Real-world prevalence of PD-L1 expression in non-small cell lung cancer: an Australia-wide multi-centre retrospective observational study

Abstract

An investigator-initiated, Australia-wide multi-centre retrospective observational study was undertaken to investigate the real-world prevalence of programmed death ligand-1 (PD-L1) expression in non-small cell lung carcinoma (NSCLC). Multiple centres around Australia performing PD-L1 immunohistochemistry (IHC) were invited to participate. Histologically confirmed NSCLC of any stage with a PD-L1 IHC test performed for persons aged ≥18 years between 1 January 2018 and 1 January 2020, and eligible for review, were identified at each centre, followed by data extraction and de-identification, after which data were submitted to a central site for collation and analysis. In total data from 6690 eligible PD-L1 IHC tests from histologically (75%) or cytologically (24%) confirmed NSCLC of any stage were reviewed from persons with a median age of 70 years, 43% of which were female. The majority (81%) of tests were performed using the PD-L1 IHC SP263 antibody with the Ventana BenchMark Ultra platform and 19% were performed using Dako PD-L1 IHC 22C3 pharmDx assay. Reported PD-L1 tumour proportion score (TPS) was ≥50% for 30% of all tests, with 62% and 38% scoring PD-L1 ≥1% and <1%, respectively. Relative prevalence of clinicopathological features with PD-L1 scores dichotomised to <50% and ≥50%, or to <1% and ≥1%, were examined. Females scored ≥1% slightly more often than males (64% vs 61%, respectively, p=0.013). However, there was no difference between sexes or age groups (<70 or ≥70 years) where PD-L1 scored ≥50%. Specimens from patients with higher stage (III/IV) scored ≥1% or ≥50% marginally more often compared to specimens from patients with lower stage (I/II) (p≤0.002). Proportions of primary and metastatic specimens did not differ where PD-L1 TPS was ≥1%, however more metastatic samples scored TPS ≥50% than primary samples (metastatic vs primary; 34% vs 27%, p<0.001). Cytology and biopsy specimens were equally reported, at 63% of specimens, to score TPS ≥1%, whereas cytology samples scored TPS ≥50% slightly more often than biopsy samples (34% vs 30%, respectively, p=0.004). Resection specimens (16% of samples tested) were reported to score TPS ≥50% or ≥1% less often than either biopsy or cytology samples (p<0.001). There was no difference in the proportion of tests with TPS ≥1% between PD-L1 IHC assays used, however the proportion of tests scored at TPS ≥50% was marginally higher for 22C3 compared to SP263 (34% vs 29%, respectively, p<0.001). These real-world Australian data are comparable to some previously published global real-world data, with some differences noted.