NEW YORK (GenomeWeb News) – In a study published online last night in PLOS Genetics, researchers found evidence for a synthetic association between a cluster of common variants associated with prostate cancer and a rare variant that dials up risk of the disease.
Authors of the study, members of an international team led by investigators at the Institute of Cancer Research, used custom arrays to do fine-mapping on a chromosome 17 locus housing HOXB, a gene suspected of contributing to prostate cancer risk based on prior research.
In the process, the researchers uncovered a small set of low penetrance common variants that appear to tag a rare, causal variant in the region. They noted that risk prediction estimates done using that common set of variants alone would likely underestimate true disease risk, highlighting a potential source of "missing heritability" in genome-wide association studies for prostate cancer and other conditions.
"[T]his appears to be an examples of a rare variant with a sufficiently large effect size to create a synthetic association signal detected through partially correlated, yet significantly more common variants," corresponding author Zsofia Kote-Jarai, with the Institute of Cancer Research, and colleagues wrote.
"Our study does not imply how widespread this phenomenon may be," Kote-Jarai said in a statement, "but it demonstrates the importance of identifying the causal genetic changes behind the many common variants that have already been shown to influence risk of disease."
For their study, the researchers directly genotyped some 700 SNPs across the HOXB-containing chromosome 17 region in 20,440 individuals with prostate cancer and 21,469 without, using Illumina's iSelect custom array. They also imputed nearly 3,200 more SNPs with the help of additional data from phase 1 of the 1000 Genomes Project and from 677 prostate cancer cases assessed through the UK Genetic Prostate Cancer Study.
From those common variants, the team tracked down four common variants upstream of the HOXB13 gene that showed significant ties to prostate cancer.
The variants spanned a stretch of sequence that also included rs138213197, a rare variant implicated in hereditary prostate cancer risk through a study of heritable prostate cancer risk that included chromosome 17 re-sequencing.
But because rs138213197 was too rare to be part of the researchers' original imputed variant set, they decided to use Sequenom and Taqman assays to take a closer look at the rare variant and its neighboring SNPs in samples from another 5,500 individuals with prostate cancer and 4,923 without.
Through further analysis of samples genotyped for both rare and common SNPs in the region, the study's authors found evidence supporting the notion that the newly detected cluster of prostate cancer-associated common SNPs likely tags the rarer causal variant at rs138213197.
Though they cautioned that there is a possibility that the common variants are making regulatory contributions to prostate cancer risk, "the much stronger evidence for significance for the rare coding variant coupled with the results of our haplotype analyses appear to indicate that this SNP is solely responsible for the detected association signal."
Such findings are useful for understanding some of the biological underpinnings of prostate cancer risk, the team noted. But they also point to the possibility of finding other situations involving synthetic associations between common and rare variants influencing specific traits or conditions.
The more common SNPs tagging this rare variant would theoretically have been unearthed through very large GWAS with enough power to pick up subtle associations between prostate cancer and common variants, the researchers noted. Even so, if those common variants had been identified first, they argued the causal variants could only have been detected with much more digging.
Moreover, the team cautioned that attempts to account for heritability in cases where common variant associations tag more pronounced, causal ties to a condition could lead to underestimates in risk effect sizes.
"Despite the fact that in this instance the rare, putative causal variant was discovered prior to that of the more common tag SNP," Kote-Jarai and co-authors wrote, "this [prostate cancer] susceptibility locus still serves as a useful illustration of the potential range of causal variation underpinning GWAS association signals, as well as the potential pitfalls of attempting to elucidate candidate causal variants."