RESUMEN
BACKGROUND: A randomised phase 2 trial of trimodality with or without induction chemotherapy (IC) in oesophageal cancer (EC) patients showed no advantage in overall survival (OS) or pathologic complete response rate. To identify subsets that might benefit from IC, a secondary analysis was done. METHODS: The trial had accrued 126 patients (NCT 00525915). Recursive partitioning and proportional hazards regression with interactions were performed. RESULTS: The median follow-up of surviving patients was 6.7 years and the median OS duration was 3.8 years (95% confidence interval (CI), 2.6-5.8 years). OS was associated with tumour length (P=0.03), cT (P=0.02), cN (P=0.04), clinical stage (P=0.01), and tumour grade (P<0.001). The effect of IC differed according to tumour grade. Among patients with well or moderately differentiated (WMD) ECs (n=59), the 5-year survival rate was 74% with IC and 50% without IC, P=0.001. IC had no effect on OS of patients with poorly differentiated (PD) ECs (31% and 28%, respectively; interaction, P=0.04; IC, P=0.03). In the multivariate reduced model, WMD with IC was an independent prognosticator for better OS (HR=0.41, 95% CI, 0.25-0.67; P=<0.001). The following four EC phenotypes emerged for OS: (1) very high risk (PD, cN2/N3), (2) high risk (PD, cN0/N1, stage cIII), (3) moderate risk (PD, cN0/N1, stage cI/II or WMD without IC), and (4) low risk (WMD with IC). The 5-year survival rates were 11%, 27%, 48%, and 74%, respectively (P<0.001). CONCLUSIONS: Our data show that IC significantly prolonged OS of WMD EC patients who undergo trimodality; prospective evaluation is needed.