Background: Therapeutic approaches for breast cancer (BC) are informed by tumor subtypes and molecular heterogeneity, both of which may be associated with particular patient ancestries. While patients of African ancestry (AA) exhibit a relatively high mortality rate compared with patients of European ancestry (EA), the biological underpinnings of this disparity are not entirely understood. Here, we present a genomic and transcriptomic comparison between BC tumors from AA and EA patients stratified by clinical features in a real-world cohort.

Methods: De-identified records from AA (n=629) and EA (n=2832) patients with BC were selected from the Tempus Oncology Database. All tumors underwent sequencing with the Tempus|xT assay, including targeted-panel DNA and full-transcriptome RNA-seq. Race metadata was obtained from abstracted records. Genetic ancestry was estimated using a set of ancestry-informative markers and a supervised version of the ADMIXTURE algorithm. Mutational prevalence, gene expression, and transcriptional signatures were compared between EA and AA groups, and stratified by stage and subtype. Statistical significance was assessed with Bonferroni correction applied to adjust for multiple testing.

Results: Genetic ancestry estimates and race metadata had strong concordance, with 97% and 91% of race data matching genetic ancestry categories in AA and EA, respectively. Compared to EA patients, AA patients were slightly younger (median age 53.8 years vs. 56.1 years) and more likely to have stage IV disease (59.0% vs. 56.5%). The percentage of triple-negative BC (TNBC) patients was much higher among AA (23.5%) compared to EA (14.3%), while the proportion of HR+/HER2− patients was higher in EA (46.4%) compared to AA (36.5%). In HR+/HER2- patients, AA patients had a higher prevalence of KMT2C (24% vs. 14%), GATA3 (21% vs. 14%), and FGFR1 (18% vs. 12%) mutations, and a lower prevalence of PIK3CA (33% vs. 42%) and PTEN (6% vs. 13%) mutations. PRSS23 expression and estrogen response late signaling were significantly downregulated in late-stage HR+/HER2- AA patients, while PTEN signaling was downregulated in stages I-III. Among TNBC, AA patients had a higher prevalence of KMT2C (23% vs. 12%) mutations. LILRA6 and PTEN were significantly downregulated in early-stage TNBC patients of AA.

Conclusions: After controlling for stage at diagnosis and subtype, we observed significant differences in BC mutational spectrums, gene expression, and relevant pathway activities between patients with genetically determined AA and EA, particularly within the TNBC and HR+/HER2- subtypes. These findings may guide future development of treatment strategies by providing data for biomarker-informed research and precision cancer care.

VIEW THE POSTER

VIEW THE PUBLICATION