Association between Genetic Variants and Diabetes Mellitus in Iranian Populations: A Systematic Review of Observational Studies

Introduction. Diabetes mellitus as the most prevalent metabolic disease is a multifactorial disease which is influenced by environmental and genetic factors. In this systematic review, we assessed the association between genetic variants and diabetes/its complications in studies with Iranian populations. Methods. Google Scholar, PubMed, Scopus, and Persian web databases were systematically searched up to January 2014. The search terms were “gene,” “polymorphism,” “diabetes,” and “diabetic complications”; nephropathy, retinopathy, neuropathy, foot ulcer, and CAD (coronary artery diseases); and Persian equivalents. Animal studies, letters to editor, and in vitro studies were excluded. Results. Out of overall 3029 eligible articles, 88 articles were included. We found significant association between CTLA-4, IL-18, VDR, TAP2, IL-12, and CD4 genes and T1DM, HNFα and MODY, haptoglobin, paraoxonase, leptin, TCF7L2, calreticulin, ERα, PPAR-γ2, CXCL5, calpain-10, IRS-1 and 2, GSTM1, KCNJ11, eNOS, VDR, INSR, ACE, apoA-I, apo E, adiponectin, PTPN1, CETP, AT1R, resistin, MMP-3, BChE K, AT2R, SUMO4, IL-10, VEGF, MTHFR, and GSTM1 with T2DM or its complications. Discussion. We found some controversial results due to heterogeneity in ethnicity and genetic background. We thought genome wide association studies on large number of samples will be helpful in identifying diabetes susceptible genes as an alternative to studying individual candidate genes in Iranian populations.


Introduction
Diabetes mellitus, as the most prevalent metabolic disorder, is characterized by chronic hyperglycemia due to defect in insulin secretion by beta cells of Langerhans islets or resistance against insulin action [1][2][3]. More than 300 million people are suffering from diabetes mellitus all over the world and studies show that population aging, changes in lifestyle and improvement in detection techniques are most important factors in increasing the numbers of cases [4]. The prevalence of type 2 diabetes mellitus (T2DM) varies in different populations from less than 6% in most populations to more than 50% in Pima Indians [5]. In 2013 it was reported that in Middle East region about 35 million people suffered from diabetes. The prevalence of diabetes has been estimated as 382 million people throughout the world while nearly 176 million of them seem to be still undiagnosed. It is predicted that this prevalence reaches to 592 million by 2035. Diabetes mellitus can also cause complications in most of organs: heart, eye, kidney, and nervous system which has resulted in high economic cost and burden [6]. Therefore, diagnosis of disease in early stages is very important.
A systematic review showed that between years 1996 and 2004 the prevalence of type 2 diabetes in Iran was 24% and the risk was 1.7% greater in women. According to this report the prevalence of T2DM in Iran seems to be 2 Journal of Diabetes Research highest amongst developing countries. Previous reports on total urban population of Middle Eastern countries show the prevalence of T2DM as 3.4% in Sudan, 20% in United Arab Emirates, 8.5% in Bahrain, and 12.1% in India [7].
Diabetes Mellitus is categorized into the following groups. Type 1 diabetes mellitus (T1DM) includes 5-10% of diabetic patients. Cellular-mediated autoimmune destruction of the beta-cells of the pancreas results in T1DM. It classically occurs in juveniles and affected patients are dependent on insulin injection in their lifetime and are very prone to ketosis [1,8,9].
T2DM includes 90-95% of patients with diabetes. Patients with type 2 diabetes may be asymptomatic for long period of time. Vascular complications such as nephropathy, neuropathy, retinopathy, and cardiovascular disease may develop in these patients. The impact of genetic component appears to be stronger in T2DM compared to T1DM [1,8,9].
Gestational diabetes mellitus (GDM) is another type which is observed during pregnancy and the prevalence may range from 1 to 14% in all pregnancies [1,8,9].
MODY (maturity onset diabetes of young) is monogenic form of diabetes comprised of several types with various features which is consisting of 1-5% of patients diagnosed as T2DM. The onset of this type of diabetes is normally before 25 years and its treatment is independent of insulin. MODY in its different forms is inherited in autosomal dominant pattern and presents as a result of mutation in transcription factors genes including HNF4 (hepatocyte nuclear factor), HNF1 , IPF1 (insulin promoter factor), and neuro-D1 [1,8,9].
There are also other types of diabetes which are considered as secondary to other conditions, for example, any damage to pancreas such as removal of pancreatic tissue, trauma, pancreatic carcinoma, and infection, or underlying diseases including endocrine diseases that alter different hormones secretion which are antagonist to insulin resulting in various clinical manifestations such as acromegaly, Cushing's syndrome, pheochromocytoma, glucagonoma, somatostatinoma, and diabetes. Diabetes (or carbohydrate intolerance) is also found in increased frequency with a large number of genetic syndromes such as Wolfram syndrome, which causes diabetes mellitus, diabetes insipidus, and other neurodegenerative disorders, MELAS (mitochondrial myopathy, encephalopathy, lactic acidosis, and stroke) which is presented with myopathy and encephalopathy caused by mitochondrial mutation, MIDD (maternally inherited diabetes and deafness) which causes diabetes, and IPEX (immune dysregulation, polyendocrinopathy, enteropathy, and X-linked syndrome) which is X-linked and alters immune system and causes multiple endocrine problems [1,8,9].
Diabetes mellitus is a multifactorial disease with both environmental and genetic causes affecting its presence and incidence. Genome wide association studies revealed the genetic heterogeneity of diabetes and the fact that difference in ethnicity can result in different susceptible genes associated with diabetes [10,11]. Studies on candidate genes related to diabetes revealed that several genes including PPRAG (peroxisome proliferator-activates receptor gamma), IRS1 and 2 (insulin receptor substrate), KCNJ11 (potassium inwardly rectifying channel), and HNFA are associated with T2DM.
Genome wide association studies (GWAS) showed that many genes including Calpain 10 and TCF7L2 (transcription factor 7-like 2) are associated with T2DM. HHEX (hematopoietically expressed homeobox), SLC30A8 (solute carrier family 30 (zinc transporter), member 8), CDK2A/B (cyclindependent kinase inhibitor 2A/B), and IG2BP2 (insulin-like growth factor 2) are other genes which have been shown to be associated with T2DM based on GWAS. Some of these genes are expressed in beta cells or involved in insulin secretion pathways [12,13]. Other candidate genes in association with T2DM include PPAR , ACE (angiotensin converting enzyme), MTHFR (methylene tetrahydrofolate reductase), FABP2 (fatty acid binding protein-2), and FTO (fat mass and obesity associated gene) [14]. GWAS was carried out in many populations including Finnish, French, and American Caucasians and showed different loci on most of the chromosomes associated with T2DM [15]. Systematic review and meta-analysis studies also confirmed the association of some genes such as PGC-1 (peroxisome proliferator-activated receptor gamma coactivator-1 ) and adiponectin with T2DM [16,17]. The genetic loci associated with T1DM have also been examined. For example, insulin (INS) gene VNTRs (variable number of tandem repeats) with a protective effect is a variants related to T1DM. In addition PTPN22 (protein tyrosin phosphatase), CTLA4, and IL2RA genes have also been shown as candidates for T1DM due to their role in Tcell signaling. Studies have also confirmed the association between IFIH1, CYP27B1, and CLEC16, which are important in immune system and T1DM [18].
There is no inclusive information for genetic association studies of diabetes in Middle Eastern population including Iranian population. In order to make a comprehensive approach, we aimed to collectively investigate and gather data for association between genetic variants and type 2 diabetes in Iranian population in a systematic review study.

Research Design and Methods
This study is reported according to PRISMA (preferred reporting items for systematic reviews and meta-analyses) guideline [19].

Data Sources and Searches.
We systematically searched international web databases: Google Scholar, PubMed, Scopus, and Persian web databases; IranMedex; and Magiran to investigate the association of genetic variants with diabetes and its complications in Iranian population up to January 2014. The search terms were "gene, " "polymorphism, " "diabetes, " "diabetes' complications, " nephropathy, retinopathy, neuropathy, foot ulcer, and CAD (coronary artery disease) and their MeSH terms and Persian equivalents with diabetes. At least three emails were sent to the corresponding author of articles which were not accessible as full text or had insufficient data. Duplicate articles and multiple publications from the same population were excluded and the most relevant data were used for the investigation. The references of all selected articles were investigated.

Study Selection.
We included all observational population-based studies ≥100 sample size, which were conducted as case-control, cohort, or cross sectional. According to WHO criteria, diabetes mellitus was defined as FBS ≥ 7 mmol/L (126 mg/dL) or 2 h plasma glucose ≥11.1 mmol/L (200 mg/dL) [20]. In addition, we excluded animal studies, clinical trials, short communications, letters to editor, dissertations, in vitro studies, review articles, and population-based studies conducted in pregnant women. Two researchers, SE and MKH, independently reviewed title, abstract, and full text of each article to assess the eligible articles according to inclusion and exclusion criteria. Inclusion and exclusion of articles were supervised by a third reviewer (OTM) in case of conflicts. There was no limitation for language.

Data Extraction and Quality
Assessment. The following data were extracted and presented in an excel sheet: author(s), year, genes and SNPs, patients' characteristics (sample size, age, and sex), city, study design, genotyping method, and significant association. Six selected items from the STROBE (strengthening the reporting of observational studies in epidemiology) checklist [21] was used for quality assessment of the included studies and assessed separately for each included article. The below items were considered: (a) clearly define the outcome and association between gene variants and diabetes; (b) give the eligibility criteria; (c) present key elements of study design; (d) report numbers or significance/nonsignificance statistically of outcome events; (e) give characteristics of study participants; and (f) describe the locations and relevant dates. All studies with quality score ≥3 were considered as high quality study and included in our systematic review.

Data Synthesis and
Analysis. Due to heterogeneity in genotyping techniques and also differences in genetic variants studied in assessed included articles performing a metaanalysis was impossible.

Search Results.
A summary of the literature review process performed in this study is presented in a flow chart in Figure 1. In final step 88 eligible studies were included in this study and assessed due to the inclusion/exclusion criteria . Table 1 shows the summary of each included investigation.
Details of the included studies are described as below.
In 2 studies the association of +49A/G variant of CTLA-4 (cytotoxic T lymphocyte associated antigen 4) was investigated and it was revealed that there was a positive significant association between AG genotype and T1DM ( = 0.01, < 0.001) [22,32]. One of these studies also revealed that there was a negative relation between AA genotype and T1DM [32].
Vitamin D receptor (VDR) was investigated in 2 studies [25,26]. Both of them assessed the association of FokI, BsmI, ApaI, and TaqI variants of VDR with T1DM. In one study which was performed on 114 subjects, Aa ( = 0.003), FF ( = 0.008) and Bb ( = 0.014) genotypes were shown to be positively associated with T1DM but no significant association was observed between TaqI polymorphism and T1DM [25]. Another study which was performed on 187 subjects showed no significant association between T1DM and FokI, BsmI, and ApaI. However TT genotype was observed to be more frequent in controls ( = 0.007) and negatively associated with T1DM [26].
Different polymorphisms of TAP2 and their association with T1DM were investigated in 191 subjects. It was shown that Ile379Val and Stop687Gln had significant association with T1DM ( = 0.001 and = 0.013, resp.). Other variants of TAP2 (transporter 2 ATP-binding cassette) including Ala565Thr, Arg651Cys, and Ala665Thr were not associated with T1DM significantly [27].

MODY.
In this category only one study was found [34] and the association of Val255Met polymorphism of HNF and diabetes was investigated in 101 subjects including MODY subjects, their relatives, and healthy controls [34]. The study was designed as a case control association study and PCR-RFLP and direct sequencing was used for genotyping of variants. Val255Met mutation in this investigation was observed in MODY subjects and their relatives but not in control group.
Association between two polymorphisms of TCF7L2 including rs7903146 (C/T) and rs12255372 (G/T) was assessed in three studies [38,48,53]. In one study investigating rs7903146 (C/T) in 466 subjects, significant difference in the frequency of CC and CT genotypes was observed between T2DM and control group ( = 0.045) [38]. However in another study this polymorphism showed a positive significant association with TT genotype in T2DM ( = 0.008) [53]. Investigation of rs12255372 (G/T) polymorphism of TCF7L2 in another study showed that among 491 subjects there was a positive significant association between TT carriers and T2DM ( = 0.014) [48].
Investigating the association of calreticulin with T2DM in 650 subjects showed two mutations, only in case group, 9 bp deletion of 397-399 codons and G>T mutation at IVSII-142 [39].
Association of estrogen receptor and its PvuII(PP, Pp,pp) and XbaI(XX,Xx,xx) polymorphism with T2DM in two studies was assessed [40,49]. In one study, the significant association of PvuII and XbaI polymorphisms with T2DM was observed ( = 0.014 and = 0.002, resp.) [40]. In another study it was revealed that pooled Pp+pp and XX+xx male carriers have positive significant association with T2DM ( = 0.001 and = 0.026, resp.) [49].
Association of Pro12Ala polymorphism of PPAR with T2DM was investigated in three studies [41,69,75]. Negative significant association of Ala/Ala genotype with T2DM was observed in two studies ( = 0.003) [41,75]. Ala/Pro carriers were shown to be positively associated in one study [41] and negatively associated in another study [75] with T2DM ( < 0.001 and = 0.003, resp.). However in the third study no significant association was observed between Pro12Ala polymorphism of PPAR and T2DM [69].
Association of different polymorphisms of ApoAI with T2DM was investigated in two studies [43,72]. In one study investigation of MSPI polymorphism showed no significant association with T2DM [43]. In another study it was shown that G-75A polymorphism of ApoAI was not associated with T2DM while a significant association was found between C+83T polymorphism and T2DM ( = 0.028) [72].
Two studies assessed the association of −156G>C polymorphism of CXCL5 with T2DM [44,65]. In one study positive significant association of GC carriers ( = 0.006) and negative association of GG carriers ( = 0.01) was found between diabetic patients and healthy controls [44] and in another study positive significant association was found between GC or CC carriers and T2DM ( = 0.009) [64].
Adiponectin and its association with T2DM were investigated in two studies [56,73]. In one study, two variants including +45T/G and −11391G/A were investigated but no significant relation was found between these two polymorphisms and T2DM [56]. However in another study investigating +45T/G polymorphism of adiponectin, it was revealed that nonobese T2DM subjects had significantly higher TT genotype ( = 0.04) [73].

Genes Related to Insulin
Resistance. In this subgroup, four studies were assessed and all were designed as casecontrol association studies. Three studies used PCR-RFL [76,78,79] and one study used real-time PCR [77] for genotyping.
Paraoxonase 1 (163T/A polymorphism) was the only variant in this subgroup which was not significantly associated with atherosclerosis risk in T2DM [80]. Investigation of another polymorphism of Paraoxonase 1 (Q192R) revealed that diabetics with CAD had significant increase in RR genotype ( < 0.05) [81].
Investigating B1 and B2 alleles of CETP TaqI B allele in 400 subjects showed that pooling of B1B1+B1B2 and B1B1 genotypes had significant association with T2DM and CAD ( < 0.01). On the other hand assessing the G894T polymorphism of eNOS showed positive significant association between TG and TT carriers and CAD ( < 0.05). It was also shown that concomitant presence of NOS3 T allele and CEPT B1 allele was significantly associated with T2DM ( = 0.004) and CAD ( = 0.002) [83].
Studying the association of 163A/G polymorphism of SUMO4 showed a positive significant association between AA carriers and diabetic nephropathy ( < 0.05) [93].
It has also been shown that interaction of eNOS (G894T polymorphisms) and MTHFR (C677T and A1298C polymorphisms), eNOS T/1298 C, and eNOS T/677 T carriers had positive significant association with macroalbuminuria. In addition significant association of eNOS GT+TT carriers with lipid profile ( < 0.05) was shown in this study [99].
ACE (I/D alleles) polymorphisms were investigated in three studies. In one study positive significant association ( < 0.01) between DD carriers and progression but not development of albuminuria was observed [104]. In another study in nephropathic subjects, association of DD carriers and T2DM ( = 0.006) but not nephropathy was revealed [105]. However another study did not show significant association of ACE with macroalbuminuria [98]. Interaction of ACE (I/D alleles) with Factor V Leiden (G1691A polymorphism) was not significant [100].
Association of interleukins including IL-4 and IL-10 with diabetic nephropathy was also investigated in two studies. Study of −592C/A polymorphism of IL-10 revealed positive significant association between CC carriers and nephropathy ( = 0.001) [94] and study of −590C/T polymorphism of IL-4 revealed positive significant association between CT carriers and nephropathy ( < 0.001) [101].

Genes Related to Diabetic
Retinopathy. All of the 3 studies in this subgroup were conducted as case-control association studies [106][107][108].
Investigating null/positive genotype of GSTM1 in a study, which used ARMS-PCR for genotyping, revealed that null genotype is significantly associated with diabetic retinopathy ( < 0.05) [106]. This result was confirmed in another study by using multiplex PCR as genotyping [107], although no significant association between GSTT1 and retinopathy was found in this study [107].

Genes Related to Diabetic Foot Ulcer.
In this subgroup only one study was found [109] which was case-control designed and used ARMS-PCR for genotyping. In this study association of −7C/T and −2578C/A polymorphisms of VEGF with diabetic foot ulcer (DFU) was investigated, and a positive significant association between AA carriers and DFU ( = 0.03) was observed. But no significant association was found between −7C/T of VEGF and DFU [109].

Discussion
Diabetes mellitus is the eighth most frequent disease leading cause of death throughout the world and now ranks the fifth, following communicable diseases, cardiovascular disease, cancer, and injuries [110]. Prevalence of diabetes mellitus is increasing worldwide [6].
Type 2 diabetes is the most frequent type of diabetes mellitus [110,111]. Multiple genes and environmental factors affect the prevalence of T2DM. Systematic review and metaanalysis studies have assessed the association between different genes and T2DM. Glutathione-S-transferase including GSTMI, GSTM1, and GSTP1 are important genes and their association with diabetes has been investigated in two metaanalysis studies [112,113]. It was revealed that null genotype of both GSTM1 and GSTT1 could be as a risk factor for diabetes while no significant association was found between GSTP1 and diabetes [112,113]. These results are in line with our finding and confirming the association of the null genotype of GSTT1 and GSTM1 with T2DM in Iranian population [50]. The relation between GSTT1 and GSTM1 and diabetic retinopathy has also been shown in Iranian population [106,107].
MTHFR is another important gene in association with T2DM and other related complications. No association of C677T polymorphism of MTHFR gene was found in Africans, Asians, and Caucasians [114]. MTHFR is a gene which is involved in DNA methylation and synthesis and regulated folate activity. It has been reported that mutant homozygote and heterozygote of C677T polymorphism of MTHFR increase plasma homocysteine which is an important factor leading to diabetic nephropathy (DN) and, as a result, C677T polymorphism of MTHFR can be associated with the development of DN. In Caucasians this association has also been observed [115]. In Iranian population, in line with other populations, no significant association was found between MTHFR polymorphisms and T2DM [63] although significant association of 677T and 1298C alleles with diabetic nephropathy was observed [87]. Interaction of two polymorphisms and also interaction of MTHFR polymorphisms with other genes such as ACE and eNOS showed an association with development of DN [96,99].
Interleukins especially IL-10 are shown to have significant association with T2DM in different ethnic groups. Significant association between −592C/A and −819C/T polymorphisms of IL-10 and T2DM risk in Africans was shown [116]. Also −1082A/G polymorphism of IL-10 seems to be a risk factor for T2DM in Asians but not in Europeans and Africans [116]. In another study it was reported that −1082GG genotype of IL-10 and −174CC genotype of IL-6 are as risk factors for T2DM in Egyptians [117]. In our study IL-4 has been studied but no significant association with T2DM was seen [66].
Investigation of different polymorphisms of adiponectin and their association with T2DM in a meta-analysis study revealed −11391G>A and −11426A>G polymorphisms of adiponectin as risk factor for T2DM in Europeans and −11377C>G variant of adiponectin as risk factor of T2DM in Europeans and Asians [110]. It was also shown that +45T>G and +276G>T variants are not associated with T2DM in Europeans, Africans, and Asians [110]. In contrast to previous studies, +45T>G and −11391G>A polymorphisms were not associated with T2DM in this study due to difference in genetic background [56].
Among candidate genes in association with T2DM, TCF7L2 is one of the strongest genes related to diabetes. In a meta-analysis study after pooling all data of European, African, and Asian populations, it has been revealed that rs12255372 polymorphism of TCF7L2 significantly increases the risk of T2DM. However no data from Iranian population as a type of Asian population was included in this metaanalysis [111]. In line with these data, in our study, we found a positive association between rs12255372 and rs7903146 variants of TCF7L2 and T2DM [38,48,53].
Pro12Ala polymorphism of PPAR is found to be as a protective variant especially in Asian population although the results were highly heterogeneous [118]. In our study the results were controversial. In one study no association was found between this polymorphism and T2DM in contrast with previous studies [69] while in other studies protective effect of this polymorphism with T2DM was confirmed in line with previous studies [41,75]. These controversial data may be due to different ethnicity in different populations being studied in Iran. Small sample size could be another reason for the controversial results.
Type 1 diabetes mellitus is another type of diabetes in which HLA (human lymphocyte antigen), IR1R1 (interleukin-1 receptor type 1), CTLA-4, and VDR are some important genes studied in association with T1DM [119]. VDR is one of the most studied genes in relation to T1DM due to its role in T-cell mediated autoimmune diseases [119,120]. Metaanalysis studies showed that BsmI and FokI polymorphisms of VDR increased the risk of T1DM in East and West Asians, respectively [119]. In our study, controversial results were found. In one study BsmI was shown to be associated with T1DM [25] while in another study no association was found [26]. FokI was not associated with T1DM in our study in contrast to previous studies which might be due to difference in ethnicity and genetic background or insufficient clinical data [25,26].
Among diabetic complications, CAD and DN are the most studied complications. Meta-analysis studies showed that rs2010963 and rs3025039 polymorphisms of VEGF, 4b/a, T-786C, and G894T polymorphisms of eNOS and ACE are associated with DN [121][122][123]. In our study, 4a/b and G894T polymorphisms of eNOS and ACE were shown to be associated with DN in line with previous studies [91,104,105]. In our study rs2010963 polymorphism of VEGF was also found to be associated with diabetic nephropathy which was in line with previous studies [95].
The most investigated genes in association with CAD include PPAR , TCF7L2, ACE, TNF-, adiponectin, and IRS1 [118]. In our study, +45T>G and +276G>T polymorphisms of adiponectin were investigated and the significant association of these variants with CAD was confirmed [82].
Overall it seems that more studies are needed to identify diabetes susceptibility genes in Iranian population. Investigation of candidate genes is one way to understand these genes, but the method with the least expenses and error is GWAS which scans a set of loci in association with a disease in many people. GWAS in different populations such as Americans, Caucasians, Australians, West Africans, and Europeans revealed multiple loci in association with T2DM [124][125][126][127][128]. In Asians especially in Middle Eastern countries GWAS is critically needed. In our investigations, we reviewed five studies which investigated the association between eNOS, ACE, MTHFR, and factor V Leiden and the interaction between them and diabetic nephropathy and reach a coherent result [91,97,99,100,104] while GWAS could be helpful and reduce time and other expenses [126].
However, there are some strength and some limitations in our study. Firstly, this systematic review, for the first time, assessed the association between genetic variants and diabetes in Iranian population. Moreover, the studies included in this systematic review were assessed for quality and selected as high quality (scored ≥ 3). For limitations it should be considered that all included studies were observational, which does not allow reliable inferences about causality. Moreover due to methodological heterogeneity, we were not able to pool the data and perform the meta-analysis.
Insufficient data might cause the conflicting results; therefore GWAS on defined population with large sample size is suggested as a more comprehensive approach answering many more questions.