Abstract
BACKGROUND AND OBJECTIVE: Expression of programmed death-ligand 1 (PD-L1) in tumour cells (TCs) is predictive of immunotherapy efficacy in non-small cell lung cancer (NSCLC). Small biopsy samples collected by bronchoscopy are often used to diagnose peripheral lung cancer. It is questionable whether these small samples from radial endobronchial ultrasonography (r-EBUS) procedures are representative of PD-L1 expression in TCs. METHODS: We retrieved data of consecutive patients who had surgery for NSCLC and previous r-EBUS biopsy sampling, from 2017 to 2019 in our centre. PD-L1 expression in tumour cells was categorised as <1%, 1%-49% and ≥50%. PD-L1 expression was compared between r-EBUS samples and surgical specimens. RESULTS: Among 1026 patients who had r-EBUS, 521 had a diagnosis of lung cancer on r-EBUS sample. PD-L1 testing was indicated in 356 cases and results were considered contributive in 325 cases (91%). 82 patients with PD-L1 expression in r-EBUS samples had subsequent surgical resection of the nodule and were included in the study. PD-L1 expression was identical between r-EBUS samples and surgical specimens in 67% of cases, with kappa 0.44 (p<0.001). 82% of patients with PD-L1≥50% in surgical specimens were identified in r-EBUS samples. Nonetheless, 31% of patients with no PD-L1 expression in r-EBUS samples had some expression in surgical specimens. CONCLUSION: Small samples obtained by r-EBUS are adequate for assessment of PD-L1 expression in tumour cells, with moderate concordance compared to surgical specimens. Reassessment of PD-L1 expression in larger samples may be useful to guide therapy in patients with no PD-L1 expression in r-EBUS samples.