930MO PD-L1 expression and outcomes of pembrolizumab and placebo in completely resected stage IB-IIIA NSCLC: Subgroup analysis of PEARLS/KEYNOTE-091

Abstract

In the PEARLS/KEYNOTE-091 study of completely resected stage IB (T ≥4 cm) to IIIA NSCLC (AJCC v7) following ≤4 cycles of adjuvant chemotherapy as recommended per local guidelines (NCT02504372), pembrolizumab (pembro) significantly prolonged DFS vs placebo in the ITT population at the second interim analysis (IA2) (HR 0.76; P = 0.0014). The DFS benefit in the PD-L1 TPS ≥50% population was not significant (HR 0.82; P = 0.14). We present more detailed results for the PD-L1 subgroups. Eligible patients were randomized 1:1 to pembro 200 mg or placebo Q3W for 18 doses. PD-L1 TPS <1% vs 1-49% vs ≥50% was a stratification factor. Dual primary end points were DFS in the ITT and TPS ≥50% populations. Data are from IA2 (median follow-up, 35.6 months). Of the 1177 randomized patients, PD-L1 TPS was <1% in 39.5%, 1-49% in 32.2%, and ≥50% in 28.3%. Baseline characteristics were generally balanced between the pembro and placebo arms in the ITT and TPS ≥50% populations and were generally similar in the ITT and TPS ≥50% populations. Grade 3-5 AE rates for pembro vs placebo were 34.1% vs 25.8% in the ITT and 37.8% vs 25.0% in the TPS ≥50% population; AEs led to discontinuation in 19.8% vs 5.9% and 23.2% vs 6.7%. Pembro improved DFS in the TPS ≥50%, 1-49%, and <1% subgroups (Table).Table: 930MOPD-L1 TPSPembroPbo≥50%N = 168N = 165Median (95% CI), moNR (44.3-NR)NR (35.8-NR)Estimates at 12 / 24 / 36 mo, %79.5 / 67.9 / 65.974.5 / 67.1 / 57.6HR (95% CI)0.82 (0.57-1.18)*1-49%N = 189N = 190Median (95% CI), mo44.2 (34.9-NR)31.3 (22.5-NR)Estimates at 12 / 24 / 36 mo, %80.5 / 65.4 / 54.668.2 / 54.3 / 44.8HR (95% CI)0.67 (0.48-0.92)<1%N = 233N = 232Median (95% CI), mo47.4 (35.0-NR)34.9 (25.5-NR)Estimates at 12 / 24 / 36 mo, %76.6 / 68.5 / 55.572.3 / 58.3 / 48.8HR (95% CI)0.78 (0.58-1.03)*Dual primary end point; P = 0.14. Open table in a new tab *Dual primary end point; P = 0.14. As expected, median and long-term DFS estimates in the pembro arm were numerically improved in the PD-L1 TPS ≥50% subgroup vs the 1-49% and <1% subgroups. The lack of statistically significant benefit for pembro in the TPS ≥50% population at IA2 likely results from placebo overperformance in this population and a smaller population size. DFS in the PD-L1-selected populations will be tested at the next IA. Overall, data support the benefit of pembro for completely resected stage IB-IIIA NSCLC and, if recommended, adjuvant chemotherapy, regardless of PD-L1 expression.