Public Health Impact of Genome-Wide Association Studies: Glass Half Full or Half Empty?

Genome-wide association studies (or GWAS) are large-scale genetic investigations of human disease that measure simultaneously hundreds of thousands of genetic variants scattered throughout the human genome. GWAS burst onto the scientific scene in the mid 2000’s. Propelled by technological advances and falling prices, GWAS have revolutionized the search for genetic influences on common diseases of major public health significance. Since 2005, >1,600 publications have identified > 2,000 replicated genetic associations with > 300 common human diseases and traits.

But what is the public health impact of GWAS?  Have GWAS findings provided clinical applications?  GWAS have many limitations, such as their inability to fully explain the genetic/familial risk of common diseases; the inability to assess rare genetic variants; the small effect sizes of most associations; the difficulty in figuring out true causal associations; and the poor ability of findings to predict disease risk.  In addition, GWAS have not fully addressed interactions of genes with disease risk factors such as diet, environmental exposures and infectious diseases. Other issues include the limited available information on impact of genomic information on health behavior, and the lack of readiness of health systems in integrating this new information into practice. These limitations have not stopped entrepreneurs in the US and around the world from packaging and marketing GWAS information into direct-to-consumer personal genomic tests.

So is the genomic medicine glass half full or half empty as a result of GWAS?  In a recent review, Dr Teri Manolio from the National Human Genome Research Institute explored current and potentially encouraging near term clinical applications of GWAS, in the areas of disease risk prediction and screening, disease classification, and drug development and toxicity.

First, for risk prediction and screening, while GWAS findings have not proven useful so far for prediction of most common diseases, this is beginning to change.  For example, in type 1 diabetes (T1D), GWAS has the potential to substantially contribute to the identification of genetically high risk individuals. Predictive models using GWAS (> 50 variants) are the highest known for any disease.  Whether or not we can reduce the incidence or delay the onset of T1D in high risk individuals is still an open question.  In addition, the recent major GWAS discoveries in selected cancers by the International Oncology Genetics Consortium have been encouraging enough for scientists and policy makers to begin a serious dialogue about using this information in population screening for common cancers. GWAS variants could be used to stratify the population by level of risk in combination with age as a threshold for cancer screening. Of course, there are major evidentiary, ethical and implementation challenges that remain before GWAS can be used in population screening.

Second, for disease classification, GWAS may become more useful to identify disease subtypes that have different causes or responses to treatment. Consider the example of maturity-onset diabetes of the young or MODY. A common form of MODY is due to mutations in the HNF1A gene. Although mutations in the gene underlying MODY were identified before the GWAS era, they could have important implications for patients and their relatives, as many patients with HNF1A‑MODY are better managed with sulphonylureas than with metformin or insulin.  The GWAS approach also demonstrated associations of common variants in HNF1A with levels of C-reactive protein, which is a potential biomarker of the condition. The use of genetic testing in clinical practice needs to be further evaluated.

Third, for drug development and toxicities, GWAS continues to provide valuable information on gene-drug interactions with the potential to develop safer and more effective drugs as well as to reduce toxicities in the clinical use of existing medications.  For example, treatment of hepatitis C, a common viral infection and a cause of liver cirrhosis, with pegylated interferon plus ribavirin is complicated by hemolytic anemia induced by ribavirin. A GWAS of change in hemoglobin levels during ribavirin treatment identified inosine triphosphatase (ITPA) variants that can protect against ribavirin-induced anemia. This not only points to a possible marker of adverse treatment response but also to possible new therapeutic agents. For existing medications, an example of GWAS success is the finding of a strong association between SLCO1B1 variants with myopathy, related to simvastatin therapy.  Myopathy occurs in 1–5% of patients treated with statins and is char­acterized by muscle pain, weakness and elevated muscle enzyme lev­els. Dosing guidelines have been issued by the Clinical Pharmacogenomics Implementation Consortium for managing simvastatin-induced myopathy risk in the context of SLCO1B1 genotyping. Nevertheless, while the test is commercially available, genotyping is not currently required in conjunction with treatment.

In summary, there have been plenty of insights learned from GWAS. Even as science moves on to other technologies such as whole genome sequencing and other “omics”, the already collected GWAS  information worldwide is truly a phenomenal amount of “big data ”. This information will be analyzed for years to come in order to gain further insight into gene-gene and gene-environment interactions and response to treatment.  But we also need to have realistic expectations about the public health impact of GWAS. We need to change our expected translation timeline to years or even decades. We cannot rush into implementation of these technologies to reap their health benefits until we evaluate their validity and utility, assess the balance of benefits and harms, and explore their added value in clinical practice. We must also find optimal and equitable ways for implementation and for measuring their real impact across the whole population.

GWAS was a giant step in genomic medicine. Yet, it is only the first step of a long translational journey from gene discoveries to improved health for individuals and populations.