2149 - Heterogeneous Survival Impact of Induction Immunotherapy in Definitive Radiotherapy for Esophageal Squamous Cell Carcinoma: A Real-World Overlap Weighting Analysis

Purpose/Objective(s):

Induction chemoimmunotherapy (iCIT) is increasingly used in esophageal squamous cell carcinoma (ESCC), but its survival benefit in patients undergoing definitive radiotherapy (RT) remains unclear. This study evaluated the overall effect of iCIT on overall survival (OS) and explored effect modification by RT parameters, dynamic neutrophil-to-lymphocyte ratio (?NLR), and clinicopathological factors.

Materials/Methods:

We identified 172 ESCC patients who received definitive RT. iCIT was defined as induction chemotherapy plus PD-1 inhibitors. Overlap weighting (OW) based on propensity scores (age, T stage, N stage, lesion location, RT dose, concurrent chemotherapy, target volume) balanced covariates. Weighted Cox models estimated the association between iCIT and OS, with interaction terms for target volume (ENI vs. IFI), RT dose (>50.4 Gy vs. =50.4 Gy), and ?NLR (post-RT minus pre-RT NLR). Subgroup analyses were performed by T stage, N stage, age, and BMI. A competing risk model assessed factors associated with grade =2 radiation pneumonitis (RP).

Results:

OW achieved balance for most covariates (SMD<0.1 for 9 of 13 variables). iCIT was not significantly associated with OS overall (HR 1.51, 95%CI 0.86-2.66, p=0.15). No interactions were observed between iCIT and target volume (p=0.85) or RT dose (p=0.77). However, iCIT significantly modified the effect of ?NLR (p-interaction=0.012). In non-iCIT patients, each 1-unit increase in ?NLR was associated with a 4.5% higher mortality (HR 1.045, 95%CI 1.024-1.066), whereas this association was absent in iCIT recipients (HR 0.97, 95%CI 0.92-1.02). In subgroup analysis, iCIT was associated with worse OS in patients with T2-3 disease (HR 2.35, 95%CI 1.29-4.29, p=0.005), but not in those with T4 disease (HR 0.56, 95%CI 0.19-1.64, p=0.29). Grade =2 RP was rare; iCIT showed no significant association (HR 2.22, 95%CI 0.63-7.88, p=0.22).

Conclusion:

In this real-world ESCC cohort, induction immunotherapy did not confer an overall survival benefit, and its effect was not modified by RT target volume or dose. Notably, iCIT eliminated the prognostic impact of ?NLR, suggesting immunotherapy may reshape host immune dynamics. The differential effect by T stage (adverse in T2-3, neutral in T4) warrants further investigation. These hypothesis-generating findings highlight ?NLR as a potential effect modifier and underscore the need for personalized strategies in immuno-radiotherapy.