Development of the combined positive score (CPS) for the evaluation of PD-L1 in solid tumors with the immunohistochemistry assay PD-L1 IHC 22C3 pharmDx.

Abstract

e14589 Background: Developing clinically relevant and highly reproducible scoring methods for PD-L1 to identify patients who will respond effectively to anti-PD-1 therapy is key in the development of companion or complementary diagnostic assays. Methods: Scoring method for PD-L1 IHC 22C3 pharmDx in NSCLC have only captured the percentage on stained tumor cells using the tumor proportion score (TPS) (Roach et al. 2016, Appl Immunohistochem Mol Morphol.) (Garon et al. 2015, N Engl J Med.). In the KN012 study 2 out of 11 responders to pembrolizumab were detected with the TPS for the gastric indication. More and more additional data indicate that in some tumor indications PD-L1 staining on both tumor and tumor-associated immune cells is associated with clinical outcome. Therefore a method evaluating both tumor and immune cells in one sitting using the combined positive score was evaluated. Results: Scoring the same patient specimens with the new combined positive score method resulted in the detection of 9 out of 11 responders. Our internal inter- and intra- observer data shows that the CPS can be scored reproducibly with an overall agreement point estimates of 93% for intra-observer and an overall agreement of 87.6% for inter-observer reproducibility in gastric carcinoma. Additionally, CPS builds on the scoring method for NSCLC as it shares the same denominator, namely total number of tumor cells. CPS is evaluated based on the number of PD-L1 positive cells (tumor, lymphocytes and macrophages) in relation to total tumor cells, and hence allows the capture of tumor and immune cells in a single read. Conclusions: Our data demonstrates that the CPS scoring method is reproducible when scored by pathologists and clinically relevant for a number of indications.