119P Characterisation of distinct subgroups of BRCA1-like triple-negative breast cancers with multiomics data analysis

Abstract

Patients with triple-negative breast cancer (TNBC) that have a BRCA1-like phenotype - a specific pattern of DNA copy number aberrations similar to that of BRCA1-mutated tumours - benefit significantly from intensified platinum-based chemotherapy (IPB-chemo) compared to those without. To further molecularly characterise BRCA1-like TNBC tumours, we set out a multiomics analysis making use of published data sets. We made use of 3 clinical data sets (TCGA, RATHER, MATADOR). Sparse generalised canonical correlation analysis and logistic regression were performed on the multiomics data sets to identify features most strongly correlated with BRCA1-like status. Next, to characterise the subgroups, differential gene expression, active-subnetwork-oriented pathway enrichment analyses and cell type deconvolution of the bulk tissue mRNA-seq data were performed. We identified n = 84/1097, 62/126 and 46/495 BRCA1-like TNBCs in the TCGA, RATHER and MATADOR cohorts respectively. Two subgroups of BRCA1-like TNBCs were discerned that are characterised by a more or less profound copy-number loss on chromosome 5q. Deconvolution analyses revealed that BRCA1-like tumours with less 5q loss (n=65/130) are associated with higher tumour-infiltrating lymphocyte levels (p=0.001) and higher immune cell type percentages (p=0.006). This BRCA1-like subgroup was characterised by a gene expression profile that strongly associates with a number of cell-mediated immunity related pathways, such as the PD-L1/PD-1 pathway (p<0.001) and natural killer cell mediated cytotoxicity (p<0.001). By analysing multiomics data from n = 192 patients with BRCA1-like triple-negative breast cancer, we found indications for the presence of distinct subgroups of which one has immunomodulatory features. How these BRCA1-like, TNBC subgroups may direct immunotherapy strategies will be addressed in follow-up studies.