The 21-gene Recurrence Score (RS) (Oncotype DX®; Genomic Health, Redwood City, CA) is the most valid and reliable multigene assay to predict prognosis or response to chemotherapy in hormone receptor-positive invasive breast cancer patients. In Japan, however, the test is not frequently used because of its expensive and no coverage by national insurance. We have developed a model to predict low recurrence risk (low-RS) using 220 patient data from St. Luke’s International Hospital, Tokyo, Japan (presented at San Antonio Breast Cancer Symposium 2016). The model with 4 factors, including the histologic type (invasive ductal or lobular), the expression level of PgR (Allred score 7,8 or < 6) and Ki67 (< 24 or > 24), and the presence of lymphovascular invasion, showed that 92% of patients with high PgR positive and Ki67 < 24 could be classified as low-RS with an AUC of 0.843 (95%CI: 0.790-0.896). The aim of this study was to validate our prediction model with external patient data.
A validation set of clinicopathological data from 77 patients who had primary invasive carcinoma surgically resected and underwent OncotypeDx® was obtained from Showa University, school of medicine, Tokyo, Japan.
According to the distribution of 4 factors between two cohorts, there was a significant difference in Ki67 level (<0.01) but not in histologic type, the expression level of PgR, and lymphovascular invasion. The prediction model of the validation cohort had still high enough discriminatory ability having an AUC of 0.705 (95%CI: 0.584-0.827).
Regardless of the inconsistency of Ki67 level between institutes, our model could provide useful information to predict low-RS in hormone receptor-positive invasive breast cancer patients. This model would be helpful to select patients who had better apply OncotypeDX®.
Yasue Tsuchida
None
All authors have declared no conflicts of interest.