Development and External Validation of a Nomogram to Predict the Benefit of Regional Node Irradiation in Patients with pT1-2N1M0 Breast Cancer

Purpose/Objective(s)To establish a nomogram for predicting the value of regional node irradiation (RNI) in reducing locoregional recurrence (LRR) in patients with pT1-2N1M0 breast cancer.Materials/MethodsA total of 4262 women with pT1–2N1M0 breast cancer treated at two institutions between 1999 and 2014 were retrospectively reviewed. All had received lumpectomy or mastectomy and axillary lymph node dissection without neoadjuvant chemotherapy. All received whole breast irradiation ± RNI if underwent breast-conserving surgery. Patients from the Cancer Hospital, Chinese Academy of Medical Sciences (CHCAMS, n=2745) formed the training set and patients from the Sun Yat-Sen University Cancer Center (SYSUCC, n = 1517) the validation set. The prognostic model for LRR was obtained from the full multivariate Cox proportional-hazard model by backward selection. The model with the smallest Akaike Information Criterion was selected as the final model. The area under the receiver operating characteristic curve (AUC) was calculated to assess the accuracy of the model.ResultsThe median age was 48 (range, 23-84) years. Four hundred and four (9.5%) patients had lumpectomy and 3858 (90.5%) patients had mastectomy. A total of 3995 (93.7%) patients received adjuvant chemotherapy. Among 3296 patients with hormone receptor (HR) positive disease, 2862 (86.8%) patients received endocrine therapy. Among 859 patients with Her2 positive disease, 233 (27.1%) patients received anti-Her2 targeted therapy. There were 576 (21.1%) patients in CHCAMS and 356 (23.5%) patients in SYSUCC who had received RNI. After median follow-up time of 70 months (range, 6-226 months), 332 (7.8%) patients had LRR. The 5-year LRR was 4.0% and 7.2% with and without RNI (P = 0.001). The nomogram incorporated seven risk factors derived from multivariable analysis of the training set including age, tumor quadrant, T stage, number of positive nodes, histological grade, HR status, and RNI. Our nomogram showed good discrimination in both the training set and the validation set (the C-indexes were 0.717 [95% CI: 0.679-0.756] and 0.712 [95% CI: 0.650-0.773], respectively). For 3-year and 5-year LRR, the AUC was 0.743 and 0.717 in the training set, and 0.723 and 0.728 in the validation set. Both in the training and validation sets, the calibration curve showed good agreement between the nomogram-based predictions and the actual observations.ConclusionThe proposed nomogram appears to accurately estimate the LRR risk of an individual case with and without RNI and could help radiation oncologists to decide on use of RNI for patients with pT1-2N1M0 breast cancer.