Nottingham clinico-pathological response index (NPRI) after neoadjuvant chemotherapy (Neo-ACT) accurately predicts clinical outcome in locally advanced breast cancer

Abstract

© 2014 AACR. Purpose: There is a need to identify more sensitive clinicopathologic criteria to assess the response to neoadjuvant chemotherapy (Neo-ACT) and guide subsequent adjuvant therapy. Experimental Design: We performed a clinicopathologic assessment of 426 patients who had completed Neo-ACT for locally advanced breast cancer (LABC) with a median follow-up of 70 months. Patients were divided into a training set treated with anthracycline combination chemotherapy (n =172); an internal validation set treated with anthracycline and taxane (n =129); and an external validation set treated with anthracycline with or without taxane (n =125). Results: A multivariate Cox regression model demonstrated the absence of fibrosis, presence of lymphovascular invasion, increasing number of lymph node metastases, and administration of hormone therapy were significantly associated with short breast cancer-specific survival (BCSS) and disease-free survival (DFS); Ps < 0.01, while reduction of tumor size was associated with DFS (P =0.022). Nottingham Clinico-Pathological Response Indexes (NPRI) were calculated, and four prognostic groups (NPRI-PG) were identified. Patients in prognostic group 2 (NPRI-PG2) for BCSS (66 of 172; 38.4%) have the same prognosis as those who achieved pathologic complete response (pCR; NPRI-PG1; 15%). Receiver-operating characteristic (ROC) curves indicated that the NPRI outperformed the currently used prognostic factors and adding the NPRI improved their performance as a predictor for both BCSS (area under the curve [AUC], 0.88) and DFS (AUC, 0.87). Conclusions: The NPRI predicts BCSS and DFS, with a higher sensitivity than pCR. The NPRI can also improve the sensitivity and specificity of clinicopathologic response as a study endpoint, for assessing response to Neo-ACT, and can serve as a valuable tool for the discovery of future predictive molecular markers. Clin Cancer Res; 21(5); 1052-62.