Programmed cell death-ligand 1 (PD-L1) Immuno-PET imaging to evaluate the tumor microenvironment (TME) of human cancer xenograft mouse models

1236 Objectives: Programmed cell death-ligand 1 (PD-L1), a cell surface immune checkpoint ligand, is overexpressed in several cancers and suppresses the host’s immune system, thereby allowing cancer cells to evade detection and proliferate. Anti-PD-L1 monoclonal antibodies (mAb) have emerged as effective therapeutics in reversing this immunosuppression in certain subsets of patients; however, in some responsive patients, tumor cell PD-L1 expression is not detected by immunohistochemistry. Therefore, selecting patients for anti-PD-L1 therapies based on histological characterization of tumor cells alone is insufficient, which may result from not accounting for PD-L1 expression in the tumor microenvironment (TME). Non-invasive imaging agents can detect and provide a real-time readout of PD-L1 expression in the TME and may be a better predictor of patient response to anti-PD-L1 therapies. Previous work has shown that an anti-PD-L1 mAb (Avelumab) can be radiolabeled with zirconium-89 (89Zr) to produce a conjugate, [89Zr]Zr-DFO-Avelumab (ZPD), which has high affinity (sub nM) for both human and murine PD-L1 [1]. ZPD previously determined that there were moderate levels of PD-L1 expression in human MDA-MB-231 tumor cells (MB) and undetectable PD-L1 expression in human MKN-45 tumor cells (MK) [1]. Herein, ZPD was further evaluated in these human cancers using xenograft mouse models to provide further insights into PD-L1 expression in the TME. Methods: ZPD was synthesized as previously described [1]. MKN-45 [MK; human gastric carcinoma; PD-L1(-)] and MDA-MB-231 [MB; human breast carcinoma; PD-L1(+)] engrafted athymic nude mice were injected with ZPD + 20 µg of Avelumab. Positron emission tomography (PET) imaging was acquired after 1 day and 3 days with in vivo biodistributions performed after 3 days. Blood and tissue uptakes were quantitated as percent injected dose per gram (%ID/g) and expressed as tissue-to-muscle (T:M) ratios. Excised tumors were then sectioned and analyzed: 1) by ex vivo autoradiography (for ZPD localization); 2) for histology and expression of murine/human PD-L1, CD45 and CD11b with immunohistochemistry (IHC). Results: Immuno-PET imaging (Figure 1A) and in vivo biodistributions of MK and MB xenograft mice demonstrated the highest ZPD uptake was present in the spleen (~25 %ID/g) and lymph nodes (~30 %ID/g) with no differences in non-target tissues. Interestingly, ZPD tumor uptake was similar between the PD-L1(-) MK tumors (7.2 %ID/g) and the PD-L1(+) MB tumors (9.8 %ID/g), and T:M ratios for MB tumors were 2-fold higher compared to MK tumors (Table 1). Ex vivo autoradiography showed that ZPD localization was heterogeneously distributed in MK tumor sections and homogeneously distributed in MB tumor sections (Figure 1B). Molecular pathology demonstrated that PD-L1 expression in MK tumors occurred in infiltrating immune cells and not tumor cells, whereas PD-L1 expression in MB tumors represented a combination of both tumor cells and infiltrating immune cell expression (Table 1). PD-L1(+) cell populations in the TME colocalize with cells expressing CD45 and CD11b, consistent with myeloid-derived suppressor cells or other infiltrating immune cells. Conclusions: These studies indicate that PD-L1 expression on both tumor cells and cells within the TME can contribute significantly to ZPD uptake, offering a potential source of variability in patient responses to anti-PD-L1 therapy that can be quantified clinically. ZPD Immuno-PET imaging may serve as a reliable diagnostic tool for selecting and monitoring patient response to anti-PD-L1 therapies. Although PD-L1(+) cell types found in xenograft mouse models are likely different from those in humans, the TME may need to be considered when predicting patients likely to benefit from anti-PD-L1 therapies.