Dukes' classification for
colorectal cancer is simple and has been widely used as a valuable prognostic
indicator. It has been used as an assessment of
gastric cancer, but it has not been evaluated for
esophageal cancer. Of 251 patients with primary
squamous cell carcinoma of the thoracic esophagus between February 1981 and April 1999, 155 patients underwent
esophagectomy with a curative intent. Clinicopathologic characteristics of those 155 patients were retrospectively investigated according to the Dukes',
tumor node
metastasis (TNM) and Japanese staging systems. Dukes' classification showed a clear correlation between
tumor stage and survival. The 3-year and 5-year survival rates of 64 Dukes' A cases were excellent (97.4% and 93.7%), good for 12 Dukes' B (75% and 75%), and poor for 79 Dukes' C (50.5% and 43.4%), respectively (P < 0.05; Dukes' A vs B, P < 0.0001; Dukes' A vs C, P < 0.10; Dukes' B vs C). TNM stage classification also showed a good correlation between
tumor stage and survival, but there were no significant differences between stage 0, I and stage IIA cases (P = 0.2678) and between stage III and stage IV cases (P = 0.8298). In the Japanese staging system, there were no significant differences among stage 0, stage 1, and stage 2 cases (P = 0.4093). Dukes' classification was significantly correlated with
tumor size, Borrmann type, histological type, and vessel invasion. Subdivision of Dukes' C according to the number of positive lymph nodes (1-4 vs > or = 5) showed a clearer correlation with survival rather than other subdivisions, such as the metastatic lymph node ratio (< 1.0 vs > 1.0) or the site of
lymph node metastasis. Dukes' classification, which incorporates the number of positive lymph nodes, correlates well with
tumor progression and provides a simple useful staging system after curative
esophagectomy for
esophageal cancer. Dukes' A
tumor could be proposed as a criterion of early esophageal
carcinoma.