Abstract / Summary
The Combined Positive Score (CPS) algorithm is used with PD-L1 IHC 22C3 pharmDx to evaluate PD-L1 expression in certain tumor types and determine patient eligibility at defined cutoffs for treatment with KEYTRUDA®(pembrolizumab). The CPS is scored by evaluating the amount of tumor cells, lymphocytes, and macrophages expressing PD-L1, divided by the total number of viable tumor cells present, then multiplying by 100. This work provides evidence that CPS evaluation using PD-L1 IHC 22C3 pharmDx is precise and reproducible across multiple tumor types and cutoffs. We retrospectively analyzed data from tumor type specific analytical validations, which assessed assay precision across four studies: combined precision (inter-instrument, inter-operator, inter-day), intra-run repeatability, inter-observer, and intra-observer precision. Tumor types assessed include basket trial cancers, biliary tract cancer (BTC), cervical cancer (CC), colorectal carcinoma (CRC), esophageal cancer (EC), gastric carcinoma (GC), head and neck squamous cell carcinoma (HNSCC), ovarian cancer (OC), prostate cancer (PC), renal cell carcinoma (RCC), triple-negative breast carcinoma (TNBC), and urothelial carcinoma (UC). Overall agreement (OA) mean point estimates for all 21-tumor types and defined cutoff pairs were above 85% across all four studies, demonstrating precision across the tested tumor types. Validation study data were then qualified to be pooled across tumor types for analysis at the CPS ≥ 1 or CPS ≥ 10 cutoff based on Fisher-Freeman-Halton heterogeneity testing. Negative, positive, and overall agreement point estimates for all four studies were above 95% indicating a high percent agreement across all tumor types and cutoffs. Our results demonstrate that the CPS algorithm is precise and reliable when used to determine PD-L1 expression across multiple tumor types and cutoffs in specimens stained with PD-L1 IHC 22C3 pharmDx.