Moving Diabetes Genetics Beyond Genome-Wide Association Studies

Ali Torkamani, PhD


September 02, 2010

The hunt for disease predisposition alleles for diabetes, especially for type 1 diabetes (T1D), has been very successful relative to genome-wide association studies (GWAS) for many other common diseases. However, most of the heritability for diabetes is yet to be explained. For already discovered loci, the challenge of translating these findings to the clinic looms large.

Investigators who gathered at the 70th Scientific Sessions of the American Diabetes Association (ADA) discussed new findings from recent GWAS in Japanese populations, efforts to identify rare variants through alternative study designs, and mechanisms of action for known susceptibility loci in hopes of illuminating entry points for therapeutic intervention.

Lessons Learned From GWAS in Japanese Populations

Most GWAS of diabetes have focused on disease predisposition in white populations. The extent to which these findings apply to nonwhite populations is an open question and is especially important in the attempt to translate findings into diverse clinical populations.

Shiro Maeda from RIKEN and colleagues in Japan[1] presented data on the effects of 23 previously identified loci for type 2 diabetes (T2D) in 3 independent Japanese populations. Only 7 loci had significant associations with susceptibility to T2D in this population, indicating that further studies for T2D associations in nonwhite populations are necessary.

The Search for Rare Variants Continues

Despite the successes in GWAS, geneticists across many fields have turned to rare genetic variants in search for unexplained heritability. At this year's ADA Scientific Sessions, investigators reported results of their efforts with copy-number variation (CNV) studies, family studies, and monogenic cases to aid in this search.

Hakon Hakonarson and investigators from the Children's Hospital of Philadelphia[2] focused on rare CNVs in childhood obesity to identify variants indirectly predisposing to T2D. Fifteen CNVs were identified as unique loci in both European-American and African American cases of childhood obesity, suggesting that rare CNVs may contribute to susceptibility to obesity in multiple ethnicities. However, the relative rarity of these CNVs highlighted some of the difficulties in associating rare variants with clinical phenotypes.

Pal Ramus Njolstad of the University of Bergen[3] furthered the search for missing heritability using CNVs in a family-based study of monogenic maturity-onset diabetes of the young (MODY). A novel copy-number gain spanning the PRSS1 gene was identified in the proband and across 4 generations of family members, with almost full segregation between CNV and MODY.

Wojciech Mlynarski of the Medical University of Lodz and colleagues[4] also reported on rare variants and familial aggregation of disease in their description of an array of glucokinase mutations in Polish patients with MODY. Of 168 unrelated patients with diabetes, 62 patients carried glucokinase mutations, including 42 different mutations and 11 novel mutations, 2 of which showed characteristics of a founder effect for MODY in Polish populations. These data present a striking case of allelic heterogeneity in a gene already known to be associated with various diabetes subtypes.

Finally, investigators presenting the results of the IRAS (Insulin Resistance Atherosclerosis) Family Study[5] fine-mapped a serum adiponectin linkage in 2 families to a nonsynonymous single-nucleotide polymorphism (SNP) in the ADIPOQ gene. This mutation is thought to lead to degradation of the protein and low serum adiponectin levels. This rare variant occurred at a frequency of 1.1% in the cohort of Hispanic-American families but 18% in the 2 linked families, demonstrating the power of family-based studies in solidifying the identification of rare functional variants.

Mechanisms, Mechanisms, Mechanisms

Although rare variants have clearly become a focus in the hunt for diabetes susceptibility loci, most talks during the genetics sessions at this year's ADA Scientific Sessions focused on mechanisms of diabetes and related comorbid conditions. Understanding the underlying mechanisms linking genetic susceptibility loci to diabetes could unveil specific opportunities for disease prevention and intervention for related comorbid conditions.

Donald W. Bowden of Wake Forest University and colleagues[6] reported findings from a GWAS for genetic associations of diabetic nephropathy in African American patients. Diabetic nephropathy is the most common cause of end-stage renal disease (ESRD). The case-control study included an initial cohort and a replication cohort of patients with T2D and ESRD as cases and persons without T2D and nephropathy as controls. Replicated SNPs were then genotyped in the non-T2D/ESRD and T2D/nonnephropathy cohorts to separate an association with individual phenotypes. Some strongly suggestive associations were reported in the T2D/ESRD cohort vs the T2D/nonnephropathy cohort, as well as in patients with T2D vs patients without T2D, but none reached statistical significance.

Investigators from the Early Growth Genetics Consortium[7] expanded their recent meta-analysis[8] on the link between fetal growth and T2D susceptibility loci. Of 23 T2D susceptibility loci known to be associated with fetal growth, variants at 4 loci were found to be associated with reduced birth weight and risk for T2D, namely ADCY5 and CDKAL1. These loci were previously identified by this group as being associated with fetal growth, as well as MTNR1B and KCNQ1. These data confirm a genetic link between reduced birth weight and increased risk for diabetes, but the mechanism behind this process is still unclear. Whether the influence of these variants is exerted through fetal or maternal genotypes is not known.

This theme of expanding on results of recent meta-analyses continued with Benjamin Voight and Inga Prokopenko,[9] who followed up on their recent demonstration of 12 novel loci associated with T2D[10] by investigating the association of these loci with fasting glycemic levels. The strongest associations suggested that KCNQ1 and CENTD2 influenced fasting glycemic levels and that CENTD2 variants were associated with indices of reduced beta-cell function, whereas variants at KLF14 were associated with reduced insulin sensitivity.

A similar type of investigation was conducted by Valeriya Lyssenko and colleagues,[11] who interrogated the genetic influence of the same 12 loci on fasting and postprandial glucose levels. However, results of this study, which was carried out in a cohort independent of that in Voight and colleagues' studies, provided a completely different set of associations. In Lyssenko and colleagues' cohort, impaired insulin secretion was associated with variants in FAM148B, MADD, and GIPR.

These data highlight the diversity of physiologic mechanisms underlying T2D susceptibility and suggest that environmental cues might turn out to be major players in unveiling risk phenotypes.

Exploring this idea, Richard Leslie and colleagues[12] performed epigenetic analyses of differential methylation to identify the mechanism through which nongenetic susceptibility to T1D could be expressed. They discovered differential methylation associated with genes involved in processes relevant to T1D autoimmunity, including inflammation and immunoglobulin secretion and degradation. However, most of these differentially methylated loci did not overlap with previously known susceptibility loci but were detected in patients before onset of T1D. Migrant studies suggest that exposure to new environmental stimuli increases risk for T1D[13]; these data suggest that changing environmental cues expose underlying risk and that the T1D phenotype is expressed only under these new conditions.

A report by Andrea Steck and colleagues from Colorado[14] bolstered this argument that changing environmental conditions could strongly influence the risk conferred by certain genotypes and T1D susceptibility. They demonstrated a stepwise decrease over time in the penetrance of the highest risk HLA-DR3/4 genotype but an increase in other HLA genotypes, suggesting that acute environmental changes influenced the development of islet autoimmunity. If these changes could be identified, they may have direct bearing on therapies to prevent T1D-associated autoimmunity.

Hints of some progress on understanding the interaction between genotype and therapeutic intervention were reported by investigators of the Diabetes Prevention Program.[15] Nonsignificant associations were seen at SNPs near PRKAG2 and metformin target AMPK, as were interactions between metformin transporter gene MATE1 and treatment response. These data suggest a linkage between response to metformin and variants of its target and regulators in the associated pathways. Although no significant results have yet been produced through these investigations, these results, if confirmed, suggest a role for patient stratification by genotype.

A Role for Master Regulators?

Two studies attempted to expose master regulators of diabetes phenotypes, suggesting that some general therapeutic strategies may be applicable to a large proportion of diabetes cases.

Nancy Cox and colleagues[16] reported their findings on a group of trans-acting regulatory SNPs that influence gene expression changes seen in insulin-resistant vs insulin-sensitive individuals. Of 41 genes previously implicated in T2D susceptibility by GWAS, 6 were differentially expressed in adipose tissue and 3 in muscle. Further analysis showed that a group of SNPs near these genes were associated with insulin sensitivity and that a small region on 1 set of genes seemed to control expression in both tissue types. These data suggest that systems-level manipulation of diabetes susceptibility loci may be a possibility.

Similarly, Struan Grant and colleagues[17] investigated transcriptional regulation in T2D. In their study, binding sites for the T2D susceptibility gene TCF7L2 were identified in a colorectal cancer cell line that is known to overexpress TCF7L2. Despite the apparent disconnect between the disease of interest and the cell-line model, previously identified T2D susceptibility loci were enriched in TCF7L2-binding sites, providing further evidence that TCF7L2 is a central node for T2D susceptibility.


Several experimental strategies were explored at the ADA's 70th Scientific Sessions, with fruitful results all around. The ever-deepening understanding of susceptibility to diabetes promises that more personalized control of diabetes could potentially be achieved by understanding the underlying genetic perturbations influencing each individual's case. Genetic studies persist as powerful tools in diabetes research and continue to unveil the complex mechanisms at play in diabetes etiology.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.