sign in sign up
<<
Home , Precision medicine, Public health genomics

1870 Diabetes Care Volume 39, November 2016

Donna K. Arnett1 and Steven A. Claas2

PRECISION Precision Medicine, , and

Diabetes Care 2016;39:1870–1873 | DOI: 10.2337/dc16-1763

Despite the advances in molecular genetics and...as genetic tests are proliferating in the U.S. population, their appropriate usage in the public health setting needs careful scrutiny. dMuin J. Khoury

The issues that Muin J. Khoury (1) identified as relevant in the intersection of genomics and public healthdrapid advances in the molecular technologies avail- able for interrogating the , the proliferation of genetic and genomic testing in the clinical setting, uncertainties surrounding the use and value of such testsdare far from surprising. These issues form the basis of much discussion in today’s scientific, professional, and popular literature. What is perhaps surprising about the statement, however, is that it was made nearly 20 years ago. As advances in molecular genetics have marched steadily forward and genetic epidemiologists have attempted to generate useful public health knowledge, paradigms have mean- while shifted and developing disciplines (e.g., epigenetics, ) now con- tribute substantively to the conversation. The challenges faced by public health practitioners in this context have stemmed from the shifting landscape upon which the discipline is built. President ’s Precision Medicine Initiative (PMI), unveiled during his 2015 State of the Union Address, represents a concerted effort to lay the scientific foundation for the careful scrutiny Khoury called for. How does this PMI impact the prevention and treatment of diabetes? As public health practitioners, we must face the current reality of the high and increasing global prevalence of diabetes. Between 1980 and 2008, the global age-standardized diabetes prevalence increased from 8.3 to 9.8% in men and from 7.5 to 9.2% in women (2). Precision medicine (PM) may offer new strategies to prevent and reduce diabetes in populations. PM seeks to integrate a bounty of data from each person’s genomedcombined with data from his or her environment and lifestyle (i.e., behaviors)dto tailor medical treatment to the individual rather than what has been characterized as “treating to the average.” We contend that PM will add todnot replaced traditional public health strategies. Prior lessons from genomic research has taught us that for common like diabetes, environmental factors have the greatest impact on prevalence. While the genome evolved over thousands of years, two key 1Dean’sOffice, College of Public Health, Univer- determinants of diabetes prevalence, diet and physical activity, have evolved very sity of Kentucky, Lexington, KY dramatically over a very short “genomic” period of time. It is this intersection be- 2Department of , School of Public tween lifestyle factors and the genome that will be critical to halt the rise in the Health, University of Alabama at Birmingham, of diabetes. As public health experts, we must do everything possible to Birmingham, AL understand and ameliorate the environmental effects on diabetes. The PMI offers Corresponding author: Donna K. Arnett, donna. the opportunity to explore both existing and new public health strategies that ad- [email protected]. dress both environmental and genomic risk factors. The PMI proposes to recruit Received 15 August 2016 and accepted 26 1 million individuals; this will identify many subjects with prediabetes and diabetes August 2016. who could be connected with known lifestyle interventions to prevent diabetes © 2016 by the American Diabetes Association. incidence or improve adherence to lifestyle or medical management. With its ap- Readers may use this article as longas the work is properly cited, the use is educational and not proach of engaging participants with their own data collection for lifestyle factors, for profit, and the work is not altered. More infor- proven clinical and population-based interventions could then be tested in high-risk mation is available at http://www.diabetesjournals individuals in relation to their genomic profiles to ascertain important –diet and .org/content/license. gene–physical activity interactions. See accompanying articles, pp. 1854, At first glance, PM, with its resolute focus on the individual patient and its 1858, 1874, 1879, 1889, 1896, 1902, apparent inevitable eminence, might alarm some public health professionals. 1909, and 1915. care.diabetesjournals.org Arnett and Claas 1871

Such concerns are largely unwarranted: people with low fasting glucose) than PRECISION PUBLIC HEALTH IN the PM paradigm is no more at odds is achieved through nonintervention or PRACTICE with public health than traditional pa- some other control condition. The left Models such as that illustrated in Fig. 1 tient care. That said, the integration of panel visualizes traditional public health are useful for conceptual orientation, genomics (and other biomolecular tech- efforts that have used that knowledge but how will public and population nologies collectively known as “omics”), to increase the number of people in health practitioners contribute in the mobile technology, electronic medical the population benefiting from the in- era of PM? Below we discuss three do- records, data analytics and warehousing, tervention (illustrated here by a single mains of public health practice that have and all the other elements of PM will drug and diet) and decrease the number played and will continue to play a role in surely impact the work of public health who could be but are not benefiting. the development of PM. professionals. Below we explore the con- Visually, this effort is represented as re- Discover and Validate New Markers ceptual and practical impact of PM on shaping the distributions of the treated of Health and public health. Finally, we discuss these and nontreated/control groups. But Discovery-based research by epidemiolo- implications in the context of diabetes note the idealized, typical distribution gists has identified associations and gen- prevention and care. of treated individuals: there is a wide erated testable hypotheses concerning range and variable response when the genomic, environmental, and lifestyle fac- PRECISION PUBLIC HEALTH IN intervention is appliedtotheentire tors influencing health and disease. For THEORY population. In contrast, the right panel example, to date, 87 type 2 diabetes ge- If PM is defined as optimizing care for represents the futuristic approach of nome-wide association studies (which well-characterized individual patients, public health interventions informed can assay millions of genetic variants in then precision public health is charac- by PM. While we retain the goal of tra- an individual DNA sample) have reported terized by discovering, validating, and ditional public health, we add to that the 640 disease phenotype–genetic variant optimizing care strategies for well-char- goal of reshaping the distribution in an- associations (3). Such discovery efforts acterized population strata. How does other dimension: by precisely targeting must increase in the era of PM. The precision public health differ from tradi- population strata that most benefit million-person cohort proposed by the tional public health? The difference is from the intervention (illustrated here PMI alone will offer unprecedented op- illustrated in Fig. 1. This figure applies by the use of omics, including pharma- portunities for discovery. The practical to precision public health in the domains cogenomics) and withholding the inter- impact of these discoveries will be the of both validating new clinical interven- vention from those strata that would not increased stratification of populations tions and monitoring the efficacy of ex- benefit, we minimize the spread in the and the shrinking of subgroups available isting health maintenance and disease tails of the distribution. In short, preci- for subsequent hypothesis testing and prevention programs. The center panel sion public health seeks to get the right validation. In some cases, this may pose describes how public health and clinical treatment or intervention to the right significant challenges in terms of achieving scientists search for and verify interven- population with the aid of detailed ge- sufficient statistical power and avoiding tions that, on the whole, produce better nomic, environmental, lifestyle, and selection bias. For example, assuming a population-level health (e.g., more other data. rough 10% prevalence of type 2 diabetes in the million-person cohort, a rare var- iant’s minor allele (1%) can be expected in only 1,000 affected individuals in the entire sample. If a type 2 diabetes dis- ease subtype stratum were the object of study, this number would be consid- erably lower. Although daunting, these are familiar challenges to researchers and professionals, and successful incorporation of PM into clinical practice will be feasible only with the assistance of public health experts who will help to validate the clinical val- idity and utility of newly identified PM markers. Monitor Population Health The emerging tools of PM will also be brought to bear on public health surveil- lance efforts. Changes will be seen in two dimensions of surveillance: the metrics being tracked and the methods used to Figure 1—Differences between traditional public health and precision public health. Hypo and gather data. Genomic datadboth sequence Hyper signify extremes of response to the intervention. genetics and epigenetic profilingdis 1872 Precision Medicine, Genomics, and Public Health Diabetes Care Volume 39, November 2016

anticipated to add much to our under- the greatest benefit. These programs to HLA-DQA1, HLA-DQB1,andHLA-DRB1 standing of disease risk; with decreas- may take a number of forms. For exam- alleles (14). Alleles in PTPN22, UBASH3A, ing costs, collection of population-level ple, there is much hope that widespread PTPN2, INS, and other have been genomic data becomes increasingly genomic profiling will enable the identi- foundtobeassociatedwithtype1diabe- possible. Already large cohort genetic fication of individuals at elevated risk for tes risk (15). Already these findings are and epigenetic studies have contrib- disease. The CDC Office of Public Health being incorporated into population-level uted to public health recommendations Genomics currently provides a Genomic research. Since 2004, The Environmental and strategies intended to reduce the Applications Toolkit for public health Determinants of Diabetes in the Young burden of diabetes (4). Wider genomic departments that includes a list of ge- (TEDDY) study has used newborn HLA surveillance may prove even more im- nomic tests that have potential public screening to identify children with high pactful. A major component of the PMI health utility (8). These “genomic appli- genetic risk of . TEDDY in- is the collection and interpretation of cations” are scrutinized by panels of vestigators are following over 6,000 chil- environmental and lifestyle data. These experts and rated according to the dren until age 15 years. Follow-up has measures may offer new surveillance application’s level of analytical and clin- discovered 225 children with type 1 dia- metrics or improve upon more traditional ical validity, clinical utility, potential betes. Within the TEDDY cohort, nested self-report data. Mobile (e.g., internet- benefits versus harms, and whether case-control studies are analyzing gene and cellular phone–based) technologies the application has an evidence-based expression, microbiomes, viromes, me- may make possible the collection of recommendation (8). Three “tier 1” (i.e., tabolomics, and dietary biomarkers. The vast amounts of environmental and life- highest level of validity, utility, and evi- TEDDY study has already published im- style data. Already these technologies dence) genomic applications have been portant findings regarding islet autoim- have been incorporated into diabetes deemed ready for pilot or demonstration munity, celiac disease, and disease risk self-monitoring (diet, activity, glucose) population-level programs (9), although and progression. In addition to this critical programs (5,6), although their patient none apply yet to diabetes. For example, discovery research, the TEDDY study is acceptability has not been established ;2–7% of breast cancers and ;10–15% providing important insights into the po- (5). Whether the use of mobile technol- of ovarian cancers are linked to mutations tential of population-level screening (16). ogies will be acceptable to participants in in BRCA1 and BRCA2 (10) (resulting in Although many genetic variants are surveillance programs also remains an ;10,000 and ;2,700 cases of breast associated with type 2 diabetes risk, the open question, but the ability to accu- and ovarian cancer, respectively, per amount of disease risk accounted for by rately detect population-level changes year, according to a 2012 estimate these variants is low (;10%), and the in health behaviors and environmental [11]). Women with clinically important mechanisms and pathways by which exposures remains an enticing possi- BRCA1 and BRCA2 mutations have a these genes exert their influence is yet un- bility. Electronic health records (EHR) 35–84% chance of developing breast clear (17). Some have reported consid- represent another possible source of cancer and a 10–50% chance of develop- erable heterogeneity in variants associated population-level data. Despite the ing ovarian cancer by age 70 years (12). with type 2 diabetes among different eth- daunting technological hurdles and un- The U.S. Preventive Services Task Force nic groups. This may significantly compli- resolved ethical issues, largely related currently recommends BRCA1 and BRCA2 cate translating discoveries in this disease to patient privacy, posed by PM’s po- genetic counseling for women with a domain into meaningful public health ef- tential surveillance measures, the U.S. family history of breast or ovarian cancer forts (18). Type 2 diabetes is itself a hetero- Centers for Disease Control and Preven- (12). Public health interventions (educa- geneous disease, and while the designation tion (CDC) 2014 surveillance strategy tion, increased screening) targeting has proven clinically useful, type 2 diabetes proposes to “accelerate the utilization of those with elevated BRCA1 and/or BRCA2 is likely a “broad umbrella” used to refer emerging tools and approaches” (7). To risk may provide measurable population- to a constellation of physiological diseases this end, the CDC appointed two senior- level benefit(9). (19). This fact may, however, prove instru- level officers to develop effective health It is feasible that PMI environmental mental to precision public health. Studies information technology policies; the and lifestyle data collected via mobile of type 2 diabetes susceptibility variants agency is also working with outside devel- technologies could be incorporated have pointed to mechanistic heterogeneity opers to create advanced surveillance into health maintenance and disease (20). A recent study used EHR data-mining technologies and tools (7). prevention programs. For example, mo- techniques in combination with genetic bile technologies could be used to collect data from over 11,000 patients to identify Prevent Disease and Maintain Health diet and activity data to identify at-risk three distinct type 2 diabetes subtypes PM’s principle of targeted intervention individuals and to mediate an interven- characterized by distinct comorbidities applies equally well to programs de- tion program. Small-scale intervention (21). This opens the possibility that geno- signed to maintain health and prevent trials have been promising (13). mic information could be used to charac- disease. These objectives fall soundly terize type 2 diabetes subtypes. Ultimately, into the public health professional’s pur- PRECISION PUBLIC HEALTH IN understanding the of these sub- view. Precision health maintenance and DIABETES PREVENTION AND CARE types may lead to subtype-specificinter- disease prevention will find ways to use Currently, our genomic understanding ventions that could be used to target the emerging technologies and tools of of type 1 diabetes outpaces that of genetically identified treatment strata. PM to efficiently target strategies to many other diseases: up to 50% of genetic To date, there are no “tier 1” genomic those population strata that will reap risk for type 1 diabetes can be attributed applications recommended for either care.diabetesjournals.org Arnett and Claas 1873

type 1 diabetes or type 2 diabetes, al- References susceptibility: recommendation statement. though there are several lower-tier ap- 1. Khoury MJ; Genetics Working Group. From Ann Intern Med 2005;143:355–361 plications for variants with evidence of genes to public health: the applications of ge- 13. Chung AE, Skinner AC, Hasty SE, Perrin EM. Tweeting to health: a novel mHealth intervention validity but insufficient evidence of util- netic technology in disease prevention. Am J Public Health 1996;86:1717–1722 using Fitbits and Twitter to foster healthy life- ity to recommend use (8). A review of 2. Danaei G, Finucane MM, Lu Y, et al.; Global Bur- styles. Clin Pediatr (Phila). 16 June 2016 [Epub evidence by the Evaluation of Genomic den of Metabolic Risk Factors of Chronic Diseases ahead of print]. DOI: 10.1177/0009922816653385 Applications in Practice and Prevention Collaborating Group (Blood Glucose). National, re- 14. Prasad RB, Groop L. Genetics of type 2 di- (EGAPP) Working Group found that ge- gional, and global trends in fasting plasma glucose abetes-pitfalls and possibilities. Genes (Basel) 2015;6:87–123 fi and diabetes prevalence since 1980: systematic netic pro ling (using 28 predictive vari- analysis of health examination surveys and epide- 15. Noble JA, Valdes AM. Genetics of the HLA ants) used to assess risk for type 2 miological studies with 370 country-years and 2.7 region in the prediction of type 1 diabetes. Curr diabetes in the general population pro- million participants. Lancet 2011;378:31–40 Diab Rep 2011;11:533–542 vided virtually no net benefit alone or in 3. National Research Institute and 16. Hagopian WA, Erlich H, Lernmark A, et al.; TEDDY Study Group. The Environmental Deter- combination with traditional risk fac- European Institute. GWAS Catalog [Internet]. Available from https://www.ebi.ac.uk/ minants of Diabetes in the Young (TEDDY): ge- tors (22). As discovery research con- gwas/. Accessed 11 August 2016 netic criteria and international diabetes risk tinues in this area (23), we expect 4. Ley SH, Ardisson Korat AV, Sun Q, et al. Con- screening of 421,000 infants. Pediatr Diabetes more novel omics discoveries in the di- tribution of the Nurses’ Health Studies to uncov- 2011;12:733–743 abetes domain. ering risk factors for type 2 diabetes: diet, 17. Fradkin JE, Hanlon MC, Rodgers GP. NIH Precision Medicine Initiative: implications for Finally, EHR have been used in na- lifestyle, biomarkers, and genetics. Am J Public Health 2016;106:1624–1630 diabetes research. Diabetes Care 2016;39: scent diabetes precision public health 5. McMillan KA, Kirk A, Hewitt A, MacRury S. A 1080–1084 studies. One study of EHR in three U.S. systematic and integrated review of mobile- 18. Kaul N, Ali S. Genes, genetics, and environ- counties identified small geographic based technology to promote active lifestyles ment in type 2 diabetes: implication in person- – areas within the counties with the in people with type 2 diabetes. J Diabetes Sci alized medicine. DNA Cell Biol 2016;35:1 12 19. Florez JC. Precision medicine in diabetes: is least effective control of diabetes. Technol 21 June 2016 [Epub ahead of print] 6. Garabedian LF, Ross-Degnan D, Wharam JF. it time? Diabetes Care 2016;39:1085–1088 Wider EHR surveillance may pinpoint Mobile phone and smartphone technologies for 20. Dimas AS, Lagou V, Barker A, et al.; MAGIC areas for neighborhood-level inter- diabetes care and self-management. Curr Diab Investigators. Impact of type 2 diabetes suscep- ventions (24). Rep 2015;15:109 tibility variants on quantitative glycemic traits 7. Centers for Disease Control and Prevention. reveals mechanistic heterogeneity. Diabetes – CONCLUSIONS Surveillance Strategy [Internet]. Centers for Dis- 2014;63:2158 2171 ease Control and Prevention, 2014. Available 21. Li L, Cheng WY, Glicksberg BS, et al. Identi- We have alluded to some of the scientific from http://www.cdc.gov/surveillance/. Accessed fication of type 2 diabetes subgroups through challenges faced by those seeking to make 15 August 2016 topological analysis of patient similarity. Sci the idea of precision public health a re- 8. Centers for Disease Control and Prevention Transl Med 2015;7:311ra174 22. Evaluation of Genomic Applications in Prac- ality. There are other challenges: translat- Prevention. Public Health Genomics Knowledge Base (v1.2) [Internet]. Centers for Disease Con- tice and Prevention. EGAPP Working Group ing omics into practice and ensuring that trol and Prevention, 2016. Available from Recommendation [article online], 2016. Avail- these novel interventions do not serve https://phgkb.cdc.gov/GAPPKB/phgHome.do. able from http://www.egappreviews.org/ only the highest socioeconomic stratum Accessed 15 August 2016 recommendations/diabetes.htm. Accessed and lead to a new form of 9. Bowen MS, Kolor K, Dotson WD, Ned RM, 11 August 2016 Khoury MJ. Public health action in genomics is 23. National Heart, Lung, and Blood Institute. disparities. While the PMI evidence base now needed beyond . Public Trans-Omics for Precision Medicine (TOPMed) Pro- is growing, we must challenge ourselves Health Genomics 2012;15:327–334 gram [article online], 2016. Available from http:// to identify and translate into action what 10. Bellcross CA, Kolor K, Goddard KA, Coates www.nhlbi.nih.gov/research/resources/nhlbi- we already know about the determinants RJ, Reyes M, Khoury MJ. Awareness and utiliza- precision-medicine-initiative/topmed. Accessed and interventions to halt the dramatic rise tion of BRCA1/2 testing among U.S. primary 15 August 2016 care physicians. Am J Prev Med 2011;40:61–66 24. Gabert R, Thomson B, Gakidou E, Roth G. of diabetes around the world. 11. Siegel R, Naishadham D, Jemal A. Cancer Identifying high-risk neighborhoods using elec- statistics, 2012. CA Cancer J Clin 2012;62:10–29 tronic medical records: a population-based ap- 12. U.S. Preventive Services Task Force. proach for targeting diabetes prevention and Duality of Interest. No potential conflicts of Genetic risk assessment and BRCA muta- treatment interventions. PLoS One 2016;11: interest relevant to this article were reported. tion testing for breast and ovarian cancer e0159227