Disclosed are methods of predicting the likelihood of long-term survival without recurrence of breast cancer for a subject having estrogen receptor-positive (ER+) breast cancer treated with adjuvant endocrine monotherapy. In various embodiments, these methods comprise performing a gene expression profile of a breast tissue sample of substantially all of the genes of the “CADER set” described herein; calculating a risk score using a regression model; and applying a double median cutoff classification to assign the subject to a sensitive, indeterminate or resistant group, wherein assignment to a sensitive group predicts longer relapse-free survival compared to the median relapse-free survival of ER+ breast cancer patients treated with adjuvant endocrine monotherapy.
