Reproducibility of PD-L1 immunohistochemistry interpretation across various types of genitourinary and head/neck carcinomas, antibody clones, and tissue types.

Programmed death-ligand 1 (PD-L1) expression by tumor cells is a mechanism for down-regulation of antitumor T-cell responses and is a target for immunotherapy in various cancers. PD-L1 status as a predictor of treatment response has led to the development of multiple platforms with different reference cutoffs. We studied 128 cases of genitourinary and head/neck carcinomas, aiming to assess the frequency of PD-L1 positivity, interobserver reliability of PD-L1 interpretation, and the concordance of PD-L1 scoring between small samples from tissue microarray and whole sections using SP263 and SP142 clones. No prostatic carcinoma (0/21) was PD-L1 positive compared with 15% to 24% PD-L1 positivity in urothelial carcinoma (UC), hypopharyngeal squamous cell carcinoma (HP-SCC), and high-grade salivary gland carcinoma. There was substantial interobserver agreement in determining overall PD-L1 positivity in UC and HP-SCC using SP263 (κ = 0.702) and SP142 (κ = 0.757) antibodies. Subgroup analysis for both antibodies showed excellent agreement in UC (κ = 0.812 and 0.827) and moderate agreement in HP-SCC (κ = 0.469 and 0.591). Moderate to substantial agreement between tissue microarray and whole sections was achieved using SP263 (overall, κ = 0.573; UC, κ = 0.424; and HP-SCC, κ = 0.667) and SP142 (UC, κ = 0.493). PD-L1 interpretation in genitourinary and head/neck carcinomas is reliable and reproducible among pathologists and across different tissue preparations. Tumor PD-L1 staining heterogeneity may lead to discrepant PD-L1 results between small biopsies and large sections from surgical resection in a subset of tumors (19% of UC and 15% of HP-SCC). Retesting in such cases may be required to determine patient suitability for anti–PD-1/PD-L1 therapy.