Obesity is one of the most common causes of morbidity and mortality among adults worldwide, with more than 13% of the world’s adult population being obese by 2014. In Mexico, its frequency among adults was estimated to be 38.8% [1]. Obesity has been associated with the development of severe chronic diseases, such as cardiovascular heart disease [2] and Type 2 Diabetes (T2D) [3]. One supported proposed mechanism has been posited to occur through a low-grade sub-clinical inflammation state [4]. T2D represents an important cause of death among adults throughout the world. Its global prevalence is estimated to be 9% among subjects aged over 18 years old, whereas in Mexico it was 9.2% in 2012 [1]. Diabetes is considered to be a polygenic disease. Many single nucleotide polymorphisms (SNPs), including those in cytokine genes such as TNF-α [5], IL-6 [6], IL-10 [7], adiponectin [7], and resistin [8,9] have been reported to increase the risk of developing T2D and obesity.

RETN encodes for resistin, a cysteine-rich C-terminal domain polypeptide secreted in humans by monocytes, macrophages, and adipocytes [10-12]. This cytokine has been proposed as a link between obesity and T2D, and has been involved in inflammation through the increase of TNF-α, IL-6 and IL-12 [13,14]. Furthermore, it has been demonstrated that adipocytes and macrophages treated with resistin enhance lipid accumulation [15-17], whereas skeletal muscle cells show a decreased fatty acid uptake under the same condition, which may further lead to lipotoxicity and insulin resistance [18]. Two polymorphisms of RETN, c.+62G>A and g.-420C>G, have been shown to impair resistin expression [19-22].

Many groups have examined the association between the c.+62G>A and g.-420C>G polymorphisms and the risk of developing either obesity or T2D. Solaleh and Motawi found in Asians and Africans [9,8], an augmented risk of developing T2D was associated with the g.-420C>G polymorphism, whereas Chi and Chen found in Asia and Northern America no risk [23,24]. In the same manner, in Africans there was a positive association between obesity and the g.-420C>G polymorphism [25,26]; whereas, Apasalamy failed to observe any change in risk [19]. Therefore, the exact effect the g.-420C>G polymorphism has toward T2D and obesity development remains inconsistent, especially in regard to ethnicity. Furthermore, in Latinos, only one study has been conducted which explored the association between the c.+62G>A and adiposity [27], suggesting the presence of this mutation was associated with overweight. The aim of the present study was to analyze whether there is an increased risk of developing obesity or T2D among Mexican adults carrying the c.+62G>A and g.-420C>G polymorphisms. Moreover, we performed a meta-analysis to analyzed the overall effect of these polymorphisms have obesity and T2D development across different populations.

Subjects

A cross-sectional study was designed, which included 164 unrelated Mexican subjects from the City of Puebla, Mexico, whose ages ranged between 18 and 70 years. The subjects were outpatients from IMSS Clinic 2 and were recruited between January 2010 and December 2013. Healthy subjects, as well as obese and diabetic subjects, were asked to participate. Subjects were excluded from the study if it was suspected that they had any acute or chronic illness that would interfere with the analysis, or if they met the criteria to be classified as pre-diabetic or overweight (BMI: 25-30 kg/m²). All the healthy and obese subjects were already used in a previous report [28]. The protocol was approved by the Scientific Research Committee of the Mexican Social Security Institute and was conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from each subject after explanation of the nature of the study.

Clinical characterization

Subjects were clinically evaluated according to a standardized questionnaire including their personal medical history. With the subjects in fasting conditions, wearing light clothing and without shoes, their height (m) and weight (Kg) were measured using a body composition analyzer (Tanita, Itabashi-Ku, Tokyo). BMI was calculated as weight/height² (kg/m²). Obesity was defined according with the International Diabetes Federation (BMI ≥30 kg/m²). Diabetes according to the America Diabetes Association criteria (Fasting Plasma Glucose: ≥126 mg/dL and/or 2-hour post-prandial glucose: ≥200 mg/dL).

Biochemical assays

Blood samples were drawn from the antecubital vein following a 10- 12 h overnight fast and collected into a Vacutainer^® Plus plastic sterile tube with BD Hemogard^™ closure containing sodium fluoride and potassium oxalate (Catalog #367925, BD Mexico City, Mexico). Subjects also underwent a second venipuncture 2 hours after drinking a glucosecontaining solution (75 g) to assess post-prandial glycemic control. The samples were kept in serum clot activator tubes at room temperature for 20 minutes. The serum fraction was then recovered and frozen at -20°C until use. Samples were used for the following endpoints: fasting plasma glucose and post-prandial glucose test. Glucose was determined using the enzymatic method/spectrophotometricglucose oxidation.

Genotyping

Blood samples for DNA extraction were stored in EDTA-containing tubes. DNA was isolated from peripheral blood leucocytes using the salting out method [29] and the DTAB-CTAB method [30]. Resistin genotypes were determined using PCR-RFLP (GeneAmp PCR System 9700, Applied Biosystems, Foster City, CA). The sequence of the RETN promoter and 3’UTR were obtained from the Eukaryotic Promoter Database (Swiss Institute of Bioinformatics, Lausanne, Switzerland), and primers were designed using the FastPCR software (Helsinki, Finland).

For the g.-420C>G polymorphism, forward and reverse primers were 5’- TGTCATTCTCACCCAGAGACA-3’ and 3 ́-GATTTGGTTAGCTGAGCCCA-5’, respectively, which produced a 533 bp PCR product. The PCR products were further digested overnight using the BbSI restriction enzyme (Catalog #R0539S, New England BioLabs Inc.). The digested products were then separated in 6% polyacrylamide gel and stained with silver nitrate. Digestion of the C allele yielded two fragments (327 and 206 bp), whereas the G allele produced a single fragment (533 bp). Thus, wild type homozygous produced two bands, those carrying the polymorphism produced a single band, and heterozygous produced three bands.

The forward and reverse primers for the c.+62G>A polymorphism were 5’-AGAGTCCACGCTCCTGTGTT-3’ and 3’-CGACCTTTATTTGGACCTCTAC-5’, respectively, which produced a 249 bp PCR product. The PCR products were digested overnight using BseRI restriction enzyme (Catalog #R0581S, New England BioLabs Inc.). The A allele, after digestion, would produce two fragments (238 and 11 bp), while the G allele produce a single fragment (249 bp). Electrophoresis patterns were as follows: wild type homozygous produced a single band, mutant homozygous produced two bands, and heterozygous produced three bands.

Publication Search

Literature was searched using PubMed, Scopus, EBSCO, Ovid and Wiley databases, until November 2015, for case-control studies conducted on humans that have determined RETN polymorphisms in diabetic or obese subjects. The keywords and related terms used were: Type 2 Diabetes or obesity, RETN, and single nucleotide polymorphism. Manual search of the references lists of the retrieved articles was performed to identify more studies.

Inclusion and exclusion criteria

Articles were eligible if they met all of the following criteria: i) Explored the association between either T2D or Obesity (BMI ≥30 Kg/ m²) and RETN polymorphisms, ii) were case-control studies, and iii) reported genotype frequencies, as well as odds ratios estimates with 95% confidence intervals or provided sufficient data to calculate them. Studies were excluded on the basis of the following criteria: i) presented data as an editorial, a review or a meeting abstract, ii) did not provide sufficient data, iii) data were presented in another more representative study from the same research group, or iv) was not written in English or Spanish. Two authors determined if each study was to be included. Afterwards, only studies focusing on RETN polymorphisms g.-420C>G and c.+62A>G were considered.

Data extraction

Two authors extracted the data independently. If there was disagreement, a third author assessed the study in question. If a single sample was believed to be used in multiple reports, these were assessed to determine which one was the most representative, or the corresponding author was contacted to resolve the issue. The data collected were: first author’s name, publication year, geographical region, T2D/obesity, T2D definition, genotyping method, source of controls, distribution of genotypes among the cases and controls, as well as the total number of subjects.

Statistical analysis

The normality of the data was assessed by the Kolmogorov-Smirnov test. Continuous variables presented normal distribution and variance homogeneity, therefore they were analyzed using the one-way ANOVA test; the Bonferroni post hoc test was performed to identify betweengroup differences. Allele frequencies were computed from genotype frequencies. Fisher’s exact test was used to search for differences in allele and genotype frequencies among groups. Deviation from Hardy- Weinberg Equilibrium (HWE) was assessed using the ψ²-test for the controls.

For the meta-analysis, the HWE for each study was determined by the ψ²-test for the controls. The crude odds ratio (OR) and the 95% confidence interval (95% CI) were calculated for each study and used to assess the level of association between the mutation and T2D or obesity. The pooled ORs were assessed for the following genetic models: homozygous (22 versus 11), heterozygous (12 versus 11), dominant (12 + 22 versus 11), recessive (22 versus 12 + 11), and allelic (2 versus 1); where 1 corresponds to C and A, and 2 corresponds to G and G for the g.-420C>G and c.+62A>G mutations, respectively. Heterogeneity was determined using the Cochran ψ²-based Q-test, and its degree was assessed by the I² value (inconsistency index). If the Q-test p-value was <0.10 or the I² value was >50%, the pooled ORs were determined using the random-effects model (DerSimmonian-Laird method) [31], otherwise the fixed-effects model (Mantel-Haenszel method) [32] was selected. The stability and sensitivity of the results were assessed by removing one study at a time and re-calculating the pooled ORs. Publication bias was evaluated by the Begg-Mazumdar’s adjusted rank correlation asymmetry test (Kendall’s tau) [33] and the Egger regression asymmetry test [34].

Statistical analyses were performed using either Statistical Package for the Social Sciences software for Windows, version 22 (SPSS, Chicago, IL), Review Manager version 5.3 (Copenhagen, DK) or StatsDirect version 3.0.147 (Cheshire, UK). Results were expressed as mean ± standard deviation, unless noted otherwise. P-values <0.05 (two-tailed) were considered statistically significant.

Subjects’ characterization

Our cohort consisted of 164 unrelated Mexican adults (45 males and 119 females). Subjects were divided into three groups according to their anthropometric and glycemic status: Control (n=49), Obese (n=70) and T2D (n=45). The anthropometric and biochemical characteristics of the three groups are presented in Table 1. Fifteen subjects had a previous diagnosis of T2D, whereas thirty of them were newly diagnosed during the protocol. The Obese group showed higher anthropometric measures, as well as fasting plasma glucose, compared with the control group (p<0.05). T2D subjects were older than those in the other two groups, and presented higher anthropometric measures than controls, but lower than the obese subjects (p<0.05). Both fasting plasma glucose and 2-hour prandial glucose levels were statistically higher in the T2D group than in the others (p<0.05).

Table 1: Characterization of the population by groups.

Lack of association of g.-420C>G or c.+62A>G polymorphisms with susceptibility to T2D or Obesity

Genotype distribution, allele frequencies, and ORs for the g.- 420C>G and c.+62A>G mutations are presented in Table 2. Genotype distribution for the controls was in accordance with the HWE. None of the genetic models showed any significant increase in the risk of developing T2D or Obesity in the presence of the either polymorphism.

Table 2: Genotype distribution and allele frequencies by group.

Characteristics of the studies from literature review

Three hundred forty-seven studies were retrieved from searching the databases and from reviewing the studies’ references lists. Two hundred ninety-five studies were excluded because they did not focus on diabetes or Obesity and RETN, focused on animals or cell lines, or were not original research articles. The remaining fifty-two studies were evaluated extensively. Five studies were not written in English or Spanish, six lacked sufficient information, seven were not case-control studies, two were duplicate studies, three considered different BMI cutoffs for obesity, and one focused on children and were therefore excluded (Figure 1). From the remaining twenty-eight studies, eighteen focused on the g.-420C>G polymorphism [8,9,19,21,23-26,35-44] and four on the c.+62A>G polymorphism [22,25,45,46]. Our current study was also considered for the meta-analysis. For the g.-420C>G polymorphism, the sample for the T2D analysis of the consisted of 8986 cases and 13288 controls, while there were 1240 cases and 2387 controls for the obesity analysis. For the c.+62A>G polymorphism, the sample for the T2D analysis consisted of 1883 cases and 1384 controls and for Obesity there were 230 cases and 215 controls. Deviations from HWE were found in five studies for the g.-420C>G polymorphism [8,25,35,38,39]. However, analyses removing these populations did not show a significant variation; therefore they were included in the metaanalysis. Detailed characteristics of the studies are listed in Table 3.

Figure 1: Flow chart for literature review of studies to be included in the meta-analysis.

Table 3: Characteristics of the studies included in the meta-analysis for g.-420C>G and c.+62G>A polymorphisms of the RETN promoter.

Meta-analysis of the association of g.-420C>G or c.+62A>G with Obesity and T2D

Overall, this meta-analysis showed an increased risk of developing Obesity associated with the g.-420C>G polymorphism for the heterozygous and dominant genetic models (OR=1.33, 95%CI: 1.01-1.75, p_{heterogeneity}=0.01, I²=64%, and OR=1.30, 95%CI: 1.00-1.68, p_{heterogeneity}=0.01, I²=64%, respectively; Figure 2 and Table 4). However, when stratifying by geographic region, the association was true for the African (OR=1.54-2.13, p<0.05) and North American regions (OR=1.42-1.49, p<0.05, Table 5). The African region was also associated with an elevated risk of developing T2D (OR=2.35-7.17, p<0.05, Table 5). No other associations between the g.-420C>G polymorphism and the risk of developing T2D was demonstrated for other regions (Table 4). None of the genetic models showed an association between the c.+62A>G mutation and the risk of developing Obesity or T2D for the overall (Figure 2 and Table 4) and regional analysis (data not shown). See supplementary information for all Forest plots.

Figure 2: The risk associated developing obesity or Type 2 Diabetes (T2D) with the RETN g.-420C>G or c.+62G>A polymorphisms. Forest Plots were constructed for the five genetic models (data shown is for the dominant model) to examined the association between a) the RETN g.-420C>G polymorphism and T2D, b) the RETN g.-420C>G polymorphism and obesity, c) the RETN c.+62G>A polymorphism and T2D, and d) the RETN c.+62G>A polymorphism and obesity.

Table 4: RETN polymorphism association with the development of Obesity and Type 2 Diabetes for the overall population.

Table 5: RETN c.-420C>G polymorphism association with the development of Obesity and Type 2 Diabetes by geographical region.

Test for sensitivity and publication bias

Publication bias was assessed by examining Begg’s funnel plot, calculating the Begg-Mazumdar´s test and Egger’s asymmetry test. The funnel plot did not indicate any major asymmetry (Figure 3). Furthermore, neither Begg-Mazumdar´s nor Egger’s test did indicate any significant bias for any genetic model (for dominant model: Begg- Mazumdar’s Kendall’s tau= 0.017, p=0.96; Egger: bias=-0.01, p=0.98). Heterozygous and dominant genetic models for the analysis of the association between the g.-420C>G mutation and Obesity were sensitive to the removal of either Engert (Northern American population), Boumaiza or El-Shal studies, resulting in the loss of significance (CG vs CC: OR=1.30, 95%CI: 0.94-1.80; OR=1.18, 95%CI: 0.96-1.46 and OR=1.28, 95%CI: 0.95-1.74, respectively). For the analysis of the association between the c.+62G>A mutation and T2D, heterozygous, dominant and allelic models showed sensitivity to the removal of Jiang’s study, resulting in a decreased risk of developing the condition (GA vs GG: OR=0.63, 95%CI: 0.51-0.78; GA+AA vs GG: OR=0.63, 95%CI: 0.51-0.78; A vs G: OR=0.66, 95%CI: 0.55-0.80). None of the other studies had a significant effect on the ORs for either polymorphisms and condition.

Figure 3: Begg’s funnel plot for publication bias test. For the RETN g.-420C>G polymorphism, no detrimental publication bias was observed for the Type 2 Diabetes analysis (a) and the obesity analysis (b). As well for the RETN c.+62G>A polymorphism, no detrimental publication bias was observed for the the Type 2 Diabetes analysis (c). Due to the sample size, no publication bias can be determined for the c.+62G>A polymorphism and obesity analysis. Each point represents a separate study for the indicated association. Funnel plots are for the dominant model.

T2D is a leading cause of morbidity and mortality among adults worldwide, and obesity has been demonstrated to be a major risk factor for its development [47]. Both conditions can feature a low-grade inflammation state. Since many cytokines have proved to play a role in this phenomenon [48-51], the RETN gene was examined. Two mutations of RETN, g.-420C>G and c.+62G>A, altered resistin expression in previous reports [19-21], however, their effect on augmenting the risk of developing obesity or T2D remains inconclusive. Here, we demonstrate that only the g.-420C>G polymorphisms augments the risk of developing obesity.

Only one study has been conducted on Mexican subjects that indirectly explored the association between the c.+62G>A mutation and overweight [27]. They determined that the A allele was associated with a low BMI, as well as with low resistin levels. A study in a Chinese population found a similar association between the A allele and a decreased risk of developing T2D [22]. In our population, however, there were no differences in either genotype distribution or allelic frequencies for both polymorphisms examined between the Control group and the obesity or the T2D groups; therefore, no associations were found in any of the genetic models. These results were in accordance with previous reports for Asians, Northern Americans, and Europe [21,23,24,36,37,39-43,45,46]. However, two studies, conducted in Asians [8] and Africans [9] population have shown an increased risk of developing T2D related to the g.-420C>G mutation. Furthermore, two other African studies detected a 2.22 and 3.06-fold increased risks of developing obesity among carriers of the g.-420G allele [25,26].

Due to the inconsistency of the data between ethnicities and because it has been demonstrated that some mutations have an effect only under certain ethnic genetic influences, this meta-analysis was performed. To our knowledge, this is the first meta-analysis which examined the effect of the c.+62G>A polymorphism on obesity or T2D. Here, we reported that the c.+62G>A polymorphism showed no association in any of the genetic models for obesity or T2D. Nor was there any association between the c.-420C>G mutation and the risk of developing T2D. The meta-analysis demonstrated, however, an increased risk of developing Obesity among carriers of the c.-420G allele. Moreover, when stratifying by geographic region, the association was true for the African and North American regions. It is possible, therefore, that subjects with an African ancestry component might also show an elevated risk of developing obesity when they carry the mutated allele. Further investigations are required to assess this association among Mexicans from regions where there is a more prominent African ancestry component, like in Guerrero state [52].

Two previous meta-analyses examined the overall effect of the g.-420C>G mutation. Wen et al. demonstrated that there was no association between the polymorphism and the risk of T2D; however, it only considered European, Asian and Northern American populations, and not African or Latin Americans [53]. Yu et al also demonstrated no association between the polymorphism and obesity. It included two populations from Europe, one from Northern America and one from Brazil (European descendants), but none from Latin America or Africa [54]. Discrepancies between these results and ours might be due to fact that they did not considered African populations, which here we demonstrate are highly susceptible.

Our study has at least three limitations. First, our subjects were not separated by gender. Some authors have found differences in anthropometric measures within male and female carriers of the g.- 420G allele [35,38,55,56]. Second, our population was formed mainly of subjects from Central Mexico, which have a predominant European ancestry component. As mentioned above, Mexico is highly diverse country with many different composition of European, African, and Native American. Third, for the meta-analysis, we used crude ORs and non-adjusted ORs. Adjusting the ORs can affect the ORs by a few tenths.

In summary, neither g.-420C>G nor c.+62G>A are associated with T2D with the overall population. Only the g.-420C>G polymorphism was associated with an increased risk of developing obesity. This effect was associated with African and North Americans. Whether subjects with an African ancestry component show the same susceptibility pattern has yet to be investigated.

We would like to express our gratitude to the participants of this study, to Maria del Carmen Sanchez-Guillen, who contributed to the development of this project, and to Maestro Ricardo Villegas-Tovar from Benemérita Universidad Autónoma de Puebla Libraries Department. This study was supported by grants from the Programa de Mejoramiento del Profesorado of Secretaría de Educación Pública (PROMEP-SEP) and the Vicerrectoria de Investigacion of Benemérita Universidad Autónoma de Puebla (VIEP-BUAP).

The authors report that there are no competing financial interests.