Small samples, big insights: PD-L1 experience of a tertiary institution

Abstract

Background PD-L1 expression guides immunotherapy decisions in non small cell lung cancer (NSCLC), yet sample type can impact assessment accuracy. This study aimed to compare PD-L1 expression between small (biopsies, cell blocks) and large (resection) NSCLC samples, assess interobserver variability, and examine whether PD-L1 scoring trends remained stable over a multi-year period.Methods A retrospective analysis was conducted on 494 NSCLC patients tested for PD-L1 (Ventana SP263) between 2018 and 2022. Sample type, tumor subtype, PD-L1 tumor proportion score (TPS), and reporting pathologist were recorded. Interobserver variability was analyzed based on routine diagnostic reports from different pathologists evaluating non-overlapping patient cohorts. Additionally, a subset of 43 patients had matched cell block and resection specimens collected from the same tumor, allowing direct comparison between preparations.Results Among the 494 NSCLC specimens, 152 were large and 342 were small samples. TPS results showed 112 samples (22%) with TPS >= 50%, 163 (34%) with TPS 1%-49%, and 219 (44%) with TPS < 1%. No significant differences in TPS categories were observed between cell blocks and tissue samples (p = 0.176) or between small and large samples (p = 0.326). TPS distributions across different pathologists (p = 0.260) and years (p = 0.250) remained consistent. In the matched 43 specimens, TPS concordance between cell block and resection was high (kappa = 0.892).Conclusion Small biopsies and cell blocks provide reliable PD-L1 results comparable to resection specimens, supporting their use for PD-L1 testing in clinical settings to enhance timely immunotherapy access for NSCLC patients.