The Wellcome Trust Centre for Human Genetics' Martin Farrall and his colleagues at the Procardis consortium linked two variants in the lipoapoprotein(a) gene to an increased risk of coronary disease and an increased level of Lp(a) in the blood. Their large-scale study was recently published in the New England Journal of Medicine. GT's Ciara Curtin caught up with Farrall to discuss his work.
Genome Technology: Why did you focus on variants associated with Lp(a) lipoprotein levels?
Martin Farrall: Like a lot of human genetic studies, we weren't particularly looking for one gene or one individual gene. We used a gene chip, an array of many SNPs — nearly 50,000 SNPs — that were scattered across the genome. These SNPs were targeted to just over 2,000 candidate genes for coronary disease. It was a bit of a surprise when signals in Lp(a) stood out so strongly. That wasn't something we were particularly expecting.
GT: Why did you take this approach?
MF: This particular study is halfway-house between studying an individual candidate gene or handful of candidate genes, and the whole GWAS approach, which is random with respect to the candidature of genes. The reason why we did this was, I suppose, opportunistic. Others had suggested this as being a cost-effective way to study the genetics of coronary disease that would supplement the information that was coming out a couple of years ago from genome-wide association studies. In fact, we think it's been successful in that objective by picking up Lp(a) with this chip, which was not particularly strong in the GWAS studies.
GT: Why did this study pick up Lp(a) and the GWAS didn't?
MF: It was luck. Luck in that the SNPs that were included in this particular cardio chip experiment were different from the ones that were included on the GWAS chips. Although we know from the HapMap that there's lots of linkage disequilibrium in the human genome, the SNPs in Lp(a) that turned out to be interesting were poorly tagged by existing GWAS chips. If those two SNPs of interest in Lp(a) hadn't been included just by chance, then we wouldn't have a story.
GT: How common are the variants?
MF: One of the SNPs has a frequency of 7 percent in Europeans and the other has a frequency of 2 percent in Europeans. They are low-frequency SNPs, but not very rare.
GT: How do you think these variants work to affect coronary disease?
MF: We interpret our data to show that these SNPs are tagging specific isoforms of the Lp(a) gene, apolipoprotein(a) gene, ApoA. It has protein motif called a kringle motif, which is repeated in a polymorphic manner in all of us, and the two SNPs tag particular isoforms of the protein that appear to be associated with a relatively high risk of coronary disease and relatively high levels of Lp(a) particles in the blood.
GT: What role do you think other variants may play in coronary disease risk?
MF: Undoubtedly there are other variants, particularly low-frequency variants, which are statistically hard to tease out from this sort of genetic epidemiology experiment.
GT: How would you go about finding them if they are hard to tease out?
MF: Lots of data, expensive experiments — or, if you like, brute force and not too much ignorance.
One lesson we've learned from this is that using GWAS SNPs from commercial panels, obviously give you very useful information, but they don't pick up everything. Sometimes you have to parse the SNP experimentally rather than guess what it is from GWAS data.