Relationship between programmed cell death ligand 1 expression and the number of biopsy specimens in advanced gastric cancer.

[PURPOSE] Programmed cell death ligand 1 (PD-L1) expression in advanced gastric cancer (AGC) exhibits spatial heterogeneity, which may lead to sampling bias during biopsies. Although multiple biopsies are believed to improve the accuracy of PD-L1 assessment, the optimal number of specimens remains uncertain. This study investigated the relationship between PD-L1 expression and biopsy specimen count in AGC.

[METHODS] We retrospectively analyzed 110 patients with AGC who underwent first-line chemotherapy and had PD-L1 combined positive scores (CPS) assessed using the 28-8 pharmDx assay. Associations between CPS and biopsy specimen count were evaluated using chi-square or Fisher's exact test. In a subgroup of 70 human epidermal growth factor receptor 2 (HER2)-negative patients treated with first-line nivolumab plus chemotherapy, survival outcomes were analyzed based on CPS status.

[RESULTS] PD-L1 CPS ≥ 5 was identified in 79 patients (71.8%). The proportion of patients with CPS ≥ 5 was significantly higher in those with ≥ 5 biopsy specimens than in those with ≤ 4 (77.5% vs. 56.7%, P = 0.03). This trend was even more pronounced in HER2-negative patients (83.6% vs. 54.5%, P < 0.01) and in those with macroscopic type 2 tumors (91.3% vs. 33.3%, P < 0.01). However, no significant differences in progression-free or overall survival were found based on CPS status, regardless of biopsy count.

[CONCLUSION] Obtaining at least five biopsy specimens enhances the detection of PD-L1 CPS ≥ 5, particularly in HER2-negative or well-circumscribed nodular AGC, potentially improving the accuracy of PD-L1 evaluation. Nevertheless, survival outcomes were unaffected, highlighting the limited predictive value of CPS.