NEW YORK – A team from the National Institutes of Health, the Johns Hopkins Bloomberg School of Public Health, and elsewhere has tracked down nearly three dozen previously unappreciated breast cancer risk loci, including a handful of sites that appear to coincide with specific breast cancer subtypes.
"These findings provide an improved understanding of genetic predisposition to breast cancer subtypes, and will inform the development of subtype-specific polygenic risk scores [PRS]," co-senior authors Montserrat Garcia-Closas, a cancer epidemiologist and geneticist at the NIH, and Nilanjan Chatterjee, a biostatistician at Johns Hopkins, and their colleagues wrote.
In a paper published in Nature Genetics on Monday, the researchers considered array-based genotyping profiles for almost 133,400 individuals with breast cancer and nearly 113,800 unaffected controls, along with more than 18,900 individuals with or without breast cancer who carried germline, cancer-related BRCA1 mutations. The search highlighted 32 new breast cancer-associated loci — a set that included at least five loci showing distinct associations with luminal and non-luminal forms of breast cancer.
By bringing together insights from the new and previously identified breast cancer-related loci, the team estimated that these risk variants explained some 54 percent of heritability when it came to risk of luminal A-like breast cancer. On the other hand, these variants appeared to account for just shy of 38 percent of susceptibility to triple-negative forms of breast cancer.
Still, the authors cautioned that "to expand the generalizability of our findings, these analyses should be replicated and expanded in multi-ancestry populations."
Using iCOGS and OncoArray custom arrays, the researchers assessed genotyping profiles for 118,474 well-characterized breast cancer cases and 96,201 unaffected controls with European ancestry who were enrolled through dozens of studies by the Breast Cancer Association Consortium (BCAC).
"To identify breast cancer susceptibility variants displaying evidence of heterogeneity, we used a novel score test based on a two-stage polytomous model that allows flexible, yet parsimonious, modeling of associations in the presence of underlying heterogeneity by estrogen receptor, progesterone receptor, HER2, and/or grade," the authors explained, noting that this analytical approach was applied exclusively to the invasive cases and controls from BCAC.
To complement those analyses, the team brought in genotypes for 17,788 controls and 14,910 more breast cancer cases without subtype information from other studies, along with data for 18,908 BRCA1 mutation carriers: 9,414 with breast cancer and 9,494 without.
Using these data, the researchers focused in on variants falling at 32 new breast cancer susceptibility loci, including variants identified with the polytomous modeling approach or through a meta-analysis focused on triple-negative breast cancer cases.
Nearly half of the susceptibility loci appeared to have heterogeneous interactions depending on the breast cancer subtype or tumor grade considered, they reported, and a handful of variants had associations going in opposite directions in specific breast cancer subtypes.
In a series of follow-up analyses, the team looked more closely at potential causal variants, variants with subtype-specific associations, and polygenic risk score clues using available regulatory sequence data from prior chromatin immunoprecipitation sequencing studies, annotation clues, linkage disequilibrium data, and more.
"These analyses show the benefit of combining standard GWAS methods with methods accounting for underlying tumor heterogeneity," the authors wrote, noting that "[t]hese methods and results may help to clarify mechanisms predisposing to specific molecular subtypes, and provide precise risk estimates for subtypes to inform the development of subtype-specific PRSs."