Homologous recombination deficiency, RB-loss gene signatures, intrinsic subtype and response to neoadjuvant treatment in HR+/HER2- early breast cancer: a correlative analysis of two phase II trials

Background: Hormone-receptor (HR)+/HER2- breast cancer (BC) is a biologically heterogeneous disease. Homologous recombination deficiency (HRD) and BRCA mutations have been previously reported to be associated with worse outcomes in HR+/HER2- metastatic BC patients receiving CDK4/6 inhibitors and endocrine therapy. Here, we assess the relation between HRD and RB-loss signatures, intrinsic subtyping, the PAM50-based chemo-endocrine score, and response to chemotherapy-based therapy and endocrine treatment in HR+/HER2- early BC. Methods: GIADA is a multicentric neoadjuvant phase II trial that treated premenopausal patients with Luminal B (LumB)-like BC (HR-positive, HER2-negative, with Ki67>20% and/or histologic Grade 3) with a combination of chemotherapy, immunotherapy and endocrine treatment. Expression of 758 genes on baseline tumor samples from all 43 patients was quantified by nCounter platform. The LETLOB phase II trial randomized postmenopausal women with clinical stage II-IIIA HR+/HER2- BC to neoadjuvant letrozole + lapatinib or letrozole + placebo for 6 months (Guarneri, JCO 2014). Gene-expression data (Affymetrix platform) from pre-treatment frozen core-biopsies was available from 66 out of 92 pts enrolled. Intrinsic subtype was assigned using a research-based PAM50 subtype predictor. A published HRD signature (Peng, Nat Commun 2014) and a signature of RB loss (RBsig), previously reported to potentially predict resistance to CDK4/6 inhibitors in HR+/HER2- BC (Malorni, Oncotarget 2016) were computed. The PAM50 based chemo-endocrine score (CES) was calculated using published definition (Prat, CCR 2017). Higher values of CES indicate increased endocrine sensitivity, while lower values indicate chemosensitivity. Association between genomic signatures was assessed through Pearson’s correlation coefficient. Association of genomic signatures with pCR was assessed through logistic regression and association with PEPI scores was assessed through Kruskal-Wallis test. Results: HRD signature levels were significantly higher in non-luminal (Basal-like and HER2-enriched) tumors as compared to Luminal (A or B) tumors (p>0.001 in the GIADA trial, p=0.021 in the LETLOB trial). Moreover, higher levels of HRD signature were associated with higher levels of RB-loss signature (Pearson correlation 0.355, p=0.020 in the GIADA trial; Pearson correlation 0.942, p< 0.001 in the LETLOB trial), higher levels of Basal-like signature (Pearson correlation 0.502, p< 0.001 in the GIADA trial; Pearson correlation 0.373, p=0.002 in the LETLOB trial) and lower levels of CES (Pearson correlation -0.422, p=0.005 in the GIADA trial; Pearson correlation -0.763, p< 0.001 in the LETLOB trial), indicative of higher chemosensitivity. In the GIADA trial, higher levels of HRD signature (p=0.018) and RBloss signature (p=0.073) and lower levels of CES (p=0.007) were associated with higher pCR rates after chemo, endocrine and immunotherapy. In the LETLOB trial, lower levels of HRD signature (p=0.068) and RBloss signature (p=0.042) and higher levels of CES (p=0.050) were associated with higher sensitivity to endocrine treatment (lower PEPI scores, 0 vs 1-3 vs 4 or more, after neoadjuvant letrozole). Conclusions: In HR+/HER2- early BC, HRD gene signatures, RB-loss gene signatures and non-luminal (especially Basal-like) intrinsic subtyping are associated with each other and associated with higher sensitivity to chemotherapy-based therapy and lower sensitivity to endocrine treatment. These observations might help correctly tailor systemic therapy, including biologic agents, in patients with HR+/HER2- early and advanced BC.