Postoperative radiotherapy after resection of esophageal small-cell neuroendocrine carcinoma: a multicenter retrospective cohort study

The role of postoperative radiotherapy after resection of esophageal small-cell neuroendocrine carcinoma remains uncertain. We evaluated associations with survival and recurrence while addressing treatment timing, confounding, and treatment-effect heterogeneity. We retrospectively analyzed 127 patients from three centers. Diagnosis required small-cell morphology plus synaptophysin and CD56 positivity. Radiotherapy and adjuvant chemotherapy were modeled as time-dependent exposures. Sensitivity analyses included a 90-day landmark design and propensity-score overlap weighting. Treatment-effect heterogeneity was assessed using interaction terms with Holm correction, and recurrence was analyzed using competing-risk methods. A cohort of 43 patients was analyzed descriptively. Radiotherapy was not associated with overall survival in the adjusted time-dependent model (hazard ratio, 1.472; 95% confidence interval, 0.820-2.641; P = 0.195), the 90-day landmark model (hazard ratio, 1.734; 95% confidence interval, 0.934-3.221; P = 0.081), or the overlap-weighted model (hazard ratio, 1.001; 95% confidence interval, 0.550-1.820; P = 0.999). Adjuvant chemotherapy showed a protective direction without statistical significance after adjustment (hazard ratio, 0.569; 95% confidence interval, 0.230-1.407; P = 0.222). No clinicopathological or biomarker-defined factor significantly modified the radiotherapy-survival association after Holm correction. First-failure patterns differed by radiotherapy status, with lower locoregional and higher distant-failure incidence among treated patients; these associations were descriptive because of residual treatment-selection bias. External-cohort estimates were nonsignificant and nonconfirmatory. This study did not demonstrate an overall survival benefit from postoperative radiotherapy and does not provide evidence supporting its routine use or patient selection based on nodal or biomarker status.