2 July 2011. According to the two largest genomewide association analyses of Parkinson’s disease to date, genetics may play a bigger role in the disorder than previously appreciated. Both papers were published in June in the open-access journal PLoS Genetics. One—a meta-analysis of five genomewide association studies (GWAS) involving some 12,000 PD cases and 21,000 controls in the U.S. and Europe—found five new risk loci and confirmed 11 reported earlier this year by the same consortium. The other study was conducted by a Silicon Valley personal genomics company that used a Web-based approach to recruit more than 3,400 PD patients and 29,000 controls. Their GWAS uncovered two novel PD loci, one of which also turned up in the European meta-analysis, and confirmed many that were identified previously. “PD does appear to be a truly complex disease caused by interaction of a significant number of more or less common variants acting in concert,” commented Robert Nussbaum, University of California at San Francisco, who was not involved in either study. The Web-based GWAS estimates that at least a quarter of the variation in PD risk derives from genes, about 75 to 85 percent of which remain to be discovered. Up until a few years ago, PD was generally viewed as being a sporadic disease.
Representing the efforts of Vincent Plagnol and Nicholas Wood, University College London, U.K., and some 140 coauthors worldwide, the first paper describes the final analysis of a two-stage study by the International Parkinson’s Disease Genomics Consortium (IPDGC). In a February Lancet paper (IPDGC, 2011; see also ARF conference story), this group reported 11 loci that achieved genomewide significance in the initial discovery phase involving some 5,000 PD patients and 12,000 controls from five international cohorts. For the second stage of analysis, the scientists genotyped an additional 7,000 cases and 9,000 controls using a custom microarray containing 1,920 single nucleotide polymorphisms (SNPs) that came up strong in the first phase. The two-step analysis generated seven additional hits beyond the 11 reported in Lancet.
While analyzing the new loci, the researchers learned of the other large GWAS led by first author Chuong Do and senior investigator Nicholas Eriksson of 23andMe in Mountain View, California. The groups agreed to share summary statistics for their top hits so these discovery-phase loci could be cross-validated on the other’s dataset. Of the seven new loci that came up in the IPDGC study, five were confirmed in the 23andMe cohort.
One of those—an SNP in the chromosome 4 region upstream of SCARB2—also met genomewide significance in the discovery phase of the 23andMe GWAS. SCARB2 has generated some buzz because it encodes a protein (LIMP-2) involved in trafficking of glucocerebrosidase (GBA), the top genetic risk factor for PD. Three of the IPDGC loci were associated with methylation and expression changes in human postmortem frontal cortex and cerebellar tissue, suggesting the SNPs could be functionally relevant.
The 23andMe study turned up another novel hit—an SNP near SREBF1/RAI1—which did not come up in the IPDGC discovery phase but did achieve nominal significance in the company’s replication analysis using the IPDGC dataset. That level of significance is considered sufficient for confirming an association that previously made the more stringent cut in a discovery GWAS.
In addition, the 23andMe team estimated that “about a quarter to a third of what’s going on in late-onset PD is due to genetics,” Eriksson said. This runs counter to some earlier PD twin studies that found little to no genetic effect (see Tanner et al., 1999; Wirdefeldt et al., 2004), and confirms others that estimated a similar genetic contribution to PD. What distinguish the 23andMe data are its tighter confidence intervals, Do told ARF. The previous studies needed pairs of twins in which either one or both people had PD—tough to find for a relatively rare disorder. The 23andMe team estimated heritability using an entirely different method based on genomewide sharing between distant relatives.
Beyond the claim that about 25 to 33 percent of PD is genetic, the next question is how much of that scientists already know. The 23andMe researchers built risk prediction models in an attempt to address this, i.e., whether “we are 95 percent of the way there, or 10 percent of the way there,” Eriksson said. “The models show that we’re closer to 10 percent there.”
Beyond the science, the 23andMe study represents a paradigm shift in terms of data collection. Virtually all PD patients were recruited in 18 months through a targeted e-mail campaign done with the Michael J. Fox Foundation, National Parkinson Foundation, The Parkinson’s Institute, and other PD patient groups and clinics. Controls came from the 23andMe customer database. Potential participants were directed via e-mail to an online form with questions about their diagnosis, symptoms, environmental risk factors, responses to medications, and so forth. Because study participants were restricted to unrelated individuals of primarily European ancestry, the 23andMe and IPDGC cohorts “were, by construction, pretty similar to each other,” Do said.
“I think the 23andMe paper is very interesting,” wrote John Hardy of University College London, a coauthor on the IPDGC study, in an e-mail to ARF. “On the science level, their findings overlap with ours and show new loci for PD. But they also show that self-referral is a valid way of identifying people with disease. We have to go through complex and slow ethics reviews. This work shows there may be a way around this laborious red tape.”
As yet, the studies “don’t tell us anything conclusive about disease mechanism,” noted Nussbaum. “We don’t know if the genes nearest to the SNPs are the ones through which the variants are acting to increase or decrease PD risk.” Nevertheless, the GWAS papers are “good for generating hypotheses that can be tested,” he wrote. “They will give basic cell biologists and neuroscientists plenty to work on for the years to come.”
In the meantime, leaders of recent PD GWAS, including the IPDGC and 23andMe teams, are joining forces for a huge meta-analysis involving some 10,000 to 15,000 cases, Eriksson said. “Our study suggests there are a lot of SNPs associated with PD out there remaining to be found.” The company recently announced it is teaming up with South San Francisco-based Genentech, Inc. in a study of AD genetics.—Esther Landhuis.
Do CB, Tung JY, Dorfman E, Kiefer AK, Drabant EM, Francke U, Mountain JL, Goldman SM, Tanner CM, Langston JW, Wojcicki A, Eriksson N. Web-based genomewide association study identifies two novel loci and a substantial genetic component for Parkinson’s disease. PLoS Genetics. 23 June 2011. Abstract
International Parkinson’s Disease Genomics Consortium et al. A two-stage meta-analysis identifies several new loci for Parkinson’s disease. PLoS Genetics. 23 June 2011. Abstract