Studies that focus on estimating disease heritability<\/a> \u2013 that is, how much genetic differences explain differences in disease risk \u2013 usually attribute a substantial fraction of disease variation to genetics. Mutations across the entire genome seem to play a role in diseases such as Type 2 diabetes<\/a>, which is about 17% heritable<\/a>, and schizophrenia<\/a>, which is about 80% heritable<\/a>. In contrast to diseases such as Tay-Sachs<\/a> or cystic fibrosis<\/a>, where mutations in a single gene cause a disease, chronic diseases tend to be polygenic<\/a>, meaning they\u2019re influenced by multiple mutations at many genes across the whole genome.<\/p>\n Every complex disease has both genetic and environmental risk factors. Most researchers study these factors separately because of technical challenges and a lack of large, uniform datasets. Although some have devised techniques to overcome these challenges, they haven\u2019t yet been applied to a comprehensive set of diseases and environmental exposures. <\/p>\n In our recently published research, my colleague Alkes Price <\/a>and I<\/a> developed tools to leverage newly available datasets to quantify the joint effects<\/a> that genetic and environmental risk factors have on the biology underlying disease.<\/p>\n To illustrate the effect gene-environment interactions have on disease, let\u2019s consider the example of aspirin use and colon cancer.<\/p>\n In 2001, researchers at the Fred Hutchinson Cancer Research Center were studying how regularly taking aspirin<\/a> decreased the risk of colon cancer. They wondered whether genetic mutations that slowed down how quickly the body broke down aspirin \u2013 meaning aspirin levels in the body would stay high longer \u2013 might increase the drug\u2019s protective effect against colon cancer. They were right: Only patients with slow aspirin metabolism had a decreased risk of colon cancer, indicating that the effectiveness of a drug can depend on a person\u2019s genetics. <\/p>\n This raises the question of how genetics and different combinations of environmental exposures<\/a>, such as the medications a patient is taking, can affect a person\u2019s disease risk and how effective a treatment will be for them. How many cases of genetic variations directly influencing a drug\u2019s effectiveness are there?<\/p>\n The gene-environment interaction of colon cancer and aspirin is unusual. It involves a mutation at a single location in the genome that has a big effect on colon cancer risk. The past 25 years of human genetics have shown researchers that these sorts of large-effect mutations are rare<\/a>.<\/p>\n For example, an analysis found that the median effect of a genetic variant on height is only 0.14 millimeters<\/a>. Instead, there are usually hundreds of variations that each have small but cumulative effects on a person\u2019s disease risk, making them hard to find.<\/p>\n How could researchers detect these small gene-environment interactions across hundreds of spots in the genome?<\/p>\n We started by looking for cases where genetic variants across the genome showed different effects on a person\u2019s biology in different environments. Rather than trying to detect the small effects of each genetic variant one at a time, we aggregated data across the entire genome<\/a> to turn these small individual effects into a large, genome-wide effect.<\/p>\n Using data from the UK Biobank<\/a> \u2013 a large database containing genetic and health data from about 500,000 people \u2013 we estimated the influence of millions of genetic variants on 33 complex traits and diseases, such as height and asthma. We grouped people based on environmental exposures such as air pollution, cigarette smoking and dietary patterns. Finally, we developed statistical tests to study how the effects of genetics on disease risk and biomarker levels<\/a> varied with these exposures. <\/p>\n We found three types of gene-environment interactions.<\/p>\n First, we found 19 pairs<\/a> of complex traits and environmental exposures that are influenced by genetic variants across the genome. For example, the effect of genetics on white blood cell levels in the body differed between smokers and nonsmokers. When we compared the effects of genetic mutations between the two groups, the strength of gene-environment interaction suggested that smoking changes the way genetics influence white blood cell counts.<\/p>\n Second, we looked for cases where the heritability of a trait varies depending on the environment. In other words, rather than some genetic variants having different effects in different environments, all of them are made stronger in some environments. For example, we found that the heritability of body mass index \u2013 the ratio of weight to height \u2013 increased by 5% for the most active people. This means genetics plays a larger role in BMI the more active you are. We found 28 such trait-environment pairs<\/a>, including HDL cholesterol levels and alcohol consumption, as well as neuroticism and self-reported sleeplessness.<\/p>\n Third, we looked for a type of gene-environment interaction called proportional or joint amplification<\/a>. Here, genetic effects grow with increased environmental exposures, and vice versa. This results in a relatively equal balance of genetic and environmental effects on a trait. For example, as self-reported time spent watching television increased, both genetic and environmental variance increased for a person\u2019s waist-to-hip ratio. This likely reflects the influence of other behaviors related to time spent watching television, such as decreased physical exercise. We found 15 such trait-environment pairs<\/a>, including lung capacity and smoking, and glucose levels and alcohol consumption.<\/p>\n We also looked for cases where biological sex, instead of environmental exposures, influenced interactions with genes. Previous work had shown evidence of these gene-by-sex<\/a>interactions<\/a>, and we found additional examples of the effects of biological sex on all three types of gene-environment interactions. For example, we found that neuroticism had genetic effects that varied across sex.<\/p>\n Finally, we also found that multiple types of gene-environment interactions can affect the same trait. For example, the effects of genetics on systolic blood pressure varied by sex, indicating that some genetic variants have different effects in men and women.<\/p>\n How do we make sense of these distinct types of gene-environment interactions? We argue that they can help researchers better understand the underlying biological mechanisms that lead from genetic and environmental risks to disease, and how genetic variation leads to differences in disease risk between people.<\/p>\n Genes related to the same function work together in a unit called a pathway<\/a>. For example, we can say that genes involved in making heme \u2013 the component of red blood cells that carries oxygen \u2013 are collectively part of the heme synthesis pathway<\/a>. The resulting amounts of heme circulating in the body influence other biological processes, including ones that could lead to the development of anemia and cancer<\/a>. Our model suggests that environmental exposures modify different parts of these pathways, which may explain why we saw different types of gene-environment interactions.<\/p>\n In the future, these findings could lead to treatments that are more personalised based on a person\u2019s genome. For example, clinicians might one day be able to tell whether someone is more likely to decrease their risk of heart disease by taking weight loss drugs or by exercising. <\/p>\n Our results show how studying gene-environment interactions can tell researchers not only about which genetic and environmental factors increase your risk of disease, but also what goes wrong in the body where.<\/p>\n (Arun Durvasula\u00a0is assistant professor of population and public health sciences, University of Southern California. Email: arun.durvasula@med.usc.ed)<\/p>\n![\"\"](\"https:\/\/th-i.thgim.com\/public\/incoming\/rhpqu8\/article69222764.ece\/alternates\/SQUARE_80\/IMG_IMG_iStock-133238207_2_1_N3B7JM3G.jpg\")
Aspirin, genetics and colon cancer<\/p>\n![\"\"](\"https:\/\/th-i.thgim.com\/public\/incoming\/52ntll\/article69352658.ece\/alternates\/SQUARE_80\/CCI_UDHindu_KSL_UH15U5SD7_R1559055694_1_e649fe3b-43d7-42c3-9703-423b9afcd3b7.jpg\")
Polygenic gene-environment interactions<\/p>\n![\"\"](\"https:\/\/th-i.thgim.com\/public\/incoming\/9chnd8\/article66957203.ece\/alternates\/SQUARE_80\/2018-08-03T074352Z_1927735421_RC1F00896910_RTRMADP_3_HEALTH-DNA.JPG\")
New gene-environment models<\/p>\n