Background: Breast cancer can be subclassified into luminal A and B, basal-like and HER2-enriched subtypes, each with distinct biology, prognosis and response to therapy. Pre-clinical studies have suggested that a basal-like subtype is associated with responsiveness to gemcitabine. Additional data suggest that patients with highly proliferating tumors (non-luminal A) might be particularly responsive to gemcitabine. We examined these hypotheses on SBG0102, a randomized trial comparing gemcitabine plus docetaxel (GD) to single agent docetaxel (D) in patients with advanced breast cancer. Secondarily, we analyzed prognosis by PAM50 subtype, which has not previously been addressed in a metastatic setting.

Patients and Methods: RNA was isolated from archival formalin-fixed, paraffin-embedded primary breast tumor tissue from patients randomly assigned to GD or D, and analyzed with NanoString Technologies' nCounter system and PAM50 intrinsic subtyping algorithm. Using time to progression (TTP) as primary endpoint, overall survival (OS) and response rate (RR) as secondary endpoints, relationships between subtypes and outcome after chemotherapy were analyzed by methods prespecified in writing as a formal prospective-retrospective clinical trial correlative study. Data analysis was performed independently by the DBCG statistical core.

Results: RNA from 270 patients was evaluable. 84 patients (31%) were classified as luminal A, 97 (36%) luminal B, 43 (16%) basal-like, and 46 (17%) as HER2-enriched. PAM50 intrinsic subtype was significantly associated with TTP (P = .0006) and OS (P = .0083) by Kaplan-Meier analysis. In the adjusted Cox model PAM50 remained independently prognostic. RR was not significantly different by subtype, and PAM50 was not a significant predictor of TTP by treatment arm. PAM50 was however a highly significant predictor of OS following GD compared to D (Pinteraction=.0016). The 43 patients with a basal-like subtype had a significant reduction in OS events (hazard ratio (HR)=0.29; 95% CI 0.15–0.57; Pinteraction=.0006), although TTP events were not significant by interaction test (HR = 0.39; 95% CI 0.19–0.82; Pinteraction=.22). Kaplan-Meier estimates revealed a gain in median OS of 10 months in the doublet arm compared to the monotherapy arm (18.7 (95% CI, 12.4–23.0) v 8.5 (95% CI, 4.2–11.8) months).

Conclusion: A significantly improved and clinical important prolongation of survival was seen from the addition of gemcitabine to docetaxel in advanced basal-like breast cancer patients. However, we could not show a similar significant reduction in TTP events. These results need validation in a second similar trial or a prospective study stratified for basal-like breast cancer to obtain convincing evidence that identification of a basal-like profile may be used to direct the use of gemcitabine in combination with docetaxel.

Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr P3-06-03.

This abstract has been revised and no longer matches the print SABCS Abstract Book.