[Surgically treated esophageal cancers: predictive model of survival]

The aim of this study was to identify prognostic factors in patients with esophageal cancer after curative resection, and to establish a predictive model of their long-term prognosis.

Eighty-nine patients operated on for neoplasia of the esophagus, who underwent a curative resection, an who did not die within one month or during the hospital stay, were included in this study. Twenty-one variables were studied by univariate analysis. The variables linked with survival were include in a Cox model. Regression coefficient of independent prognostic factors allowed to compute a score.

Life table analysis of the entire population, showed 2 and 5 year survival rates of 48% and 28%, respectively. In univariate analysis, 5 out of 21 factors were statistically linked with survival. In multivariate analysis (Cox model), 4 independent factors were linked with survival: age (P = 0.02), the American Society of Anesthesiologist classification (P = 0.01), parietal invasion (P = 0.03), and lymph node invasion (P = 0.009). The score established with these 4 factors allowed to distinguish 3 sub-groups, discriminated by their long term prognosis. Life table analysis of the 3 sub-groups were at 2 and 5 years 83%, 55%, 20% and 60%, 32%, 0%, respectively.

This model may be useful for the assessment of prognosis in patients with esophageal cancer after curative surgical treatment.