A BRCA1-Like, 25-Gene Assay Predicts for Anthracycline-Chemosensitivity in Sporadic Triple-Negative Breast Cancer.

Abstract

Abstract Background Studies have shown higher chemosensitivity to anthracyclines in BRCA1-associated breast cancer (BABC) when compared to sporadic triple-negative breast cancers (TNBC), possibly due to differences in DNA repair function. We hypothesized that a subset of TNBC with acquired BRCA1 deficiency and defective DNA repair function will benefit most from DNA-damaging agents, such as anthracyclines. Methods We applied a previously published BRCA1 gene expression signature that differentiates BABC from sporadic TNBC to three datasets of sporadic TNBC from Baylor College of Medicine (BCM, n=68), GSE2034 (n=49), and the Netherlands Cancer Institute (NKI2, n=40). The signature separated the sporadic TNBC samples into those with a gene expression profile similar to BABC, or BRCA1-like, versus those with an expression pattern similar to sporadic TNBC, nonBRCA1-like. A list of 92 genes was obtained from the overlap of the most differentially expressed genes between the BRCA1-like samples and nonBRCA1-like samples in each of the three datasets. We then confirmed a subset of the 25 most differentially expressed genes by quantitative RTPCR. We validated the predictive value of this BRCA1-based, 25-gene assay in anthracycline response in three neoadjuvant studies of fluorouracil, epirubicin, and cyclophosphamide (FEC 6 cycles, n=53), doxorubicin and cyclophosphamide (AC 4 cycles, n=12), and T-FAC (paclitaxel-FAC, n=16). Results We determined gene expression of the 92 candidate genes by RT-PCR on 30 available samples of the BCM database. 25 genes were found to have the highest correlation between the microarray and RTQPCR gene expression. Gene expression profile using these 25-gene assay was obtained for three databases which included neoadjuvant anthracycline response data. The 25-gene assay predicted for anthracycline response in sporadic triple-negative breast cancers. In a neoadjuvant FEC study, this assay predicted for pathologic complete response (pCR) in 14/25 patients with BRCA1-like pattern, vs. 7/25 with sporadic-like pattern, p<0.05. In the AC study, 6/9 patients in the BRCA1-like group achieved pCR, vs. 0/3 in nonBRCA1-like group, p<0.05. Finally, in the T-FAC study, 5/7 patients in the BRCA1-like group achieved pCR vs. 3/9 patients in the nonBRCA1-like group, p=0.15. Analysis of the microarray data of triple negative breast cancer revealed higher PARP1 expression levels in the BRCA1-like group when compared to nonBRCA1-like group. Conclusion We present a promising BRCA1-based 25-gene assay that can be used on formalin-fixed paraffin-embedded tissue that may guide therapy in triple- negative breast cancer. The assay differentiates TNBC that are very sensitive to anthracyclines, and it should now be tested and validated prospectively in clinical trials with anthracyclines, other DNA-damaging agents, and PARP1 inhibitors. Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 110.