The First Genome-Wide Association Study for Type 2 Diabetes in Youth: The Progress in Diabetes Genetics in Youth (ProDiGY) Consortium

Type 2 diabetes (T2D) is a global epidemic and an acknowledged significant population health issue in adults (1). However, until recently, T2D was not thought to be relevant to youth. This paradigm has changed with the rise in obesity among youth worldwide (2). In the U.S., the population-based study SEARCH for Diabetes in Youth showed that the unadjusted incidence rates of T2D in youth 10–19 years of age increased by 7.1% annually between 2002 and 2012, with the increase more dramatic in minority race and ethnic groups (3). It has become apparent that youth-onset T2D displays an aggressive course and is associated with greater mortality, earlier complications, and unfavorable cardiovascular disease risk when compared with adult-onset T2D or youth-onset type 1 diabetes (4–6). These trends, coupled with the limited treatment options for the management of T2D in youth, makes this an emerging health concern (7).

T2D is a complex disease, influenced by the combination of genetic, epigenetic, environmental, and behavioral factors and their interactions (8). The heritability of T2D in youth is highlighted by the high concordance rates of T2D in identical twins, the presence of a strong family history of T2D in affected youth, and the disproportionate prevalence of T2D in certain racial and ethnic groups, such as Native American, Hispanic, and African American populations (7,9). However, while there have been rapid advances in the understanding of the genetics of T2D in adults, our understanding of the genetics underlying youth-onset T2D has lagged behind (10). While a few studies have examined candidate genes for T2D in youth, the genetics of T2D in youth remain largely understudied (11,12).

Progress in Diabetes Genetics in Youth (ProDiGY) is a collaboration of two pediatric T2D studies, Treatment Options for Type 2 Diabetes in Adolescents and Youth (TODAY) and SEARCH for Diabetes in Youth, along with the Type 2 Diabetes Genetics Exploration by Next-generation sequencing in multi-Ethnic Samples (T2D-GENES) consortium, a large collaborative effort to find genetic variants that influence risk of T2D in adults (10). In this report, we set out to comprehensively examine the genetics of T2D in youth in ProDiGY by comparing subjects from TODAY and SEARCH with two control cohorts: adults older than 50 years of age without T2D and youth younger than 18 years of age without diabetes. We also examined the effect of known genetic variants associated with T2D in adults in our youth-onset diabetes cohort.

ProDiGY is a multiethnic resource that includes data from >3,000 case subjects with youth-onset T2D and 6,000 diabetes-free adult control subjects with both genome-wide genotyping and whole-exome sequence data. The collaboration brings together data from 449 youth with T2D from the TODAY study as well as data from >2,000 youth with T2D from a TODAY ancillary genetics study. It also includes data from 468 youth with T2D from SEARCH for Diabetes in Youth, as well as access to data from >10,000 adult subjects and 10,000 control subjects from T2D-GENES (13). The TODAY and SEARCH studies are described in detail elsewhere (14,15).

Briefly, TODAY was a multicenter randomized controlled trial that enrolled 699 participants (aged 10–17 years) with T2D between 2004 and 2009. Participants were overweight or had obesity (BMI ≥85th percentile for age, sex, and height), with negative pancreatic autoantibodies (GAD-65 and tyrosine phosphatase) and a fasting C-peptide concentration >0.6 ng/mL. Participants were randomized to one of three arms—metformin alone, metformin plus rosiglitazone, and metformin plus lifestyle intervention—and followed on average for 3.86 years. The results of the TODAY clinical trial have been described elsewhere (7).

TODAY Genetics is an ancillary study of TODAY distinct from the original clinical trial. To qualify, participants must have been diagnosed with T2D prior to 18 years of age and have a documented BMI ≥85th percentile at the time of diagnosis. Data and sample collection occurred at a one-time research visit at 1 of 25 clinical sites. During this visit, a self-report questionnaire on family and medical history was administered and blood samples were drawn for DNA extraction and for analysis of glucose, C-peptide, and autoantibodies.

SEARCH is a population-based prospective registry study launched in 2000 that ascertained diabetes in youth diagnosed at <20 years of age in the U.S. at study centers located in five states: South Carolina, Ohio, Colorado, California, and Washington. Youth with T2D were identified by physician report.

In this study, we examined a total of 3,006 youth from TODAY, TODAY Genetics, and SEARCH. Additionally, 856 non-Hispanic White and African American youth without diabetes recruited from Ohio, Colorado, and South Carolina as part of the SEARCH case-control study were used in this analysis (16). The TODAY and SEARCH protocols were approved by the institutional review boards of each participating institution. Participants provided written informed parental consent and child assent, including consent and assent specifically for genetic testing.

Samples for the adult control subjects in T2D-GENES were drawn from 12 studies from T2D-GENES and are described fully in Supplementary Table 1. T2D status was determined according to study-specific criteria. A total of 6,061 adult control subjects older than 50 years of age and free of diabetes were selected from non-Hispanic White, Hispanic, and African American race/ethnicities to match the race/ethnicity distribution in the youth subjects. All individuals provided informed consent, and all samples were approved for use by their institution’s institutional review board or ethics committee.

Samples for ProDiGY were genotyped on the Infinium genome-wide association study (GWAS) array, a complement to the Nexome platform by the Genetic Analysis Platform at the Broad Institute. The directly genotyped data were called by using the Autocall algorithm. We imputed our genotyped data on the Michigan Imputation Server against the 1000 Genomes Phase 3 v5 panel as the reference. All quality control steps were implemented in PLINK2 and R-3.4. Samples were filtered for sex discrepancies, low sample call rate, and close relatedness; single nucleotide polymorphisms (SNPs) were filtered for minor allele frequency, low SNP call rate, and lack of compliance with Hardy-Weinberg equilibrium. The imputation threshold (R²) was set at 0.5. Principal component analysis was performed on the genome-wide identity-by-descent pairwise distances in conjunction with complete linkage clustering of individual after merging with 1000 Genomes data. After cleaning, ∼10 million variants were available for analyses (17).

We performed association tests for T2D using a generalized linear mixed model (GLMM) with a genetic relationship matrix to account for population substructure. Covariates included sex and BMI when available. We used the Efficient and Parallelizable Association Container Toolbox (EPACTS) to run GLMM tests to obtain P value estimates and Wald tests to obtain odds ratio (OR) estimates within each racial or ethnic group. A total of 898 of the ProDiGY participants had BMI z-score information. For the purpose of our analyses, BMI z score was calculated in adults according to BMI-for-age charts at 20 years using LMS parameters from https://www.cdc.gov/growthcharts/percentile_data_files.htm/. To account for the putative effect of BMI, a sensitivity analyses was performed adjusting for sex and BMI z score in the GLMM. A binomial test was performed to compare effects of known SNPs associated with T2D in ProDiGY with data previously reported in adults. A GLMM was applied for P value estimates by using EPACTS in each race and ethnic group. Meta-analyses were then conducted by using METAL to combine results from each racial and ethnic group. A threshold of P < 5 × 10⁻⁸ was used to define genome-wide significance. To fine-map the novel SNP, a credible set analysis was performed using a Bayesian approach (18). This method is described in further detail in the Supplementary Material.

The data sets generated analyzed in the current study are available in dbGap (dbGaP Study Accession: phs001511.v1.p1, https://www.ncbi.nlm.nih.gov/projects/gap/cgi-bin/study.cgi?study_id=phs001511.v1.p1). No applicable resources were generated or analyzed during the current study.

Baseline demographics of case and control subjects are summarized in Table 1. The analyses included 3,006 youth with T2D and 6,061 adult control subjects who were selected to be older than 50 years old and free of diabetes. In secondary analyses, we also conducted a GWAS with 856 youth control subjects without diabetes who were recruited as part of the SEARCH case-control study. Mean age of the affected youth was 15.1 ± 2.8 years compared with 54.2 ± 12.4 years for the adult control subjects. The majority of participants were female and non-White. As expected, the youth had high levels of obesity, with a mean BMI z score of 2.17 ± 0.6 compared with 1.08 ± 0.8 for the adult control subjects. Matching of youth subjects with adult control subjects was successful, as evidenced by little systematic deviation (λ_GC = 1.08) of the observed distribution from the expected distribution under the null hypothesis of no association in the quantile-quantile (QQ) plot (Fig. 1A); the QQ plot for youth subjects versus youth control subjects is comparable and shown in Fig. 1B (λ_GC = 1.09).

Figure 2 shows the Manhattan plot from the transethnic meta-analysis. We identified seven genome-wide significant findings, including variants in or near TCF7L2, MC4R, CDC123, KCNQ1, IGF2BP2, PHF2, and SLC16A11 (Table 2). The association signal in PHF2 is novel, whereas the remaining six variants have been associated with T2D in adults; relative effect sizes are compared in Table 2. To fine-map the region of the novel PHF2 signal, a credible set analysis was performed (reference). The 99% credible set contained a total of 38 possible causal variants. The 99% credible set of variants for this region is summarized in Supplementary Table 2. In the subset of 898 participants in whom BMI data were available, signals were attenuated but remained nominally significant after adjustment for BMI (Table 3). Association tests for youth subjects versus youth control subjects identified a novel genome-wide significant variant, rs2604566 in CPEB2 (P = 3.2 × 10⁻⁸; OR 2.1) (Fig. 3). This variant has been shown to be nominally associated with T2D in adults (OR 1.11; P = 0.014 in Joint T2D-CHD GWAS) (19).

Our lead SNP rs10992863 in PHF2 is also associated with height and BMI (https://t2d.hugeamp.org/variant.html?variant=rs10992863). However, Locus Zoom plots for height and BMI at this locus demonstrate that rs10992863 is not the top SNP for height and BMI and is in modest disequilibrium (r² = 0.1282) with rs9650755, the top SNP for height and BMI in this locus. This suggests that the youth-onset T2D-associated SNP rs10992863 may also be associated with BMI, but it is not the same signal. To confirm this, we conducted a colocalization analysis that shows that rs10992863 (the most significant SNP for T2D) and rs9650755 (the most significant SNP in this locus for BMI) are independent of each other (Supplementary Fig. 1). To test whether the association of rs10992863 with BMI is primarily driven by the BMI-associated variant rs9650755, we tested the association of rs10992863 with BMI in the UK Biobank, conditioning on rs9650755 (Supplementary Table 3). Our results confirm that rs10992863 is associated with BMI (P = 5.1 × 10⁻⁷; β = 0.062), but the association is significantly weakened in the presence of rs9650755 (P = 2.5 × 10⁻²; β = 0.030). However, rs9650755 still remained highly significant after conditioning on rs10992863, with a threefold strengthening of the effect size (P = 1.9 × 10⁻¹¹; β = 0.090). To further validate these findings, we performed conditional analyses of the above SNPs in ProDiGY (Supplementary Table 3). When we conditioned rs10992863 on rs9650755, the results remained significant (P = 4.4 × 10⁻⁸; β = 0.319). However, rs9650755 lost the association with T2D when conditioned on rs10992863 (P = 0.34; β = 0.046), ruling out that its BMI association is driving the observed effect on T2D. Altogether, our results confirm that the association of rs10992863 with T2D in ProDiGY is independent of the association with BMI and that the association of rs10992863 with BMI is just a shadow of the association of the main BMI SNP rs9650755 driven by its moderate linkage disequilibrium with rs10992863.

Finally, we examined whether variants previously associated with T2D in adults at genome-wide significant levels were also associated with youth-onset T2D in ProDiGY. Association results in ProDiGY for 303 published variants in adults (20) is shown in Supplementary Table 2. This analysis revealed substantial consistency in direction of effects, with 86% of the SNPs showing the same direction of effect for the analyses of youth subjects versus adult control subjects (binomial test P < 2.2 × 10⁻¹⁶) and 71% of the SNPs showing the same direction of effect for the analyses of youth subjects versus youth control subjects (binomial test P < 1.3 × 10⁻¹³) despite limited power to achieve similar levels of statistical significance. Sensitivity analyses done with non-Hispanic White samples also showed a consistent direction of effect (binomial test P < 2.2 × 10⁻¹⁶ for analyses with adult control subjects and P < 1.2 × 10⁻¹³ for analyses with youth control subjects).

To further explore the relationship of genetic variants with youth-onset T2D, we constructed polygenic risk scores in ProDiGY using risk alleles and effect sizes from known T2D genetic variants previously identified in adults (20) (Supplementary Fig. 2). We compared associations of the polygenic risk score between case subjects with youth-onset T2D and adult control subjects, youth case and youth control subjects, and adult subjects with T2D from T2D-GENES with adult control subjects. In all three scenarios, we found that the polygenic risk score was significantly associated with T2D with the same direction of effect (Table 4). Additionally, we found that the OR for T2D was higher in the youth case subject versus adult control analyses compared with the all adult analyses, without overlapping CIs. Altogether, these findings validate our initial findings related to T2D in youth as well as highlight the higher aggregate genetic risk burden of diabetes variants in youth when compared with adults with T2D.

Our findings in ProDiGY provide the first large-scale evaluation of the genetics of T2D in youth. Through GWAS, we discovered seven genome-wide significant variants associated with T2D in youth, including a novel variant in PHF2 shown to be nominally associated with T2D in adults (21). PHF2 encodes histone demethylase plant homeodomain finger 2, a transcriptional coactivator of the transcription factor ChREBP. Through mouse and in vitro studies, PHF2 has been shown to be involved in adipogenesis and fat storage through the regulation of CEBPα and peroxisome proliferator–activated receptor γ transcriptional activities in adipose tissue (22,23). Additionally, mice with targeted disruption of Arid5b (AT-rich interactive domain D), a specific PHF2 coactivator partner, show a reduction in their white adipose tissue mass as a result of reduced peroxisome proliferator–activated receptor γ activity, suggesting that PHF2 could play a role in the regulation of glucose and lipid homeostasis, potentially by exerting its effect during adipocyte development. PHF2 overexpression in mouse liver leads to hepatic steatosis via epigenetic effects on ChREBP that lead to increased monounsaturated fatty acid production (24).

In GTEx (https://gtexportal.org/home), our lead SNP rs10992863 is also not an expression quantitative trait locus (eQTL) for PHF2 in any tissue, although a potential effect on expression of PHF2 at specific developmental stages has not been ruled out. Interestingly, it is an eQTL in many tissues, including adipose, for RP11–165J3.6 (all increased expression per copy of the A allele, the risk allele for T2D). This region appears to have no coding exons, a single noncoding transcript, and sits upstream of PHF2. Also, rs10992675, which is ∼482 kb upstream of rs10992863 and not in linkage disequilibrium with rs10992863 in any population, is an eQTL for RP11–526D8.11, a pseudogene region near RP11–165J3.6. On chromosome 9, regions are positioned as RP11–526D8.11—rs10992675—RP11–165J3.6—PHF2—rs10992863. Taken together, there may be a regulatory role for rs10992863 of upstream elements, which in turn may regulate expression of PHF2 in adipose tissue.

The variants in TCF7L2, MC4R, CDC123, KCNQ1, IGF2BP2, and SLC16A11 have all been previously associated with T2D in adults (20,25). In addition, a prior study in SEARCH reported that genetic variation in TCF7L2 is associated with an increased risk of T2D in African American youth, with the OR for diabetes stronger in African American than in non-Hispanic White youth (12). The divergent ethnicity-based results did not replicate in ProDiGY, suggesting that the earlier findings in SEARCH might be due to statistical fluctuations in the context of smaller sample sizes.

The analyses comparing findings in ProDiGY with known variants associated with T2D in adults showed that for the majority of variants, the direction of effect was consistent between adults and youth with T2D. The fact that only 6 of the ≥400 genetic variants that have been associated with T2D in adults achieved genome-wide significance in ProDiGY is likely explained by the relatively small sample size compared with adult cohorts (20). Overall, these results demonstrate that there is significant overlap between the genetics of T2D in adults and youth.

As obesity is a risk factor for the development of T2D, it is important to evaluate the role of BMI in genetic associations of T2D to assess whether a particular genetic variant that influences T2D risk is mediated by change in BMI. We had BMI data available in 898 youth participants with T2D from TODAY and SEARCH, in addition to the adult and youth control subjects. BMI attenuated the T2D association signals in youth, with only rs7903146 in TCF7L2 remaining genome-wide significant. However, the direction of effect remained consistent for our top findings, despite the reduction in sample size. One of the top findings in ProDiGY was rs72982988 in MC4R, a gene well-known to influence obesity with the effect size for T2D in youth greater than reported in adults (20). rs72982988 is associated with T2D in adults at genome-wide level of significance (OR 1.05; P = 1.4 × 10⁻¹⁰), even after adjustment for BMI (26). It is possible that the role of obesity-related genes in youth varies in comparison with adults and is a topic to be explored further.

In secondary analyses, we conducted a separate GWAS comparing our 3,006 youth subjects with 856 youth control subjects free of diabetes from non-Hispanic White and African American ancestries. We identified a genome-wide significant variant, rs2604566 in CPEB2 on chromosome 4 (P = 3.2 × 10⁻⁸; OR 2.1), that has not been previously reported in the literature. Given that this finding was only uncovered in the youth analyses, the likely significance is that genetic variation in CPEB2 is only active in youth and displays an age-dependent phenomenon that occurs early in development. CPEB2 encodes cytoplasmic polyadenylation element binding protein 2, an mRNA binding protein that regulates cytoplasmic polyadenylation of mRNA. CPEB2-knockout mice show reduced uncoupling protein 1 level and impaired thermogenesis in brown adipose tissue (27). Functional studies are needed to further explore the role of this gene on T2D risk in youth.

Clinically, youth-onset T2D displays a more aggressive disease course than adult-onset diabetes, with more rapid β-cell decline (4–6). We therefore hypothesized that there might be significant genetic differences between youth and adult-onset diabetes or that genetic effects might be more prominent in youth. Our results suggest that there is significant overlap in the genetic architecture of T2D in youth and adults. In our cohort, genetic effect sizes are stronger in youth for some of our top findings; however, CIs around the ORs in youth and adults largely overlap due to the smaller sample size of the youth GWAS. Another potential difference compared with adults is that several of the top findings in our analysis seem to influence T2D risk through obesity, potentially suggestive of the stronger impact that obesity-mediated genetic effects have on diabetes risk in the younger population. This also supports the potential of these variants influencing obesity and accelerating T2D onset at a younger age in these youth.

Further work is needed to understand the genetic determinants of T2D in youth and to explore reasons for differences compared with adults. In addition, the genetic determinants of medication response in the TODAY study could be examined. All ProDiGY samples have also undergone whole-exome sequencing to enable the analysis of rare variants, which is a planned next step. However, large-scale studies in adults have shown that rare variants only explain a small fraction of the heritability of T2D and that extremely large sample sizes are needed to uncover exome-wide significant findings, which presents a challenge considering our sample size (13). In addition, other potential modulators, such as gene × environment interactions and epigenetic modifiers, remain to be explored.

The biggest strength of our study is that it brings together two large available youth-onset T2D studies, TODAY and SEARCH, along with the TODAY Genetics ancillary study and the large adult T2D consortium T2D-GENES, compiling the largest collection of subjects with youth-onset T2D and adult control subjects known to us. While previous smaller scale genetic studies of youth-onset T2D have been limited in their genetic diversity, our study is multiethnic, a key advantage considering that the disease predominantly affects non-White youth. A weakness of the study is the limited BMI data because BMI was not collected in the TODAY Genetics cohort. Additionally, whereas our sample size is large compared with other pediatric studies, it is still limited from a GWAS standpoint, and further collaborations are needed to increase our sample size for more effective genetic exploration. Another limitation is that in line with the usual practice in GWAS for T2D in adults, case subjects with monogenic diabetes were not screened out. As part of a separate ongoing project, we have estimated that the prevalence of these subjects is only 2% in ProDiGY, and thus, it is unlikely that removal of such a small number of subjects would drastically influence results.

In conclusion, in ProDiGY, we have discovered seven genome-wide significant associations with T2D in youth, one novel and the remaining known, providing initial insight into the genetic architecture of T2D in youth. Importantly, ProDiGY has established a cohort of subjects with youth-onset T2D who serve as a valuable resource for future genetic investigation.

Acknowledgments. The TODAY and SEARCH for Diabetes in Youth studies are indebted to the many youth and families and the health care providers, whose participation made this study possible.

Funding. S.S. is supported by National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) grant 1K23DK120932-01A. J.M.M. is supported by American Diabetes Association Innovative and Clinical Translational Science Award 1-19-ICTS-068. SEARCH 3/4 was supported by the South Carolina Clinical & Translational Research Institute at the Medical University of South Carolina, National Institutes of Health (NIH)/National Center for Advancing Translational Sciences (NCATS) grants UL1 TR000062 and UL1 TR001450; Seattle Children’s Hospital and the University of Washington, NIH/NCATS grant UL1 TR00423; University of Colorado Pediatric Clinical and Translational Research Center, NIH/NCATS grant UL1 TR000154; the Barbara Davis Center for Childhood Diabetes at the University of Colorado at Denver, Diabetes Endocrinology Research Centers NIH grant P30 DK57516; the University of Cincinnati, NIH/NCATS grants UL1 TR000077 and UL1 TR001425; and the Children with Medical Handicaps Program managed by the Ohio Department of Health. TODAY and TODAY Genetics was completed with funding from NIDDK and the NIH Office of the Director through grants U01-DK61212, U01-DK61230, U01-DK61239, U01-DK61242, and U01-DK61254; the National Center for Research Resources General Clinical Research Centers Program grants M01-RR00036 (Washington University School of Medicine), M01-RR00043-45 (Children’s Hospital Los Angeles), M01-RR00069 (University of Colorado Denver), M01-RR00084 (Children’s Hospital of Pittsburgh), M01-RR01066 (Massachusetts General Hospital), M01-RR00125 (Yale University), and M01-RR14467 (University of Oklahoma Health Sciences Center); and the National Center for Research Resources Clinical and Translational Science Awards grants UL1-RR024134 (Children’s Hospital of Philadelphia), UL1-RR024139 (Yale University), UL1-RR024153 (Children’s Hospital of Pittsburgh), UL1-RR024989 (Case Western Reserve University), UL1-RR024992 (Washington University in St. Louis), UL1-RR025758 (Massachusetts General Hospital), and UL1-RR025780 (University of Colorado Denver). SEARCH for Diabetes in Youth (SEARCH 3) is funded by the Centers for Disease Control and Prevention (PA numbers 00097, DP-05-069, and DP-10-001) and supported by the NIDDK. The SEARCH for Diabetes in Youth Cohort Study (1UC4DK108173) (SEARCH 4) is funded by the NIH/NIDDK and supported by the Centers for Disease Control and Prevention. The Population Based Registry of Diabetes in Youth Study (1U18DP006131, U18DP006133, U18DP006134, U18DP006136, U18DP006138, and U18DP006139) is funded by the Centers for Disease Control and Prevention and supported by the NIH/NIDDK. The sites for SEARCH 1–4 and funding are as follows: Kaiser Permanente Southern California (U18DP006133, U48/CCU919219, U01 DP000246, and U18DP002714), University of Colorado Denver (U18DP006139, U48/CCU819241-3, U01 DP000247, and U18DP000247-06A1), Cincinnati Children’s Hospital Medical Center (U18DP006134, U48/CCU519239, U01 DP000248, and 1U18DP002709), University of North Carolina at Chapel Hill (U18DP006138, U48/CCU419249, U01 DP000254, and U18DP002708), Seattle Children’s Hospital (U18DP006136, U58/CCU019235-4, U01 DP000244, and U18DP002710-01), and Wake Forest University School of Medicine (U18DP006131, U48/CCU919219, U01 DP000250, and 200-2010-35171). Sequencing for T2D-GENES cohorts was funded by NIDDK grant U01DK085526 (Multiethnic Study of Type 2 Diabetes Genes) and National Human Genome Research Institute grant U54HG003067 (Large Scale Sequencing and Analysis of Genomes). Sequencing for ProDiGY cohorts was funded by NIDDK grant U01DK085526. Analysis was supported by NIDDK grant U01DK105554 (AMP T2D-GENES Data Coordination Center and Web Portal). The Mount Sinai Institute for Personalized Medicine Biobank Program is supported by The Andrea and Charles Bronfman Philanthropies. The research from the Korean cohort was supported by a grant of the Korea Health Technology R&D Project through the Korea Health Industry Development Institute, funded by the Ministry of Health and Welfare, Republic of Korea (grants HI14C0060 and HI15C1595). The Framingham Heart Study is funded by NIH contracts N01-HC-25195 and HHSN268201500001I and NIH grants U01 DK085526 and R01 DK078616. Additional support for the Framingham Heart Study is provided by the National Heart, Lung, and Blood Institute (NHLBI) (N01‐HC‐25195 and grant R01 NS17950) and National Institute on Aging (AG08122 and AG033193). This research was conducted in part using data and resources from the Framingham Heart Study of the NHLBI of the NIH and Boston University School of Medicine. The analyses reflect intellectual input and resource development from the Framingham Heart Study investigators participating in the SNP Health Association Resource project. This work was partially supported by the NHLBI’s Framingham Heart Study (contract number N01-HC-25195) and its contract with Affymetrix, Inc. for genotyping services (contract number N02-HL-6-4278). This research was partially supported by NIDDK grant 1R01DK8925601. A portion of this research used the Linux Cluster for Genetic Analysis. The Mexico City Diabetes Study has been supported by the following grants: RO1 HL24799 from the NHLBI; Consejo Nacional de Ciencia y Tecnología 2092, M9303, F677-M9407, 251M, 2005-C01-14502, and SALUD 2010-2151165; and Consejo Nacional de Ciencia y Tecnología (Fondo de Cooperación Internacional en Ciencia y Tecnología C0012-2014-01-247974). The Diabetes in Mexico Study was supported by Consejo Nacional de Ciencia y Tecnología grant S008-2014-1-233970 and by Instituto Carlos Slim de la Salud, A.C.

The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention or the NIDDK. This study includes data provided by the Ohio Department of Health, which should not be considered an endorsement of this study or its conclusions. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. S.S. assisted in analysis of results, interpretation of data, wrote the first draft of the manuscript, and edited and approved the manuscript. L.C. and J.M.M. conducted the analysis and edited and approved the manuscript. J.T., J.D., S.G., S.C., R.A.G.-K., M.M.K., R.S., M.H.B., L.E.W., A.M., J.F., G.I., J.M.M., and D.D. assisted in forming the analyses plan, interpretation of data, and edited and approved the manuscript. J.C.F. supervised analyses and interpretation of data and edited and approved the manuscript. J.C.F. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. This study were presented in oral form at the 78th Scientific Sessions of the American Diabetes Association, Orlando, FL, 22–26 June 2018 and the Annual Meeting of the Pediatric Endocrine Society, Baltimore, MD, 24 April–1 May 2019.