The purpose of this study is to identify immunohistochemical evidence of lymph-node micrometastasis in histologic node-negative gastric cancer patients and to evaluate the prognostic significance of lymph-node micrometastasis.

A retrospective study of 50 gastric cancer patients who underwent curative resections from October 1990 to November 1994 was performed. Two consecutive sections were prepared: one for ordinary hematoxylin and eosin staining, and the other for immunohistochemical staining with Pan cytokeratin antibody (Novocastra, UK). In the univariate analysis, the survival rate was calculated using the Life Table Method, and the multivariate analysis was determined using a Cox Proportional Hazards Model. The statistical analyses of the relationships between the clinicopathologic factors and micrometastases were performed by using a Chi-square test.

Of 2522 harvested lymph nodes, 81 (4.1%) nodes and 19 (38%) of 50 patients were identified as having lymph- node micrometastases by using immunohistochemical staining for cytokeratin. The incidence of lymph-node micrometastases was significantly higher in diffuse type carcinomas (54%, P=0.024) and in patients with serosal invasion (52.2%, P=0.05). For patients with lymph-node micrometastases (n=19), the 5-year survival rate was significantly decreased (73.7%, P=0.015). The Lauren's classification (P= 0.021) and the depth of invasion (P=0.035) were shown by multivariate analysis to have a significant relationship with the presence of micrometastases. Multivariate analysis revealed that lymph-node micrometastasis was independently correlated with survival in histologic node-negative gastric cancer patients.

The presence of cytokeratin detected lymph-node micrometastases correlates with the worse prognosis for patients with histologic node-negative gastric cancer.