Programmed Death Ligand 1 Expression in Paired Non–Small Cell Lung Cancer Tumor Samples

Abstract

BackgroundProgrammed death ligand 1 (PD-L1) expression may predict response to anti–programmed death 1 (anti–PD-1) or anti–PD-L1 treatment. There is limited information on changes in PD-L1 expression over time in patients with non–small cell lung cancer (NSCLC). Patients and MethodsEligible patients with NSCLC who received surgery or underwent biopsy at Samsung Medical Center, Seoul, Republic of Korea, and Aarhus University Hospital, Aarhus, Denmark, between February 2004 and April 2012 were included. PD-L1 expression in paired tumor tissue samples from the same patients at different dates and lesions was measured using a laboratory-developed prototype immunohistochemistry assay (22C3 antibody). PD-L1 positivity was defined as tumor cell membrane positivity in ≥ 1% of tumor cells (proportion score). Concordance of PD-L1 expression was analyzed by treating proportion score as categoric or continuous variables. ResultsNinety-one patients were included in the analysis. The median interval between the 2 tumor collection dates was 20 months, with 91% of paired samples collected > 3 months apart. The concordance rate for PD-L1 classification between paired samples was 67% (95% confidence interval, 57%-77%). When treating the immunohistochemistry proportional score as a continuous variable, a significant correlation of PD-L1 expression was observed between the paired samples (Pearson correlation coefficient, 0.61; P < .001). ConclusionThere are good correlations of PD-L1 expression from paired NSCLC samples. For patients whose PD-L1 status is negative, it may be valuable to obtain additional tissue samples for retesting PD-L1 expression when anti–PD-1 immunotherapy is considered.