Perioperative chemotherapy and long-term cure in resected grade 3 gastroenteropancreatic neuroendocrine carcinomas: The GETNE SOUTH-NEC Study

Abstract

BACKGROUND: In resected stage I-III gastroenteropancreatic neuroendocrine carcinomas (GEP-NEC), whether perioperative chemotherapy eradicates residual disease or merely delays relapse is unknown, since Cox models cannot distinguish these effects. METHODS: We analysed 162 patients with resected stage I-III GEP-NEC from the GETNE registry. Event-free (EFS), recurrence-free (RFS) and overall survival (OS) were modeled by Cox regression with chemotherapy as a time-dependent exposure, adjusted for age, sex, performance status, stage, Ki-67 and site, and by a Bayesian promotion-time cure model that separates a clonogenic (cure) component from a kinetic (relapse timing) component. RESULTS: Seventy-four patients (46%) received chemotherapy; median follow-up was 93 months. Chemotherapy was associated with longer EFS (hazard ratio [HR], 0.61; 95% confidence interval [CI], 0.40-0.96) and RFS (HR, 0.59; 95% CI, 0.37-0.93), but not clearly with OS (HR, 0.73; 95% CI, 0.46-1.15). On EFS, the cure model attributed the association to a lower clonogenic burden (ratio, 0.67; 95% credible interval [CrI], 0.40-1.11; posterior probability, 94%) rather than to slower kinetics (ratio, 0.93); on RFS both components contributed. The population cure fraction was 22.7% (95% CrI, 15.9-29.6) for EFS and 28.6% (21.9-35.7) for RFS, matching the Kaplan-Meier plateaus. No Ki-67 by chemotherapy interaction was detected. CONCLUSION: In resected grade 3 GEP-NEC, perioperative chemotherapy is associated with longer EFS and RFS and, on the composite endpoint, with a higher probability of long-term cure rather than delayed recurrence. These associations are observational and the cure-versus-delay decomposition is descriptive. Cure models add information beyond conventional survival metrics.