Premium
Risk factors and a predictive nomogram for lymph node metastasis of superficial esophagogastric junction cancer
Author(s) -
Zhu Min,
Cao Bin,
Li Xiao,
Li Peng,
Wen Zixian,
Ji Jiafu,
Min Li,
Zhang Shutian
Publication year - 2020
Publication title -
journal of gastroenterology and hepatology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.214
H-Index - 130
eISSN - 1440-1746
pISSN - 0815-9319
DOI - 10.1111/jgh.15004
Subject(s) - nomogram , medicine , receiver operating characteristic , confidence interval , odds ratio , oncology , logistic regression , multivariate analysis , metastasis , area under the curve , lymph node , cancer , radiology , surgery
Background and Aim No predictive model for lymph node metastasis (LNM) of superficial esophagogastric junction (EGJ) cancer exists. This study aimed to evaluate incidence, identify risk factors, and develop a predictive nomogram for LNM in patients with superficial EGJ cancers. Methods Data were extracted from the Surveillance, Epidemiology, and End Results database for model development and internal validation. Another data set was obtained from two hospitals for external validation. A nomogram was developed based on independent risk factors that resulted from a multivariate logistic regression analysis. Internal and external validations were performed to assess the performance of nomogram model by receiver operating characteristic and calibration plot. Results Prevalence of LNM was 11.41% for intramucosal cancer and increased to 26.50% for submucosal cancer. On the multivariate analysis, large tumor size (odds ratio [OR] = 1.42; P < 0.001), moderately and poorly/un‐differentiated pathological type (OR = 5.62 and 7.67; P = 0.024 and 0.008, respectively), and submucosal invasion (OR = 2.73; P = 0.004) were independent risk factors of LNM. The nomogram incorporating these three predictors demonstrated good discrimination (area under the estimated receiver operating characteristic curve [AUC]: 0.74; 95% confidence interval [95%CI]: 0.68, 0.80) and calibration (mean absolute error was 0.012). Moreover, the discrimination in the internal and external validation sets was good (AUC: 0.73 [95%CI: 0.66, 0.81] and 0.74 [95%CI: 0.60, 0.89], respectively). Nomogram provided better clinical usefulness as assessed by a decision curve analysis. Conclusions Prevalence of LNM in superficial EGJ cancer was high. The first risk‐predictive nomogram model for LNM of superficial EGJ cancer may help clinicians to decide optimal treatment option preoperatively.