Thirty-gene pharmacogenomic test correlates with residual cancer burden after preoperative chemotherapy for breast cancer.

PURPOSE We examined whether the response predicted by a 30-gene pharmacogenomic test correlated with the residual cancer burden (RCB) after preoperative chemotherapy with paclitaxel, 5-fluorouracil, doxorubicin, and cyclophosphamide (T/FAC). EXPERIMENTAL DESIGN Gene expression profiling was done at diagnosis in 74 patients with stages I to III breast cancer and was used to calculate a pharmacogenomic score and predict response to chemotherapy [pathologic complete response (pCR) or residual disease (RD)]. All patients received 6 months of preoperative T/FAC. Following pathologic review, a RCB score was calculated based on residual tumor and lymph node features. Four RCB classes were assigned; RCB-0 (pCR), RCB-I (near-PCR), RCB-II (moderate RD), and RCB-III (extensive RD). The correlations between the pharmacogenomic score, predicted pathologic response, RCB score, and RCB class were examined. RESULTS Thirty-three patients were predicted to have pCR, and 40 were predicted to have RD. Observed responses were RCB-0: n=20 (27%); RCB-I: n=5 (7%); RCB-II: n=36 (49%); and RCB-III: n=13 (16%) patients. Pharmacogenomic and RCB scores were correlated (Pearson's R=-0.501, P<0.0001). There was no difference between the mean genomic predictor scores for RCB-0/I groups (P=0.94), but these were different from the mean scores of the RCB-II/III groups (P<0.001). Among the 25 patients with RCB-0/I response, 19 (76%) were predicted to achieve pCR. The pharmacogenomic test correctly predicted RD in 92% of the patients with RCB-III, which corresponds to chemotherapy-resistant disease. CONCLUSIONS The 30-gene pharmacogenomic test showed good correlation with the extent of residual invasive cancer burden measured as both continuous and categorical variables.