Integrative Molecular and Pathologic Profiling of Triple Negative Breast Cancer

Abstract

Introduction Breast cancer (BC) is a heterogeneous disease presenting with variable morphologies, molecular profiles, clinical behaviours and responses to therapy. Triple negative breast cancer (TNBC), characterised by the absent expression of the Oestrogen Receptor (ER), Progesterone Receptor (PR) and HER2 overexpression, is an aggressive class of BC for which targeted therapies are lacking. Patients with TNBC comprise approximately 17% of BC patients and show variable responses to chemotherapy, which reflects the transcriptomic and genomic heterogeneity of the disease. In this study, we hypothesised that deciphering TNBC molecular heterogeneity can pave the way for the identification of novel therapeutic targets and the refinement of classification. We investigated large, well annotated cohorts of TNBC using a variety of molecular techniques to: 1) identify genetic predictors of clinical outcomes in TNBC, 2) investigate the mechanistic and prognostic role of novel targets in TNBC including periplakin (PPL), 3) further evaluate the mechanisms of cell adhesion in TNBC. Methods Next generation sequencing was applied to a group of TNBC cases (n=126) with long term follow-up data. Data mining using the supervised neural artificial network (ANN) was performed to identify genes or pathways associated with distant metastasis-free survival (DMFS) and breast cancer-specific survival (BCSS). The prognostic ability of these genes was validated using multiple public domain datasets such as the Breast Cancer Gene-Expression Miner v4. 0 and Genotype 2 outcome datasets. Multivariate Cox regression analysis identified a prognostic gene signature that was independently associated with poor prognosis. Tissue microarrays (TMAs) were constructed to examine the protein expression using immunohistochemistry (IHC). PPL was identified by ANN analysis as a strong prognostic marker with respect to DMFS and BCSS. The prognostic ability of PPL was also validated using multiple public domain datasets such as the METABRIC, Breast Cancer Gene-Expression Miner v4.0, KM plotter and Genotype 2 outcome datasets. This was followed by the evaluation of its protein expression using IHC. The transient knockdown of PPL using siRNA was used to assess its role in proliferation, migration and invasive capabilities of TNBC cell lines (MB-MDA 231 and MB-MDA 436). E-cadherin protein expression was determined using TMAs (n=813). IHC for E-cadherin staining using full-face sections was carried out to confirm the identity of E-cadherin and its role on the global gene expression profile. RNA-seq analysis was completed on invasive ductal carcinoma of the breast (IDC) tumours with either E-cadherin positive or negative/low E-cadherin expression. Genes that differentially expressed E-cadherin were assessed using the Robina implementation of EdgeR. Results ANN identified high expression of a set of genes (ACSM4, SPDYC, PPL, DCTN1-AS1, RP11-29H23.5, RPS10P18, AC079305.10 and PAXBP1-AS1) that can individually be used as indicators of poor prognosis and shorter outcomes in TNBC (p<0.05). Adjusting for clinicopathological factors including the patient age, grade, nodal stage, tumour size and lymphovascular invasion using multivariate Cox regression analysis yielded a two-gene prognostic signature (ACSM4 and SPDYC) which was associated with poor prognosis (p<0.05), independent of other prognostic variables. To validate these findings, the protein expression of these two genes was significantly associated with patient outcomes in both an independent and a combined manner (p<0.05). High PPL mRNA expression was significantly associated with shorter BCSS and DMFS (both P<0.01). PPL mRNA expression was also an independent, poor prognostic predictor for BCSS and DMFS regardless of clinicopathological factors including patient age, grade, nodal stage, tumour size and lymphovascul