CheckMate 274 met the primary endpoints of improved disease-free survival (DFS) with nivolumab (NIVO) v placebo (PBO) both in the intent-to-treat (ITT) population and in patients (pts) with programmed death-ligand 1 (PD-L1) expression ≥1% assessed by tumor cell score (TC). We report updated analysis of DFS by PD-L1 expression as assessed by both TC and combined positive score (CPS; measuring PD-L1 on both tumor and immune cells) with minimum follow-up in the ITT population of 31.6 months. CheckMate 274 (NCT02632409) is a phase 3, randomized, double-blind, trial of adjuvant NIVO v PBO in pts with muscle-invasive urothelial carcinoma who are at high risk of recurrence after radical resection. Pts were randomized 1:1 to NIVO 240mg or PBO every 2 weeks intravenously for ≤1 year. Primary endpoints are DFS in the ITT population and in pts with TC PD-L1≥1%. CPS was determined retrospectively from the previously stained immunohistochemistry slides. This analysis only included pts with PD-L1 quantifiable by both CPS and TC. 630 pts had quantifiable PD-L1 by TC and CPS; 250 (40%) had TC≥1% (NIVO, n=125; PBO, n=125), 380 (60%) had TC<1% (NIVO, n=191; PBO, n=189), 558 (89%) had CPS≥1 (NIVO, n=282; PBO, n=276), and 72 (11%) had CPS<1 (NIVO, n=34; PBO, n=38). In pts with TC<1%, 81% (n=309) had CPS≥1. Table shows the number of pts and DFS outcomes in pts with TC≥1% and CPS≥1. In pts with TC<1% who also had CPS≥1, median DFS (95% CI) was 19.2 (16.1-25.6) months with NIVO vs. 10.4 (8.2-19.4) months with PBO; HR for NIVO v PBO in these pts was 0.79 (95% CI, 0.60-1.05).Table 28PMedian DFS (95% CI), moHR (95% CI)DFS probability, % 24-moDFS probability, % 33-moTC ≥ 1%NIVO (n=125)52.6 (25.8–NE)0.48 (0.34–0.69)6258PBO (n=125)8.4 (5.6–17.9)3632CPS ≥ 1NIVO (n=282)25.6 (19.3–41.8)0.64 (0.51–0.80)5147PBO (n=276)8.5 (7.8–15.2)3835HR, hazard ratio; mo, months; NE, not estimable. Open table in a new tab HR, hazard ratio; mo, months; NE, not estimable. With extended follow-up, this exploratory analysis of PD-L1 expression by CPS showed that most pts with TC<1% had CPS ≥1. In pts with TC<1% and CPS≥1, median DFS with NIVO was nearly double that with PBO. These results support the interpretation that most pts with TC<1% also benefit from adjuvant NIVO.

CheckMate 274 (NCT02632409), a phase 3 trial of adjuvant nivolumab (NIVO) vs placebo (PBO) in high-risk muscle-invasive urothelial carcinoma (MIUC) after radical resection, demonstrated improved disease-free survival (DFS) with adjuvant NIVO vs PBO in the intent-to-treat (ITT) and tumor PD-L1 expression ≥ 1% populations. We report exploratory analyses (minimum follow-up, 11.0 months) of associations between pre-treatment tumor and immune features and DFS to identify patients (pts) who may potentially benefit most from adjuvant NIVO. Pts were randomized 1:1 to NIVO 240 mg IV every 2 weeks or PBO for ≤ 1 year of adjuvant treatment. Pre-treatment tumor tissue from the most recently resected site or the transurethral resection yielding MIUC diagnosis was provided for biomarker analyses. Tumor mutation burden (TMB; mutations/MB) was measured by whole exome sequencing. CD8+ immune cell infiltration was measured with digital immunohistochemistry (IHC). Gene signature scores were computed from RNA-seq data. Cox proportional hazards models were used to assess association between DFS and biomarkers (continuous scale) and estimate treatment-effect hazard ratios (HRs; NIVO vs PBO) for biomarker tertile subgroups. Of 709 randomized pts, 458, 445, and 323 were evaluable for TMB, CD8 IHC, and RNA-seq. TMB, and CD8 T cell infiltration and IFN-γ signature scores (Table) were positively associated with DFS in the NIVO arm. The associations appeared more pronounced in pts with PD-L1 ≥ 1%. Of these 3 biomarkers, evidence that association with DFS differed between arms was strongest for the IFN-γ signature. Biomarkers associated with preexisting antitumor immunity were associated with improved DFS with NIVO vs PBO, reinforcing the mechanism for benefit with immunotherapy and extending prior findings in metastatic UC to the adjuvant setting. Multivariable modeling is ongoing. Further validation of these exploratory analyses is warranted.Table: 1737MOPopulationDFS HR (95% CI)ITT; n = 7090.70 (0.57–0.85)BiomarkerTertile lowTertile mediumTertile highCD8 IHCOverall; n = 445a0.86 (0.58–1.29)0.78 (0.51–1.21)0.55 (0.33–0.90)PD-L1 ≥ 1%; n = 1790.60 (0.28–1.25)0.44 (0.21–0.92)0.36 (0.17–0.78)PD-L1 < 1%; n = 2641.01 (0.62–1.65)1.12 (0.64–1.95)0.81 (0.41–1.59)IFN-γ gene signatureOverall; n = 323a1.05 (0.64–1.72)0.71 (0.42–1.20)0.28 (0.14–0.54)PD-L1 ≥ 1%; n = 1280.54 (0.20–1.46)0.57 (0.23–1.41)0.12 (0.04–0.41)PD-L1 < 1%; n = 1931.36 (0.76–2.43)0.78 (0.40–1.51)0.57 (0.24–1.33)aTwo patients were not evaluable for PD-L1.CI, confidence interval; IFN-γ, interferon gamma. Open table in a new tab