Comparison of programmed death ligand-1 expression in preoperative transbronchial lung biopsy and resected specimens in non-small cell lung cancer.

Purpose: Programmed death-ligand 1 expression is heterogeneous in non-small cell lung cancer, and small specimens may not accurately represent the entire tumor. The current study investigated the discordance in programmed death-ligand 1 expression between preoperative biopsy samples and resected specimens.

Methods: We retrospectively collected data of patients with non-small cell lung cancer who underwent surgical resection from May 2022 to June 2024. The programmed death-ligand 1-positive tumor proportion score was evaluated for each case.

Results: In total, 118 patients were included in this study. Programmed death-ligand 1 expression was discordant between the biopsy and resected specimens in 34 cases (28.8%), and it was underestimated in 25 (21.2%) biopsy specimens. The concordance according to Cohen's kappa was κ = 0.410 (95% confidence interval: 0.243-0.577). The number of discordant cases decreased as the number of tumor cells in biopsy specimens increased. In the group with > 400 tumor cells, agreement rate was 100%. The least absolute shrinkage and selection operator model identified never smoker, small tumor size, clinical stage II-IV and ≤ 200 tumor cells in biopsy specimens as predictors of underestimation. The area under the receiver operating characteristic curve using those four factors was 0.773 (0.663-0.884).

Conclusions: Programmed death-ligand 1 expression in biopsy and resected specimens is often discordant, often being underestimated in biopsy specimens. Discordance is more likely when tumor cell counts are low in the biopsy samples. Therefore, caution is advised when treatment decisions are made based on programmed death-ligand 1 assessments of small specimens.