Correlation between miRNA and gene expression profiles and response to neoadjuvant chemotherapy in patients with locally advanced and inflammatory breast cancer.

Abstract

548 Background: Gene expression profiling may predict for pathologic complete response (pCR) after neoadjuvant therapy (NT). Identification of both miRNA and gene expression may yield novel targets and improved NT. Methods: Patients (pts) with HER2− locally advanced/inflammatory breast cancer (BC) were randomized to receive docetaxel, doxorubicin, cyclophosphamide (TAC, arm A), or A and C given every 2 weeks × 4, followed by carboplatin and nab-paclitaxel (arm B). Pts with HER2+ BC received NT similar to arm B, with trastuzumab (arm C). Core biopsies were snap-frozen prior to NT, and RNA was extracted for gene array analysis (MammaPrint, BluePrint: basal, luminal, and HER2 profiling, and Agilent 44K microarray), and for deep sequence miRNA analysis with the Solexa/Illumina platform. Pathological assessment included grade, ER, HER2 expression, and evidence of residual BC vs.pCR. Unsupervised hierarchical clustering analysis and Fisher’s exact tests were utilized to assess for correlation between miRNA and gene expression, and pathologic assessment. Significant analysis of microarrays (SAM) was used to identify miRNAs differentially expressed and predictive for response (pCR vs. other) to NT. Results: miRNA expression was measured in 72 of 119 enrolled pts; 487 of the 882 miRNAs sequenced, were expressed with raw counts of > 10 in at least 3 pts. miRNA expression patterns revealed three distinct pt clusters, which were significantly associated with the 3 BluePrint subtypes and ER status (Fisher’s exact test, P-values < 1e-5), and moderately associated with HER2 status and MammaPrint classes at borderline significance (P-value: 0.08 and 0.05). SAM identified 6 up and 9 down-regulated miRNAs associated with pCR in HER2− patients (N=43, 4 pCRs) with ≥2 fold changes at false discovery rate (FDR) < 10%. No significant signals of miRNA expression associated with pCR were found in HER2+ pts (N=28; 13 pCRs) at the same FDR level. Conclusions: Combined assessment of miRNA and gene expression patterns is feasible, and pending validation, may yield information leading to individualized and improved NT.