How extensive should lymph node dissection be for cancer of the thoracic esophagus?

From 1985 to 1992, 171 patients with cancer of the thoracic esophagus underwent esophagectomy with systematic dissection of regional lymph nodes including cervical nodes. The hospital mortality rate was 5.3%. The dissected nodes were classified into four groups: the deep cervical (C), upper mediastinal and cervical paratracheal (U), middle and lower mediastinal (L), and upper perigastric (G) groups. The U group mainly consisted of nodes beside the recurrent laryngeal nerves. The phase of cancer infiltration of lymph nodes was evaluated by the total number and the distribution of involved nodes. Of cases with nodal involvement, only 37% were in the late phase, in which more than seven nodes or in which the U, L, and G groups were all involved. Of cases in the earliest phase in which only one node was involved, 93% had either the U or G group involved. The C group of nodes was infrequently involved until the late phase. Cancer had metastasized to the U and G groups across a considerable anatomic distance even in earlier phases. Outcomes of the cases with nodal involvement not in the late phase were satisfactory; the cumulative survival was 60% at 3 years and 54% at 5 years. Systematic nodal dissection would benefit even cases with nodal involvement, unless the disease is in the late phase. Nodes beside the recurrent nerves and upper perigastric nodes should be dissected with higher priority, though they are located anatomically distant.