Abstract P5-04-04: Identification of AR driven tumors within TNBC using a novel gene signature

Abstract

Abstract Background The androgen receptor (AR) is an emerging biomarker and favourable therapeutic target in breast cancer (BC), especially in Triple negative BC (TNBC) which lacks a targeted treatment and shows high molecular heterogeneity. Proportion of AR positive tumours are known to vary (10 to 43%) within TNBC and recent multi-institutional study on AR positive TNBC, showed AR protein status alone is not a reliable prognostic marker for therapeutic response. A deeper examination of AR regulated pathways is necessary to derive AR regulated gene signatures which could identify TNBC tumours truly driven by AR and amenable to anti androgen therapies. In this study, we attempted to develop AR regulated gene signature using gene expression data derived from cell lines under the influence of androgens using systematic bioinformatic approach and tested the ability of the signature to identify AR regulated tumours within TNBC. Methods Microarray data of AR positive BC cell lines (GSE61368 & GSE28305) was used to derive AR regulated genes by a bioinformatic pipeline, using integrated method by incorporating protein interaction networks, topological parameters, and semantic similarity to arrive at final set of 16 genes. Correlation of these genes with AR protein status by reverse phase protein array (RPPA) was evaluated in 447 tumors of TCGA dataset. A 3rd quartile cut-off in the AR expression (by RPPA) was used to classify them as AR positive and negative and a logistic regression model was constructed using positive AR as determinant to generate AR probability score with minimum set of genes. 3rd quartile cut-off in the probability score was used to classify tumors as AR high and low. Distribution of AR probability score was examined in 144 and 187 TNBC tumors in TCGA and METABRIC, respectively. Comparisons between AR high and low tumors were done for clinical parameters, mutational spectrum, expression of ER regulated genes (FOXA1, GATA3 and TFF1) and EMT gene signatures (Byers et al., 2013, Tan et al., 2014). Luminal androgen receptor (LAR) subtype identification by TNBCtype (Lehmann et. al) was done in both datasets to examine the overlap with our method of identification of AR driven tumors within TNBC. Results Logistic regression model identified 5 (SOCS2, ID1,PIP,GADD45 and ZNF689) of the 16 AR regulated genes as predictors for AR probability score which ranged from 0.009 to 0.68 in 144 TNBC tumors within TCGA data. 41/144 (29%) TNBC tumors were categorized as having AR high at cut off of 0.13. The AR probability score ranged from 0.04 to 0.64 in 187 TNBC tumors within METABRIC dataset and 47/187 (25%) were classified as AR driven tumors at cut off of 0.22. Though differences in the clinical variables like age , tumor size and lymph node(LN) status were not different between the AR high and low groups in the TCGA dataset, a significant increase in age, LN positive tumors was observed in the AR high groups in METABRIC. A significant higher PIK3CA mutations was observed in the AR high tumors both in the TCGA and METABRIC (p=0.012 and p<0.0001 respectively) with a proportionate lower p53 mutations in the AR high tumors in both data sets (p=0.03 and p=0.35 respectively). Further, AR high tumors showed significant higher expression of ER regulated genes (TCGA-p<0.05 & METABRIC, p<0.05), were more epithelial by EMT score (TCGA-p<0.05 & METABRIC, p<0.05). Analysis by TNBCtype showed that 30% of AR high tumors in TCGA and 40% of AR high tumors in METABRIC were identified as belonging to LAR subgroup (Lehmann et al). Conclusion Identification of AR driven tumors within TNBC has both prognostic and predictive utility. Methods using limited number of AR regulated genes could be easily applied to larger BC cohorts to identify TNBC tumours driven by AR signalling and may respond well to anti-androgen therapies. Citation Format: Savitha Rajarajan, Snijesh V.P, Maalavika Pillai, Mohit Kumar Jolly, Jyothi S Prabhu. Identification of AR driven tumors within TNBC using a novel gene signature [abstract]. In: Proceedings of the 2021 San Antonio Breast Cancer Symposium; 2021 Dec 7-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2022;82(4 Suppl):Abstract nr P5-04-04.