posters International Association for Breast Cancer Research 2014

SOX10 is an independent marker of poor prognosis in triple-negative and basal-like breast cancer (#91)

Ashwini Raghavendra 1 , Jodi M. Saunus 1 , Ana Cristina Vargas 1 , Chanel E. Smart 1 , Sarah Song 2 , Jason Madore 1 , Lynne Reid 1 , Mythily Mariasegaram 1 , Colleen Niland 1 , Darina Vuong 3 , Sarah Sim 3 , Margaret C. Cummings 1 3 4 , Peter T. Simpson 1 , Sunil R. Lakhani 1 3 4
  1. UQ Centre for Clinical Research, University of Queensland, Herston, Qld, Australia
  2. Institute for Molecular Bioscience, University of Queensland, St Lucia, Qld, Australia
  3. Pathology Queensland, Qld Health, Royal Brisbane and Women's Hospital, Qld, Australia
  4. UQ School of Medicine, University of Queensland, Herston, Qld, Australia

SOX10 belongs to the SOX family of transcription factors. It is an important regulator of early neural crest development, melanocyte differentiation and melanoma biology. Triple-negative (TN)/basal-like breast cancer is a clinically heterogeneous group lacking in prognostic biomarkers and therapeutic targets. Recent studies reported that SOX10 is over-expressed in these tumours. The aims of the current study were: i) to investigate prognostic significance of SOX10 expression in a large cohort of invasive breast cancers (n=449) with long-term clinical follow up data (25 years); ii) to investigate SOX10 mRNA expression in published datasets (e.g. TCGA); and iii) to evaluate SOX10 protein expression in normal and tumour-adjacent normal breast. 

At the mRNA level, SOX10 expression is enriched in the basal-like molecular subtype and shows a bimodal distribution. At the protein level, we found SOX10 was predominantly expressed in subsets of TN (44%; 32/73) and basal-like (49%; 28/57) breast cancers, and associated with markers of poor outcome (e.g. tumour grade, lymph node status). SOX10 expression predicted poor survival (p<0.0001; Kaplan Meier analysis) and was the most powerful prognostic indicator in a multivariate (stepwise Cox Proportional Hazards) model after excluding TN status (p=0.0004). Critically, it also predicted poor survival within the TN/basal-like group (p=0.0168; n=54). In both normal (reduction mammoplasty; n=18) and tumour-associated normal (n=19) breast, SOX10 localised to luminal or myoepithelial compartments, or both, but this was mutually exclusive with whole lobules displaying the same expression pattern. Interestingly, this in situ analysis revealed a striking correlation between SOX10 and c-kit expression, and an inverse association with ER. Future studies will focus on investigating whether SOX10 may play a role in regulating luminal progenitor cell function.