IntroductionWhereas in mucosal carcinoma of the esophagus endoscopic local treatment is undisputably the recommended treatment, submucosal esophageal carcinoma reveals a higher percentage of lymph node metastases and consequently oncologic esophageal resection with systematic lymph node dissection is indicated. Increasingly, these carcinomas are treated by means of endoscopic submucosal dissection (ESD), although the diagnostic methods for the detection of potential lymph node metastases (endoscopic ultrasound, computed tomography and PET) are marked by an insufficient sensitivity and specifity. It was the aim of our study to produce a model for the prediction of lymph node metastases in early esophageal carcinoma.MethodsWe analyzed the clinical and histopathological data of 68 consecutive patients with a pT1-esophageal carcinoma in whom an oncological resection was performed in the Department of General and Abdominal Surgery, Mainz, Germany. Submucosal carcinomas (pT1b) were classified according to sm level I-III. In a first step we made a univariate analysis (one-way ANOVA) to test the predictive value of the following potential parameters with regard to the lymph node status (positive/negative): age, sex, histologic tumor type, tumor size, grading, mucosa-/submucosa-type, R-classification, pL, pV, localization of the tumor, and surgical technique (transhiatal / transthoracic approach).All significant variables of the univariate analysis were included in the multivariate analysis. For this purpose we used a general discrimination analysis (forward stepwise). In a third step, the Kruscal-Wallis test with post-hoc comparisons was applied to define the cut-off value of the parameter tested as significant which displays the risk category of a lymph node positivity. ResultsIn the univariate analysis, tumor size (p=0.004), grading (p=0.034), pL (p=0.008), pV (p=0.05) and sm-type (p=0.035) proved to be statistically significant. The variables which attained significance in the multivariate analysis were tumor size (p=0.02) and pV (p= 0.034). In the Kruscal-Wallis test with post-hoc comparisons, the cut-off value of the tumor size was 1 cm (model p=0.014 and between the categories p=0.05).ConclusionThe risk category of lymph node positivity in early esophageal cancer exists according to the prediction model on the basis of the tumor size of >1 cm and a venous infiltration. The hitherto usual sm-level I-III classification of submucosal carcinomas of the esophagus plays a minor role with regard to the prediction of potential lymph node metastases and consequently the indication for endoscopic or surgical therapy.