CT and Pathological Control Study on Small Cluster Lymph Nodes Metastases of Esophageal Cancer and Construction of Nomogram Prediction Model

Abstract

Aim of this study was to explorer the predictor of small cluster lymph nodes metastasis of esophageal cancer and construction of nomogram prediction model. We retrospectively analyzed the 202 patients with pathologically confirmed esophageal cancer from Fujian Provincial Tumor Hospital between January 2008 and December 2009. The clinical and imaging features of preoperative enhanced CT of small clustered lymph nodes in the mediastinal and supraclavicular region were recorded in detail, and compared with postoperative pathological data. We have identified using single factor χ2 test or Fisher's exact test to explorer the predictors of small cluster lymph nodes metastasis of esophageal cancer, and using Logistic multivariate analysis. Based on binary Logistic regression analysis results, a nomogram model was established by using R software. The predicting accuracy of the nomogram was evaluated by C-index and calibration curve. The total of 961 small lymph node stations of 202 patients met with inclusion criteria were observed. The pathological positive rates of small cluster lymph nodes group and small non-clustered lymph nodes group were 55.3%(187/338) and 16.7%(104/623),The difference was statistically significant (P<0.05).Univariate χ2 test or Fisher's exact test showed that histological grading, vascular invasion and the short-axis diameter of lymph nodes were significantly associated with small cluster lymph node metastasis of esophageal cancer in supraclavicular area and mediastinal (all P < 0.05). Logistic multivariate analysis showed that histological grading, vascular invasion and the short-axis diameter of lymph nodes (≥0.545cm) were independent predictor factors for small cluster lymph node metastasis (P < 0.05).After the predictor factors were include into the nomogram, and its AUC was 0.5931 (95%CI 0.5363-0.6498).The calibration curve show that the predicted and actual lymph node metastasis were consistent. Histological grading, vascular invasion and the short-axis diameter of lymph nodes (≥0.545cm) were independent predictor factors for small cluster lymph node (≥ 3 lymph nodes clustered and nonfusion) metastasis of esophageal cancer patients in the supraclavicular and mediastinal and construction of nomogram prediction model.