Skip to main content
Premium Trial:

Request an Annual Quote

Several Cancer Types Significantly Tied to Polygenic Risk Scores in New PheWAS

NEW YORK (GenomeWeb) – Results from a phenome-wide association study (PheWAS) suggest polygenic risk scores (PRS) made up of cancer-associated variants are significantly associated with prior cancer diagnoses for at least half a dozen cancer types.

With the help of a longitudinal biorepository called the Michigan Genomics Initiative, investigators at the University of Michigan and elsewhere analyzed genotype and phenotype data for nearly 28,300 individuals of European ancestry. Nearly half of those participants had at least one cancer diagnosis in their medical history, based on electronic health record (EHR) data.

Using the PheWAS approach, the team searched for associations between PRS and a dozen cancer phenotypes or diagnoses, uncovering significant links to breast, prostate, and thyroid cancers as well as melanoma, basal cell carcinoma, and squamous cell carcinoma. The results were published online today in the American Journal of Human Genetics.

"Our results demonstrate that PRS, a summary score constructed based on results of large population-based GWASs, can be potentially useful for cancer risk stratification among patients in an academic medical center," corresponding author Bhramar Mukherjee, a biostatistics, epidemiology, and cancer researcher at the University of Michigan, and her colleagues wrote.

The genetic risk scores coincided with some non-cancer traits as well, though further analysis demonstrated that at least some of the secondary associations stemmed from conditions that were set off by cancer diagnoses or treatment such as those for urinary incontinence in men treated for prostate cancer.

Such observations led the team to perform what they called additional exclusion PRS PheWAS that left out individuals already diagnosed with conditions related to each PRS, prompting a closer inspection of the order in which individuals were diagnosed with the PRS-associated primary and secondary phenotypes.

For their initial PheWAS, the researchers used custom Illumina Infinium CoreExome-24 bead arrays to genotype blood samples from 37,412 individuals enrolled in the Michigan Genomics Initiative. After their quality control steps, they were left with data for 28,260 individuals of recent European ancestry, including 13,490 who had one or more prior cancer diagnoses in their electronic health records.

For the same participants, the team tracked down more than 10,300 distinct International Classification of Disease codes, providing the phenotypic basis for the PheWAS. It also came up with PRS sets that encompassed 253 SNPs by compiling and integrating cancer-associated variants reported in the National Human Genome Research Institute-European Bioinformatics Institute GWAS catalog.

Each of the 12 cancer phenotypes selected for the PheWAS had been linked to five or more independent risk variants in the past, the researchers noted, and at least 250 of the study's participants were affected by each.

The researchers carried out the PheWAS using these data — detecting associations with primary cancer phenotypes for six of the 12 PRS — as well as a genome-wide association study that highlighted new and known risk variants for the cancers considered. They also used available EHR data to explore secondary PheWAS associations, ranging from pre-cancerous conditions to clinical circumstances that can arise after cancer treatment.

While some of the secondary associations disappeared when the team performed exclusion PRS PheWAS, others remained. For example, a PheWAS done with the breast cancer PRS in the absence of individuals diagnosed with breast cancer obviated ties between that PRS and abnormal mammogram. But ties between the thyroid cancer PRS and hypothyroidism remained in a PheWAS done without thyroid cancer patients. 

"Our approach could directly discard trait associations driven by the primary cancer diagnosis," the authors explained, "and has the potential to identify clinically useful diagnostic traits among many that are conveniently measured in panel tests of biomarkers."