Background: In the phase III KEYNOTE-010 study (NCT01905657), pembrolizumab (pembro; MK-3475) provided superior OS over docetaxel (doce) in patients (pts) with previously treated, PD-L1–positive advanced NSCLC (P = 0.0008 for 2 mg/kg and P< 0.0001 for 10 mg/kg in the total [tumor proportion score (TPS) ≥ 1%] population). We assessed outcomes in KEYNOTE-010 when PD-L1 was further categorized as a TPS of 1%-24%, 25%-49%, 50%-74%, and ≥ 75%. Methods: Pts with progression after ≥ 2 cycles of platinum-doublet chemotherapy and PD-L1 TPS ≥ 1% assessed by IHC (22C3 antibody) were randomized to pembro 2 or 10 mg/kg Q3W or doce 75 mg/m² Q3W for 24 mo or until progression, intolerable toxicity, or other reason. PD-L1 TPS 1%-49% vs ≥ 50% was a randomization stratification factor. Response was assessed centrally per RECIST v1.1 every 9 wk. Survival was assessed every 2 mo. Primary end points were OS and PFS; ORR was secondary. Pembro doses were pooled for this analysis. Results: Of the 2222 pts screened for PD-L1, 33.6% had TPS < 1% and were ineligible. Of the 1034 pts who met all eligibility criteria and enrolled, PD-L1 TPS was 1%-24% in 45.6% of pts, 25%-49% in 11.6%, 50%-74% in 15.3%, and ≥ 75% in 27.5%. Doce outcomes were similar regardless of TPS (Table). For pembro, OS, PFS, and ORR generally increased along with TPS, with the longest OS and PFS and highest ORR in pts with TPS ≥ 75% (Table). Pembro provided longer OS than doce in all 4 TPS categories evaluated. Conclusions: Increasing PD-L1 expression was associated with more favorable outcomes with pembro, but not with doce, verifying PD-L1 as a predictive biomarker for pembro in NSCLC. Pembro improved OS over doce even at the lowest TPS category assessed. Clinical trial information: NCT01905657

*P values are nominal.