Content area
Abstract
Background:Despite advancements in the early detection of breast cancer (BC) and improved access to therapeutic options, the global mortality rate due to BC among women remains high. BC progression is a complex and multistep process. Lymphovascular invasion (LVI), one of the earliest stages of metastasis in BC, is controlled by specialised molecular pathways and is strongly associated with poor outcomes. LVI signifies the expansion of tumour cells towards the peritumoural region inside lymphatic spaces and blood vessels. A successful LVI event requires the involvement of a variety of cellular, biophysical, and biochemical pathways. Because the underlying molecular mechanisms of LVI are complex and involve connections with other oncogenic and metastatic pathways, no targeted treatments are currently available to prevent the progression of LVI and metastatic recurrence. Investigation into the molecular determinants of LVI is critical in elucidating the molecular mechanisms underlying LVI, discovering novel prognostic factors, and identifying possible treatment targets for invasive BC. The genes that play a role in the pathogenesis of LVI in invasive BC have received little attention; it was therefore hypothesised that by conducting bioinformatic analyses, key genes related to LVI in BC could be identified, some of which may have prognostic value, serve as potential therapeutic targets, and have roles in the various mechanisms regulating the development of LVI. This study aimed to identify genes associated with LVI, evaluate the clinicopathological and prognostic significance of the identified genes at the mRNA and protein levels, and investigate their functional roles.
Methodologies:Using artificial neural network (ANN) analysis and the weighted average difference (WAD) method on two transcriptome cohorts with well-characterised LVI status (Molecular Taxonomy of Breast Cancer International Consortium [METABRIC] and the Cancer Genome Atlas [TCGA]), key genes were identified. The candidate genes MX1, IDH2, HMGB3, RRM2, CCNB1,and CCNB2were each assessed at the transcriptome level using the METABRIC and TCGA data (n=1980 and n=854, respectively) and at the proteome level via immunohistochemical (IHC) analysis and tissue microarray (TMA) sections derived from a large, well-characterised invasive BC cohort (n=3,173). The expression of these genes was evaluated with clinicopathological parameters that included LVI, tumour grade, lymph node (LN) stage, tumour size, patient outcome, BC molecular subtypes, and relevant biological markers. In a different approach, in vitrofunctional assays using a panel of BC cell lines were performed to investigate the effect of transient siRNA knockdown of HMGB3and CCNB2on cell proliferation, migration, cell cycle, and apoptosis. Furthermore, the effect of silencing HMGB3and CCNB2on the adherence and transmigration of BC cell lines across dermal lymphatic endothelial cells (DLECs) and human vascular endothelial cells (HUVECs) was evaluated to assess their role in LVI development.
Results:In this study, a 99-gene set associated with LVI status in invasive BC was generated using WAD analysis. In addition, ANN analysis produced a list of 100 genes of interest. From these gene lists, MX1, IDH2,HMGB3, RRM2, CCNB1,and CCNB2were selected for further investigation. All selected genes were significantly associated with LVI positivity at the mRNA level in both the METABRIC and TCGA cohorts, but MX1, RRM2, and CCNB1were not associated with LVI at the protein level. All biomarkers included in this study had a significant association with poor clinicopathological parameters, such as high tumour grade, poor NPI, hormonal receptor negativity, and HER2 positivity at the mRNA and/or protein levels. The results of the survival analysis varied among these biomarkers. High expression of these genes at the mRNA level was significantly associated with poor outcomes; however, only IDH2, HMGB3, and CCNB2expression at the protein level was associated with shorter BC-specific survival (BCSS). In multivariate analysis, the expression of MX1, IDH2, HMGB3, and CCNB2were found to be factors that predicted patient survival independently from the other variables, including tumour size, histological grade, LN stage, and LVI. Additionally, all the selected biomarkers were significantly associated with EMT- and MMP-related markers, such as E-cadherin, N-cadherin, P-cadherin, TWIST2, MMP1, MMP9, and MMP12 at the transcriptome and proteome levels.
