Indirect comparison of nivolumab-ipilimumab versus chemotherapy (Chemo) and/or immunotherapy in dMMR/MSI-H metastatic colorectal cancer (mCRC) using reconstructed individual patient data (IPD).

Abstract

3579 Background: In patients with dMMR/MSI-H mCRC, phase III randomized trials have established immune checkpoint (IO) inhibition as a key first-line strategy. In KEYNOTE-177, pembrolizumab demonstrated superior outcomes vs chemo, supporting an IO-first approach. In CheckMate 8HW, nivolumab plus ipilimumab (Nivo+Ipi) outperformed both nivolumab and chemo, demonstrating benefit from dual checkpoint blockade. In contrast, COMMIT evaluated FOLFOX plus bevacizumab plus atezolizumab and reported improved outcomes vs atezolizumab monotherapy. Despite these advances, a direct head-to-head comparison between Nivo+Ipi and chemo-IO remains unavailable. We addressed this unmet question by reconstructing IPD from published Kaplan-Meier (KM) curves to enable an indirect cross trial comparison. Methods: Published PFS KM curves were digitized to pseudo-IPD. To mitigate cross-trial differences, arms were divided into four groups: Nivo+Ipi (CheckMate 8HW; n=171), chemo-IO (COMMIT; n=38), IO monotherapy [pembrolizumab (KEYNOTE-177; n=153), nivolumab (CheckMate 8HW; n=170), atezolizumab (COMMIT; n=40); n=363], and chemo (standard arms in KEYNOTE-177/CheckMate 8HW; n=238). The primary estimand was PFS restricted mean survival time (RMST) at τ=24 months. As a late-benefit sensitivity analysis, RMST was also estimated at τ=48 months where ≥10% patients remained at risk, to limit instability from sparse tails. KM comparisons (log-rank and Cox HR) were derived from reconstructed pseudo-IPD. Results: At τ=24 months, Nivo+Ipi improved PFS RMST vs chemo (18.87 vs 10.33 months; ΔRMST +8.54, 95% CI 6.78-10.31; p<0.001) and vs IO monotherapy (18.87 vs 14.61 months; ΔRMST +4.26, 95% CI 2.54-5.99; p<0.001). Chemo-IO improved RMST vs chemo (16.92 vs 10.33 months; ΔRMST +6.59, 95% CI 3.59-9.60; p<0.001). In the key indirect comparison, Nivo+Ipi vs chemo-IO did not show a statistically significant RMST difference (18.87 vs 16.92 months; ΔRMST +1.95, 95% CI -1.14-5.04; p=0.217). In contrast, the late-benefit (τ=48 months) analysis demonstrated a statistically significant advantage for Nivo+Ipi (RMST 35.00 vs 27.71 months; ΔRMST +7.29, 95% CI 0.08-14.49; p=0.048). KM-derived estimates were consistent with these patterns: Nivo+Ipi outperformed chemo (p<0.001) and IO monotherapy (p<0.001), and chemo-IO outperformed chemo (p<0.001). Compared with chemo-IO, Nivo+Ipi was associated with numerically longer PFS (NE vs 30.2 months; HR 0.62, 95% CI 0.36-1.05; p=0.073). Conclusions: Nivo+Ipi and chemo IO substantially improved 24-month PFS RMST vs chemo and IO monotherapy, with similar outcomes. Nivo+Ipi showed a significant advantage over chemo-IO at 48 months, consistent with a sustained late PFS effect. Findings are hypothesis-generating given cross-trial heterogeneity and reconstruction-related bias.