Outcomes of single-agent PD-(L)-1 versus combination with chemotherapy in patients with PD-L1-high (≥ 50%) lung cancer.

Abstract

9052 Background: Single agent PD-(L)-1 blockade (IO) and PD-(L)-1 blockade combined with chemotherapy (ChemoIO) are both standard first-line treatments for patients with PD-L1-high (≥ 50%) metastatic non-small cell lung cancer (NSCLC). These regimens have not been compared prospectively, so comparative effectiveness is unclear. It is also unknown if clinical and molecular tumor characteristics differentially associate with benefit to IO vs ChemoIO in patients with NSCLC with high PD-L1 tumor expression. Methods: All patients with metastatic NSCLC treated with IO or ChemoIO at two institutions (Memorial Sloan Kettering Cancer Center and Dana-Farber Cancer Institute) were reviewed. Patients with EGFR or ALK alterations, PD-L1 expression < 50%, treated with IO or ChemoIO in > 1st line setting were excluded. Overall survival (OS), progression-free survival (PFS), and objective response rate (ORR) were compared between the IO vs ChemoIO groups and association with clinical, pathologic, and molecular features was examined. To account for NGS panel differences, tumor mutational burden (TMB) values were harmonized using a z-score conversion as previously described (Vokes et al, 2019). Results: Of 639 patients with stage IV EGFR/ ALK wild-type NSCLC and PD-L1 ≥50% treated in the 1st line setting, 504 received IO and 135 received ChemoIO. Baseline ECOG performance status (p = 0.3), median PD-L1 % (p > 0.9) and TMB (p = 0.2) were similar between the IO and ChemoIO groups. For patients receiving IO vs ChemoIO, there was no significant difference in OS (HR 0.8, 95% CI 0.6 to 1.08; p = 0.2). Median PFS was shorter (HR 0.7, 95% CI 0.6 to 0.9; p = 0.004) and ORR was lower (40% IO vs 55% ChemoIO, p = 0.002) in the IO group. Among patients with durable responses (> 6 months), never smokers were less common in the IO group (6% vs 18%, p < 0.001), but there was no difference in PD-L1 expression, TMB, or mutational ( KRAS, STK11, or KEAP1) profile to suggest differential predictors of benefit to IO or ChemoIO. Conclusions: In patients with PD-L1 high NSCLC, there was no survival benefit associated with the addition of chemotherapy to IO. There were also no clear differences in PD-L1 expression or molecular features associated with durable response to IO vs ChemoIO. These findings have implications for treatment selection in this population.