Free
Genetics  |   February 2015
The −308G>A Polymorphism of the TNF Gene Is Associated With Proliferative Diabetic Retinopathy in Caucasian Brazilians With Type 2 Diabetes
Author Affiliations & Notes
  • Luís F. C. Sesti
    Laboratory of Human Molecular Genetics, Universidade Luterana do Brasil (ULBRA), Canoas, Brazil
  • Daisy Crispim
    Endocrine Division, Hospital de Clínicas de Porto Alegre (HCPA), Porto Alegre, Brazil
  • Luís H. Canani
    Endocrine Division, Hospital de Clínicas de Porto Alegre (HCPA), Porto Alegre, Brazil
  • Evelise R. Polina
    Laboratory of Human Molecular Genetics, Universidade Luterana do Brasil (ULBRA), Canoas, Brazil
  • Jakeline Rheinheimer
    Endocrine Division, Hospital de Clínicas de Porto Alegre (HCPA), Porto Alegre, Brazil
  • Patrícia S. Carvalho
    Laboratory of Human Molecular Genetics, Universidade Luterana do Brasil (ULBRA), Canoas, Brazil
  • Jorge L. Gross
    Endocrine Division, Hospital de Clínicas de Porto Alegre (HCPA), Porto Alegre, Brazil
  • Kátia G. Santos
    Laboratory of Human Molecular Genetics, Universidade Luterana do Brasil (ULBRA), Canoas, Brazil
    Cardiology Division, Hospital de Clínicas de Porto Alegre (HCPA), Porto Alegre, Brazil
  • Correspondence: Kátia G. Santos, Programa de Pós-Graduação em Biologia Celular e Molecular Aplicada à Saúde, Universidade Luterana do Brasil (ULBRA), Av. Farroupilha, 8001, Prédio 22, 5o andar, 92425-900, Canoas, RS, Brazil; kgsantos2010@gmail.com
Investigative Ophthalmology & Visual Science February 2015, Vol.56, 1184-1190. doi:10.1167/iovs.14-15758
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to Subscribers Only
      Sign In or Create an Account ×
    • Get Citation

      Luís F. C. Sesti, Daisy Crispim, Luís H. Canani, Evelise R. Polina, Jakeline Rheinheimer, Patrícia S. Carvalho, Jorge L. Gross, Kátia G. Santos; The −308G>A Polymorphism of the TNF Gene Is Associated With Proliferative Diabetic Retinopathy in Caucasian Brazilians With Type 2 Diabetes. Invest. Ophthalmol. Vis. Sci. 2015;56(2):1184-1190. doi: 10.1167/iovs.14-15758.

      Download citation file:


      © 2016 Association for Research in Vision and Ophthalmology.

      ×
  • Supplements
Abstract

Purpose.: We tested the hypothesis that tumor necrosis factor (TNF) gene polymorphisms are associated with diabetic retinopathy (DR) in Caucasians with type 2 diabetes mellitus.

Methods.: In a case-control study, the −238G>A (rs361525), −308G>A (rs1800629), and −857C>T (rs1799724) polymorphisms of the TNF gene were genotyped in 745 outpatients with type 2 diabetes, including 331 subjects without DR, 246 with nonproliferative DR (NPDR), and 168 with proliferative DR (PDR).

Results.: Genotype and allele frequencies of the −238G>A, −308G>A, and −857C>T polymorphisms in subjects with NPDR were not significantly different from those of subjects without DR (P > 0.05 for all comparisons). However, the A allele of the −308G>A polymorphism was more frequent in subjects with PDR than in those with no DR (18.1% vs. 11.5%, corrected P = 0.035). Multivariate logistic regression analysis showed that the −308A allele was independently associated with an increased risk of PDR, under a dominant model (adjusted odds ratio [aOR], 1.82; 95% confidence interval [CI], 1.11–2.98). The combined analysis of the three polymorphisms also showed that haplotypes containing the −308A allele were associated with an increased risk of PDR (aOR, 2.36; 95% CI, 1.29–4.32).

Conclusions.: This study detected, for the first time to our knowledge, an independent association of the −308G>A polymorphism in the TNF gene with PDR in Caucasian Brazilians with type 2 diabetes. This finding suggests that TNF is a potential susceptibility gene for PDR.

Introduction
Diabetic retinopathy (DR) is a severe chronic complication of diabetes mellitus and the leading cause of blindness among working adults in the Western world.1 Diabetic retinopathy also is associated with an increased risk of cardiovascular disease2,3 and all-cause mortality3 in diabetic subjects. Although poor glycemic control and longer duration of diabetes are strong risk factors, genetic factors also have been recognized as contributing to the development and progression of DR.4 
Inflammation has a critical role in the development of the early and late stages of DR.5,6 Tumor necrosis factor (TNF), formerly called TNF-α, is a proinflammatory cytokine that promotes the upregulation of adhesion molecule expression, leukocyte recruitment, apoptosis induction, and monocyte chemo-attraction. It also is responsible for the amplification of the immune response through stimulation of the expression of various transcription factors, growth factors, and other inflammatory mediators.5 Experimental studies in diabetic animals have shown that TNF is involved in capillary degeneration, pericyte loss, and permeability, which all are characteristic of DR.6 Levels of TNF are increased in the retinas or vitreous humor of diabetic animals and patients. Moreover, the vitreal and serum concentrations of TNF are higher in diabetic subjects with DR or proliferative DR (PDR) than in controls.6,7 Similarly, serum levels of soluble TNF receptors 1 and 2 are highly correlated with the severity of DR in subjects with type 2 diabetes.8 
Recently, the −238G>A polymorphism in the gene encoding TNF (TNF) was associated with the serum levels of TNF in Eastern Indians with prediabetes9 and with the progression of prediabetes to type 2 diabetes in Eastern Indians,9 and PDR in Bengali Hindu.10 The −308G>A polymorphism was associated with the progression of impaired glucose tolerance to type 2 diabetes in Finns,11 type 2 diabetes in Asians,12,13 macrovascular complications in Scandinavians,14 and renal disease in Poles and Chinese with diabetes.15,16 In the Japanese population, the −857C>T polymorphism was associated with insulin resistance,17 type 2 diabetes,18 and carotid plaque formation in subjects with type 2 diabetes.19 Therefore, these findings prompted us to test the hypothesis that the −238G>A (rs361525), −308G>A (rs1800629), and −857C>T (rs1799724) polymorphisms in the TNF gene are associated with DR in Caucasian Brazilians with type 2 diabetes. 
Subjects and Methods
Study Population
A total of 745 Caucasian Brazilian type 2 diabetic outpatients was included in this case–control study. Diabetic subjects were enrolled from four centers located at general hospitals in the state of Rio Grande do Sul, namely, Hospital de Clínicas de Porto Alegre, Grupo Hospitalar Conceição, Hospital São Vicente de Paulo, and Hospital Universitário de Rio Grande. Type 2 diabetes was diagnosed according to the American Diabetes Association20 criteria. Subjects were subjected to a standardized clinical and biochemical evaluation comprising a physical examination and a battery of laboratory tests (glycated hemoglobin, serum creatinine, lipid profile, and urinary albumin excretion), as described previously in detail.21 A questionnaire was used to collect data from medical records regarding age, sex, age at diabetes diagnosis, smoking behavior, presence of arterial hypertension, use of medication, and other comorbidities. 
The diagnosis of DR was based on ophthalmoscopic examination through dilated pupils, and fluorescein angiography was conducted when indicated. Diabetic retinopathy was graded as absent (no fundus abnormalities), nonproliferative (NPDR; microaneurysms, hard exudates, retinal hemorrhages, and intraretinal microvascular abnormalities), or PDR (new vessels within one disc diameter of the disc and/or new vessels originating elsewhere).22 The DR classification was based on the most severe degree of retinopathy in the worse affected eye and was performed in each institution by a single ophthalmologist who was unaware of the genotypic data of the subjects. Subjects with DR were defined as case subjects (n = 414), and subjects without any degree of DR were defined as control subjects (n = 331). Among the subjects with DR, 246 had NPDR and 168 had PDR. 
To estimate the allele frequencies of the −238G>A, −308G>A, and −857C>T polymorphisms in the general population, we also genotyped 174 Caucasian Brazilians recruited from healthy volunteer blood donors (72% male, with a mean age of 44 years) from Hospital de Clínicas de Porto Alegre (Porto Alegre, Brazil). All subjects of this study were of European descent (primarily from Portugal, Spain, Italy, and Germany). The tenets of the Declaration of Helsinki were followed, the hospital ethics committees approved the study, and all diabetic subjects and blood donors gave written informed consent. 
Genotyping
We extracted DNA from peripheral blood leukocytes by the salting-out procedure.23 The genotypes of the −238G>A polymorphism in the TNF gene were determined by real-time PCR with the probes and primers provided in the Human Custom TaqMan Genotyping Assay (Life Technologies, Carlsbad, CA, USA). The plates were loaded into a real-time PCR thermal cycler and heated for 10 minutes at 95°C, followed by 40 cycles of 95°C for 15 seconds and 63°C for 1 minute. The fluorescence data files from each plate were analyzed by automated allele-calling software (SDS 2.1; Life Technologies). 
Gene fragments containing the −308G>A and −857C>T variant sites were amplified by PCR using the primers and conditions described by Wilson et al.24 and Du et al.,25 respectively. The PCR products were digested with the appropriate restriction enzymes under the conditions recommended by the manufacturer (New England Biolabs, Ipswich, MA, USA). The digested fragments were visualized on 8% polyacrylamide gels stained with silver nitrate. To improve genotyping accuracy, samples with known genotypes were used in each run, and the laboratory personnel were blinded to the clinical characteristics of the subjects. 
Statistical Analyses
Continuous data are presented as mean ± SD or median (25th–75th percentile). Categorical variables are reported as number (percentage) or percentage. The Kolmogorov-Smirnov test was used to verify the normality of quantitative variables. Continuous variables were compared among groups of diabetic subjects by the ANOVA test for normally distributed variables or the Kruskal-Wallis test for variables with a skewed distribution. Categorical variables were compared by the χ2 test. Allele frequencies were determined by gene counting, and departures from Hardy-Weinberg equilibrium were verified with the χ2 test. Allele and genotype frequencies were compared among groups of subjects with the χ2 test or Fisher's exact test. Statistical analyses were performed using SPSS (version 18.0; SPSS, Inc., Chicago, IL, USA) and PEPI (version 4.0)26 statistical software. 
Multiple logistic regression analyses were performed to evaluate the association of TNF polymorphisms with NPDR or PDR, and estimates of odds ratios (ORs) and 95% confidence intervals (CI) were adjusted for the demographic and clinical variables that were associated with these outcomes in the univariate analyses. Because of the low frequency of the minor alleles, the analyses were performed assuming a dominant genetic model (for the −238G>A and −308G>A polymorphisms, the carriers of the A allele were compared to subjects with the GG genotype, and for the −857C>T polymorphism, the carriers of the T allele were compared to subjects with the CC genotype). The dominant model was chosen to maximize the statistical power and does not necessarily reflect the pathophysiological mechanism of DR. The association analyses were performed using the SPSS package. Power calculations were estimated in the PEPI program. 
Pairwise linkage disequilibrium among the TNF polymorphisms was calculated and expressed in terms of D' and r2.27 Haplotype frequencies were estimated by a Bayesian method using the PHASE program (version 2.1).28,29 We also used PHASE to compare the distribution of different TNF haplotypes between the cases and controls, computing P values by a permutation test with 1,000 random replicates. 
The P values were corrected for multiple comparisons with the Bonferroni correction formula (corrected P [Pc] = 1 − (1 − P)n; n = number of comparisons) according to the number of polymorphisms and correlated phenotypes (NPDR and PDR). The P values <0.05 were considered statistically significant. 
Results
Description of Type 2 Diabetic Subjects
The demographic and clinical characteristics of the type 2 diabetic subjects are summarized in Table 1 according to the presence and severity of DR. Subjects with NPDR or PDR had a longer duration of diabetes and higher proportion of insulin therapy compared to subjects without DR. Moreover, subjects with PDR more often were men and had higher levels of serum creatinine, lower levels of high-density lipoprotein cholesterol, and a higher proportion of renal disease than those without this complication. Although glycemic control did not differ according to the presence and severity of DR, approximately 47% of the diabetic subjects had glycated hemoglobin levels higher than 7.0% (range, 3.3%–14.4%). 
Table 1
 
Demographic and Clinical Characteristics of 745 Type 2 Diabetic Subjects According to the Presence and Severity of DR
Table 1
 
Demographic and Clinical Characteristics of 745 Type 2 Diabetic Subjects According to the Presence and Severity of DR
Variable No DR, n = 331 NPDR, n = 246 PDR, n = 168 P
Age, y 59.9 ± 9.6 61.2 ± 9.8 62.1 ± 8.6 0.062
Sex, % males 43.2a 50.8a,b 60.7b 0.001*
Duration of diabetes, y 11.8 ± 7.0a 14.2 ± 8.9b 15.7 ± 8.8b <0.001*
Glycated hemoglobin, % 7.2 ± 2.3 7.0 ± 1.8 6.7 ± 1.7 0.347
Hypertension, % 69.6 75.1 76.7 0.209
Systolic blood pressure, mm Hg 143.1 ± 22.5 143.6 ± 25.2 148.2 ± 23.9 0.048*
Diastolic blood pressure, mm Hg 86.3 ± 14.2 85.9 ± 12.7 85.4 ± 12.9 0.839
Body mass index, kg/m2 29.9 ± 9.1 33.4 ± 19.0 38.9 ± 26.1 0.503
Insulin use, % 29.0a 47.9b 57.8b <0.001*
Serum creatinine, μM 79.6 (70.7–97.2)a 88.4 (70.7–106.1)a 132.6 (79.6–380.1)b <0.001*
Total cholesterol, mM 5.3 ± 1.2 5.4 ± 1.3 5.3 ± 1.3 0.974
HDL cholesterol, mM 1.15 ± 0.30a 1.12 ± 0.31a,b 1.07 ± 0.32b 0.013*
Triglycerides, mM 1.7 (1.3–2.6) 1.8 (1.2–2.5) 1.8 (1.3–2.8) 0.791
Renal disease, % 47.5a 54.7a 81.0b <0.001*
Polymorphisms of TNF in Blood Donors and Type 2 Diabetic Subjects
The genotype frequencies were in agreement with those predicted by Hardy-Weinberg equilibrium for all TNF polymorphisms in healthy blood donors and type 2 diabetic subjects, with the exception of the −238G>A polymorphism among diabetic subjects, in whom there was a lower frequency of heterozygotes than expected (expected frequency = 10.3% versus observed frequency = 9.5%, P = 0.032). The genotype and allele frequencies for the −238G>A, −308G>A, and −857C>T polymorphisms did not differ significantly between type 2 diabetic subjects and blood donors (Table 2). Of eight expected haplotypes, seven were observed in both groups of subjects, and the haplotype frequencies were similar in the subjects with type 2 diabetes and blood donors, as shown in Supplementary Table S1. The three polymorphisms were in weak linkage disequilibrium, as indicated by the low r2 coefficients (Supplementary Table S2). 
Table 2
 
Frequencies of Genotypes and Alleles of TNF Polymorphisms in Blood Donors and Subjects With Type 2 Diabetes
Table 2
 
Frequencies of Genotypes and Alleles of TNF Polymorphisms in Blood Donors and Subjects With Type 2 Diabetes
Polymorphism Genotype/ Allele Blood Donors, n (%) Type 2 Diabetes, n (%) P*
−238G>A n = 169 n = 695
GG 147 (87.0) 624 (89.8) 0.551
GA 22 (13.0) 66 (9.5)
AA 0 (0.0) 5 (0.7)
G 316 (93.5) 1314 (94.5) 0.903
A 22 (6.5) 76 (5.5)
−308G>A n = 170 n = 726
GG 129 (75.9) 535 (73.7) 0.983
GA 38 (22.3) 176 (24.2)
AA 3 (1.8) 15 (2.1)
G 296 (87.1) 1246 (85.8) 0.941
A 44 (12.9) 206 (14.2)
−857C>T n = 165 n = 657
CC 125 (75.8) 475 (72.3) 0.420
CT 40 (24.2) 169 (25.7)
TT 0 13 (2.0)
C 290 (87.9) 1119 (85.2) 0.563
T 40 (12.1) 195 (14.8)
Polymorphisms of TNF in Type 2 Diabetic Subjects According to the Presence and Severity of DR
Table 3 shows the distribution of the genotype and allele frequencies of the −238G>A, −308G>A, and −857C>T polymorphisms in type 2 diabetic subjects according to the presence and severity of DR. There were no significant differences in the genotype and allele frequencies between subjects with NPDR and without DR for the three polymorphisms, indicating that they are not associated with the presence of NPDR. 
Table 3
 
Frequencies of Genotypes and Alleles of TNF Polymorphisms in 745 Type 2 Diabetic Subjects According to the Presence and Severity of DR
Table 3
 
Frequencies of Genotypes and Alleles of TNF Polymorphisms in 745 Type 2 Diabetic Subjects According to the Presence and Severity of DR
Polymorphism Genotype/Allele No DR, n (%) NPDR, n (%) PDR, n (%)
−238G>A n = 308 n = 228 n = 159
Genotypes GG 277 (89.9) 202 (88.6) 145 (91.2)
GA 28 (9.1) 24 (10.5) 14 (8.8)
AA 3 (1.0) 2 (0.9) 0 (0.0)
P >0.999* 0.999†
Alleles G 582 (94.5) 428 (93.9) 304 (95.6)
A 34 (5.5) 28 (6.1) 14 (4.4)
P >0.999* 0.993†
OR (95% CI) for the A allele, dominant model
 Univariate 1.15 (0.66–2.00)‡ 0.86 (0.44–1.67)§
 Multivariate 1.04 (0.55–1.96)‡ 1.12 (0.50–2.51)§
−308G>A n = 322 n = 241 n = 163
Genotypes GG 251 (78.0) 174 (72.2) 110 (67.5)
GA 68 (21.1) 61 (25.3) 47 (28.8)
AA 3 (0.9) 6 (2.5) 6 (3.7)
P 0.631* 0.076†
Alleles G 570 (88.5) 409 (84.9) 267 (81.9)
A 74 (11.5) 73 (15.1) 59 (18.1)
P 0.421* 0.035†‖
Odds ratio (95% CI) for the A allele, dominant model
 Univariate 1.36 (0.93–2.00)‡ 1.70 (1.12–2.59)§‖
 Multivariate 1.41 (0.92–2.15)‡ 1.82 (1.11–2.98)§‖
−857C>T n = 270 n = 230 n = 157
Genotypes CC 195 (72.2) 167 (72.6) 113 (72.0)
CT 67 (24.8) 60 (26.1) 42 (26.7)
TT 8 (3.0) 3 (1.3) 2 (1.3)
P 0.970* 0.986†
Alleles C 457 (84.6) 394 (85.7) 268 (85.4)
T 83 (15.4) 66 (14.3) 46 (14.6)
P >0.999* >0.999†
Odds ratio (95% CI) for the T allele, dominant model
 Univariate 0.98 (0.66–1.45)‡ 1.01 (0.65–1.57)†
 Multivariate 1.04 (0.68–1.60)‡ 1.00 (0.57–1.74)†
However, the data show that the frequency of the A allele of the −308G>A polymorphism was higher in subjects with PDR than in those with no DR (18.1% vs. 11.5%, Pc = 0.035). Under a dominant model, the −308A allele was associated with an increased risk of PDR, even after adjusting for age, sex, duration of diabetes, systolic blood pressure, body mass index, use of insulin, serum creatinine, and high-density lipoprotein cholesterol levels (Table 3). 
As seen in Table 4, there were no significant differences in the haplotype frequencies according to the presence and severity of DR. However, the frequency of subjects carrying haplotypes with the A allele of the −308G>A polymorphism were higher in the PDR group than in the group with no DR (30.9% vs. 21.4%, uncorrected P = 0.027). After controlling for the covariates associated with this outcome, the multivariate analysis confirmed that, under a dominant model, haplotypes containing the −308A allele were associated with an increased risk of PDR (Table 4). 
Table 4
 
Frequencies of the Haplotypes of TNF Polymorphisms in 745 Type 2 Diabetic Subjects According to the Presence and Severity of DR
Table 4
 
Frequencies of the Haplotypes of TNF Polymorphisms in 745 Type 2 Diabetic Subjects According to the Presence and Severity of DR
Haplotype No DR, n = 646 NPDR, n = 488 PDR, n = 330
−238G/−308G/−857C 0.6911 0.6568 0.6531
−238G/−308G/−857T 0.1430 0.1389 0.1312
−238G/−308A/−857C 0.1080 0.1414 0.1626
−238G/−308A/−857T 0.0039 0.0034 0.0102
−238A/−308G/−857C 0.0446 0.0553 0.0354
−238A/−308G/−857T 0.0068 0.0010 0.0049
−238A/−308A/−857C 0.0026 0.0032 0.0026
−238A/−308A/−857T
P 0.272* 0.130†
OR (95% CI) for the haplotypes carrying the −308A allele, dominant model
 Univariate 1.34 (0.91–1.97)‡ 1.65 (1.08–2.52)§‖
 Multivariate 1.51 (0.96–2.35)‡ 2.36 (1.29–4.32)§‖
Discussion
This study detected, for the first time to our knowledge, an independent association of the −308G>A polymorphism in the TNF gene with PDR in Caucasian Brazilians with type 2 diabetes. With respect to the presence of any degree of DR, our findings are in accordance with previous studies, including a recent meta-analysis,30 which showed no association of the −308G>A polymorphism with DR in Indian,31 Chinese,16,32,33 and Japanese34 subjects with type 2 diabetes. However, with respect to PDR, our data conflicted with those obtained from the Scandinavian (n = 878),14 Indian (n = 493 and 196),10,35 and Japanese (n = 251)34 subjects with type 2 diabetes, which showed no association between the −308G>A polymorphism and PDR. 
The combination of the low frequency of the −308A allele, the definition of the diabetic control group (with or without NPDR) and the relatively small sample size of some of the previous studies (patients with PDR, n = 39 and 104, respectively)34,35 may be the main reason for this discrepancy. The frequency of the −308A allele ranges from 0.07 to 0.22 depending on the population analyzed.30,36,37 Even within the same country, the frequency of the −308A allele varies. Such is the case in India, where the frequency of the −308A allele was reported as being 0.06 in the North-West population from Punjab31 and 0.19 in the Bengali Hindu population.10 Considering that the −308A allele has a low frequency in most populations and that its effect in the development or progression of DR may be modest, such an association can only be detected in samples with a large population size. Indeed, the present study identified a moderate association of the −308A allele with an increased risk of PDR in Brazilian Caucasians with type 2 diabetes, even after adjusting for age, sex, duration of diabetes, systolic blood pressure, body mass index, use of insulin, serum creatinine, and high-density lipoprotein cholesterol levels. 
The three polymorphisms analyzed in our study were selected based on their previously reported association with vascular complications of type 2 diabetes or based on the putative functional role of these gene variants. In fact, our findings related to the −308G>A polymorphism corroborate earlier in vitro studies that showed a 2- to 5-fold increase in TNF gene expression in the presence of the −308A allele compared to the −308G allele in cell lines stimulated with phorbol myristate acetate, retinoic acid, or lipopolysaccharide, alone or in combination.38 Based on the assumption that the overexpression of TNF can generate proapoptotic and proinflammatory signals7 leading to leukocyte adherence in retinal blood vessels, blood–retinal barrier breakdown, pericyte loss, capillary degeneration, and increased vascular permeability,6 the increased expression of the TNF gene due to the presence of the −308A allele could be responsible, at least in part, for the abnormalities that occur during the progression of PDR. Recently, systemic levels of soluble TNF receptors 1 and 2 were shown to be directly associated with increasing severity of DR in Hispanics with type 2 diabetes, supporting the idea that inflammation and insulin resistance may have a critical role in the development of DR.8 
In our study, the frequencies of the −308A allele and of haplotypes carrying the −308A allele tended to increase according to the presence and severity of DR (P values for trend = 0.004 and 0.007, respectively). The haplotype analysis confirmed that the carriership of the −308A allele is associated with the presence of PDR, regardless of which alleles are present at the −238 and −857 nucleotide positions. This finding remained significant even after the adjustment for the demographic and clinical variables also associated with PDR in the multivariate analysis. The higher prevalence of the −308A allele among those with PDR suggests that the A allele could exert a greater effect on disease progression than on disease development. In fact, studies have identified a stronger genetic susceptibility for the more advanced form of DR, rather than the presence of any degree of DR.4 
With respect to the −238G>A polymorphism, a study on the Bengali Hindu population showed that the −238A allele had a frequency of 0.23 in type 2 diabetic subjects and was associated with an increased risk of PDR under dominant, additive, and allelic models.10 In our study, the frequency of the −238A allele was low (0.06) and quite similar in subjects with or without DR. Despite the low frequency of the −238A allele, power calculations indicated that our study had a power of approximately 85% at a significance level of 5% to detect an association of the same magnitude as that found by Paine et al.10 under the dominant model for PDR. As these are the only two studies to date that have investigated the association of the −238G>A polymorphism with PDR, whether these conflicting results may be mainly attributed to the ethnicity of the subjects or to other factors cannot be determined at present. 
In the Japanese population, the −857C>T polymorphism was associated with insulin resistance,17 type 2 diabetes in obese subjects,18 higher serum low-density lipoprotein cholesterol levels, and carotid plaque formation in subjects with type 2 diabetes.19 However, to our knowledge no previous study investigated the relationship between this polymorphism and diabetic microvascular complications. In the present study, the genotype and allele frequencies for the −857C>T polymorphism were similar among groups of subjects, indicating that this gene variant is not related to the pathogenesis of DR. 
One limitation of our study is the method used to grade DR. Although the mydriatic ophthalmoscopy is accessible in most clinical settings, clinical ophthalmoscopy has reduced sensitivity to grade lower levels of DR (mild NPDR).39 If some diabetic subjects with mild NPDR were included in the control group (no DR), this may have underpowered the analysis of the association of the −308G>A polymorphism with NPDR. In addition, the fact that the diagnosis of DR relied on a single ophthalmologist at each of the four institutions may have resulted in interobserver differences. Again, it might have affected the analysis of NPDR, but not the association with PDR. 
Another concern that arises is whether the undetected population stratification may have confounded our analysis, as the Brazilian population is genetically admixed. We think this is unlikely, as all the subjects included in our study were born in the State of Rio Grande do Sul and had at least two European grandparents. Rio Grande do Sul State was peopled by Europeans immigrants, mainly from Portugal, Spain, Italy, and Germany. Data from official census and genetic studies of ancestry-informative markers show that more than 80% of the population of the state are white and have a predominant European ancestry.40 Thus, despite some degree of genetic admixture in the self-declared Caucasian Brazilian population, it is unlikely that genetic heterogeneity could be a confounding factor in our analysis. 
Considering the low frequency of the minor alleles, we assumed a dominant genetic model to assess the association of TNF polymorphisms with DR to ensure adequate statistical power. We also analyzed a potential association of the −308G>A polymorphism with DR under recessive, additive, and allelic models. In these models, the −308A allele was associated with an increased risk of PDR (data not shown). However, controlling for confounding factors was not possible because the frequency of homozygotes for the −308A allele was extremely low. Therefore, the discovery of an independent association under the dominant model does not necessarily mean that the dominant model is the best one to describe the relationship between the −308G>A polymorphism and PDR. 
In conclusion, the A allele of the −308G>A polymorphism in the TNF gene was independently associated with the increased risk of PDR in Caucasian Brazilians with type 2 diabetes. Our findings provide suggestive evidence that polymorphisms in the TNF gene may be potentially involved in the pathogenesis of PDR and should be investigated further. 
Acknowledgments
Supported by grants from Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq), Brasília, Brazil, and Fundo de Incentivo à Pesquisa e Eventos do HCPA (FIPE-HCPA), Porto Alegre, Brazil. 
Disclosure: L.F.C. Sesti, None; D. Crispim, None; L.H. Canani, None; E.R. Polina, None; J. Rheinheimer, None; P.S. Carvalho, None; J.L. Gross, None; K.G. Santos, None 
References
Sivaprasad S Gupta B Crosby-Nwaobi R Evans J. Prevalence of diabetic retinopathy in various ethnic groups: a worldwide perspective. Surv Ophthalmol. 2012; 57: 347–370. [CrossRef] [PubMed]
Kawasaki R Cheung N Islam A Is diabetic retinopathy related to subclinical cardiovascular disease? Ophthalmology. 2011; 118: 860–865. [CrossRef] [PubMed]
Kramer CK Rodrigues TC Canani LH Diabetic retinopathy predicts all-cause mortality and cardiovascular events in both type 1 and 2 diabetes: meta-analysis of observational studies. Diabetes Care. 2011; 34: 1238–1244. [CrossRef] [PubMed]
Kuo JZ Wong TY Rotter JI. Challenges in elucidating the genetics of diabetic retinopathy. JAMA Ophthalmol. 2014; 132: 96–107. [CrossRef] [PubMed]
Adamis AP Berman AJ. Immunological mechanisms in the pathogenesis of diabetic retinopathy. Semin Immunopathol. 2008; 30: 65–84. [CrossRef] [PubMed]
Tang J Kern TS. Inflammation in diabetic retinopathy. Prog Retin Eye Res. 2011; 30: 343–358. [CrossRef] [PubMed]
Ejaz S Chekarova I Ejaz A Importance of pericytes and mechanisms of pericyte loss during diabetes retinopathy. Diabetes Obes Metab. 2008; 10: 53–63. [PubMed]
Kuo JZ Guo X Klein R Systemic soluble tumor necrosis factor receptors 1 and 2 are associated with severity of diabetic retinopathy in Hispanics. Ophthalmology. 2012; 119: 1041–1046. [CrossRef] [PubMed]
Dutta D Choudhuri S Mondal SA Tumor necrosis factor alpha −238G/A (rs 361525) gene polymorphism predicts progression to type-2 diabetes in an Eastern Indian population with prediabetes. Diabetes Res Clin Pract. 2013; 99: e37–e41. [CrossRef] [PubMed]
Paine SK Sen A Choudhuri S Association of tumor necrosis factor α, interleukin 6, and interleukin 10 promoter polymorphism with proliferative diabetic retinopathy in type 2 diabetic subjects. Retina. 2012; 32: 1197–1203. [CrossRef] [PubMed]
Kubaszek A Pihlajamäki J Komarovski V Finnish Diabetes Prevention Study. Promoter polymorphisms of the TNF-alpha (G-308A) and IL-6 (C-174G) genes predict the conversion from impaired glucose tolerance to type 2 diabetes: the Finnish Diabetes Prevention Study. Diabetes. 2003; 52: 1872–1876. [CrossRef] [PubMed]
Liu ZH Ding YL Xiu LC A meta-analysis of the association between TNF-a −308G>A polymorphism and type 2 diabetes mellitus in Han Chinese population. PLoS One. 2013; 8: e59421. [CrossRef] [PubMed]
Zhao Y Li Z Zhang L The TNF-alpha −308G/A polymorphism is associated with type 2 diabetes mellitus: an updated meta-analysis. Mol Biol Rep. 2014; 41: 73–83. [CrossRef] [PubMed]
Lindholm E Bakhtadze E Cilio C Association between LTA, TNF and AGER polymorphisms and late diabetic complications. PLoS One. 2008; 3: e2546. [CrossRef] [PubMed]
Buraczynska K Koziol-Montewka M Majdan M Genetic determination of TNF and myeloperoxidase production in dialyzed patients with diabetic nephropathy. Ren Fail. 2004; 26: 633–639. [CrossRef] [PubMed]
Wang Y Ng MC So WY Association between tumour necrosis factor-α G-308A polymorphism and risk of nephropathy in obese Chinese type 2 diabetic patients. Nephrol Dial Transplant. 2005; 20: 2733–2738. [CrossRef] [PubMed]
Ohara M Maesawa C Takebe N Different susceptibility to insulin resistance and fatty liver depending on the combination of TNF-α C-857T and adiponectin G+276T gene polymorphisms in Japanese subjects with type 2 diabetes. Tohoku J Exp Med. 2012; 226: 161–169. [CrossRef] [PubMed]
Kamizono S Yamada K Seki N Susceptible locus for obese type 2 diabetes mellitus in the 5′-flanking region of the tumor necrosis factor-alpha gene. Tissue Antigens. 2000; 55: 449–452. [CrossRef] [PubMed]
Yamashina M Kaneko Y Maesawa C Association of TNF-alpha gene promoter C-857T polymorphism with higher serum LDL cholesterol levels and carotid plaque formation in Japanese patients with type 2 diabetes. Tohoku J Exp Med. 2007; 211: 251–258. [CrossRef] [PubMed]
American Diabetes Association. Diagnosis and classification of diabetes mellitus. Diabetes Care. 2005; 28 (suppl): S37–S42. [PubMed]
Canani LH Capp C Ng DPK The fatty acid–binding protein-2 A54T polymorphism is associated with renal disease in patients with type 2 diabetes. Diabetes. 2005; 54: 3326–3330. [CrossRef] [PubMed]
Wilkinson CP Ferris FL III Klein RE ; for Global Diabetic Retinopathy Project Group. Proposed international clinical diabetic retinopathy and diabetic macular edema disease severity scales. Ophthalmology. 2003; 110: 1677–1682. [CrossRef] [PubMed]
Lahiri DK Nurnberger JI Jr. A rapid non-enzymatic method for the preparation of HMW DNA from blood for RFLP studies. Nucleic Acids Res. 1991; 19: 5444. [CrossRef] [PubMed]
Wilson AG di Giovine FS Blakemore AI Duff GW. Single base polymorphism in the human tumour necrosis factor alpha (TNF alpha) gene detectable by NcoI restriction of PCR product. Hum Mol Genet. 1992; 1: 353. [CrossRef] [PubMed]
Du T Guo X-H Zhu X-L Association of TNF-alpha promoter polymorphisms with the outcomes of hepatitis B virus infection in Chinese Han population. J Viral Hepat. 2006; 13: 618–624. [CrossRef] [PubMed]
Abramson JH Gahlinger PM. Computer Programs for Epidemiologic Analyses: PEPI Version 4.0. Salt Lake City, UT: Sagebrush Press; 2002.
Lewontin RC. On measures of gametic disequilibrium. Genetics. 1988; 120: 849–852. [PubMed]
Stephens M Scheet P. Accounting for decay of linkage disequilibrium in haplotype inference and missing data imputation. Am J Hum Genet. 2005; 76: 449–462. [CrossRef] [PubMed]
Stephens M Smith NJ Donnelly P. A new statistical method for haplotype reconstruction from population data. Am J Hum Genet. 2001; 68: 978–989. [CrossRef] [PubMed]
Meng N Zhang Y Li H Ma J Qu Y. Association of tumor necrosis factor alpha promoter polymorphism (TNF-α 238 G/A and TNF-α 308 G/A) with diabetic mellitus, diabetic retinopathy and diabetic nephropathy: a meta-analysis. Curr Eye Res. 2014; 39: 194–203. [CrossRef] [PubMed]
Sikka R Raina P Matharoo K TNF-α (g.-308 G > A) and ADIPOQ (g. + 45 T > G) gene polymorphisms in type 2 diabetes and microvascular complications in the region of Punjab (North-West India). Curr Eye Res. 2014; 39: 1042–1051. [CrossRef] [PubMed]
Lee SC Pu YB Thomas GN Tumor necrosis factor alpha gene G-308A polymorphism in the metabolic syndrome. Metabolism. 2000; 49: 1021–1024. [CrossRef] [PubMed]
Wang N Huang K Zou H No association found between the promoter variants of TNF-alpha and diabetic retinopathy in Chinese patients with type 2 diabetes. Curr Eye Res. 2008; 33: 377–383. [CrossRef] [PubMed]
Yoshioka K Yoshida T Takakura Y Relationship between polymorphisms 804C/A and 252A/G of lymphotoxin-alpha gene and −308G/A of tumor necrosis factor alpha gene and diabetic retinopathy in Japanese patients with type 2 diabetes mellitus. Metabolism. 2006; 55: 1406–1410. [CrossRef] [PubMed]
Prasad P Tiwari AK Kumar KM Association of TGFβ1, TNFα, CCR2 and CCR5 gene polymorphisms in type-2 diabetes and renal insufficiency among Asian Indians. BMC Med Genet. 2007; 8: 20. [CrossRef] [PubMed]
Visentainer JE Sell AM da Silva GC TNF, IFNG, IL6, IL10 and TGFB1 gene polymorphisms in South and Southeast Brazil. Int J Immunogenet. 2008; 35: 287–293. [CrossRef] [PubMed]
Van Dyke AL Cote ML Wenzlaff AS Cytokine SNPs: comparison of allele frequencies by race and implications for future studies. Cytokine. 2009; 46: 236–244. [CrossRef] [PubMed]
Abraham LJ Kroeger KM. Impact of the −308 TNF promoter polymorphism on the transcriptional regulation of the TNF gene: relevance to disease. J Leukoc Biol. 1999; 66: 562–566. [PubMed]
Lin DY Blumenkranz MS Brothers RJ for the Digital Diabetic Screening Group. The sensitivity and specificity of single-field nonmydriatic monochromatic digital fundus photography with remote image interpretation for diabetic retinopathy screening: a comparison with ophthalmoscopy and standardized mydriatic color photography. Am J Ophthalmol. 2002; 134: 204–213. [CrossRef] [PubMed]
Pena SDJ Di Pietro G Fuchshuber-Moraes M The genomic ancestry of individuals from different geographical regions of Brazil is more uniform than expected. PLoS One. 2011; 6: e17063. [CrossRef] [PubMed]
Table 1
 
Demographic and Clinical Characteristics of 745 Type 2 Diabetic Subjects According to the Presence and Severity of DR
Table 1
 
Demographic and Clinical Characteristics of 745 Type 2 Diabetic Subjects According to the Presence and Severity of DR
Variable No DR, n = 331 NPDR, n = 246 PDR, n = 168 P
Age, y 59.9 ± 9.6 61.2 ± 9.8 62.1 ± 8.6 0.062
Sex, % males 43.2a 50.8a,b 60.7b 0.001*
Duration of diabetes, y 11.8 ± 7.0a 14.2 ± 8.9b 15.7 ± 8.8b <0.001*
Glycated hemoglobin, % 7.2 ± 2.3 7.0 ± 1.8 6.7 ± 1.7 0.347
Hypertension, % 69.6 75.1 76.7 0.209
Systolic blood pressure, mm Hg 143.1 ± 22.5 143.6 ± 25.2 148.2 ± 23.9 0.048*
Diastolic blood pressure, mm Hg 86.3 ± 14.2 85.9 ± 12.7 85.4 ± 12.9 0.839
Body mass index, kg/m2 29.9 ± 9.1 33.4 ± 19.0 38.9 ± 26.1 0.503
Insulin use, % 29.0a 47.9b 57.8b <0.001*
Serum creatinine, μM 79.6 (70.7–97.2)a 88.4 (70.7–106.1)a 132.6 (79.6–380.1)b <0.001*
Total cholesterol, mM 5.3 ± 1.2 5.4 ± 1.3 5.3 ± 1.3 0.974
HDL cholesterol, mM 1.15 ± 0.30a 1.12 ± 0.31a,b 1.07 ± 0.32b 0.013*
Triglycerides, mM 1.7 (1.3–2.6) 1.8 (1.2–2.5) 1.8 (1.3–2.8) 0.791
Renal disease, % 47.5a 54.7a 81.0b <0.001*
Table 2
 
Frequencies of Genotypes and Alleles of TNF Polymorphisms in Blood Donors and Subjects With Type 2 Diabetes
Table 2
 
Frequencies of Genotypes and Alleles of TNF Polymorphisms in Blood Donors and Subjects With Type 2 Diabetes
Polymorphism Genotype/ Allele Blood Donors, n (%) Type 2 Diabetes, n (%) P*
−238G>A n = 169 n = 695
GG 147 (87.0) 624 (89.8) 0.551
GA 22 (13.0) 66 (9.5)
AA 0 (0.0) 5 (0.7)
G 316 (93.5) 1314 (94.5) 0.903
A 22 (6.5) 76 (5.5)
−308G>A n = 170 n = 726
GG 129 (75.9) 535 (73.7) 0.983
GA 38 (22.3) 176 (24.2)
AA 3 (1.8) 15 (2.1)
G 296 (87.1) 1246 (85.8) 0.941
A 44 (12.9) 206 (14.2)
−857C>T n = 165 n = 657
CC 125 (75.8) 475 (72.3) 0.420
CT 40 (24.2) 169 (25.7)
TT 0 13 (2.0)
C 290 (87.9) 1119 (85.2) 0.563
T 40 (12.1) 195 (14.8)
Table 3
 
Frequencies of Genotypes and Alleles of TNF Polymorphisms in 745 Type 2 Diabetic Subjects According to the Presence and Severity of DR
Table 3
 
Frequencies of Genotypes and Alleles of TNF Polymorphisms in 745 Type 2 Diabetic Subjects According to the Presence and Severity of DR
Polymorphism Genotype/Allele No DR, n (%) NPDR, n (%) PDR, n (%)
−238G>A n = 308 n = 228 n = 159
Genotypes GG 277 (89.9) 202 (88.6) 145 (91.2)
GA 28 (9.1) 24 (10.5) 14 (8.8)
AA 3 (1.0) 2 (0.9) 0 (0.0)
P >0.999* 0.999†
Alleles G 582 (94.5) 428 (93.9) 304 (95.6)
A 34 (5.5) 28 (6.1) 14 (4.4)
P >0.999* 0.993†
OR (95% CI) for the A allele, dominant model
 Univariate 1.15 (0.66–2.00)‡ 0.86 (0.44–1.67)§
 Multivariate 1.04 (0.55–1.96)‡ 1.12 (0.50–2.51)§
−308G>A n = 322 n = 241 n = 163
Genotypes GG 251 (78.0) 174 (72.2) 110 (67.5)
GA 68 (21.1) 61 (25.3) 47 (28.8)
AA 3 (0.9) 6 (2.5) 6 (3.7)
P 0.631* 0.076†
Alleles G 570 (88.5) 409 (84.9) 267 (81.9)
A 74 (11.5) 73 (15.1) 59 (18.1)
P 0.421* 0.035†‖
Odds ratio (95% CI) for the A allele, dominant model
 Univariate 1.36 (0.93–2.00)‡ 1.70 (1.12–2.59)§‖
 Multivariate 1.41 (0.92–2.15)‡ 1.82 (1.11–2.98)§‖
−857C>T n = 270 n = 230 n = 157
Genotypes CC 195 (72.2) 167 (72.6) 113 (72.0)
CT 67 (24.8) 60 (26.1) 42 (26.7)
TT 8 (3.0) 3 (1.3) 2 (1.3)
P 0.970* 0.986†
Alleles C 457 (84.6) 394 (85.7) 268 (85.4)
T 83 (15.4) 66 (14.3) 46 (14.6)
P >0.999* >0.999†
Odds ratio (95% CI) for the T allele, dominant model
 Univariate 0.98 (0.66–1.45)‡ 1.01 (0.65–1.57)†
 Multivariate 1.04 (0.68–1.60)‡ 1.00 (0.57–1.74)†
Table 4
 
Frequencies of the Haplotypes of TNF Polymorphisms in 745 Type 2 Diabetic Subjects According to the Presence and Severity of DR
Table 4
 
Frequencies of the Haplotypes of TNF Polymorphisms in 745 Type 2 Diabetic Subjects According to the Presence and Severity of DR
Haplotype No DR, n = 646 NPDR, n = 488 PDR, n = 330
−238G/−308G/−857C 0.6911 0.6568 0.6531
−238G/−308G/−857T 0.1430 0.1389 0.1312
−238G/−308A/−857C 0.1080 0.1414 0.1626
−238G/−308A/−857T 0.0039 0.0034 0.0102
−238A/−308G/−857C 0.0446 0.0553 0.0354
−238A/−308G/−857T 0.0068 0.0010 0.0049
−238A/−308A/−857C 0.0026 0.0032 0.0026
−238A/−308A/−857T
P 0.272* 0.130†
OR (95% CI) for the haplotypes carrying the −308A allele, dominant model
 Univariate 1.34 (0.91–1.97)‡ 1.65 (1.08–2.52)§‖
 Multivariate 1.51 (0.96–2.35)‡ 2.36 (1.29–4.32)§‖
Supplementary Tables
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×