Predicting patients at low probability of requiring postmastectomy radiation therapy.

Abstract

Postmastectomy radiation therapy (PMRT) is recommended for patients with four or more positive lymph nodes (LN+). Given the ramifications of PMRT for immediate reconstruction, we sought to create a model using preoperative and intraoperative factors to predict which patients with a positive sentinel lymph node will have less than four LN+. The database from a prospective multicenter study of 4,131 patients was used for this analysis. Patients with one to three positive sentinel lymph nodes (SLN) and tumors < 5 cm (n = 1,133) in size were randomly divided into a training set (n = 580) and a test set (n = 553). Multivariate logistic regression was used on the training set to create a prediction rule that was subsequently validated in the test set. Median patient age was 57 (range, 27-100) years, and median tumor size was 2.0 (range, 0.2-4.8) cm. In the training set, factors associated with having four or more LN+ on multivariate analysis were: tumor size [odds ratio (OR) = 2.087; 95% confidence interval (CI): 1.307-3.333, P = 0.002), number of positive SLN (P < 0.0005), and proportion of positive SLN (OR = 3.602; 95% CI: 2.100-6.179, P < 0.005). A predictive model was established with a point assigned to each positive SLN, T2 (vs. T1), and if proportion of positive SLN was > 50%, for a maximum of five points. In both the training and test sets, patients with one point had a low probability of having four or more LN+ (3.8% and 3.3%, respectively). Tumor size, number of positive SLN, and the proportion of positive SLN influence whether patients will have four or more LN+. A simple model can predict the probability of requiring PMRT.