Survival predictors in unresectable stage III NSCLC treated with induction chemoimmunotherapy: a real-world multicenter study of treatment cycles, response, and maintenance duration

Background Optimal integration of induction chemoimmunotherapy with definitive chemoradiotherapy (CRT) and maintenance immunotherapy for unresectable stage III NSCLC remains undefined. Methods This multicenter retrospective study included 321 patients who received induction chemoimmunotherapy followed by CRT with or without maintenance immune checkpoint inhibitors (ICIs). Landmark survival from radiotherapy completion was used to avoid guarantee-time bias. Results Patients receiving 1–2 induction cycles showed numerically longer median PFS (28.0 months) than those with 3–4 (24.6) or >4 cycles (14.2; log-rank p = 0.179), but this trend was confounded by higher concurrent CRT rates in the 1–2 cycle group (74.5%vs. 47.1% vs. 33.7%). After propensity-score matching, the response-related survival advantage was attenuated (HR for PFS 0.70, p = 0.079). In matched analyses, maintenance ICI was associated with improved PFS (25.2 vs. 8.9 months; HR 0.47) and OS (NR vs. 22.0 months; HR 0.24) (both p < 0.001). An exploratory data-driven analysis suggested that >9 maintenance cycles yielded further PFS (HR 0.23) and OS (HR 0.17) benefits. Conclusion In this real-world cohort, induction cycle number did not significantly affect survival; the apparent advantage of 1–2 cycles was likely confounded. Early radiographic response was not an independent predictor after matching. Maintenance ICI was associated with improved outcomes, and >9 cycles showed additional benefit, though with potential bias. These findings provide hypotheses for prospective trial design.