P2.01-027 A Comparison of Five Different Immunohistochemistry Assays for Programmed Death Ligand-1 Expression in Non-Small Cell Lung Cancer Samples

Abstract

Randomized trials have shown treatment with programmed death-1 (PD-1)/programmed death ligand-1 (PD-L1) inhibitors can provide a survival benefit to patients with advanced stage non-small cell lung cancer (NSCLC). PD-L1 expression, determined by immunohistochemistry (IHC) using different protocols, antibodies and thresholds for positivity for different inhibitors, has been reported to be potentially predictive of clinical outcome. The objective of this study was to compare the staining patterns of prominent PD-L1 IHC assays in clinically relevant NSCLC samples. Consecutive full sections of 20 NSCLC samples, comprising five each of resection, core biopsy, cytology, and pleural fluid samples, underwent IHC with the following anti-PD-L1 antibodies/autostainers: 22C3/Link 48, 28-8/Bondmax, SP142/Bondmax, SP263/Benchmark XT, E1L3N/Benchmark XT according to publicly-available protocols. PD-L1 expression were scored manually by pathologists according to the percentage of tumor cells (%TC) stained on a continuous scale. Using published tumor cell percentage thresholds for 22C3, 28-8, SP142 and SP263 of ≥50%, ≥1%, ≥5%, and ≥25%, the frequency of PD-L1 positive cases were 10%, 15%, 70%, and 15% of cases respectively. When a ≥1% threshold was applied, the corresponding frequencies were 70%, 15%, 95%, 65% respectively, and 55% for E1L3N. Using published thresholds, cases were positive according to 1, 2, 3, 4 and 5 antibodies in 15%, 25%, 25%, 0% and 5% of cases respectively. Sorting of cases according to increasing %TC staining revealed a similar order of cases between antibodies, albeit with differences in %TC quanta and occasional exceptions to the order. Spearman rho analysis indicated %TC staining significantly (p<0.05) correlated between most antibody pairs, except 28-8 and 22C3, 28-8 and SP142, and 28-8 and E1L3N. Unsupervised hierarchical clustering revealed two subgroups, comprising of SP142/SP263 and 22C3/28-8/E1L3N. The classification of cases as PD-L1 positive can vary significantly according to the antibody and protocol used. Differences were more likely due to protocol dependent staining intensities and nominated thresholds for positivity, rather than differences in antibody affinity for different epitopes.