High accuracy gene signature for chemosensitivity prediction in breast cancer

Abstract

Neoadjuvant chemotherapy is usually offered to breast cancer patients to shrink the tumor and any involved lymph nodes. Pathologic complete response is a way of measuring response to this treatment, which is normally defined as the absence of residual invasive disease in the breast and in the axillary lymph nodes at the completion of the treatment. Clinically patients who achieve pCR tend to have a very good prognosis. As a standard treatment for breast cancer, neoadjuvant chemotherapy offers several benefits to patients. But not all patients benefit equally from this treatment. Therefore it is of great value to investigate what molecular information obtained from a patient’s tumor could be used to determine if the patient would benefit from a particular chemotherapy. In practice the estrogen receptor status could be used to guide the decisions on hormonal therapy. Further, the gene expression data that reflect subtle differences in tumors could be utilized to build a predictor of response to cancer drugs. In this study we sought to use gene expression data to predict who might achieve pCR to sequential anthracycline paclitaxel preoperative chemotherapy. Our aim was to uncover one gene signature for developing predictors of pCR to neoadjuvant chemotherapy that have a much higher accuracy than DLDA-30, so it might have the potential for clinical applications. With the ability to account for multiple gene interactions, the multivariable techniques, such as genetic algorithm and SLR, have demonstrated their utility in identifying robust gene signatures of clinical relevance.