Precision beyond Genomics: Environment, Exposures, and Social Background

While genomics captures a portion of risk of complex chronic disease over one’s lifetime, it cannot be properly captured without lifetime cumulative environmental interactions: Risk = G + E, where G = Genetics and E = Environment. Assessing lifetime risk requires defining population health risk factors, controlling for cumulative or confounding factors, and measuring, as best as possible, the quantitative summary exposure. One such effort to do this is the ‘Global Burden of Diseases, Injuries, and Risk Factors (GBD) Study’ (Robison 2018). Each year, the GBD collaborator group

Provision and quality assessment of actionable genomic information over time

FIGURE 6.3 Provision and quality assessment of actionable genomic information over time. This framework illustrates how improving genomic resolution for individuals, populations, and disease cohorts can impact the clinic and be assessed in the academic literature.

publishes quantitative risk estimations for behavioral, occupational, and metabolic risks for 195 countries. While useful for calculating population-level risk, it can be difficult for estimation of individual-level risk. For precision medicine, efforts are underway to define, measure, and interpret individual environmental interactions using health and population records, and clinical measurements. This effort can best be summed up in characterizing the ‘exposome’.

• The National Institute for Occupational Safety and Health (NIOSH) defines the exposome as the measure of all the exposures of an individual in a lifetime and how those exposures relate to health (Dudbridge 2013; NIOSH 2019).

The exposome can be divided into three categories (Bradburne & Lewis 2018): The general external exposome represents an individual’s social, economic, psychological, and geographic background. The specific external exposome is an external event or events, such as a toxicological exposure, particular diet, radiation event, or lifestyle factors, such as smoking. The internal exposome represents the body systems affected by the external exposomes. These include organ systems, gut microflora, adducts, oxidative stress, and others.

Within the exposome, three classes of biomarkers can be described. The first class is biomarkers of susceptibility. These would include genetic variants tied to increased risk of susceptibility to a toxicant or exposure. The second class is biomarkers of exposure, which describe a marker, such as a DNA adduct, whose presence indicates an exposure event. Examples include adducts that are indicative of aflatoxin exposure, such as the presence of AFB-1 guanine in urine or blood. Interestingly, the presence of AFB-1 guanine in urine is typical of an ‘acute’ exposure to aflatoxin in the past 24hours. It can persist for months in blood, however, so its presence there is more indicative of a chronic exposure event. The last class is biomarkers of effect. These are markers whose magnitude is indicative of the level of exposure effect. An example is bladder cell micronuclei which can be observed in urine using a simple assay. In general, the higher the proportion of micronuclei, the more the amount of arsenic that the subject may have ingested from a contaminated water source (Bradburne & Lewis 2018).

Getting More Personal: The Microbiome as an Interface

At the interface of nearly all human systems with the environment is the microbiome. Mucosal lung, skin, gastrointestinal tract, mouth, vaginal, and cochlear surfaces all contain microbiomes that interface chemical exposures. These communities can modify drug interactions, activate (or de-activate) xenobiotics, and induce inflammatory responses by the presence or absence of key taxa (termed dysbiosis). Efforts are underway to define normal versus abnormal microbiota conditions for various health effects, including for environmental exposures. Gut microbiota alone have been shown to affect obesity, depression, quality of life, and can be heritable (Breitwieser et al. 2017). Specific taxa can be used for biomarkers or as sentinels for exposures, but it is not well understood how to incorporate them into overall measures of quantitative risk.

Models of Risk and Exposure

There are two basic ways to establish evidence that can inform actionable information. The first is establishing mechanisms that can be assessed, or even detected, which inform the presence and progression of a disease resulting from an exposure. An example of this is the use of a diagnostic to detect a biomarker indicative of a disease, such as the detection of lactate dehydrogenase (LDH) circulating in blood for melanoma or a variety of other pathologies (Weighill et al. 2019). For exposure research, many mechanisms are either not known or they do not have a direct diagnostic test. One solution for this is the development of Adverse Outcome Pathways’, or AOPs (Bradburne & Lewis 2018). This involves classing mechanistic toxicological effects that are common to particular chemical exposure. For example, a molecular initiating event could be chemical absorption on the skin, followed by intermediate steps such as cellular interactions, cellular effects, and organ-level effects, and lastly, the adverse outcome such as skin inflammation. By using an AOP, even if the offending chemical or exposure is not known, one can look at the mechanistic effects and make some inference of the exposure and the treatment.

The second way to build actionable information for exposures is to develop quantitative risk measurements. In genomic medicine and exposure science, there are some basic statistical tools that can be used to calculate risk. Odds ratios are used to measure the effectiveness of a particular diagnostic test (e.g., a genomic assay or a serum biomarker test). It is the probability of a test being positive if a patient has a disease, versus the odds of a test being positive if a patient does not have a disease. Absolute risk is the probability of a health effect occurring under specific conditions, while relative risk is the likelihood of a health effect occurring in a group of people compared to a separate group of people with different backgrounds or in different environments. The GBD group utilizes relative risk as the primary method for calculation of global disease burden. The quantitative measurement utilized is the summary exposure value (SEV), which represents the continuous relative risk accumulated over time.

Using this methodology, the GBD provides a quantitative measure of risk- attributable burdens in life years, which are adjusted to remove the influence of pre-existing disability as a confounding factor (disability-adjusted life years, or DALYs). The 2017 GBD report claims to quantitate risks that combined add up to 61% of all deaths and 48% of DALYs worldwide, represented in categories such as unsafe water, air pollution, lead exposure, and others.

Likelihood Ratio (LR)

Likelihood ratios (LRs) have utility in combining quantitative risk from a disparate source, such as from the SEV, with the pre-existing risk from genetic variants. An LR can be defined as the ratio of the probability that a result is correct to the probability that the result is incorrect. A positive LR = sensitivity/(l - specificity), whereas a negative LR = (1 - sensitivity)/specificity. LRs have been used for over 100 years to provide confidence that the risk for a disease resulting from a test is ‘overlayed’ onto the preexisting risk. For example, in the nomogram in Figure 6.4, if a pre-test probability for

Application of LRs from a genetic test to inform risk for a chronic disease. A nomogram showing the relationship between the pre- and post-test probabilities when applying an LR

FIGURE 6.4 Application of LRs from a genetic test to inform risk for a chronic disease. A nomogram showing the relationship between the pre- and post-test probabilities when applying an LR.

a disease or trait is 25% and the LR is 1, indicating no change in risk, the probability remains at 25%. However, if the LR is 2, indicating a higher probability that the outcome is correct, the resulting (post-test) probability is ~40%. In the same way, genomic risk scores for variants have adopted LRs to predict pre- and post-test probabilities. This may provide a seamless way to obtain a quantitative, combinatory risk score between risk generated from exposure science and risk generated from genetic variants.

Social Determinants of Health (SDOH)

It is well known that overall chronic disease risk does not come from the presence or absence of a variant or an exposure alone, but rather, is placed over a background of lifestyle, demographic, and economic factors that influence health effects and disease


Social Determinants of Health* That May Be Tracked in the EHR and Could Eventually Be Incorporated into Chronic Disease Risk Scores for Environmental Exposures

SDOH* Domains


Alcohol use

Drink frequency; binge (six or more) drinking frequency


Malaise; hopelessness


Highest school level; highest degree

Financial-resource strain

Difficulty in meeting basic needs

Income (neighborhood median household income)

Entered or median income for location through census data

Intimate-partner violence

Humiliation/emotional abuse; fear; sexual exploitation; violence

Physical activity

#Days/week of exercise; duration of each exercise period

Race or ethnic group

Non-Hispanic White, Hispanic/Latino, Asian; self-identification

Residential address

Address, zip code

Social connection/isolation

Frequency of social contact; religious affiliation and attendance, frequency; organization attendance, frequency


Anxiety and inability to feel comfortable, rest, or sleep

Tobacco use

Cumulative use; repetitive use

Modified from Bradburne and Lewis (2018),

outcomes. This information has been recognized for years to be predictive in the public health arena. For example, it is well known that variables such as your education level, ethnicity, and even zip code can influence lifespan. Efforts are now underway to collect this information, incorporate it into the EHR, and use it for individualized patient health care. Table 6.1 shows typical SDOH data (Adler & Stead 2015; Bradburne & Lewis 2018) that are being collected through surveys in various health care systems.

Precision Medicine and Environmental Health for the Military

The Million Veterans Program

The primary study for associating environmental and occupational hazards w'ith active duty service members is the Million Veteran Program (MVP). The pool of study subjects represents the best look at the phenotypes for long-past chronic exposures. The MVP has also been coupled w'ith a similarly-themed civilian effort, the All of Us study. Study goals and integration of All of Us and MVP scientific goals, and resources are reviewed in National Academies of Science (2014). The military will be able to utilize this cohort for GWAS studies, and pre- and post-exposure samples are available for biomarker discovery in the Department of Defense Serum Repository (Bradburne & Lewis 2017). Tracking resources are available to associate exposure sources w'ith geographic locations, and an effort is underway to establish an Individual Longitudinal Exposure Record (ILER) for each service member (Bradburne et al. 2015; National Academies of Sciences 2014). These efforts and resources should allow opportunities to associate sources with health effects, establish population GWAS studies to determine individual genomic susceptibility, evaluate serum samples for biomarkers of exposure and effect, and incorporate SDOH for a more comprehensive and quantitative understanding of risk of individuals and groups to environmental and occupational exposures.

Ethical, Legal, and Social Issues (ELSI)

Determining individual genetic susceptibility in a military service member has been considered since the 1960s and 1970s when individual Mendelian genetic tests were utilized (De Castro et al. 2016). Military populations share most of the same ethical, legal, and social issues as civilian populations, with a significant exception for the military as the insurer is also the health care provider, and any clinical test for genetic susceptibility is establishing what could be a pre-existing risk that could impact insurability or even service status. The primary action taken by the military at this time has been to adopt, as policy, The Genetic Information Non-discrimination Act (GINA). GINA states that a service member cannot be discriminated against based on this genetic information (De Castro et al. 2016). However, an important distinction is that GINA is not law for DoD personnel (as it is for civilians), but only policy, which perhaps leaves it easier to change in the future.


Personalized environmental health for the military is emerging as genomic medicine emerges, providing new tools to associate exposed populations with genetic markers and discover new biomarkers of exposure and biomarkers of effect. Many of the tools for incorporating genomic medicine into medicine and preventive medicine use techniques for estimating quantitative risk that are amenable to incorporating environmental health risk. Lastly, grouped ontologies of toxicologic effect and direct mechanistic correlative biomarkers are on the horizon for direct interrogation using emerging genomic tools. These should provide better resolution for association studies, preventive medicine, and source mitigation in the future.


Adler, N., Stead, W.W. “Patients in context—EHR capture of social and behavioral determinants of health,” New England Journal of Medicine, vol. 372, pp. 698-701, 19 February 2015.

Bradburne, C., Graham, D., & Kingston, H.M., et al. “Overview of ‘omics technologies for military occupational health surveillance and medicine,” Military Medicine, vol. 180, no. suppl_10, pp. 34-48, 2015.

Bradburne, C., & Lewis, J.A. “Personalizing environmental health: At the intersection of precision medicine and occupational health in the military,” Journal of Occupational and Environmental Medicine, vol. 59, no. 11, pp. e209-2214, 2017.

Bradburne, C., & Lewis, J.A. The U.S. Military and the Exposome. In: Dagnino S., Macherone A. (eds) Unraveling the Exposome - A Practical View, pp. 63-85. Cham: Springer Nature, 2018.

Breitwieser, F.P., Lu, J., & Salzberg, S.L. “A review of methods and databases for metage- nomic classification and assembly,” Briefings in Bioinformatics, vol. 18, pp. 1125-1136, 23 September 2017.

Buniello, A., MacArthur, J.A.L., & Cerezo, M.. et al. “NHGRI-EBI GWAS Catalog of Published Genome-Wide Association Studies, Targeted Arrays and Summary Statistics 2019”. [Online]. Available: [Accessed 9 September 2019].

Chen, W„ Stambolian, D., & Edwards, A.O., et al. “Genetic variants near TIMP3 and high- density lipoprotein-Associated loci influence susceptibility to age-related macular degeneration,” Proceedings of the National Academy of Sciences of the United States of America, vol. 107, no 16. pp. 7401-7406, 20 April 2010.

De Castro, M., Biesecker, L.G., & Turner, C., et al. “Genomic medicine in the military,” NPJ Genomic Medicine, vol. 1. p. 15008, 2016.

Dudbridge, F. “Power and predictive accuracy of polygenic risk scores,” PLoS Genetics, vol. 9. no. 3, p. el003348, 21 March 2013.

Fisher, R. “The correlation between relatives on the supposition of Mendelian inheritance,” vol. 52. no. 2. p. 3990433, 1918.

Green, E.D., Guyer, M.S.. and the National Human Genome Research Institute. “Charting a course for genomic medicine from base pairs to bedside,” Nature, vol. 470, pp. 204-213, 2011.

Manolio, T.A., Collins, F.S., & Cox, N.J., et al. “Finding the missing heritability of complex diseases,” Nature, vol. 461, pp. 747-753, 8 October 2009.

National Academies of Sciences. “Recommended Core Domains and Measures,” in Capturing Social and Behavioral Domains and Measures in Electronic Health Records: Phase 2, NAS. 2014. pp. 227-236.

National Human Genome Research Institute. “Genomics and Medicine,” [Online]. Available: [Accessed 9 September 2019].

National Institute for Occupational Safety and Health (NIOSH). “Exposome,” [Online], Available: [Accessed 9 September 2019].

Rappaport, S.M. “Genetic factors are not the major causes of chronic diseases,” PLoS One, vol. 11, no. 4. p. eO 154387, 22 April 2016.

Raychaudhuri, S., Plenge, R.M., & Rossin, E.J., et al. “Identifying relationships among genomic disease regions: Predicting genes at pathogenic SNP associations and rare deletions,” PLoS Genetics, vol. 5. no. 6. p. el000534. 26 June 2009.

Robison, K. “Omics! Omics!,” November 2018. [Online]. Available: http://omicsomics. l/nanopore-community-rneeting-2018-clive.html [Accessed 9 September 2019].

Sudmant, R, Rausch, T., & Gardner, E.J., et al. “An integrated map of structural variation in 2,504 human genomes.” Nature, vol. 526, pp. 75-81, 1 October 2015.

U.S. National Library of Medicine. “Genetics Home Reference,” 2019. [Online]. Available: [Accessed 9 September 2019].

Venter, J.C., Adams, M.D., & Myers, E.W., et al. “The sequence of the human genome,” Science, vol. 291, no. 5507, pp. 1304-1351, 2001.

Weighill, D., Macaya-Sanz, D., & DiFazio, S.P., et al. “Wavelet-based genomic signal processing for centromere identification and hypothesis generation,” Frontiers in Genetics, vol. 10, p. 487, 2019.

< Prev   CONTENTS   Source   Next >