Insights into the Genetic Architecture of Diabetic Nephropathy

Abstract

Diabetic nephropathy (DN) is a devastating complication of type 1 and type 2 diabetes and leads to increased morbidity and premature mortality. Susceptibility to DN has an inherent genetic basis as evidenced by familial aggregation and ethnic-specific prevalence rates. Progress in identifying the underlying genetic architecture has been arduous with the realization that a single locus of large effect does not exist, unlike in predisposition to non-diabetic nephropathy in individuals with African ancestry. Numerous risk variants have been identified, each with a nominal effect, and they collectively contribute to disease. These results have identified loci targeting novel pathways for disease susceptibility. With continued technological advances and development of new analytic methods, additional genetic variants and mechanisms (e.g., epigenetic variation) will be identified and help to elucidate the pathogenesis of DN. These advances will lead to early detection and development of novel therapeutic strategies to decrease the incidence of disease.

Introduction

Diabetic nephropathy (DN) has a multifactorial etiology resulting from the combined influence of genetic and environmental factors. Two coding variants in the apolipoprotein L1 gene (APOL1) play a major role in susceptibility to several related non-diabetic kidney diseases in individuals with African ancestry; however, current evidence does not support a similar single major locus effect in DN. Although there is a clear genetic component, it is likely constituted by multiple genetic variants each of nominal effect. Nonetheless, remarkable progress has been made in understanding the genetic architecture of DN. This article provides an overview of the methods used and highlights the resulting observations that have advanced our understanding of the genetic determinants in DN.

Natural History of DN

Our understanding of the natural history of DN continues to evolve. The onset of type 1 diabetes (T1D) is typically clinically apparent and abrupt; this is not the case in type 2 diabetes (T2D). Therefore the natural history of DN has been more intensively studied in T1D. It is clear that microalbuminuria does not uniformly precede falling glomerular filtration rate (GFR) or loss of kidney function in all patients [1, 2]. In addition, urinary albumin excretion has proven to be a continuous risk factor for DN and cardiovascular disease (CVD) [3]. Most patients with advanced DN have heavy proteinuria. Widespread use of angiotensin converting enzyme inhibitors and angiotensin-receptor blockers limit the usefulness of albuminuria as an intermediate phenotype for DN and has altered the natural history of disease.

Genetic Determinants of DN

The etiology of DN is multifactorial, yet clearly has an inherent genetic basis. Allowing for a permissive environment of hyperglycemia, family history of kidney disease appears to be among the strongest risk factors for initiation of diabetes-associated nephropathy [4–7]. Familial aggregation of nephropathy in T2D has been reported in numerous populations, e.g. European Americans [8], Canadians [9], Europeans [10], Asians [11], Brazilians [12], Indians [13] and Japanese [14], but appears to be most prominent in African Americans [15, 16] and Pima Indians [17, 18]. The variable risk of DN in groups with different population ancestries suggests that differences in lifestyle and environment played roles in selecting alleles with differential disease risk.

Insights into the Genetic Architecture of DN

Evidence of an important genetic component to DN has stimulated extensive efforts to decipher the genetic architecture of disease in multiple populations. DN susceptibility loci can be identified in a variety of ways (Table 1). Three broad analytic approaches have typically been applied: candidate gene studies, linkage analysis and genome-wide association studies, each identifying different putative susceptibility loci (Table 2). We review each approach and recent observations, in turn.

Table 1.

Approaches to Gene Identification

Approach	Description
Candidate Gene Approach	Assessment of genetic variation, typically single nucleotide polymorphisms (SNPs), in one or more genes with a plausible biological role in the disease/trait of interest. The goal of this approach is to demonstrate a significant difference in allele frequencies between case and control subjects or mean trait values by genotype.
	Pros: targeted evaluation of biologically relevant genes studies of this type are common and well suited for meta-analyses thereby increasing power through an increase in sample size
	Cons: biased selection as candidates are based on “known” biological pathways small individual study sizes resulting in underpowered analyses and inconsistent results targeting of a small number of genetic polymorphisms resulting in failure to comprehensively test candidate genes
Linkage Analysis	Interrogation of hundreds of markers, typically microsatellite markers, spaced evenly across the genome to identify regions co-inherited with a disease/trait of interest in families.
	Pros: unbiased analysis of the entire genome; i.e. potential for identification of novel biological candidates and pathways family-based studies are immune to population stratification
	Cons: analysis is affected by linkage disequilibrium patterns and allele frequency differences linkage regions are typically large encompassing megabases of DNA that require additional genotyping for refinement of the linkage signal
Genome Scan Meta-analysis (GSMA)	An exploratory, quantitative data analysis method used to synthesize linkage results from independent studies.
	Pros: increased sample size due to combining cohorts, thus increasing power
	Cons: heterogeneity among phenotypes and between populations can reduce statistical significance in the overall analysis
Mapping by Admixture Linkage Disequilibrium (MALD)	Genome-wide analysis approach employing1500–2000 genetic markers, typically SNPs, exhibiting marked frequency differences between ancestral populations (ancestry informative markers or AIMs). These markers are used to identify regions of the genome in admixed populations with differential disease risk in their ancestral populations resulting in differential linkage disequilibrium (LD) patterns surrounding loci linked to the disease/trait of interest, i.e. increased LD in regions linked to disease.
	Pros: unbiased analysis of the entire genome economically advantageous as it requires fewer markers than GWAS
	Cons: the disease/trait of interest has to have a measureable frequency difference between the two ancestral populations requires identification of AIMs that differentiate the contribution of the ancestral populations identified regions are typically large encompassing megabases of DNA that require additional genotyping for refinement of the signal
Genome-Wide Association Study (GWAS)	Analysis of a large number of genetic markers, typically SNPs, in DNA samples from multiple individuals with the goal of detecting common (minor allele frequency >5%) genetic variation associated with the disease/trait of interest.
	Pros: performed in population-based samples allowing for recruitment of large sample sizes unbiased scanning of the entire genome for susceptibility loci imputation allows for increased coverage
	Cons: coverage is currently limited to common variation current arrays are not as informative in non-European-derived populations

Table 2.

Genes Associated with Diabetic Nephropathy

Gene Symbol	Gene Name	References
Candidate Gene Association
ACE	Angiotension converting enzyme	[26]
NOS3	Endothelial nitric oxide synthase	[22–25]
EPO	Erythropoietin	[21]
PRKCB	Protein kinase C beta	[20]
Linkage
CNDP1	Carnosinase 1	[31, 32, 30]
Genome Wide Association
ACACB	Acetyl-coenzyme A carboxylase beta	[60, 61]
CARS	Cysteinyl-tRNA synthetase	[63]
ELMO1	Engulfment and cell motility 1 gene	[56, 57, 63]
FRMD3	4.1 protein ezrin, radixin, moesin [FERM] domain containing 3	[63]
NCALD	Neurocalcin delta	[59]
PVT1	Plasmacytoma variant translocation 1	[62]
Mapping by Admixture Linkage Disequilibrium (MALD)
MYH9	Non-muscle myosin heavy chain 9	[68, 69]

Candidate Gene Studies

The candidate gene approach involves assessment of genetic variation, typically single nucleotide polymorphisms (SNPs), in one or more genes with plausible physiological roles in DN. The goal is to demonstrate a significant difference in allele frequencies between cases with DN and control subjects [19]. In some studies, controls had longstanding diabetes without evidence of nephropathy, while in others, controls lacked diabetes and nephropathy. Limitations of this approach include that candidate gene studies are frequently based on a poor understanding of pathways involved, small numbers of cases and controls resulting in underpowered analyses, failure to comprehensively test candidate genes by targeting a small number of genetic polymorphisms and ethnic-specific genetic variability.

Numerous candidate gene studies have been reported; however, the results have largely been inconsistent. Three important examples of successful candidate gene studies in DN (with subsequent replication) are the protein kinase C β gene (PRKCB) association in T2D nephropathy among Hong Kong Chinese [20], a SNP in the erythropoietin gene (EPO) promoter with proliferative diabetic retinopathy and DN in individuals with European-ancestry [21] (not in African Americans; DW Bowden personal communication), and endothelial nitric oxide synthase gene (NOS3) association in DN [22–25]. These reports demonstrate the potential usefulness of a candidate gene approach when appropriate genes and pathways are targeted.

Given these limitations, individual candidate gene studies can contribute to larger, well-powered meta-analysis efforts. Meta-analysis is a statistical method for combining results from independent studies. Their major advantage is a larger sample size resulting in increased power to detect more nominal effects. A large meta-analysis assessed the angiotensin 1-converting enzyme insertion-deletion (ACE I/D) polymorphism in 26,580 ethnically diverse DN cases (type 1 and type 2 diabetes) and control subjects [26]. For DN patients with T2D, the ACE D allele was associated with DN risk with an odds ratio (OR) in the range of 1.25–1.57 in the Asian subgroup, an effect size consistent with that observed for common variants associated with complex disease. However, no significant effects were detected for the Caucasian subgroup.

Mooyaart et al. [27] took the innovative approach of applying a combined bioinformatic and meta-analysis to evaluate genetic associations with DN. Based on literature searches, 671 genetic association studies investigating DN were identified, among which 34 replicated genetic variants were identified with 21 significantly associated with DN in a random-effects meta-analysis including ACE, AKR1B1, APOC1, APOE, EPO, NOS3, HSPG2, VEGFA, FRMD3, CARS, UNC13B, CPVL, CHN2, and GREM1. A limitation of this approach is publication bias, as only published data were included resulting in overestimation of effect size. Despite these challenges, these loci represent putative targets for future studies. The meta-analysis approach will have increasing utility as genetic studies of DN are published. Meta-analyses, while well powered, also have limitations. This approach is limited to the number of genetic variants evaluated, i.e. typically not a comprehensive survey. In addition, study samples are not always collected in a uniform fashion and may lack precise phenotype characterization resulting in phenotypic heterogeneity and loss of power.

Linkage

Moving from focused efforts to evaluate candidate genes, a central theme of genetic studies has been to comprehensively survey the genome in an unbiased fashion for loci involved in disease susceptibility. This approach interrogates hundreds of markers spaced evenly across the genome in an effort to identify regions of the genome co-inherited with disease or trait of interest in families. A major advantage of this approach is that family-based studies are immune to population stratification although they can be affected by linkage disequilibrium patterns and allele frequency differences. However, linkage regions are typically large encompassing megabases of DNA that require subsequent detailed analysis to refine the linkage signal.

The first genome-wide linkage scan was performed to identify nephropathy genes in Pima Indian sibling pairs with T2D and retinopathy in 1998 with the strongest evidence for linkage on chromosome 7 with a LOD (log of the odds) score of 2.7; approximately 500:1 odds that this region of the genome was linked or co-inherited with nephropathy [28]. Current studies have extended the evaluation of linkage peaks by identifying positional candidate genes within linkage intervals for subsequent genetic evaluation. In 2002, a linkage scan in a Turkish kindred with multiple DN-affected individuals identified a major linkage peak on chromosome 18 with a LOD score 6.6, i.e. 10^6.6:1 odds for linkage [29]. Subsequently, variations in the carnosinase 1 gene (CNDP1) were implicated as the basis of the linkage peak [30], an observation replicated in European Americans [31]. Carnosine has antioxidant properties functioning as a scavenger of oxygen free radicals and may inhibit advanced glycosylation end-product formation. Among the variants associated with DN was a variable trinucleotide repeat (D18S880) located in exon 2 encoding a leucine repeat in the leader peptide of the carnosinase-1 precursor. Under a recessive model, the 5 leucine-5 leucine (5L-5L) gene variant was protective and associated with decreased serum carnosinase levels. Ahluwalia et al. [32] more broadly implicated the CNDP1 and CNDP2 loci through a three SNP haplotype in a large sample of European DN.

The Family Investigation of Nephropathy and Diabetes (FIND) study performed a linkage study for DN [33]. FIND targeted predominantly sibling pairs concordant or discordant for DN (n>9,000 individuals) from European American, African American, Mexican American, and American Indian families. Initial analyses in 378 pedigrees (n=1,227) assessed linkage to DN and the quantitative intermediate phenotypes of albuminuria and GFR [34, 35]. When combined with previous reports, these findings reinforced the evidence for DN genes on chromosomes 3q, 7q, 10p, and 18q [28, 29, 36]. As an extension of these findings, He et al. [37] reported that the non-catalytic region of tyrosine kinase (NCK1) and transmembrane protein 22 (TMEM22) loci underlying the chromosome 3q linkage peak may contribute to risk of DN; NCK1 plays a role in podocyte biology linking phosphorylated nephrin with the actin cytoskeleton. Vionnet et al. [38] also reported association between DN in T1D with variants in the adiponectin gene (ADIPOQ) under the 3q linkage peak.

The initial FIND report was followed by a larger whole-genome linkage analysis (n>5,500 SNPs; average spacing of 0.6cM) of 1,235 nuclear and extended pedigrees encompassing 3,972 diabetic participants ascertained for DN [39]. As a result, evidence for linkage of DN to chromosome 6p (LOD 3.09, P=8.0×10 ⁻⁵) in European American families was detected with suggestive evidence on chromosome 7p in American Indian families. For albuminuria, regions on chromosomes 3p and 16q in African Americans, 7q in European Americans, and 22q in Mexican Americans displayed suggestive evidence of linkage. The linkage peak observed on chromosome 22q overlaps the APOL1 gene region; a locus previously implicated in non-diabetic nephropathy in African Americans [40–42].

As a result of these and previous studies, there have been numerous investigations targeting a range of kidney-related phenotypes. Studies have identified multiple linkage peaks lacking genome-wide significance and possessing little between-study consistency. Despite disparate results, Rao et al. [43] used genome scan meta-analysis (GSMA) as an exploratory data analysis method [44–46] to synthesize linkage results. This quantitative method was used to combine linkage results from 22 independent studies and assess their concordance. This collaborative genome scan included surrogate measures of kidney disease, i.e. albuminuria, GFR, serum creatinine concentration, and creatinine clearance, from European, African American, Mexican American, and American Indian populations. While this approach has been successfully applied to other phenotypes [47–49], none of the regions highlighted by GSMA reached genome-wide statistical significance. An advantage of this approach is increased sample size due to combining cohorts, thus increasing power. However, the lack of significance could be attributed to diverse linkage signals across studies due to heterogeneity among phenotypes and between populations. The heterogeneity among phenotypes, the authors cite, could be attributed to ethnicity, as race is used in estimating equations to determine GFR; reversibility of albuminuria and lack of correlation with kidney disease; medications targeting the renin-angiotensin-aldosterone system and their resulting impact on GFR and albuminuria; and expected day to day variation in renal phenotypes.

The utility and significance of family-based linkage studies to identify the genetic determinants of DN remains unclear. There are few examples where linkage studies successfully identified loci underlying complex traits such as DN. CNDP1 is an exception however, the overall implications remain unclear. Family studies will likely have new utility in testing the role of rare exonic variants in DN.

Genome-wide Association Studies (GWAS)

As technology has led the way for development of methodologies to capture variation across the genome we have entered a new era in the search for genetic contributors to DN with genome-wide association studies (GWAS). The GWAS approach involves rapidly scanning many genetic markers (usually SNPs) in DNA samples from multiple individuals to detect common (minor allele frequency >5%) genetic variations associated with disease [50]. In contrast to family-based linkage, these studies are often performed in population-based samples, frequently comparing cases and controls. GWAS have evolved in complexity, now scanning up to five million SNPs spanning the genome. Moreover, additional genotypes can be imputed [51] using known haplotype structure from HapMap samples [52–55].

The first GWAS for DN was performed in a Japanese population using a low density array genotyping ~80,000 gene-based SNPs [56]. Several replicated DN susceptibility loci were identified in this landmark GWAS, including the engulfment and cell motility 1 gene (ELMO1) [57, 58, 56], neurocalcin delta (NCALD) [59] and acetyl-coenzyme A carboxylase beta (ACACB) loci [60, 61]. The initial Japanese GWAS was soon followed by a GWAS in Pima Indians with DN where a DNA pooling approach was utilized to investigate ~115,000 SNPs [62]. The plasmacytoma variant translocation 1 locus (PVT1) was associated with DN in Pima Indians.

A GWAS in Genetics of Kidneys in Diabetes (GoKinD) participants included 820 cases with T1D and DN and 885 controls with longstanding T1D lacking nephropathy [63]. GoKinD identified SNPs in the 4.1 protein ezrin, radixin, moesin [FERM] domain containing 3 (FRMD3) and cysteinyl-tRNA synthetase (CARS) loci as DN-associated, with replication based on time to development of nephropathy in participants in the Diabetes Control and Complications Trial (DCCT)-Epidemiology of Diabetes Interventions and Complications (EDIC) Study. These associations failed to achieve strict statistical evidence of genome-wide association. FRMD3 association was also present in African Americans with T2D and nephropathy [64].

More recently, GWAS of DN have utilized higher density arrays in an effort to capture common genomic variation across the genome. In 2011, the first GWAS of DN in African Americans was performed using 965 T2D patients with end-stage renal disease (ESRD) and 1,029 controls without diabetes or nephropathy [65]. From the GWAS, the top 724 SNPs with evidence of association to DN-associated ESRD were targeted for genotyping in a replication cohort of 709 DN-ESRD cases and 690 controls. Twenty five SNPs with consistent evidence of association in both the GWAS and replication study were further evaluated in an additional 1,246 T2D patients lacking kidney disease and 1,216 non-diabetic ESRD patients to differentiate candidate loci for DN, T2D, and/or all-cause ESRD. Although genome-wide significance was not found for any of these variants, several loci, e.g. RPS12, LIMK2, and SFI1, were identified as candidates for DN. A combined analysis of all nephropathy patients (n=2,890) highlighted significant association for SNPs in LIMK2 and SFI1 with p-values up to 2.2 × 10⁻⁶ suggesting a contribution to all-cause ESRD. This GWAS also revealed moderate evidence for association with ELMO1 (242 genotyped and 535 imputed SNPs, with 15 (P=0.005–0.05) and 57 (P=0.0038–0.05) associated, respectively (DW Bowden, unpublished). The authors acknowledge that study design may not be ideal given that studies of DN have typically used controls with T2D lacking nephropathy. However, African Americans with T2D and normal albuminuria after 10 or more years of disease are ideal but uncommon [66].

Although GWAS have identified several novel genes as potential contributors to DN, the approach is not without limitations. GWAS can be confounded by population admixture (combining multiple races for analysis) resulting in association due to differences between genetically distinct population. In addition, GWAS to date have only evaluated common genetic variation using arrays which are not as comprehensive in their coverage of non-European populations.

Potential Roles of MYH9 and APOL1 in DN

Loci associated with non-diabetic forms of nephropathy make among the strongest contributions to complex human disease. Mapping by admixture linkage disequilibrium (MALD) successfully led to identification of the APOL1 and non-muscle myosin heavy chain 9 (MYH9) gene associations with a spectrum of previously unrelated kidney diseases in individuals with African ancestry [41, 42]. APOL1-associated diseases include focal segmental glomerulosclerosis (FSGS), HIV-associated collapsing glomerulosclerosis, arteriolar nephrosclerosis (histologically manifesting as focal global glomerulosclerosis, interstitial fibrosis and microvascular changes), and hypertension-attributed ESRD [40]. These disorders occur far more commonly in African Americans, relative to European Americans.

MALD is a genome-wide analysis approach useful in admixed populations with differential disease risk in ancestral populations [67]. Compared to GWAS, MALD is economical; employing 1500–2000 genetic markers exhibiting marked frequency differences between ancestral populations (ancestry informative markers or AIMs). Genome-wide MALD markers were evaluated in African American cases with FSGS and non-nephropathy controls. A region on chromosome 22q contained 10% excess African ancestry in nephropathy cases [68, 69]. Fine mapping initially implicated MYH9; however markers from the 1000 Genomes Project clarified that the adjacent APOL1 gene was the major nephropathy risk locus in the region [41]. Two coding variants in APOL1, termed G1 and G2, demonstrated autosomal recessive inheritance with an association odds ratio (OR) of 29 in HIVAN, 17 in FSGS, and 7.3 in non-diabetic “hypertension-attributed” ESRD [41, 70]. These variants are virtually absent in European-derived populations where MYH9 remains nephropathy-associated with an OR of approximately 1.3 [71–73]. This does not prove that MYH9 contributes to disease, as this region demonstrates marked linkage disequilibrium and strong evidence of selective forces. Therefore, MYH9 could reflect association with as yet undetected markers in the nearby APOL1 locus, as well as in the APOL2-APOL6 region. APOL1 nephropathy risk variants were conserved in the population since they protect from African sleeping sickness in sub-Saharan Africa, a parasitic disease caused by Trypanosoma brucei rhodesiense and transmitted by the sting of an infected tse-tse fly.

APOL1 and MYH9 are predominantly associated with non-diabetic kidney disease; however, they may play roles in susceptibility to DN and have proven to be important for investigating the pathogenesis of DN in African Americans. It was subsequently shown that stratifying African Americans with clinically diagnosed DN based on the presence of one or zero (versus two) APOL1 risk variants enriched for cases with DN [64]. Individuals possessing fewer than two chromosome 22q risk variants demonstrated significant association with FRMD3, initially detected in a GWAS evaluating T1D and DN in European Americans [63]. Association with FRMD3 was not observed in African Americans with two APOL1 risk variants (likely with FSGS) or in a prior GWAS for diabetic ESRD in African Americans [65]. This suggests that genetic dissection of non-DN from DN was achievable in individuals with African ancestry based on MYH9 and APOL1 variants. This strategy has great clinical importance; African Americans with two APOL1 risk variants and T2D are more likely to have FSGS-related kidney disease than DN. Future clinical trials evaluating DN treatments will need to exclude or stratify for these individuals to ensure a more homogeneous population with DN is evaluated.

An interesting observation is that diabetes appears to impact the renal effects of APOL1 risk variants in African Americans. In a large population-based study, stronger association of APOL1 with nephropathy was evident in non-diabetic African Americans [74]. This is somewhat surprising as hyperglycemia clearly contributes to diabetes-associated glomerulosclerosis. A similar observation was made in first-degree relatives of African Americans with non-diabetic ESRD [75]. Close relatives of these index cases who had diabetes and two APOL1 risk variants were somewhat less likely to demonstrate association with proteinuria or reduced GFR, compared to non-diabetic relatives. The mechanisms where hyperglycemia may paradoxically alter or weaken nephropathy risk in those with two APOL1 risk variants remains to be determined.

The Future of Genetics Studies of DN

The search for genes associated with DN relies on continued innovation. While progress is being made with recent technological advances in GWAS; objectively, genetic variants have not been identified that unambiguously define DN genes. Thus, DN appears typical of common complex diseases. Compared to contemporary GWAS in T2D and coronary heart disease, sample sizes accessible for DN remain small and underpowered to detect more nominal effects (OR<1.3).

Major research questions remain regarding the underlying genetic architecture of DN. It remains unclear whether shared genetic contributors exist for T1D and T2D associated nephropathy. In addition, as suggested by variable population prevalence rates, the impact of genetic heterogeneity between populations needs to be addressed, i.e. replication across well-powered, ethnically diverse samples to assess impact. It is noteworthy that contemporary studies of DN and other common complex diseases have focused on common genetic variants captured by array technology that inefficiently tag variation in minority populations. It is enlightening that common variants identified to date only explain a modest proportion of disease risk in many common diseases [76]. Geneticists are now speculating about the contribution of alternative modes of genetic transmission, e.g. rare variants, copy number variation, gene-gene or gene-environment interactions and epigenetic mechanisms.

With current progress, researchers are gaining appreciation of the immense genetic heterogeneity underlying common disease [77] and while whole genome sequencing remains costly and a bioinformatics challenge, exome sequencing offers a glimpse of the coding genome that is overlooked by GWAS targeting common variation in the population. Exome sequencing is a complementary approach to GWAS with the possibility of identifying rare coding variants with large effect. This technology has facilitated the creation of large databases of low frequency variants allowing researchers to evaluate these variants as genetic determinants of DN.

Moving beyond sequence variation, evaluation of epigenetic mechanisms of disease is being performed. Epigenetics broadly refers to heritable changes in gene expression without change to the primary nucleotide sequence [78]. These changes result from molecular modification of DNA or chromatin through phenomenon such as histone modifications or DNA methylation, the latter being extensively investigated in common disease. This represents an attractive mechanism by which the hyperglycemic environment could mediate renal disease [79]. Epigenetic studies of DN are now emerging [80] although additional studies are necessary to evaluate the utility of traditional DNA sources, i.e. peripheral blood, as a viable surrogate for sources more proximal to renal phenotypes.

Conclusions

There is an inherent genetic risk for DN, evidenced through familial aggregation, ethnic-specific prevalence rates and emerging results from genetic studies. Candidate gene studies have previously focused on existing biology, targeting candidates involved in known pathways. Until recently, these studies have evaluated small numbers of patients and typically lacked ethnic diversity. This approach has been strengthened by technological advances allowing for cost effective, unbiased surveys of the entire genome, i.e. GWAS, with the hope of identifying novel genes which further elucidate the underlying biology of disease. Moving forward, researchers are beginning to assess the role of uncommon variation and the impact of epigenetic mechanisms of disease risk. There remains a critical need to accelerate research, with extension of the findings to multiple ethnic groups and diverse disease types with the development of novel screening tests and treatments for DN. The true benefit of these findings would be translational application with a reciprocal relationship for basic scientists to provide new tools for patient assessment while clinical researchers make novel observations about the nature and progression of disease to clarify phenotypic heterogeneity. Overall, these efforts could elucidate the molecular basis of DN and lead to early detection and development novel therapeutic strategies to decrease the incidence of DN and attenuate the associated morbidity and mortality.