CHAPTER 14 GENETICS OF TYPE 2 DIABETES Jose C. Florez, MD, PhD, Miriam S. Udler, MD, PhD, and Robert L. Hanson, MD, MPH Dr. Jose C. Florez is Chief of the Diabetes Unit and an investigator in the Center for Genomic Medicine, Massachusetts General Hospital, Boston, MA, and Co-Director of the Program in Metabolism and Institute Member in the Broad Institute, Cambridge, MA, and Associate Professor in the Department of Medicine, Harvard Medical School, Boston, MA. Dr. Miriam S. Udler is Clinical Fellow in the Diabetes Unit and Center for Genomic Medicine, Massachusetts General Hospital, Boston, MA, and Postdoctoral Fellow in the Programs in Metabolism and Medical & Population Genetics, Broad Institute, Cambridge, MA, and Research Fellow in the Department of Medicine, Harvard Medical School, Boston, MA. Dr. Robert L. Hanson is Clinical Investigator and Head, Genetic Epidemiology and Statistics Unit in the Diabetes Epidemiology and Clinical Research Section, National Institute of Diabetes and Digestive and Kidney Diseases, Phoenix, AZ. SUMMARY Type 2 diabetes is thought to result from a has been found to have a stronger effect insufficient sample sizes to detect small combination of environmental, behavioral, than the rs7903146 SNP in TCF7L2, which effects, a nearly exclusive focus on popu- and genetic factors, with the heritability itself has only a modest effect (odds ratio lations of European descent, an imperfect of type 2 diabetes estimated to be in the ~1.4). Nonetheless, GWAS findings have capture of uncommon genetic variants, range of 25% to 72% based on family and illustrated novel pathways, pointed toward an incomplete ascertainment of alternate twin studies. Since early 2007, genome- fundamental biology, drawn attention to (non-SNP) forms of genetic variation, wide association studies (GWAS) have led the role of beta cell dysfunction in type 2 and the lack of exploration of additional to an explosion of data for the genetics diabetes, confirmed prior epidemiologic genetic models. of type 2 diabetes and related traits. observations, and provided possible These GWAS have occurred on the back- targets for pharmacotherapy and pharma- As the community embraces complemen- ground of genotyping arrays populated by cogenetic clinical trials. tary approaches that include systematic common single nucleotide polymorphisms fine-mapping, custom-made replication, (SNPs), deployed in various cohorts that On the other hand, the causal variants denser genotyping arrays, platforms that have coalesced to form large international have only been identified for a handful of focus on functional variation, next-gen- consortia. As a result, a list of genetic these loci, a substantial proportion of the eration sequencing techniques, systems loci that influence type 2 diabetes and heritability of these phenotypes remains biology approaches, and expansion to quantitative glycemic traits has begun unexplained, and this has tempered non-European populations, the coming to accumulate. Over 100 type 2 diabe- expectations with regard to their use in years will witness exponential growth in the tes-associated loci have been identified, clinical prediction. Together, the approx- understanding of the genetic architecture in addition to others involved in deter- imately 100 loci associated with type 2 of metabolic phenotypes. Whether these mining quantitative glycemic traits, such diabetes thus far explain ~10%–15% of findings prove useful in disease prediction as insulin resistance. However, no variant the genetic predisposition to the disease. or therapeutic decision-making must be that is widely shared across populations Limitations of early GWAS included tested in rigorously designed clinical trials. TYPE 2 DIABETES AS A GENETIC DISEASE The explosive parallel growth in the prev- 2 diabetes is greater for monozygotic by mutations in single genes (hence, alence of the related metabolic disorders twins (who share 100% of their DNA termed monogenic or Mendelian), prove of obesity and type 2 diabetes in much of sequence) than for dizygotic twins (who, that single base pair changes in the the developed and developing worlds over like siblings, share approximately 50% of coding regions of key genes, which lead the past few decades is almost certainly their DNA sequence) (1,2,3,4,5). Second, to alterations in protein sequence and driven by environmental and behavioral the incidence of diabetes is much higher function, are sufficient to cause hyper- factors, since genetic components do in certain racial/ethnic groups, despite an glycemia in the diabetic range (11,12). not change in an appreciable manner environment that is relatively comparable Consistent with this notion, the heritability over such a short time period. However, to that of neighboring populations (6,7,8). of type 2 diabetes estimated in a set of several lines of evidence suggest that vari- Third, family history is an independent risk Scandinavian families ranges from 25% ation in DNA sequence does contribute factor for the development of diabetes to 69% (13), and a large international to type 2 diabetes risk. First, twin studies in population studies (9,10). And fourth, meta-analysis of twin studies has reported have shown that concordance for type rare familial forms of diabetes, caused a heritability estimate as high as 72% (14). Received in final form April 4, 2016 14–1 DIABETES IN AMERICA, 3rd Edition Taken together, these observations illus- Factors for Type 2 Diabetes) that the risk of pathogenic mechanisms from which ther- trate that rapid changes in the global type 2 diabetes differs in the various ethnic apeutic windows may emerge. Because epidemiology of type 2 diabetes are likely groups that compose the U.S. population, germline genetic variation always predates caused by environmental and behavioral and the presumption is that some of these the onset of disease, the arrow of time factors overlaid on a background of differences are genetic in nature (15). establishes a causal relationship that is genetic predisposition. This genetic predis- not evident with other biologic associa- position may vary across populations, in Why is genetic exploration relevant? tions. Thus, the genetic approach has a some measure due to their divergent Regardless of whether genetic predictors unique opportunity to shed light on the genetic history and unequal selection become useful markers of disease onset pathophysiology of diabetes in its various pressures in specific geographic regions. or progression in clinical practice, the manifestations, helping unravel its clinical Thus, it is well known and described else- identification of genetic variants asso- heterogeneity and potentially refine thera- where in this volume (see Chapter 13 Risk ciated with type 2 diabetes illuminates peutic strategies. DISCOVERY OF TYPE 2 DIABETES GENES Before the sequencing of the human leukocyte antigen (HLA) region on type genes that encode antihyperglycemic drug genome was accomplished, genetic 1 diabetes (20,21), by and large, linkage targets: the p.Pro12Ala polymorphism mapping was dependent on the genera- analysis did not yield reproducible positive in the peroxisome proliferator-activated tion of anonymous genetic markers and results for type 2 diabetes. This is because receptor gamma 2 (encoded by PPARG) their anchoring on specific locations in in type 2 diabetes there is not a single (22) and the p.Glu23Lys polymorphism the genome. This task, first achieved genetic locus that exerts a very strong in the islet ATP-dependent potassium with restriction fragment length polymor- effect in the general population or even channel Kir6.2 (encoded by KCNJ11) phisms and then with other markers, such in individual family pedigrees. Thus, the (22,23). A third locus, a noncoding variant as microsatellites or sequence tag sites, effect of genetic variation is probabilistic in the transcription factor 7-like 2 gene enabled the introduction of whole-genome rather than deterministic; a substantial (TCF7L2), was discovered by large-scale linkage analysis and positional cloning, proportion of people with some risk vari- association testing in areas of sugges- which proved extremely useful in the iden- ants may be disease-free, whereas others tive linkage (24). The common intronic tification of genetic mutations that cause who carry protective alleles may instead rs7903146 polymorphism had the stron- monogenic disease. The linkage approach, have type 2 diabetes, due to a constel- gest statistical association (though with which depends on the cosegregation of lation of other factors. In such situations, a modest odds ratio ~1.4) and the most a causal mutation with the anonymous the amount of information provided by widespread effect on type 2 diabetes risk marker along the lines of inheritance in meioses within families, on which the (Figure 14.1) (25,26), albeit with an inter- pedigrees composed of affected and power of linkage analysis depends, is esting exception in some Native American unaffected members, is particularly useful greatly reduced, and the number of fami- populations (27). for traits where disease-causing alleles lies required can be inordinately large. are highly penetrant: that is, the presence The panorama changed dramatically with of the genetic variant virtually always To demonstrate the effect of genetic varia- the advent of genome-wide association co-occurs with disease, and its absence tion on human phenotypes, an alternative studies (GWAS) (28). Several factors co-occurs with absence of the disease. As approach