Clinicopathological variables associated with lymph node metastasis in submucosal invasive gastric cancer

Background We aimed to elucidate clinicopathological variables associated with lymph node metastasis of submucosal invasive gastric cancer. Methods Specimens were surgically resected from 201 patients who had primary submucosal gastric cancer. We studied 39 consecutive patients with lymph node metastasis and 162 patients without lymph node metastasis. We compared the following clinicopathological characteristics of the patients in relation to lymph node metastasis: age, sex, tumor size, histology, extent of submucosal invasion, lymphatic and venous invasion, and ulceration of the tumor. Submucosal invasion was divided subjectively into sm1, sm2, and sm3 (representing invasion of the upper-, middle-, and lower-third of the submucosa, respectively). We also studied the relationship between lymph node metastasis of submucosal gastric cancer and immunohistochemistry for p53, Ki67, vascular endothelial growth factor (VEGF), α-fetoprotein, sLe^a, and dendritic cells (DCs). Results In terms of conventional pathological factors, lymph node metastasis in submucosal gastric cancer was related to tumor size (P = 0.002), depth of submucosal invasion (P = 0.001), lymphatic invasion (P < 0.0001), and venous invasion (P = 0.012). Lymph node metastasis in sm1 gastric cancer was significantly related to VEGF expression (P = 0.047). Also, lymph node metastasis in sm3 gastric cancer was significantly correlated with DC expression (P = 0.016). Multivariate analysis showed that tumor size, tumor invasion depth in the submucosal layer, and lymphatic invasion were independent predictors of nodal metastasis in submucosal gastric cancer. Conclusion Conventional pathological factors, such as tumor size, depth of submucosal invasion, and lymphatic invasion, have a significant influence on lymph node metastasis. VEGF expression and DC expression may be helpful predictors of lymph node metastasis in patients with sm1 and sm3 gastric cancer, respectively.