Race Associated With Variable Risk of Lymph Node Involvement in Endometrial Cancer

Abstract

Nomograms for predicting lymph node involvement in endometrial adenocarcinoma (EAC) typically include various tumor features, including depth of invasion, grade, and lymphovascular invasion (LVI). Whether demographic factors such as age and/or race impact the risk of nodal involvement is not clear. The purpose of this study was to investigate demographic, clinical, and pathologic predictors of LN involvement among patients with EAC. All EAC patients between 2004-2015 (n = 429,084) in the National Cancer Database (NCDB) were initially identified. Patients with non-adenocarcinoma histology, AJCC tumor stage other than primary pathologic T1b or T2, fewer than 10 LN sampled at the time of surgery, undifferentiated histology, or missing data regarding LVI, tumor grade, tumor size, or LN count were excluded. This left 15,117 patients for analysis. Logistic regression identified predictors of LN involvement. Chi-square tests were performed to assess for statistically significant differences in frequencies between groups. Significance was set at a p value of < 0.05. From 2004-2014, 15,117 patients with EAC were included, of whom 2,756 had positive pelvic and/or paraaortic LN identified at the time of surgical staging. On multivariate analysis, predictors of LN involvement identified as significant on univariate analysis were included, and all factors remained significant, including increasing age per year (odds ratio [OR] 0.986, 95% confidence interval [CI] 0.982-0.991, p<0.001), black race relative to white race (OR 1.32, 95% CI 1.12-1.57, p = 0.001), pathologic T2 relative to T1b (OR 1.65, 95% CI 1.49-1.82, p<0.001), poorly differentiated relative to well differentiated tumor grade (OR 1.35, 95% CI 1.20-1.53, p<0.001), presence versus absence of LVI (OR 5.17, 95% CI 4.71-5.68, p<0.001), increase in tumor size per millimeter (OR 1.004, 95% CI 1.003-1.005, p<0.001), and number of LNs sampled at the time of surgical staging (OR 1.015, 95% CI 1.011-1.020, p<0.001). Multivariate logistic regression of 1,087 black women revealed that age (OR 0.981, 95% CI 0.963-0.998, p = 0.03), tumor grade (OR 3.75, 95% CI 2.25-6.50, p<0.001), and LVI (OR 4.13, 95% CI 3.04-5.68, p<0.001) remained significant predictors of LN involvement. Multivariate logistic regression of 14,030 white women revealed that age (OR 0.979, 95% CI 0.973-0.985, p<0.001), tumor grade (OR 1.44, 95% CI 1.29-1.63, p<0.001), and LVI (OR 5.32, 95% CI 4.85-5.85, p<0.001) remained significant predictors of LN involvement. In addition to more traditional pathologic risk factors for LN involvement, patient race was also found to be a significant risk factor for positive LNs. The odds ratio of pathologic risk factors predicting involved lymph nodes also varied by race, and investigation into the underlying etiology of this difference warranted.