A novel [89Zr]-anti-PD-1-PET-CT to assess response to PD-1/PD-L1 blockade in lung cancer

Background Harnessing the anti-tumor immune system response by targeting the program cell death protein (PD-1) and program cell death ligand protein (PD-L1) axis has been a major breakthrough in non-small cell lung cancer (NSCLC) therapy. Nonetheless, conventional imaging tools cannot accurately assess response in immunotherapy-treated patients. Using a lung cancer syngeneic mouse model responder to immunotherapy, we aimed to demonstrate that [89Zr]-anti-PD-1 immuno-PET is a safe and feasible imaging modality to assess the response to PD-1/PD-L1 blockade in NSCLC. Materials and methods A syngeneic mouse model responder to anti-PD-1 therapy was used. Tumor growth and response to PD-1 blockade were monitored by conventional 2-deoxy-2-[18F]fluoro-D-glucose ([18F]-FDG) PET scans. Additionally, tumor lymphocyte infiltration was analyzed by the use of an [89Zr]-labeled anti-PD-1 antibody and measured as 89Zr tumor uptake. Results Conventional [18F]-FDG-PET scans failed to detect the antitumor activity exerted by anti-PD-1 therapy. However, [89Zr]-anti-PD-1 uptake was substantially higher in mice that responded to PD-1 blockade. The analysis of tumor-infiltrating immune cell populations and interleukins demonstrated an increased anti-tumor effect elicited by activation of effector immune cells in PD-1-responder mice. Interestingly, a positive correlation between [89Zr]-anti-PD-1 uptake and the proportion of tumor-infiltrating lymphocytes (TILs) was found (Cor = 0.8; p = 0.001). Conclusion Our data may support the clinical implementation of immuno-PET as a promising novel imaging tool to predict and assess the response of PD-1/PD-L1 inhibitors in patients with NSCLC.


Introduction
Lung cancer is the leading cause of cancer deaths (1).Over the last two decades, the development of targeted therapies against certain oncogenic drivers first, and more recently, the use of immune check-point inhibitors (ICIs) have significantly improved the outcomes of metastatic non-small cell lung cancer (NSCLC) patients (2,3).More specifically, immune modulation through the blockade of the program cell death protein (PD-1)/program cell death ligand protein (PD-L1) axis has obtained the best long-term survival rates ever, with more than 30% of patients being alive at five years (4).ICIs, such as PD-1/PD-L1 monoclonal antibodies (mAbs), reactivate the antigen-specific effector T cells, thus boosting the anti-tumor immune response.Nevertheless, tumor response assessment has become a challenge in NSCLC patients receiving immunotherapy-based systemic regimens.
Although immunohistochemical PD-L1 expression in NSCLC is a predictive biomarker of response to PD-1/PD-L1-inhibiting mAbs, other potential predictive biomarkers such as PD-1 expression in tumor-infiltrating lymphocytes (TILs) or the ability of the PD-1 antibody to reach its target have not been evaluated (5).Moreover, the discovery of new combination therapies are emerging to further improve the efficacy of ICIs (6).Recent reports have demonstrated how targeted therapies can modulate the antigenicity of tumor cells and enhance T cell immune recognition, resulting in a potentially synergistic improvement of the efficacy of this therapeutic approach (7,8).
Inhibitor of differentiation-1 (Id1) is a negative transcription regulator that belongs to the Id (Id1-Id4) gene family (9,10).In NSCLC patients, Id1 has been associated with poor response and prognosis, as it plays a central role in tumorigenesis, tumor angiogenesis, metastasis, and tumor progression, suppressing the antitumor immune response (11)(12)(13)(14).Moreover, Id1 has been described as an immunosuppressor factor involved in the generation of an immunosuppressive tumor microenvironment during tumor progression.In advanced melanoma, Id1 upregulation through tumor growth factor b (TGF-b) has been shown to promote the differentiation of dendritic cells (DCs) to myeloid-derived suppressor cells (MDSCs) and to suppress CD8 + T-cell proliferation (12).Furthermore, a recent analysis of Id1 expression from peripheral blood mononuclear cells of stage III and IV melanoma patients, strongly associates high Id1 levels with the presence of phenotypic and immunosuppressive markers in monocytic MDSCs, whereas low Id1 levels are associated with a more immunogenic myeloid phenotype (15).More recently, our group has shown that the combined blockade of Id1 and PD-1/PD-L1 displays synergistic therapeutic activity in KRAS-mutant lung cancer in mouse models.Id1 downregulation enhanced PD-L1 expression on lung cancer cells surface and increased CD8 + T cell infiltration, sensitizing lung tumors that do not respond to PD-1/ PD-L1 mAbs (16).Interestingly, we have also demonstrated how a MEK1/2 inhibitor can modulate the immunosuppressive tumor microenvironment of KRAS-mutant lung adenocarcinoma (LUAD) tumors though Id1 downregulation (unpublished data).
Advanced imaging methods, specifically computed tomography (CT), positron-emission tomography (PET) and magnetic resonance imaging (MRI), have been established as powerful tools for the staging of lung cancer and the accurate assessment of therapeutic response (17,18).PET is a well-established 3dimensional molecular imaging platform that enables noninvasive quantification of the relevant biologic tumor characteristics, using isotope-labelled tracers (19).In NSCLC, the conventional use of 2-deoxy-2-[ 18 F]fluoro-D-glucose ([ 18 F]-FDG) PET is useful for proper initial staging and response monitoring of patients on systemic treatment.However, this imaging tool may be suboptimal in immunotherapy-treated patients in whom a metabolic uptake increase does not necessarily mean disease progression (20).Response to ICIs is characterized by different patterns, such as progression prior to treatment response (pseudoprogression), hyperprogression, and dissociated responses following treatment.These patterns, however, are not reflected in the Response Evaluation Criteria in Solid Tumors version 1.1 (RECIST 1.1), which is standard for response assessment in oncology.As such, new response evaluation tools are required.
The use of radiotracers other than [ 18 F]-FDG is emerging as a non-invasive method to monitor in real time the immune landscape of patients receiving ICIs.This approach is currently under evaluation and may potentially enter routine clinical practice if proven effective (21).Monoclonal antibody-based PET (immuno-PET) is another potential biomarker to 1) verify optimal delivery of targeted agents to tumors and, 2) measure target expression (22,23).
On these premises, we created an in vivo immuno-PET model to profile the immune landscape in a lung cancer mouse model exposed to PD-1/PD-L1 axis blockade.Here we provide key evidence on the preclinical implementation of immuno-PET as a novel imaging tool that may detect antitumor effector immune cells.This strategy could be used to predict and assess the response of PD-1/PD-L1 inhibitors in patients with lung cancer.

Gene expression
Quantification of interleukins: interleukin-1b (Il-1b); tumor necrosis factor alpha (Tnf-a) and interferon gamma (Ifn-g) gene expression was determined by real-time quantitative PCR as previously described (13,16).Gapdh was used as an endogenous control.The primers designed for RT-PCR are listed in Table S1.

Murine models
All animal procedures were approved by the institutional Committee on Animal Research and Ethics (regional Government of Navarra) under the protocol number CEEA 054-19E1.

Antibody conjugation and zirconium-89 radiolabeling
Antibody radiolabeling with zirconium-89 was carried out using a slightly modified version of the protocol described by Vosjan MJ (24).Briefly, a buffer exchange was performed in a small fraction of the monoclonal antibody with a 0.1M bicarbonate buffer (pH = 9), which was then incubated with a 3-fold molar excess of the chelator deferoxamine (DFO) dissolved in 20 µL of dimethyl sulfoxide (Sigma-Aldrich, Saint Louis, MO, USA).After 30 minutes of conjugation, the reaction mixture was purified using a disposable PD-10 desalting column (Healthcare Life Sciences, Eching, Germany).Then 111 MBq of zirconium-89 was added to the anti-PD-1 (RMP1-14, BioXCell, Lebanon, NH, USA) antibody solution (buffered at pH = 7 with HEPES [Lonza, Basel, Switzerland], oxalic acid and sodium bicarbonate) and the reaction was left at room temperature for 30 minutes.Finally, the solution was purified to eliminate any possible non-chelated zirconium-89 by purification with a PD10 column (Healthcare Life Sciences, Eching, Germany).

In vivo PET imaging with [ 18 F]-FDG and [ 89 Zr]-anti-PD-1
All PET images were acquired on a Mosaic (Philips, Amsterdam, The Netherlands) small animal dedicated tomograph and reconstructed applying dead time, decay, random and scattering corrections into a 128×128 matrix with a 1 mm voxel size.Additionally, on the days that [ 89 Zr]-anti-PD1 images were acquired, CT images were performed in a U-SPECT6/E-class (MILabs, Duwboot, The Netherlands) system to obtain the corresponding anatomical correlate of the tumors.
To obtain PET [ 18 F]-FDG images, mice were fasted overnight with ad libitum access to drinking water.On the day of the study, a dose of 9.3 ± 0.8 MBq was injected intravenously in the tail vein.After 50 minutes, the animals were anesthetized with 2% isoflurane in 100% O 2 gas and placed prone on the scanner bed for a 15minute image acquisition.Images with [ 89 Zr]-anti-PD-1 were acquired during 30 minutes and 24, 72 and 144 hours postinjection of an intravenous single dose (3.8 ± 0.02 MBq).
PET data were exported and analyzed using the PMOD software (PMOD Technologies Ltd., Adliswil, Switzerland) and transformed to standardized uptake value (SUV) units using the formula SUV = [tissue activity concentration (Bq/cm3)/injected dose (Bq)] × body weight (g).[ 18 F]-FDG and 89 Zr uptake in the tumors was analyzed by drawing volumes of interest (VOI) manually containing the entire tumor, guided by CT when available.Semi-automatic segmentation was then performed including voxels with a value greater than 50% of the maximum value of the tumor.Finally, the average of the SUV values within the semi-automatic VOI was calculated (SUV mean).

Statistical analysis
A Shapiro-Wilk test was conducted to analyze the normality of the samples.Statistical significance was assessed using a Mann-Whitney U test (for comparisons between two groups) and one-way ANOVA followed by a post hoc test, or Kruskal-Wallis followed by a post hoc test (for comparisons between different groups).The relationship between zirconium-89 uptake and immune cell infiltration was analyzed using Pearson correlations.A p-value of <0.05 was considered statistically significant.Statistical analyses were performed using Prism software version 8.0 (GraphPad, San Diego, CA, USA).

Results [ 18 F]-FDG-PET scan fails to identify PD-1/ PD-L1 blockade antitumor response
In order to explore the limitations of conventional [ 18 F]-FDG-PET scans for accurately assessing the response to immunotherapy treatments, we used a lung cancer syngeneic mouse model exposed to PD-1/PD-L1 axis blockade, as previously published (16).
Taken together, these results suggest that the [ 18 F]-FDG-PET scan fails to assess antitumor immune response upon PD-1/PD-L1 blockade in this animal model.
[ 89 Zr]-anti-PD-1 uptake correlates with tumor-infiltrating CD8 + T cells Given the limitations observed with conventional [ 18 F]-FDG-PET to monitor PD-1/PD-L1 blockade antitumor response, we used an additional novel radiotracer labeling protocol based on the use of anti-PD-1 mAb with 89 Zr.PD-L1 is widely expressed in both immune cells, as well as in tumor cells, whereas PD-1 is mainly expressed on the surface of activated T cells, B cells, and monocytes (25).Therefore, we hypothesized that [ 89 Zr]-anti-PD-1 signal may predict and correlate with tumor infiltrating lymphocytes.
Collectively, the Zr uptake and its correlation with TILs in anti-PD-1 responder mice, suggest that [ 89 Zr]-anti-PD-1 immuno-PET allows an accurate evaluation of tumor response to immunotherapy.
Collectively, these data show that Id1 absence in the host microenvironment can induce tumor TIL infiltration.More importantly, we also demonstrate that Id1 inhibition may enhance proinflammatory interleukin expression implicated in immune T cell activation.limitations of conventional [ 18 F]-FDG-PET scans in assessing antitumor responses.Our results show that an [ 89 Zr]-anti-PD-1-PET-CT could accurately assess tumor response to ICIs and may constitute a potential biomarker which directly labels effector T cells and predicts the efficacy of PD-1 inhibition in real time, non-invasively and safely.
[ 18 F]-FDG-PET-CT is a powerful tool for monitoring lung cancer initial staging and antitumor response due to its ability to detect small metastatic lesions and regional lymph node tumor spread more accurately than the conventional CT and MRI imaging methods (26, 27).A number of clinical studies have shown that alterations in metabolic activity, expressed as changes in SUV during induction therapy or at interim evaluation, are associated with tumor response.In NSCLC, a reduction of SUVmax below 2.5 after 2-4 conventional chemotherapy cycles has been considered a predictor of future response associated with a substantially higher median time to recurrence (28).Similarly, when the treatment studied is a targeted therapy, such as the EGFR tyrosine-kinase  inhibitor erlotinib, a reduction in SUVmax measured by [ 18 F]-FDG-PET is also associated with durable therapeutic responses in NSCLC patients (29).In contrast with conventional chemotherapy or targeted therapies, the response pattern of patients treated with ICIs may be substantially different, with some patients developing hyperprogression or pseudoprogression (30).[ 18 F]-FDG is a reliable radiotracer for monitoring glucose metabolism.However, changes in glucose metabolism are not restricted to tumor cells, and anti-tumor immune-related cells can similarly show major changes in the glucose metabolism when they are externally activated.ICIs reactivate effector T cells, boosting immune and natural inflammatory response and conventional [ 18 F]-FDG-PET scans have shown inaccuracies when examining responses to these drugs (31).We also observed limitations with [ 18 F]-FDG-PET in our lung cancer murine model when assessing tumor response to PD-1/PD-L1 axis blockade among mice responding to anti-PD-1 blockade and Id1 inhibition.In previous studies, we have shown that Id1 complete abrogation at both host microenvironment and tumor cells sensitized lung tumors to anti-PD-1 therapy, substantially reducing tumor growth (16).However, no significant differences were observed at the SUVmax 1 [ 18 F]-FDG uptake level between responders and non-responders to immunotherapy.According to RECIST, the tumor burden may transiently increase and then decrease as treatment continues, due to an immune reaction between tumor cells and host immune cells (32).The immune-related Response Evaluation Criteria in Solid Tumor (irRECIST) has been proposed as an update to the RECIST criteria for the assessment of response to ICIs (33).For PET response evaluation, different response criteria have also been proposed, such as the EORTC (European Organisation for Research and Treatment of Cancer) and PERCIST (Positron Emission Tomography Response Criteria in Solid Tumors) (34,35).Nevertheless, the specific mechanism of action of ICIs has created unique [ 18 F]-FDG-PET-CT response patterns, which make it difficult to determine these responses using the PERCIST criteria.This has led to the elaboration of new multiple criteria specifically addressing ICIs response, such as PECRIT (Early Prediction of Response to Immune Checkpoint Inhibitor Therapy) or PERCIMT (Response Evaluation Criteria for Immunotherapy) (36).However, the lack of harmonization hampers the wider adoption of molecular imaging as a more accurate and reliable tool for response assessment to immune-modulating agents in cancer patients.
Similarly, in our lung cancer mouse model we show that [ 89 Zr]anti-PD-1 uptake in tumor lesions was significantly higher among treatment-responding Id1 knock-out mice, than among nonresponding mice maintaining a constitutive Id1 expression.More importantly, the novelty of our study relies on the demonstration of a clear correlation between that [ 89 Zr]-anti-PD-1 uptake and the proportion of TILs (CD3 + T and effector CD8 + T cells) infiltrating t u m o r l e s i o n s a n d w i t h a p r o i n fl a m m a t o r y t u m o r microenvironment.As we previously demonstrated, Id1 and PD-1 combine blockade synergy was generated mainly through CD8 + T cells infiltration (16).We also showed in our mouse model an increase in the presence of CD8 + T cells and the correlation with the [ 89 Zr]-anti-PD-1 uptake in anti-PD-1 responder mice.Consistently, we found a significant upregulation of proinflammatory cytokines in responding mice, such as Ifn-g and Tnf-a, that are involved in CD8 + T cell tumor killing activity (41,42).Moreover, the expression of Ifn-g and Il-1b in tumor samples of anti-PD-1 responder mice, might suggest the infiltration by other antitumor immune system cells (43, 44).Therefore, taken together these data we provide a potential mechanistic explanation for the enhanced tumor response observed.
The main limitation of our study is the relatively small tumor sample size analyzed, since most mice with combined complete Id1 genetic abrogation (Id1-deficient mice injected with Id1-silenced tumor cells) and PD-1 blockade showed complete regression of their syngeneic lung cancer tumors, so they could not be measured.Moreover, our experiments were only performed in immunologically competent murine models, so these results should be confirmed in humanized murine models.
We have demonstrated the utility of [ 89 Zr]-anti-PD-1 immuno-PET as a novel imaging tool for the non-invasive, real-time detection of antitumor effector TILs in a lung cancer mouse model responder to anti-PD-1 therapy.In this regard, [ 89 Zr]-anti-PD-1 (immuno-PET) may probably perform similarly well in other lung cancer subtypes for which immunotherapy has been approved in frontline treatment or as subsequent lines (KRAS or BRAF mutant lung tumors).

Conclusions
This study proposes the potential use of 89 Zr]-anti-PD-1 immuno-PET as a safe and feasible imaging modality to monitor PD-1/PD-L1 inhibitor antitumor response in NSCLC.We show that [ 89 Zr]-anti-PD-1 uptake is significantly higher in anti-PD-1responding mice as compared to non-responding mice.Moreover, our data may confirm the capacity of [ 89 Zr]-anti-PD-1 immuno-PET to label PD-expressing immune cells, such as effector CD8 + TILs.Nevertheless, further investigation is warranted.

4 Id1
FIGURE 4 Id1 inhibition at the tumor-microenvironment promotes proinflammatory interleukin expression and T cell infiltration.(A) Representative IHC images illustrating CD3 + T cells, CD8 + T cells and Id1 + cells of LLC cells inoculated in C57 and IDKO mice.Scale bar: 200mm.(B) Left: Quantification of proportion of relative stained area of CD3 + T cells.Right: Quantification of the proportion of relative stained area of CD8 + T cells of tumor samples illustrated in (A).(C) Representative images of multiplex immunofluorescence staining panel with nuclei (white), Id1 (red), CD3 (green), CD8 (yellow) of LLC cells inoculated in C57 and IDKO mice.Scale bar: 200mm.(D) Left: Quantification of multiplex immunofluorescence staining of CD3 + T cells.Right: Quantification of multiplex immunofluorescence staining of CD8 + T cells of tumor samples illustrated in (C).(E) Relative mRNA expression levels of Il-1b, Ifn-g and Tnf-a in LLC tumors in C57 and IDKO mice.Asterisks denote significance (*p < 0.05, ***p < 0.001), and error bars denote SD.

TABLE 2
Immunohistochemical analysis of markers of tumor-infiltrating lymphocytes from all mice treated with [ 89 Zr]-anti-PD-1.

TABLE 3
Immunohistochemical and multispectral immunophenotyping analysis of markers of tumor-infiltrating lymphocytes from anti-PD-1 nonresponding and responding mice.