Expression of programmed death ligand 1 assessed on histologic samples is a confirmed predictive biomarker for anti-PD-1 immunotherapy, but its evaluation is not approved for immunocytochemistry. We investigated if PD-L1 expression shows comparable results on paired cytologic and histologic tumor specimens and interobserver variability. Percentage of PD-L1-positive tumor cells of 247 paired samples of non-small cell lung cancer was evaluated by three independent investigators. Samples were compared on the basis of the continuous values and also categorized with the tumor proportion score (TPS). Concordance was defined if continuous values were both within a deviation of 10% and if categorized values were identically grouped. Interobserver variability was assessed by the standard deviation of the mean. Based on continuous values between paired samples, perfect concordance rate was approximately 53%. With categorization of PD-L1 expression based on TPS, category was identical in 74.1%. However, defining the continuous values of PD-L1 expression between paired samples within a deviation of 10% as concordant, concordance rate was 82%. Interobserver variability was significantly higher in evaluation of cytologic specimens. Evaluation of PD-L1 expression in paired histologic and cytologic tumor specimens shows comparable results if a deviation of 10% between the values is tolerated. Interobserver variability demonstrates a much more challenging interpretation of PD-L1 expression for cytologic samples.
Keywords: Cytologic; Expression; Histologic; NSCLC; Paired samples; Programmed death ligand 1 (PD-L1).