Evaluating PD-L1 in Head & Neck Squamous Cell Carcinoma: concordance between 22C3 PharmaDx and SP263 assays on whole sections from a multicenter study

AIMS The introduction of immunotherapy in head and neck squamous cell carcinoma (HNSCC) raises the need for harmonization between different types of antibodies and immunohistochemistry platforms for evaluating the expression of (Programmed death-ligand 1) PD-L1 with Combined Positive Score (CPS) in this tumor. We compare the expression of PD-L1, using the CPS and two widely used antibody (22C3 pharma Dx and SP263 assay) in a cohort of HNSCC. METHODS AND RESULTS We analyzed forty-three whole sections of HNSCC with two different anti-PD-L1 antibodies, 22C3 pharma Dx and SP263 assay. Results, expressed as CPS, were evaluated by ten trained pathologists and statistical analyses, including interobserver agreement, were performed. We found a very similar distribution for PD-L1 expression between 22C3 pharmaDx assay and SP263 assay in our cohort and a strong significant correlation between the two assays for all specimens (Spearman r= 0.945; p<0.0001). The interobserver reliability among pathologists for the continuous scores of CPS with ICC and the correlation between the two assays was both good. Moreover, the agreement rate between assays was high at all cut-offs and was best for the most relevant cut-off of CPS ≥1, while the kappa values were always in the range of almost perfect. CONCLUSIONS Two different assays (22C3 pharmaDx assay and SP263 assay) for PD-L1 in HNSCC showed high agreement. This data suggests the interchangeability of these two antibodies in the selection of patients with HNSCC for immunotherapy.