Fig 1.
AR status is significantly different between race groups and among molecular subtypes.
(A). AA women have more AR-negative tumor types in each molecular subtype. (B). Within the AR-negative subtypes, there are significantly higher proportions of TNBC basal-like. (C). All TNBC samples were subjected to “Vanderbilt” subtypes. AAs, compared to White AR-negative QNBC patients, had more BL1 (24% v 17%), BL2 (16% v 12%), and IM (24% v 19%) subtypes. Inversely, AR-negative White QNBC patients had more mesenchymal (M) (25% v 20%), mesenchymal stem-like (MSL) (12% v 8%), and unstable (UNS) (14% vs 8%) subtypes compared to AR-negative QNBC AA TNBC patients.
Table 1.
Patient characteristics of TCGA population.
The TCGA invasive breast cancer dataset had the largest patient set of RNA-seq data (primary breast cancers for 180 AAs and 745 Whites) was used to quantify distributions of AR expression across patient groups in order to calculate a suitable threshold to stratify the entire dataset/population as AR-positive or AR-negative categories, based on highest and lowest tertiles, exclusively.
Fig 2.
AR tumor status is associated with younger ages and AR-negative patients have a significantly higher rate of disease progression.
(A). Density plot of ages for each AR status. The mean age of AR positive subtypes is 59, and the mean age of AR negative subtypes is 56. (B). Sub-stratifying ages by race groups indicates that there is a significant difference in the age for AR negative AA (p = 0.034) as compared to AR positive category. (C). AR negative patients compared AR positive patients have a higher rate of disease progression as determined by cumulative survival analysis. (D) AR negative AA patients with basal subtypes have a higher rate of disease progression, even compared to whites with the same tumor subtype. Log-Rank test was used to calculate P values, and significance was determined p<0.05. (E). Kaplan Meier plot shows the overall survival probability in AR-positive and AR-negative patients. (F). Kaplan Meier plot shows the overall survival probability in Whites and African American AR-negative patients.
Table 2.
PAM50 and Vanderbilt gene expressions subtypes stratified by AR expression in total population (quartile).
Fig 3.
Expression heatmap showing comparison between PAM50 genes subtypes the AR-positive vs AR-negative tumors.
Heat Map using PAM50 gene signature compared to AR status. Genes that are enriched in ER and AR-negative subtypes (blue bar) show complete absence of expression in a subset of the AR-positive subtypes that is traditionally categorized as ‘unclassified’ subtypes (blue arrow). Genes enriched in the Hormone Receptor (HR) subtypes are typically decreased in the AR-negative subtype (red bar), including some samples that would normally considered ‘unclassified’ or HR-positive.
Fig 4.
A. Genes most highly associated (bivariate cutoff 1.0E-07) with AR expression across the TCGA dataset were used to determined novel gene expression signatures associated with AR tumor status. Distinct subgroups of genes with shared expression trends were identified using K-means cluster analysis and separated into 5 nodes of genes with expression trends that are either upregulated or downregulated in the AR-negative tumors. B. A subset of genes related to the Immunomodulatory TNBC subtype display statistically significant differences in expression between AA vs White patients when comparing expression in AR-high and AR-low categories.