By Richard Jefferys
On January 8, the Office of AIDS Research convened a two-day workshop to discuss genome-wide association scan (GWAS) studies in HIV infection. GWAS involve studying known variations in the human genetic code (called single nucleotide polymorphisms or SNPs) to find out if these variants impact an outcome of interest, such as the rate of disease progression in HIV-infected individuals. The overall goal of the workshop was to prioritize which facets of HIV infection should be investigated using the GWAS approach, as in addition to evaluating outcomes involving disease progression there are many other possibilities, such as looking for associations with resistance to HIV infection among individuals who remain seronegative despite repeated exposures to the virus. The workshop also discussed key issues related to the conduct of GWAS studies, such as the availability of appropriate cohorts and control groups, and the question of whether potential cohorts have appropriate samples and informed consent.
The workshop opened with a presentation by Francis Collins, the genetics luminary who presided over the Human Genome Project. Collins described some of the pros and cons of GWAS: one major advantage is that they are unbiased by any assumptions regarding which genetic polymorphisms might be relevant to the condition being studied. Cons include the fact that so many different polymorphisms are analyzed that strict statistical penalties must be applied to the results to ensure that associations are not found simply by chance. The threshold of significance is usually a p value of less than 0.05, but in most GWAS studies a result is not considered significant unless the p value is less than 10-8. Collins explained that since the effects of specific polymorphisms are often modest, huge sample sizes are typically required for statistically significant results to emerge. He gave the example of the Fusion study of diabetes, which was only able to demonstrate significant associations when combined with two other larger cohorts for a meta-analysis. To give a sense of the whittling down process that is involved in GWAS, this research identified six new places in the genome (called loci) with polymorphisms associated with type 2 diabetes, but only after following up on 69 possible associations identified from over 2 million analyzed SNPs.
Another limitation of GWAS is that only the most common variations are currently known (they are documented in a database called the International HapMap). Variants present in less than 5% of the population are not captured, suggesting that polymorphisms associated with relatively rare phenomena (such as elite control of HIV infection) could easily be missed. Data are also currently heavily skewed toward genetic polymorphisms seen in people of European descent, and information on other ethnicities lags far behind.
Three presenters discussed GWAS studies in HIV: David Goldstein from Duke University and the Center for AIDS Vaccine Immunology (CHAVI), Ioannis Theodorou from the French National Agency for AIDS Research, an Paul Bakker from Harvard who is working with Bruce Walker on the International HIV Controllers Study. The overall theme of these presentations was that the vast majority of associations and possible associations (“signals” in GWAS lingo) with control of viral load and rate of disease progression show up in the Human Leukocyte Antigen (HLA) region of chromosome 6. HLA genes are extremely diverse and complex, making it a challenge to interpret these data. Although HLA genes generally exert their influence via adaptive immune responses – T cells and B cells – because they control how antigens are presented to these cells, there are also interactions between molecules encoded by HLA genes and the innate immune system. This further adds to the difficulty of trying to understand the mechanistic link between polymorphisms in HLA regions and disease outcome. All three presenters echoed Francis Collins regarding the need to combine cohorts for meta-analyses in order to increase the statistical power and robustness of any findings.
The workshop also highlighted additional topics that need to be considered in the context of genomics research. Sunil Ahuja discussed how gene copy numbers can influence disease prognosis, citing his work on the gene for the chemokine CCL3L1, which showed that possession of more gene copies is associated with reduced susceptibility to HIV infection. Mary Carrington outlined the importance of carefully selecting cases and controls for GWAS studies due to the possibility of bias in the selection process impacting the study outcome (for example, the short survival of rapid progressors leads to their under-representation in cohort studies). Michael Katze from the University of Washington delivered a provocative talk on the role of systems biology in infectious disease research; this relatively new scientific discipline takes a very different tack from traditional reductionist approaches that hone in on the role of single factors. Instead, systems biology combines large amounts of information in order to identify connections lurking amidst the complexity. Katze delivered a twist on the oft-used metaphor of the man searching for his keys under a lamppost, noting that new techniques like GWAS can illuminate thousands of metaphorical keys. The challenge therefore becomes identifying which set of keys you’re looking for.
The workshop also heard reports from three previously convened working groups (Susceptibility to HIV Infection, Disease Outcomes, and Pediatrics & Adolescents) on GWAS study priorities in their respective bailiwicks, along with a summary of known HIV cohorts and considerations for obtaining appropriate informed consent (including consent that would facilitate depositing research findings in the new GWAS study database at the National Library of Medicine, dbGAP).
These reports were discussed on the second day of the workshop by a panel moderated by Francine McCutchan, with the goal of offering recommendations to NIAID regarding short-term and long-term priorities for GWAS studies in HIV infection. Suggested short-term priorities included meta-analyses of existing cohorts to look for associations with disease progression and resistance to HIV infection. The panel also recommended the development of a database of extant cohorts and the promulgation of guidance regarding appropriate sample collection for GWAS and other new investigative approaches (such as proteomics and systems biology). Longer-term goals included the development of new cohorts, particularly in Africa and Asia, and expanding the information available in genomics databases to cover a broader array of populations. Full details of the workshop recommendations will be made available in a forthcoming report from the Office of AIDS Research*.
* A link to the OAR report will be added when it is available online.