are the ratios per lesion, black vertical lines show 95% CI; white dots within black lines and values below the violin plot the actual geometric means.

EXTENDED DATA FIGURE 6: 89ZED88082A tumour uptake in relation to response. (a) Relationship between pretreatment 89ZED88082A-uptake and best overall response; red dots show geometric mean SUVmax per patient. Violin plot areas show actual distribution of SUVmax at the metastasis level per category, black vertical lines are 95% CIs of geometric mean SUVmax (PD: 149 lesions in 19 patients; SD: 6 lesions in two patients, PR: 41 lesions in eight patients; CR: 36 lesions in four patients). (b) Relationship between pretreatment 89ZED88082A-uptake in patients with progressive disease and those that did not progress. (c-d), Changes in tumour lesion SUVmax between the pre- and on-treatment PET scans. Patients are grouped per best overall response (PD, or no-PD). Violin plots show actual distribution →

Whole-body CD8+ T-cell visualization before and during cancer immunotherapy

→ of individual lesions. Baseline (BL) to response scan 1 (RS1, on-treatment) trajectories of individual lesions are shown with grey lines, projected at 30 days, when PET scan was regularly performed; red lines and dots (geometric means) present per-patient aggregated data; white dots are the overall geometric means with black 95% CI bars. (c) Compared to pre-treatment, patients with SD, PR, or CR show a lowered uptake on the on-treatment PET scan than those with PD (Pinteraction = 0.018). (d) Same change in SUVmax, projected at 5 cc tumour volume to adjust for volume changes (Pinteraction = 0.71).

EXTENDED DATA FIGURE 7: PET image examples of 89ZED88082A uptake in non-malignant sites.

EXTENDED DATA FIGURE 7: Continued.

Axial (a-d, i), coronal (e, f), and sagittal (g, h) views of 89ZED88082A PET scans with low dose CT. (a) High uptake in urinary bladder diverticulum with urolithiasis and accompanying inflammation. (b) Uptake in an aortic atherosclerotic plaque, also detectable on non-attenuated corrected images (data not shown). (c) Uptake post-surgery in a patient after inguinal lymph node dissection with a seroma. (d) Bilateral uptake in subcutaneous inflammation reaction on heparin injections. (e) Hashimoto's thyroiditis with high uptake before treatment and (f) Increased uptake during treatment experiencing a flare-up. (g-h) Two patients who received radiotherapy to the spine; arrows indicate the border of the radiation field; the insert shows the radiation field. The irradiated bone marrow in the spine, shows less uptake than non-irradiated bone marrow. (g) Patient who received 5x4 Gy to the spine for painful bone metastases 3 months before 89ZED88082A PET scan. (h) Patient who received 25x1.8 Gy on para-aortal lymph nodes alongside the spine 12 months before the 89ZED88082A PET scan. (i) Uptake of tracer in normal appendix on PET (right), with corresponding diagnostic CT (left).

Whole-body CD8+ T-cell visualization before and during cancer immunotherapy

EXTENDED DATA FIGURE 8: 89ZED880082A/CED88004S pharmacokinetics and integrity. (a-b) Time-concentration profiles (mean ± standard deviation) of 89ZED88082A/CED88004S serum protein and radioactivity following a single IV infusion of (a) 4 mg (n = 3) and (b) 10 mg (n = 6). (c) Timeconcentration profile (mean ± standard deviation) of 89ZED88082A/CED88004S pretreatment and during treatment in the presence of immune checkpoint inhibitor antibodies. (d) Representative integrity assay (SDS-PAGE combined with autoradiography) of 89ZED880082A in urine and serum in samples drawn after 30 min and 2, 4, and 7 days after tracer injection (n = 3).

Supplementary information

SUPPLEMENTARY INFORMATION FIGURE S1: Flow charts indicate the number of tumour tissues available for analysis at baseline and during treatment. IHC, immunohistochemistry. VOI, volume of interest.

Whole-body CD8+ T-cell visualization before and during cancer immunotherapy

SUPPLEMENTARY INFORMATION FIGURE S2: Flow cytometry gating strategy used in internalization experiments, supplementary to Extended Data Fig. 3. Peripheral blood mononuclear cells (PBMCs) were gated in a forward scatter (FSC) versus side scatter (SSC) dot plot. Lymphocytes were gated in an FSC versus SSC dot plot, doublets were excluded by plotting FSC height (FSC-H) versus area (FSC-A). CD3 positive T-cells were gated on the anti-human peridinin chlorophyll protein complex-cyanine5.5 (PerCP/Cy5.5)-CD3 staining. Binding/internalization of CED88004S was detected using an anti-human allophycocyanin-IgG F(ab')2 fragment within the total PBMC population (blue) or CD3-positive cell population (red). At least 10,000 events were measured within the CD3-positive cell population. CED88004S membrane binding is expressed as mean fluorescent intensity (MFI), and no cell sorting was applied. Samples were measured in duplicate and corrected for background fluorescence and non-specific antibody binding.

SUPPLEMENTARY INFORMATION TABLE S1: Release specifications of 89ZED88082A. Test Method Specification

Appearance

Visual inspection Colourless to light yellow pH Ph. Eur. 5.2 5.8

Filter integrity

Strength

Bubble point test < 20%

Radioactive concentration 8.88 13.32 MBq mL 1

Purity SE UPLC

RCP, main peak: ≥ 85.0% RCP, radioactive HMWF: ≤ 10.0% RCP, radioactive LMWF: ≤ 10.0%

Protein purity main peak: ≥ 95.0%

Protein concentration

UV VIS 0.30 0.50 mg mL 1

Immunoreactive fraction SE UPLC ≥ 70%

Bacterial endotoxins

Sterilitya

Endosafe < 2.5 EU mL 1

Sterility test No growth

aPost release test; Results 14 days after inoculation. EU mL 1: Endotoxin units per millilitre, HMWF: High molecular weight fraction, LMWF: Low molecular weight fraction, Ph. Eur.: European Pharmacopeia, RCP: Radiochemical purity, SE UPLC: Size exclusion ultra performance liquid chromatography.

Summary and future perspectives

Summary

Cancer immunotherapy increased survival of patients with advanced stages of several tumor types, although not all patients respond. Several immune checkpoint inhibitors targeting cytotoxic T-lymphocyte-associated protein-4 (CTLA-4), programmed cell death protein (PD-1)/programmed death-ligand 1 (PD-L1) and lymphocyte activation gene-3 (LAG-3) have been approved for the treatment of patients with various tumors. Furthermore, many novel immunotherapeutic agents are being developed. These agents have very different molecular structures and characteristics which may influence their pharmacokinetics. Immune targets are expressed in tumor tissues and by various components of the immune system. Therefore, the whole-body distribution of novel cancer immunotherapies is often unknown. Information on whether these therapies reach the tumor in a sufficient dose may be a biomarker of response. For this reason, there is growing interest in understanding the in vivo behavior of immunotherapeutic agents.

PD-L1 expression, microsatellite-instability (MSI)/defective mismatch repair (dMMR), and tumor mutational burden (TMB) are clinically used biomarkers. However, the immune response to checkpoint inhibition is highly dynamic and complex, and currently there is no biomarker available that predicts with high certainty the effectiveness of immune checkpoint inhibitor therapy. This may be partly explained by the fact that treatment decisions are often made using information obtained from immunohistochemistry analysis in a single tumor biopsy. Heterogeneity in target expression and drug uptake between tumor lesions within one patient are not considered, hampering the selection of patients that may benefit from immunotherapies. Here, we demonstrate how non-invasive molecular imaging of immunotherapeutic drugs radiolabeled with a suitable positron emission tomography (PET) isotope such as zirconium-89 (89Zr) may help. PET imaging using 89Zr-radiopharmaceuticals can provide insight into tumor target heterogeneity and whether tumors are reached in sufficient dose. Therefore, the research performed in this thesis aimed to develop 89Zrradiopharmaceuticals to advance the development of novel cancer immunotherapies and explore their use as a biomarker of response. Chapter 1 introduces the studies described in this thesis.

Chapter 2 comprises a literature review exploring the current clinical use of radiopharmaceuticals based on antibodies. We identified 58 ongoing clinical trials that studied radiolabeled antibodies or antibody-based constructs, including full-sized monoclonal antibodies and antibody fragments such as nanobodies, minibodies, Fab-fragments, and bispecific T cellengaging antibodies (BiTEs). A range of radioisotopes is used to study whole-body distribution and tumor-targeting. 89Zr was the most frequently applied positron emitter for radiolabeling of antibodies. This may be due to its availability, radioactive half-life that matches the time most

Summary and future perspectives

antibodies need to reach the tumor, and feasibility for production and radiolabeling under good manufacturing practice (GMP) conditions. Most clinical trials in this field are conducted in relatively small groups of patients within one center. This precludes translation of results from clinical trials to daily practice. We found that 11 out of 26 antibodies or antibody-related drugs radiolabeled with PET isotopes were investigated in the multicenter setting. Larger studies will require the harmonization of radiolabeling and imaging procedures across centers. To facilitate this, manufacturing procedures for several 89Zr-labeled antibodies were validated to obtain GMP grade 89Zr-radiopharmaceuticals for administration to patients in chapters 3, 4, 6, 8 and 9

Radioimmunotherapeutic (RIT) agents are antibody-based radiopharmaceuticals loaded with a therapeutic α- or β -emitting radionuclide to selectively eradicate tumors. The RIT agent lutetium-177 (177Lu)-labeled NNV003 consists of an antibody that targets the CD37 receptor expressed on B cells in non-Hodgkin’s lymphoma (NHL) patients. In chapter 3, we developed 89Zr-NNV003 and investigated whether 89Zr-NNV003 PET imaging of tumor-bearing mice can predict whole-body distribution of 177Lu-NNV003 radioimmunotherapy. PET imaging revealed 89Zr-NNV003 accumulation in REC1 human B cell NHL tumors over time. This was not observed for indium-111 (111In)-labeled IgG control molecule, indicating that the observed tumor uptake is specific for CD37. Also, ex vivo quantification showed a 2.8-fold higher tumor uptake and 4.8fold higher tumor-to-blood ratio for 89Zr-NNV003 than for 111In-IgG. In RAMOS human Burkitt’s lymphoma tumor-bearing mice, we found higher 89Zr-NNV003 tumor uptake compared with 111In-IgG tumor uptake at 10 μg, 25 μg and 100 μg protein doses. In the RAMOS tumor model, 89Zr-NNV003 and 177Lu-NNV003 showed similar ex vivo normal organ and tumor uptake. 89ZrNNV003 imaging can therefore help to predict the whole-body distribution of 177Lu-NNV003 accurately. This data may enable clinical evaluation of 89Zr-NNV003 PET imaging as a tool to select patients eligible for 177Lu-NNV003 radioimmunotherapy.

Visualizing the in vivo behavior of immune checkpoint-targeting antibodies may help to better predict response to immune checkpoint therapy. In chapter 4, we radiolabeled PD-1-targeting antibody pembrolizumab with 89Zr. This allowed us to study its whole-body distribution with PET imaging in humanized mice bearing human A375M melanoma tumors. PET imaging and ex vivo analysis showed high 89Zr-pembrolizumab uptake in murine tissues containing human immune cells, including spleen, lymph nodes and bone marrow. This uptake was reduced 6.0fold in the spleen by supplementation with unlabeled pembrolizumab, indicating saturation of PD-1 receptors in these tissues and specific uptake. We found modest 89Zr-pembrolizumab tumor uptake: Lower than uptake in lymphoid tissues, but higher than uptake in other organs. 89Zr-pembrolizumab in the blood pool was increased by unlabeled pembrolizumab, while tumor uptake was not affected. Similarly, autoradiography revealed blockade of signal by unlabeled

pembrolizumab in the spleen, but not in the tumor. Accordingly, immunohistochemistry showed PD-1 positive cells present in the spleen, while tumor tissue did not show PD-1 expression. In conclusion, 89Zr-pembrolizumab PET imaging captured PD-1-mediated uptake in both tumor and normal tissues.

In chapter 5, we evaluated 89Zr-pembrolizumab imaging as a non-invasive approach to assess tumor response to PD-1 blockade in patients with melanoma and non-small cell lung cancer (NSCLC). The optimal protein dose for PET imaging was 5 mg 89Zr-pembrolizumab, and the optimal timepoint for scanning was day 7. We found that tumor 89Zr-pembrolizumab uptake correlated with tumor response to anti-PD-1 antibody treatment, progression-free, and overall survival. Tumor maximum standardized uptake value (SUVmax) was similar in melanoma and NSCLC patients. PET imaging revealed 89Zr-pembrolizumab uptake at 5 mg was highest in the spleen, and there was 89Zr-pembrolizumab uptake in Waldeyer’s ring, normal lymph nodes, and at sites of inflammation. This study shows that 89Zr-pembrolizumab PET imaging in patients with metastatic melanoma and NSCLC is safe and feasible, but validation in larger studies is required to prove its impact on patient selection for anti-PD-1 therapy.

Combining immune checkpoint inhibitors improves the survival of patients with advanced stages of several tumor types, but can increase immune-related adverse events. The CX072 Probody targets PD-L1 and is activated in vivo by proteases specifically present in the tumor microenvironment, thereby potentially reducing PD-L1-mediated toxicities in normal tissues. In chapter 6, we developed 89Zr-CX-072 and investigated its conditional activation and whole-body distribution by PET imaging. 89Zr-CX-072 showed accumulation human MDAMB-231 triple-negative breast cancer (TNBC) tumors of immune-compromised mice and 2.1fold higher tumor-to-blood ratios than 89Zr-labeled non-binding control Probody. Tumor tissue autoradiography revealed high 89Zr-CX-072 uptake in high PD-L1-expressing regions. Activated CX-072 species were detected in these tumors, with 5.3-fold lower levels found in the spleen. 89Zr-CX-072 uptake by lymphoid tissues of immune-competent mice bearing syngeneic MC38 colon carcinomas was limited compared to 89Zr-labeled normal anti-PD-L1 antibody (not protease-activatable). This preclinical data supports the idea that CX-072 accumulates specifically in PD-L1-expressing tumors.

Next, we investigated whether the tumor-specific activation we observed in mouse models is translatable to patients. In chapter 7, we describe a first-in-human study with PET imaging to evaluate 89Zr-CX072’s whole-body distribution and tumor uptake. A 10 mg protein dose resulted in sufficient 89Zr-CX072 blood pool levels and was therefore considered most optimal. PET imaging showed tumor uptake was present in all patients, and the highest tumor uptake was found on day 7. After treatment with CX-072, one patient experienced stable disease,

Summary and future perspectives

and two patients had a partial response. PD-L1 tumor expression levels measured by IHC were high (90%) in one patient and low (≤1%) in the other patients. Highest normal tissue uptake was found in the spleen, although less than the unconditionally activated anti-PD-L1 antibody 89Zr-atezolizumab. A few patients demonstrated uptake in normal lymph nodes (axillary: 57.1%, inguinal: 62.5%) and Waldeyer's ring (50%). Furthermore, 89Zr-CX072 was intact (inactivated) in serum and plasma. These findings demonstrate evident 89Zr-CX-072 tumor uptake, even in lesions with ≤1% PD-L1 expression, and modest uptake in normal lymphoid organs, with no unexpected uptake in other healthy tissues.

In addition to predicting response, PET imaging can be used to study the effect of different binding affinities on distribution to tumors and immune tissues of a bispecific antibody. In chapter 8, we evaluated the in vivo distribution of 89Zr-labeled bispecific antibody ERY974, which simultaneously targets CD3 on T cells and glypican 3 (GPC3) on tumor cells. In immunodeficient mice bearing GPC3-expressing tumors, 89Zr-ERY974 tumor uptake was GPC3dependent and specific over 89Zr-labeled control antibodies, bispecific antibody targeting CD3 and the non-mammalian protein keyhole limpet hemocyanin (KLH) or KLH-targeting antibody.

In humanized mice, 89Zr-ERY974 tumor uptake was ~3.5-fold higher than in immunodeficient mice, suggesting that human immune cells facilitate 89Zr-ERY974 tumor uptake. We found a preferential distribution of 89Zr-ERY974 to tumor areas containing CD3-expressing T cells by using autoradiography. Additionally, high 89Zr-ERY974 uptake was observed in spleen and lymph nodes. We concluded that 89Zr-ERY974 can be used to study ERY974’s whole-body distribution in patients to support its clinical development.

Furthermore, PET imaging can provide information on tumor immune status. For example, the amount of tumor infiltrating CD8+ T cells due to immune checkpoint inhibition may be a predictive factor of response. In chapter 9, we performed a first-in-human PET imaging study with 89ZED88082A to visualize CD8+ T cells in patients with solid tumors. With an optimal 89ZED88082A protein dose of 10 mg, spleen uptake was observed within 1 h. On day 2, there was 89ZED88082A uptake in all normal lymphoid tissues and tumor lesions across the body, which was variable between and within patients. We found that 89ZED88082A tumor uptake was predictive for progressive disease versus no progressive disease and overall survival. Also, SUVmax was higher in dMMR tumors and in lesions with an immunohistochemical stromal/ inflamed phenotype. Furthermore, tumor autoradiography correlated with CD8 expression. After 2 cycles of atezolizumab treatment, PET imaging showed no change in the geometric mean of 89ZED88082A tumor uptake, but revealed temporal heterogeneity in individual lesions, independent of tumor response. This study showed an excellent correlation between 89ZED88082A tumor uptake and CD8 expression upfront PD-L1 immune checkpoint inhibitor therapy, and suggests large heterogeneity in CD8 presence during treatment.

Future perspectives

Preclinical development of 89Zr-radiopharmaceuticals

Studying the in vivo behavior in a mouse model may be informative when developing a radiopharmaceutical that can predict drug distribution in patients. A preclinical study allows for the use of appropriate control molecules, frequent blood sampling, serial PET scans, precise measurements of ex vivo uptake per organ and a controlled environment. In this thesis (chapters 5 and 7) we showed for two 89Zr-radiopharmaceuticals, 89Zr-pembrolizumab and 89Zr-CX-072 respectively, how clinical study design might benefit from observations in mouse models.

It is challenging to select a mouse model in which, besides tumor targeting, immune cells can be studied. The translatability of results from mouse studies will increase if immunotherapeutic antibodies are evaluated that are cross-reactive with mouse and human targets. However, most immunotherapeutic antibodies are specific to human targets. The use of surrogate molecules reactive with murine target can help to overcome this issue. Alternatively, immunecompromised mice can be reconstituted with human immune cells by engraftment with human peripheral blood mononuclear cells (PBMCs) or human CD34+ hematopoietic stem cells (HSCs). In this thesis (chapters 4 and 8), we demonstrated the utility of humanized mice to study the whole-body distribution of immune-targeting antibodies. While this model likely reflects the presence of multiple hematopoietic cell lineages, homing of these cells to immune tissues is hampered. The development of customized CRISPR/Cas9-based mouse models that allow for knock-in or knock-out of specific immune target(s) may support the creation of a more suitable mouse model and advance the clinical translation of results obtained in such a model.

Radiopharmaceuticals to study drug behavior

Cancer immunotherapies may be more effective when multiple immune targets are combined. Therefore, antibody-based constructs in development are getting more complex. For example, next to full-sized monospecific antibodies, multispecific antibodies such as bispecific antibodies are gaining interest. These include bifunctional (bivalent, trivalent or tetravalent) and trifunctional antibodies, and trispecific antibodies. In this thesis (chapter 8), we demonstrated how PET imaging could provide insight into the in vivo behavior of 89ZrERY974, a bispecific antibody simultaneously engaging T cells and tumor cells with different binding affinities. Furthermore, we used PET imaging to reveal how the PD-L1 targeting prodrug CX-072’s whole-body distribution is affected by its in vivo activation (chapter 6 and 7).

For these engineered biomolecules, in vivo behavior and pharmacokinetics are not readily predicted based on molecular structure. Our studies showed that whole-body distribution

Summary and future perspectives

of immune-targeting antibodies do not just depend on target expression in the tumor, but is primarily determined by their presence in healthy organs of the immune system. In addition, target affinity, Fc receptor affinity, target internalization after antibody-binding, and molecular weight are all considered determinants of whole-body distribution for a specific immunotherapeutic drug. In this setting, PET imaging may be a valuable tool to support immunotherapeutic drug development.

Radiopharmaceuticals to monitor response to immunotherapy

The development of novel immunotherapeutic drugs is costly, and many patients have to participate in clinical studies. Furthermore, not all patients respond to immunotherapy, while they are all at risk for developing adverse events. Therefore, strategies to optimally select those patients most likely to benefit from immunotherapy are urgently needed. PET imaging of radiolabeled immunotherapeutic drugs can help visualize drug distribution throughout the body and inform on whether it reaches the tumor. Earlier, 89Zr-atezolizumab imaging was able to predict tumor response, progression-free survival and overall survival in a small clinical study. We found that 89Zr-pembrolizumab imaging also predicted response to PD-1 blockade and survival in a small number of patients, as described in chapter 5. Larger studies are required to obtain the definitive impact of this approach.

The presence of T cells and potentially other immune cells in the tumor may be a biomarker of response. We have limited tools available for monitoring the immune status of metastatic cancers, and current methods include blood- or biopsy-based measurements of T cells. Still, these techniques do not reflect the dynamic and spatial information that is required to monitor immune responses to therapeutic intervention. In this thesis (chapter 9), we non-invasively visualized CD8+ T cells in cancer patients with one-armed antibody 89ZED88082A using PET imaging. 89ZED88082A tumor uptake before anti-PD-L1 antibody treatment correlated with disease outcome (progressive vs non-progressive) and overall survival. This indicates that PET imaging is a suitable strategy for visualizing tumor immune status and may be used for patient selection.

Immunotherapeutic drugs targeting T cells are gaining interest. They include T cellexpressed inhibitory receptors: T cell immunoglobulin and mucin domain-3 (TIM-3) and T cell immunoglobulin and ITIM domain (TIGIT), as well as inhibitory ligands in the B7 family: B7H3, B7-H4 and B7-H5. Furthermore, immune cell-based therapies such as chimeric antigen receptor (CAR) T cells are being evaluated in clinical trials. Knowledge of both target and drug distribution in tumors and normal tissues will increase our understanding of the antitumor immune response. Compiling this data in a warehouse of clinical trials results may help to select the most optimal strategy for patient stratification and prediction of treatment response

to immunotherapy.

Implementation of 89Zr-radiopharmaceuticals in clinical practice

To enable the administration of 89Zr-radiopharmaceuticals to patients, a GMP-compliant manufacturing process needs to be developed. This process must be robust and reproducible, and quality control of the radiolabeled product must comply with a predefined list of specifications. In this thesis (chapters 5, 7 and 9) we showed that insight into the whole-body distribution of a new specialized designed medicine can be obtained already in a small number of patients. However, validation in larger studies is often required to prove the predictive value of PET imaging for patient selection. To achieve this, radiolabeling and imaging procedures must be standardized across centers. Sharing GMP-compliant manufacturing procedures for 89Zr-radiopharmaceuticals will support clinical implementation. Also, setting up the transport of 89Zr-radiopharmaceuticals to other centers is essential.

There is a growing need for patient selection, and 89Zr-radiopharmaceuticals can potentially serve as a companion diagnostic to guide the administration of cancer immunotherapies to patients and help with clinical decision making. Knowledge from multiple disciplines, including medical specialists, pharmacists, nuclear physicians, and radiochemists, needs to be combined to organize the development of 89Zr-radiopharmaceuticals from bench to bedside. Technological advances may allow broad application of 89Zr-radiopharmaceuticals. For example, a whole-body PET scanner provides images with higher resolution and requires a lower radioactive dose, enabling potential use in the detection of small lesions, long-term response monitoring and follow-up, and administration to children.

Nederlandse samenvatting

Immuuntherapie heeft de overleving van patiënten met vergevorderde stadia van verschillende soorten kanker enorm verbeterd. Helaas responderen niet alle patiënten op behandeling met deze vorm van therapie. Immuuntherapeutische geneesmiddelen hebben met elkaar gemeen dat ze het eigen immuunsysteem stimuleren. Met name T-cellen spelen een essentiële rol bij een tumor-specifieke immuunrespons. Tumorcellen kunnen het immuunsysteem ontwijken via het inschakelen van zogenaamde immuuncheckpoints. Een behandeling met immuuncheckpoint-remmers blokkeert deze immuuncheckpoints, waardoor immuuncellen de tumor weer kunnen herkennen en opruimen.

Meerdere immuuncheckpoint-remmers zijn inmiddels goedgekeurd door de European Medicines Agency (EMA) en de Food and Drug Administration (FDA) voor de behandeling van verschillende typen kanker, namelijk antilichamen gericht tegen cytotoxisch T-lymfocytgeassocieerd antigeen-4 (CTLA-4), geprogrammeerde celdood-1 (PD-1), geprogrammeerde celdood-ligand 1 (PD-L1) en lymfocyt-activatie gen-3 (LAG-3) immuuncheckpoints. Met deze antilichamen en nieuwe immuuncheckpoint-remmers, zoals antilichamen gericht tegen en T-cel immunoglobuline en ITIM-domein (TIGIT), worden momenteel klinische trials verricht. Daarnaast is een groot aantal immuuntherapeutische geneesmiddelen met veelal innovatieve werkingsmechanismen in ontwikkeling. Antilichamen gericht tegen receptoren op T-cellen krijgen hierbij veel aandacht, waaronder T-cel immunoglobuline en mucine domein-3 (TIM-3) en remmende liganden uit de B7 family: B7-H3, B7-H4 en B7-H5.

In de kliniek bestaat de behoefte om farmacokinetiek van nieuwe immuuntherapeutische geneesmiddelen inzichtelijker te maken en meer kennis te hebben over welke factoren een rol spelen bij het ontwikkelen van een tumorrespons. Expressie van PD-L1 in de tumor en microsatelliet-instabiliteit (MSI)/DNA mismatch reparatie deficiëntie (dMMR) en het aantal mutaties (TMB) in tumor DNA zijn biomarkers die toegepast worden om patiënten met een bepaald type tumor te selecteren voor behandeling met een immuuncheckpointremmer. Echter is de tumor-specifieke immuunrespons die volgt na blokkade van een immuuncheckpoint zeer dynamisch en complex, waardoor de huidige biomarkers geen precieze voorspelling geven. Vaak is er sprake van heterogeniteit in target expressie tussen de verschillende tumor laesies van één patiënt. Daarnaast ontbreekt voor de meeste nieuwe immuuntherapeutische geneesmiddelen kennis over hoe ze zich over het menselijk lichaam verdelen, of ze in voldoende mate de tumor bereiken en of ze worden opgenomen door gezonde weefsels. Al deze informatie kan niet worden verkregen door middel van analyse in een enkel tumor biopt.

Het doel van de studies die in dit proefschrift worden beschreven is onderzoeken of moleculaire beeldvorming met positron emissie tomografie (PET) hierbij op twee mogelijke

manieren kan ondersteunen, namelijk 1) bij de ontwikkeling van nieuwe geneesmiddelen en 2) bij het optimaliseren van behandeling met immuuntherapie. Door een immuuntherapeutisch geneesmiddel radioactief te labelen (radiolabelen) met een geschikt isotoop, zoals zirconium-89 (89Zr), kunnen tumoropname en verdeling van het geneesmiddel over het gehele lichaam niet-invasief in beeld worden gebracht. In dit proefschrift beschrijven we de ontwikkeling van 89Zr-gelabelde antilichamen, ook wel 89Zr-radiofarmaca, evalueren we de lichaamsverdeling en bestuderen we of PET beeldvorming met deze 89Zr-radiofarmaca een effectieve biomarker kan zijn voor respons. Figuur 1 geeft een overzicht van de 89Zr-radiofarmaca die worden beschreven in dit proefschrift. Hoofdstuk 1 introduceert de verschillende studies die zijn uitgevoerd.

Hoofdstuk 2 beschrijft een literatuuronderzoek waarbij we de huidige klinische toepassing van radiofarmaca die zijn gebaseerd op antilichamen of antilichaamconstructen uiteen hebben gezet. Er werden 58 lopende klinische trials geïdentificeerd. Hiertoe behoorden zowel intacte monoklonale antilichamen als antilichaamfragmenten, zoals nanobodies, minibodies, Fabfragmenten en bispecifieke T-cel-bindende antilichamen (BiTEs). Daarbij werd een reeks aan radioisotopen gebruikt om tumoropname en lichaamsverdeling te bestuderen. 89Zr bleek het meest frequent gebruikte isotoop te zijn voor de radiolabeling van antilichamen. Dit komt waarschijnlijk door de beschikbaarheid en fysische halfwaardetijd van 89Zr, welke overeenkomt met de tijd die antilichamen nodig hebben om te tumor te bereiken. Daarnaast is 89Zr-radiolabeling van antilichamen relatief eenvoudig en onder ‘goede manier van produceren’ (GMP) condities te organiseren. Dit maakt toepassing van 89Zr-radiofarmaca voor PET beeldvorming in patiënten mogelijk.

Naast een aantal radioisotopen dat met name geschikt is voor diagnostiek met behulp van moleculaire beeldvorming, bestaat er een aantal isotopen dat kan worden ingezet als therapie. Lutetium-177 (177Lu)-gelabeld NNV003 is een zogenaamd radio-immuuntherapeutisch antilichaam gericht tegen de CD37 receptor op B-cellen en is gekoppeld aan 177Lu. Dit geneesmiddel heeft als doel om, na binding aan CD37 in B-cel non-Hodgkinlymfoom (NHL), tumorcellen selectief te elimineren door middel van β -straling. In hoofdstuk 3 ontwikkelden

we een 89Zr-gelabelde variant van dit antilichaam en onderzochten we of 89Zr-NNV003 PET in tumor-dragende muizen de lichaamsverdeling van 177Lu-NNV003 therapie op een juiste manier reflecteert. 89Zr-NNV003 liet accumulatie zien in humaan REC1 B-cel NHL en het indium-111 (111In)gelabelde IgG controle antilichaam niet, wat betekent dat 89Zr-NNV003 tumoropname specifiek is voor CD37. Ook toonde ex vivo kwantificering een 2,8 keer hogere tumoropname en 4,8 keer hogere tumor-tot-bloed ratio aan voor 89Zr-NNV003 dan voor 111In-IgG. In muizen geïnoculeerd met humaan RAMOS Burkitt lymfoom vonden we tevens een hogere tumoropname voor 89ZrNNV003 dan voor 111In-IgG bij drie verschillende antilichaam doseringen (10 μg, 25 μg en 100 μg).

FIGUUR 1: NNV003, pembrolizumab, CX-072, ERY974 en CED88004S worden radioactief gelabeld met 89Zr om zo meer inzicht te krijgen in de verdeling over het gehele lichaam, opname in de tumor en tumorrespons na behandeling met immuuncheckpoint-remmers.

1: 89Zr-NNV003 antilichaam gericht tegen de CD37 receptor op normale B cellen en B cel tumoren, 2: 89ZrCX-072 Probody kan na activatie door proteases binden aan het PD-L1 immuuncheckpoint op tumorcellen, NK-cellen en macrofagen, 3: 89Zr-pembrolizumab gericht tegen het PD-1 immuuncheckpoint dat tot expressie komt op T-cellen en in mindere mate op NK-cellen, 4: 89Zr-ERY974 bispecifiek antilichaam kan gelijktijdig binden aan zowel de glypican-3 receptor op tumorcellen als de CD3 receptor op T-cellen, 5: 89ZED88082A gericht tegen de CD8 receptor op CD8 positieve T-cellen.

In het RAMOS-tumormodel lieten 89Zr-NNV003 en 177Lu-NNV003 vergelijkbare tumoropname en verdeling over normale organen zien. 89Zr-NNV003 PET zou daarom potentieel kunnen worden ingezet om de lichaamsverdeling van 177Lu-NNV003 nauwkeurig te voorspellen in patiënten.

Door de lichaamsverdeling van immuuncheckpoint-remmers in beeld te brengen, wordt mogelijk meer inzicht verkregen in de tumorrespons van patiënten die worden behandeld met deze immuuntherapeutische geneesmiddelen. In hoofdstuk 4 hebben we pembrolizumab, een anti-PD-1 antilichaam, radioactief gelabeld met 89Zr. In gehumaniseerde muizen met

humane A375M melanomen vonden we met PET hoge opname van 89Zr-pembrolizumab in muizenweefsels die humane immuuncellen bevatten, waaronder de milt, lymfeklieren en beenmerg. Opname in de milt nam 6,0-voudig af na suppletie met ongelabeld pembrolizumab, wat wijst op verzadiging van PD-1 receptoren en specifieke opname. 89Zr-pembrolizumab tumoropname was lager dan in milt, lymfeklieren en beenmerg, maar hoger dan in andere organen. De 89Zr-pembrolizumab concentratie in het bloed werd hoger door suppletie met ongelabeld pembrolizumab, waarbij tumoropname niet werd beïnvloed. Dit werd bevestigd met autoradiografie op ex vivo weefsel, waarbij het signaal in de milt, maar niet in de tumor, werd geblokkeerd door ongelabeld pembrolizumab. Ook immunohistochemische analyse toonde aan dat er PD-1 positieve immuuncellen aanwezig waren in de milt, terwijl het tumorweefsel geen PD-1 expressie vertoonde. Uit deze data kunnen we concluderen dat 89Zrpembrolizumab PET in gehumaniseerde muizen zowel PD-1 gemedieerde tumoropname als opname in normale weefsels in beeld brengt.

Daarnaast hebben we in hoofdstuk 5 89Zr-pembrolizumab PET beeldvorming verricht in patiënten met melanoom en niet-kleincellig longcarcinoom. De optimale antilichaam dosis was 5 mg 89Zr-pembrolizumab en het optimale tijdstip voor scannen was dag 7. Tumoropname uitgedrukt als de maximale gestandaardiseerde opname waarde (SUVmax) was vergelijkbaar in patiënten met melanoom en niet-kleincellig longcarcinoom. 89Zrpembrolizumab tumoropname correleerde met tumorrespons op behandeling met anti-PD-1 immuuncheckpoint-remmers en overleving van de patiënten. Op de PET scans was opname in de milt het hoogst en zagen we regelmatig opname in de ring van Waldeyer, normale lymfeklieren en op locaties van inflammatie. Uit deze studie blijkt dat PET beeldvorming met 89Zr-pembrolizumab in patiënten met gemetastaseerd melanoom en niet-kleincellig longcarcinoom veilig en haalbaar is, echter is validatie in een grotere groep patiënten nodig om te beoordelen wat de impact van deze benadering op de selectie van patiënten voor anti-PD-1 therapie kan zijn.

89Zr-PET kan mogelijk bijdragen aan inzicht in de lichaamsverdeling van nieuwe immuuntherapeutische geneesmiddelen. CX-072 Probody is een innovatief antilichaam dat is ontwikkeld om in de tumor micro-omgeving geactiveerd te worden door proteases, waardoor binding aan het PD-L1 immuuncheckpoint mogelijk is. Op deze manier wordt PD-L1gemedieerde toxiciteit in normale weefsels potentieel verminderd. We hebben in hoofdstuk 6 de conditionele activatie en lichaamsverdeling van CX-072 onderzocht door middel van PET beeldvorming met 89Zr-gelabeled CX-072. 89Zr-CX-072 liet accumulatie zien in immuungecompromitteerde muizen met humane MDA-MB-231 triple-negatieve borstkankers (TBNC), waarbij de tumor-tot-bloed verhouding 2,1 keer hoger was dan voor 89Zr-gelabeled controle Probody (een antilichaam dat niet aan PD-L1 kan binden). Met behulp van autoradiografie

vonden we hoge 89Zr-CX-072 opname in gebieden met veel PD-L1 expressie. De geactiveerde vorm van 89Zr-CX-072 werd met Western blot in 5,3 keer hogere hoeveelheden in deze tumoren gedetecteerd dan in de milt. 89Zr-CX-072 opname in milt, lymfeklieren en beenmerg van immuun-competente muizen geïnoculeerd met MC38 coloncarcinoom was bovendien beperkt in vergelijking met 89Zr-gelabeled normaal anti-PD-L1 antilichaam (niet activeerbaar door proteases). Deze preklinische data ondersteunen de hypothese dat CX-072 specifiek accumuleert in tumoren die PD-L1 tot expressie brengen.

Vervolgens hebben we in hoofdstuk 7 onderzocht of de tumor-specifieke activatie van CX-072 die we vonden in muismodellen ook plaatsvindt in patiënten. In patiënten met verschillende soorten kanker werden 89Zr-CX072 lichaamsverdeling en tumoropname geëvalueerd met PET. Een antilichaam dosis van 10 mg resulteerde in voldoende 89Zr-CX-072 spiegels in bloed. Deze dosis werd daarom als het meest optimaal beschouwd. PET toonde aan dat 89Zr-CX-072 tumoropname bij alle patiënten aanwezig was, waarbij de hoogste tumoropname werd gevonden op dag 7. In normaal weefsel werd de hoogste 89Zr-CX-072 opname gevonden in de milt. Deze opname was lager dan in een eerdere studie geobserveerd voor 89Zr-atezolizumab. Atezolizumab is een anti-PD-L1 antilichaam dat geen activatie vereist om te kunnen binden. In een deel van de patiënten werd op de PET scan opname in normale lymfeklieren (axillair: 57.1%, inguinaal: 62.5%) en de ring van Waldeyer gezien (50%). Bovendien was 89Zr-CX-072 intact (inactief) aanwezig in serum en plasma. Concluderend laat 89Zr-CX-072 PET opname zien in verschillende typen tumoren, bescheiden opname in normale lymfoïde organen en geen onverwachte opname in andere gezonde weefsels.

89Zr-PET kan tevens worden ingezet om het effect van de verschillende bindingsarmen van een bispecifiek antlichaam op de distributie naar tumoren en immuunweefsels te bestuderen. Het innovatieve ERY974 antilichaam kan gelijktijdig binden aan CD3 op T-cellen en glypican 3 (GPC3) op tumorcellen, met als doel om de antitumor immuunrespons te stimuleren. In hoofdstuk 8 vonden we GPC3-gemedieerde 89Zr-ERY974 tumoropname in immuun-deficiënte muizen met tumoren die GPC3 tot expressie brengen. Deze opname was specifiek ten opzichte van 89Zrgelabeled bispecifiek controle antilichaam gericht tegen CD3 en keyhole limpet hemocyanine (KLH), een eiwit dat niet voorkomt in zoogdieren, en 89Zr-gelabeled controle antilichaam gericht tegen KLH. In muizen geïnoculeerd met humane CD34-positieve hematopoietische stamcellen was 89Zr-ERY974 tumoropname ~3,5 keer hoger dan in immuun-deficiënte muizen. Dit suggereert dat 89Zr-ERY974 tumoropname afhankelijk is van de aanwezigheid van deze humane immuuncellen. Autoradiografie toonde een preferentiële distributie van 89Zr-ERY974 naar tumorgebieden waar zich T-cellen met CD3 expressie bevinden aan. Bovendien werd er hoge opname van 89Zr-ERY074 in de milt en lymfeklieren waargenomen. Deze bevindingen laten zien dat PET beeldvorming met 89Zr-ERY974 potentieel kan worden ingezet om de

distributie van ERY974 naar tumoren en normale organen in beeld te brengen.

Tenslotte laten we in hoofdstuk 9 zien hoe 89Zr-PET inzicht kan geven in de distributie van CD8positieve T-cellen voorafgaand en tijdens immuuncheckpoint-blokkade in patiënten met solide tumoren. Hiertoe hebben we een studie uitgevoerd met 89ZED88082A, een éénarmig antilichaam gericht tegen CD8. Een antilichaam dosis van 10 mg was optimaal voor tumor visualisatie, waarbij al binnen 1 uur opname in de milt werd waargenomen. Op dag 2 was er op de PET scans 89ZED88082A opname zichtbaar in alle normale lymfoïde weefsels en in tumorlaesies. Tumoropname was sterk variabel tussen laesies in dezelfde patiënt en tussen patiënten. 89ZED88082A tumoropname uitgedrukt in SUVmax was voorspellend voor overleving van de patiënten. Daarnaast vonden we een hogere SUVmax in dMMR-deficiënte tumoren en in laesies met, volgens immunohistochemische analyse, stromaal en geïnflammeerde fenotypen. Met autoradiografie op tumorbiopten brachten we de 89ZED88082A distributie op macroschaal in beeld, welke correleerde met immunohistochemisch bepaalde CD8 expressie. Na 2 cycli atezolizumab was met PET geen verandering in totale 89ZED88082A tumoropname te zien. Individuele leasies lieten na behandeling echter heterogene opname zien, onafhankelijk van tumorrespons. Deze studie toonde een duidelijke correlatie aan tussen 89ZED88082A tumoropname en CD8 expressie voorafgaand aan therapie met PD-L1 immuuncheckpointremmers en suggereert dat er grote heterogeniteit bestaat in de aanwezigheid van CD8 op het moment van de tweede 89ZED88082A PET scan tijdens deze behandeling.

De meeste klinische studies met radiofarmaca worden gedaan in relatief kleine groepen patiënten binnen één ziekenhuis of centrum. Dit maakt translatie van resultaten uit deze studies naar toepassing in de dagelijkse praktijk vaak lastig. Om grootschaliger onderzoek in meerdere ziekenhuizen of centra laagdrempeliger te maken, moet harmonisatie van radiolabeling procedures plaatsvinden. GMP-validatie van radiofarmaca productie en het delen van deze informatie met andere centra kan hier mogelijk aan bijdragen. Voor de klinische studies die worden beschreven in dit proefschrift, hebben we de productie van verschillende 89Zr-radiofarmaca, waaronder 89Zr-NNV003, 89Zr-pembrolizumab, 89Zr-CX-072, 89Zr-ERY974 en 89ZED88082A, onder GMP-condities gevalideerd. In hoofdstuk 10 worden de studies uit dit proefschrift samengevat en de resultaten bediscussieerd.

Samenvattend: In dit proefschrift laten we zien hoe PET beeldvorming met 89Zr-radiofarmaca inzicht kan geven in de farmacokinetiek van verschillende immuuntherapeutische antilichamen die nog in (klinische) ontwikkeling zijn of reeds zijn geregistreerd voor patiënten met bepaalde vormen van kanker. Tevens laten we in een beperkt aantal patiënten zien dat tumorrespons op behandeling met immuuncheckpoint-remmers zich mogelijk laat voorspellen door tumoropname op de 89Zr-PET scan. In de toekomst zou PET beeldvorming

Nederlandse samenvatting

met 89Zr-radiofarmaca de selectie van patiënten die potentieel baat hebben bij behandeling met immuuntherapie, en daarmee klinische besluitvorming, kunnen ondersteunen.

Appendices

Dankwoord Publications

Curriculum vitae

Appendices

Dankwoord

Een combinatie van kennis uit meerdere disciplines is nodig om 89Zr-radiofarmaca studies preklinisch en in patiënten te organiseren. Ik heb ontzettend veel geleerd van alle samenwerkingen die ik tijdens mijn promotietraject aan heb mogen gaan. Een aantal collega’s wil ik hieronder in het bijzonder bedanken.

Ik wil graag mijn promotores prof. dr. Lub-de Hooge en prof. dr. de Vries bedanken voor de fijne begeleiding, jullie enthousiasme en alle kennis die jullie met me wilden delen. Marjolijn, na onze eerste kennismaking tijdens mijn masterproject wist ik zeker dat een promotietraject iets voor mij zou zijn. Jij hebt me laten zien hoe je het werk als (ziekenhuis) apotheker kunt combineren met het doen van onderzoek en ik heb dan ook veel bewondering voor hoe goed dit je afgaat. Ik heb veel van je mogen leren over de ontwikkeling van 89Zrradiofarmaca en alles wat daarbij komt kijken. We praatten tijdens ons overleg regelmatig over de toepassing van verschillende tracers en welke aspecten interessant zouden zijn om nog uit te zoeken. Wanneer het tegenzat, had jij altijd een pragmatische oplossing. Dit zorgde ervoor dat ik enthousiast en met frisse moed je kantoor verliet. Je nam regelmatig de tijd voor hulp of uitleg en daar wil ik je enorm voor bedanken. Ik ben erg blij dat we ook in mijn huidige baan als apotheker samen mogen werken. Liesbeth, via Marjolijn kwam ik terecht bij jouw onderzoeksgroep binnen de Medische Oncologie en Nucleaire Geneeskunde. Meteen was ik onder de indruk van het grote aantal samenwerkingsverbanden en de mogelijkheid om met veelal interessante moleculen, met soms nog interessantere werkingsmechanismen, te werken. Ik ben ontzettend dankbaar dat ik hier een bijdrage aan heb mogen leveren. Je gaf me een kijkje in de wereld van het onderzoek en hebt me laten zien dat hierin altijd wat nieuws te ontdekken valt. In een patiënt gedragen de meeste geneesmiddelen zich toch net even anders dan in een preklinisch model en ik heb dan ook veel mogen leren van jouw kennis en ervaring uit de kliniek. Daarbij zorgde jouw kritische blik ervoor dat ik het beste uit mijzelf wilde halen.

Ik wil alle leden van de leescommissie, Prof. dr. S. de Jong, Prof. dr. A. W. J. M. Glaudemans en Prof. dr. N. H. Hendrikse, graag bedanken voor het lezen en beoordelen van mijn proefschrift.

Tevens gaat mijn dank uit naar alle coauteurs voor hun waardevolle bijdrage aan de studies die worden beschreven in dit proefschrift.

De 89Zr-imaging groep stond altijd voor mij klaar. Dr. Pool, Martin, je was mijn begeleider tijdens mijn masterproject en ook in het onderzoek werkten we regelmatig samen, met als resultaat een aantal mooie hoofdstukken in dit proefschrift. Je hebt me geïntroduceerd in de 89Zr-radiofarmacie, waarbij ik veel heb mogen leren van je

doelgerichtheid en analytische blik. Ik ken niemand met zoveel parate algemene kennis (met name tijdens een pubquiz) als jij. Nu we beide werkzaam zijn binnen de radiofarmacie, zit er wellicht in de toekomst nog een samenwerking in. Dr. Waaijer, Stijn, wat hebben we een hoop tijd samen doorgebracht op het lab. Ik heb veel bewondering voor jouw toewijding en enthousiasme in het onderzoek. Je was altijd bereid om te helpen en nam regelmatig de tijd om al mijn vragen te beantwoorden. Ik ben blij dat we samen hebben mogen werken aan hoofdstuk 8. Dr. Suurs, Frans, niet alleen als kamergenoot maar ook als sportbuddy heb ik veel tijd met je mogen doorbrengen. Bedankt voor onze vermakelijke gedachtewisselingen over het doen van onderzoek, eten en sportprestaties. Onze reis naar New York om een nieuwe methode van radiolabeling te leren zal ik niet snel vergeten.

Dr. van der Veen, Elly, jij hebt me laten zien dat je bij een tegenslag in het onderzoek zeker niet bij de pakken neer moet gaan zitten; met voldoende doorzettingsvermogen en amibitie kom je heel ver. Jouw enthousiasme zorgde altijd voor een goede sfeer op het lab. Bedankt voor de fijne samenwerking aan hoofdstuk 3 en 4.

Linda Broer, ik ben blij dat ik je heb leren kennen tijdens mijn promotietraject en dat we samen aan hoofdstuk 5 hebben mogen werken. Naast een fijne collega, was jij er altijd voor een kopje koffie en gezelligheid, waardoor ik even tot rust kon komen.

Linda Pot-de Jong, je was intensief betrokken bij de validatie en productie van een groot aantal 89Zr-radiofarmaca en altijd bereid om te helpen waar nodig, waarvoor ik je ontzettend dankbaar ben. In de loop der jaren heb jij je ontwikkeld als spil binnen de 89Zr-groep, waaruit jouw nuchterheid en humor niet meer weg te denken zijn. Ik ben blij dat we nog steeds collega’s zijn.

Claudia van Winkel, ondanks dat we relatief kort samen hebben gewerkt, heb ik je leren kennen als een enthousiaste en slimme PhD student. Ik ben erg nieuwsgierig hoe je onderzoek verder zal verlopen en hoop dat we zeker nog contact zullen houden.

Arts-onderzoekers van de afdeling Medische Oncologie waarmee ik samen heb mogen werken, dr. Kirsten Moek, dr. Iris Kok, Laura Kist-de Ruijter, Jahlisa Hooiveld-Noeken en Pim van de Donk, veel dank voor het delen van al jullie kennis over het opzetten en uitvoeren van een studie in patiënten. Onze overleggen over interpretatie van klinische data heb ik als erg leerzaam ervaren en hebben mijn onderzoek een extra dimensie gegeven.

Ik wil alle PhD studenten en analisten van het Medische Oncologie lab (MOL) bedanken voor de interessante discussies tijdens onze meetings, maar ook voor alle gezelligheid buiten het lab. Dankzij jullie kijk ik terug op mijn promotietraject als een ontzettend leuke tijd.

Dr. Timmer en dr. Meijer, Hetty en Coby, jullie deur stond altijd open voor advies wanneer een

experiment weer eens iets mislukte. Ik heb veel gehad aan onze discussies tijdens de imaging meetings; deze waren altijd erg leerzaam en constructief. Als lab managers hebben jullie een grote bijdrage geleverd aan mijn onderzoek, waarvoor ik jullie enorm wil bedanken.

Prof. dr. de Jong, Steven, ik heb veel plezier beleefd aan onze wetenschappelijke discussies. Jouw input tijdens onze meetings heeft bij veel experimenten een cruciale rol gespeeld.

Dr. Jorritsma-Smit, Annelies, als apotheker en ‘qualified person’ was jij betrokken bij validatie en productie van de 89Zr-radiofarmaca die worden besproken in dit proefschrift. Zonder jou was translatie naar de kliniek niet mogelijk geweest.

Ik wil het Medische Oncologie secretariaat, Gretha Beuker en Hilda Tooi, tevens bedanken voor alle ondersteuning die ik tijdens mijn promotietraject heb mogen ontvangen.

Dr. Funke, Anouk, ook jij stond als projectmanager altijd klaar om me te helpen, waarvoor ik je ontzettend wil bedanken.

Ik wil de afdeling Nucleaire Geneeskunde en Moleculaire Beeldvorming (NGMB), in het bijzonder prof. dr. Rudi Dierckx, prof. dr. Erik de Vries, prof, dr. Philip Elsinga, dr. Hendrikus Boersma, dr. Inês Farinha Antunes en Marianne Schepers, graag bedanken voor het faciliteren van de vele uren die ik door heb mogen brengen op jullie laboratoria.

Rolf Zijlma, bedankt voor je hulp bij verschillende analyses. Bram Maas en Klaas-Willem Sietsma, bedankt voor jullie technische ondersteuning. Jurgen Sijbesma, ontzettend bedankt voor je hulp bij het gebruik van de microPET scanner en de verwerking van beeldmateriaal.

Graag bedank ik alle medewerkers van de Centrale Dienst Proefdieren (CDP) voor alle ondersteuning die ik heb gekregen bij mijn preklinische studies.

Dr. Francien Talens en dr. Elly van der Veen, wat fijn dat jullie mijn paranimfen willen zijn. Lieve Francien, we leerden elkaar kennen in 2013 tijdens ons masterproject, de rest is geschiedenis. We weten precies wat we aan elkaar hebben en dat waardeer ik enorm. Ik ben ontzettend dankbaar dat ik zo af en toe mag profiteren van je nuchtere visie. Je bent niet alleen mijn paranimf in het onderzoek, maar vooral een hele fijne vriendin.

Lieve Elly, als mede-farmaceut delen we natuurlijk onze interesse in geneesmiddelen, maar ik heb je de afgelopen jaren ook mogen leren kennen als goede vriendin. Ik wil je enorm bedanken voor je mentale support, immer vrolijke persoonlijkheid en de vele momenten van ontspanning tijdens het bezoeken van festivals of het zingen van liedjes bij de après ski.

Tijdens mijn promotietraject voelde ik me tevens gesteund door vrienden en familie. Zij verdienen dan ook een bijzonder woord van dank.

Lieve Francien en Stijn, we hebben elkaar leren kennen als guppies in de Vissenkom. Inmiddels dragen jullie de titel ‘Doctor’ al en mag ik me daar binnenkort bij aansluiten. Bijna wekelijks kwamen we bij elkaar over de vloer voor het beoordelen van ieders kookkunsten en het doen van fanatieke spelletjes; maar wat hebben we daarnaast een boel samen meegemaakt de afgelopen jaren. Nu we zijn herenigd in Groningen, kijk ik uit naar wat we de komende tijd nog samen gaan beleven.

Lieve bewoners van kamer G0.10, Gerda, Margot, Martin en Arjan, veel dank dat ik als aangetrouwde roomie deel mocht uitmaken van jullie avonden met culinaire hoogstandjes.

Lieve farmaceuten, Myrthe, Berdien, Stijn en Ken Ho, bedankt voor alle gezelligheid tijdens weekendjes weg in Groningen, Leeuwarden, Maastricht, Nijmegen en Amsterdam. Voor farmacotherapeutische vraagstukken weet ik jullie te vinden.

Lieve chicks, Inge, Cindy en Joline, bij jullie kan ik altijd terecht voor advies en ontspanning. Ik ben ontzettend dankbaar voor onze vriendschap.

Lieve mama, jij hebt me geleerd in mezelf te geloven en daar ben ik je zeer dankbaar voor. Zonder jouw steun en vertrouwen was ik de studie Farmacie niet zo goed doorgekomen, laat staan dit promotietraject. Mijn doorzettingsvermogen heb ik van jou. Ik weet dat papa trots zou zijn geweest.

Lieve Robert, je was er altijd voor me als luisterend oor en daar ben ik je enorm dankbaar voor.

Ik kan me geen betere broer wensen.

Lieve Bertus en Lucie, Marcel en Nadia, Matthijs en Clarissa. Jullie hebben altijd interesse getoond in mijn onderzoek en de voortgang van mijn promotie; ontzettend bedankt daarvoor.

Ik prijs me zeer gelukkig met jullie als schoonfamilie.

Tot slot, lieve Arjan, mijn nuchtere Groninger, rots in de branding. Onwijs dankbaar ben ik voor jouw onvoorwaardelijke liefde en steun. Ik kon bij jou altijd terecht wanneer ik in paniek raakte omdat het allemaal weer eens anders ging dan gepland. Ik kijk enorm uit naar alles wat ons nog te wachten staat als gezin, samen met onze zoon Milan. Ik hou van jullie.

Appendices

Publications

This thesis:

Giesen D *, Moek KL *, Kok IC, de Groot DJA, Jalving M, Fehrmann RSN et al. Theranostics using antibodies and antibody-related therapeutics. J Nucl Med. 2017;58(Suppl 2):83S-90S.

Waaijer SJ, Giesen D, Ishiguro T, Sano Y, Sugaya N, Schröder CP et al. Preclinical PET imaging of bispecific antibody ERY974 targeting CD3 and glypican 3 reveals that tumor uptake correlates to T cell infiltrate. J Immunother Cancer. 2020;8(1):e000548.

Giesen D, Broer LN, Lub-de Hooge MN, Popova I, Howng B, Nguyen M et al. Probody therapeutic design of 89Zr-CX-072 promotes accumulation in PD-L1-expressing tumors compared to normal murine lymphoid tissue. Clin Cancer Res. 2020;26(15):3999-4009.

van der Veen EL, Giesen D, Pot-de Jong L, Jorritsma-Smit A, de Vries EGE, Lub-de Hooge MN. 89Zrpembrolizumab biodistribution is influenced by PD-1-mediated uptake in lymphoid organs. J Immunother Cancer. 2020;8(2):e000938.

Kist de Ruijter L, Hooiveld-Noeken JS, Giesen D, Lub-de Hooge MN, Kok IC, Brouwers AH et al. First-in-human study of the biodistribution and pharmacokinetics of 89Zr-CX-072, a novel immunoPET tracer based on an anti-PD-L1 Probody. Clin Cancer Res. 2021;27(19):5325-5333.

Kok IC, Hooiveld JS *, van de Donk PP *, Giesen D, van der Veen EL, Lub-de Hooge MN et al. 89Zr-pembrolizumab imaging as a non-invasive approach to assess clinical response to PD-1 blockade in cancer. Ann Oncol. 2022;33(1):80-88.

Giesen D, Lub-de Hooge MN, Nijland M, Heyerdahl H, Dahle J, de Vries EGE et al. 89Zr-PET imaging to predict tumor uptake of 177Lu-NNV003 anti-CD37 radioimmunotherapy in mouse models of B cell lymphoma. Sci Rep. 2022;12(1):6286.

* These authors contributed equally to this work

Other:

Pool M, Terwisscha van Scheltinga AGT, Kol A, Giesen D, de Vries EGE, Lub-de Hooge MN. 89Zr-onartuzumab PET imaging of c-MET receptor dynamics. Eur J Nucl Med Mol Imaging. 2017;44(8):1328-1336.

Giesen D, Lub-de Hooge MN, Brouwers AH. Moleculaire beeldvorming met 89Zr-immunoPET in translationale studies op het gebied van immuuntherapie. TvNG 2020;43(4): 2576-2582.

Broer LN, Knapen DG, Suurs FV, Moen I, Giesen D, Waaijer SJH et al. 89Zr-3,2-HOPO-mesothelin

antibody PET imaging reflects tumor uptake of mesothelin targeted 227Th-conjugate therapy in mice. J Nucl Med. 2022;online ahead of print.

Curriculum vitae

Danique Giesen was born on December 7, 1990 in Zwolle. In 2009 she obtained her preuniversity (VWO) diploma at Thomas à Kempis College. Subsequently, she started the bachelor's degree in pharmacy at the University of Groningen. In 2016 she graduated cum laude from the master's program in pharmacy.

Early on in her pharmacy education, Danique discovered that she had a great affinity with doing research. During her master's research within the departments of Medical Oncology and Nuclear Medicine of the University Medical Center Groningen under the supervision of Prof. Dr. Marjolijn Lub-de Hooge, she first came into contact with radiopharmaceuticals and other imaging techniques. She researched the development of a fluorescently-labeled antibody that can bind to both T cells and tumor cells simultaneously, in order to stimulate the antitumor immune response.

A project internship at the end of 2015 at GE Healthcare's radiopharmacy in Zwolle further encouraged Danique's aspiration to conduct research as a pharmacist in the field of radiopharmaceuticals, preferably within oncology. During this internship she researched the validation and implementation of 68Ga-labeled radiopharmaceuticals.

In 2016, Danique started her PhD research at the University Medical Center Groningen, again within the departments of Medical Oncology and Nuclear Medicine, under the supervision of Prof. Dr. Liesbeth de Vries and Prof. Dr. Marjolijn Lub-de Hooge. During her PhD research she was involved in the development, validation and clinical translation of multiple 89Zr-labeled radiopharmaceuticals. To this end, she worked closely with international pharmaceutical companies. The results of her research are described in this thesis.

During her PhD research, Danique has been committed to sharing knowledge on 89Zratezolizumab production for administration to patients with researchers, doctors and pharmacists from international hospitals. In February 2019, she was invited to give an oral presentation at the ESMO Targeted Anticancer Therapies Congress in Paris about her research on the development of an 89Zr-labeled prodrug antibody (Probody) targeting the immune checkpoint PD-L1.

Since February 2021, Danique has been working as a radiopharmacist and qualified person in training at the Nuclear Medicine department of the University Medical Center Groningen.