Timing of irAE and survival in NSCLC.

Abstract

e20659 Background: Immune-related adverse events (irAEs) during immune checkpoint inhibitor (ICI) therapy are associated with improved outcomes in non-small cell lung cancer (NSCLC). However, factors influencing outcomes among patients with irAEs are not well established. This study assesses whether timing of irAEs is associated with outcomes in advanced NSCLC. Methods: This single-center retrospective cohort study included patients with advanced NSCLC treated with ICIs who developed ≥1 irAE. Kaplan–Meier estimates and multivariable Cox models (adjusted for age, sex, ECOG, PD-L1 status, smoking status, and line of therapy) were used to compare PFS and OS between patients with early-onset irAE (≤30 days after ICI start) versus late-onset irAE (>30 days). Immortal time bias was addressed via prespecified landmark analyses at 180 days and a modified 180-day landmark restricted to irAE onset <180 days. A post hoc analysis assessed outcomes following ICI discontinuation due to an irAE, comparing post-discontinuation PFS and OS (indexed from the date of ICI discontinuation) in patients with early versus late irAEs. Results: Of 117 patients (87% Stage IV), early-onset irAE predicted worse PFS (5.0 vs 22.0 mo; adjusted hazard ratio [aHR] 4.17; 95% CI 2.51-6.95) and OS (10.3 vs 35.2 mo; aHR 3.90; 95% CI 2.27-6.70). In the 180-day landmark analysis, early-onset irAE remained correlated with poorer PFS (13.6 vs 24.6 mo; aHR 2.20; 95% CI 1.06-4.57) and OS (18.4 vs 41.7 mo; aHR 2.65; 95% CI 1.36-5.16). Modifying the 180-day landmark analysis to restrict to patients with irAE onset <180 days attenuated the association for PFS (13.6 vs 18.3 mo; aHR 2.30; 95% CI 0.98–5.41), while OS remained significant (18.4 vs 30.9 mo, aHR 2.54; 95% CI 1.24-5.20). In the subgroup analysis of patients who discontinued ICI due to an irAE (n = 64), median ICI duration was 0.7 vs 6.9 mo for early and late irAEs, respectively (p < 0.001). Early irAEs were linked with markedly shorter median post-discontinuation PFS (2.5 vs 19.2 mo; aHR 4.00; 95% CI 1.82-8.80) and post-discontinuation OS (4.6 vs 27.8 mo; aHR 4.10; 95% CI 1.75-9.61). Conclusions: Among patients with advanced NSCLC who develop irAEs, onset within 30 days of ICI is associated with inferior survival, including in landmark analyses limiting immortal time bias. Rapid disease progression after early treatment discontinuation suggests that abbreviated ICI exposure may contribute to these outcomes. These findings support further investigation of strategies to extend ICI exposure after early irAEs. Analysis N Outcome Early irAE(median mo) Late irAE(median mo) aHR (95% CI) No landmark 117 PFS 5.0 22.0 4.17 (2.51-6.95) 117 OS 10.3 35.2 3.90 (2.27-6.70) > 180-day landmark 90 PFS 13.6 24.6 2.20 (1.06-4.57) 99 OS 18.4 41.7 2.65 (1.36-5.16) > 180-day modified landmark w/ irAE <180 days 52 PFS 13.6 18.3 2.30 (0.98-5.41) 61 OS 18.4 30.9 2.54 (1.24-5.20) ICI discontinuation due to an irAE 64 PFS from ICI stop 2.5 19.2 4.00 (1.82-8.80) 64 OS from ICI stop 4.6 27.8 4.10 (1.75-9.61)