Association between diabetic retinopathy and genetic variations in α2β1 integrin, a platelet receptor for collagen

Informed consent was obtained from all subjects enrolled into the study. Normal subjects (n = 169; 135 males, 34 females), recruited at Hibiya Medical Center (Tokyo, Japan), Fuji Electronics (Tokyo, Japan), and Keio University Hospital (Tokyo, Japan) for their regular checkups, were analyzed. The mean age was 47.1 ± 5.9 years (mean ± SD). They had no clinical or laboratory evidence of either past vascular disorders or any form of diabetes.

Patients were genetically unrelated 227 Japanese subjects who had had a diagnosis of type II diabetes mellitus, as defined by World Health Organization criteria,25 and were followed up on a regular basis at an outpatient clinic of Saitama Social Insurance Hospital (Saitama, Japan). They were divided into 2 groups (Table1). The first group included 119 patients with retinopathy (35 patients with retinopathy without nephropathy, group B; and 84 patients with both retinopathy and nephropathy, group C). The second group consisted of 108 patients without retinopathy or nephropathy (group A). The group A patients were selected to match groups B and C patients in terms of age at diagnosis of diabetes, sex, and disease duration after diagnosis of diabetes. Diabetic retinopathy was diagnosed by independent diabetic ophthalmologists using standard fundus photos and was classified as simple, preproliferative, and proliferative (either treated with photocoagulation or not). Diabetic nephropathy was diagnosed using the following criteria: (1) urinary excretion of albumin per gram (urinary albumin index; UAI) ≥ 30 mg/g and (2) the presence of retinopathy, which indicates diabetic microangiopathy. It was classified as microalbuminuria (UAI, 30-299 mg/g), overt proteinuria (UAI ≥ 300 mg/g), and chronic renal failure. Diabetic patients with UAI ≤ 20 mg/g were classified as “no nephropathy.” Those with UAI between 21 and 29 mg/g were assumed to be in the “gray zone” and were excluded from the study. Those with UAI ≥ 30 mg/g but without retinopathy were also excluded from the study because proteinuria is observed in diabetic patients with reasons other than diabetic nephropathy.26 Clinical data, including age at diagnosis of diabetes, disease duration after diagnosis of diabetes, body mass index (BMI) at diagnosis of diabetes, hemoglobin (Hb) A_1C, antihypertensive drug treatment, and hyperlipidemia were collected from medical records of the patients.

Genomic DNA was isolated from peripheral blood leukocytes in 227 patients with type II diabetes mellitus and 169 normal subjects, as described previously.27 For genotyping of the Bgl II polymorphism, a 600-bp DNA fragment that contains the Bgl II (+/−) polymorphism was amplified using polymerase chain reaction (PCR) with a DNA thermal cycler (Perkin Elmer, Takara Biomedicals, Chiba, Japan). Briefly, the reaction was performed in a final volume of 100 μL containing 1 μg genomic DNA, 10 mM Tris (pH 8.3), 50 mM KCl, 1.5 mM MgCl₂, 0.001% gelatin, 0.2 mM each of deoxynucleotide triphosphate, and 2.5 U Taq polymerase.

The 2 oligonucleotide primers7 were 5′-GATTTAACTTTCCCGACTGCCTTC (nucleotide number 2789-2812) and 5′-CATAGGTTTTTGGGGAACAGGTGG (nucleotide number 3346-3369). PCR conditions of 34 cycles in this study were as follows: the first 2 cycles were 94°C for 1 minute, 69°C for 1 minute, and 72°C for 1 minute; the second 2 cycles were 94°C for 1 minute, 67°C for 1 minute, and 72°C for 1 minute; the remaining 30 cycles were 94°C for 1 minute, 65°C for 1 minute, and 72°C for 1 minute. Amplified DNA was digested with a restriction enzyme Bgl II (Takara Shuzo, Osaka, Japan) at 37°C for 3 hours, followed by an electrophoresis on a 2% agarose gel. The PCR products containing Bgl II (+) would be cut into visible fragments of 200 bp and 400 bp, whereas those containing Bgl II (−) would not be cut.

Genotyping of the PLA1/A2 polymorphism was performed essentially as described.24 

Student's t test was used to compare age at diagnosis of diabetes, disease duration after diagnosis of diabetes, BMI at diagnosis of diabetes, and HbA_1C at diagnosis of diabetes between subgroups of patients. Analyses of genotype frequency counts were performed using the chi-square test. Multiple logistic regression analysis was performed to evaluate the relationship between those with versus those without retinopathy (categorical variable, yes or no) and other variables. Independent variables included in the analysis were Bgl II genotype (categorical variable, +/+, +/−, −/−), sex (categorical variable, male or female), age at diagnosis of diabetes (quantitative variable), disease duration from diagnosis of diabetes (quantitative variable), and BMI at diagnosis of diabetes (quantitative variable). Analysis of variance (ANOVA) was used for the comparison of the mean values of current HbA_1Clevels among the 3 Bgl II genotypes. All statistical analyses were performed using Statview (version 5.0, for Macintosh, Abacus Concepts, Berkeley, CA). Variability in sampling associated with the estimated odds ratio (OR) was assessed by 2-sided 95% confidence intervals (CI). An OR (95% CI) greater than 1 was considered to be significant. AP value less than 0.05 was considered to be statistically significant.

To explore the clinical significance of the Bgl II polymorphism, we performed a cross-sectional study comparing diabetic patients with (n = 119) and without (n = 108 retinopathy. Genotyping was performed also on 169 nondiabetic subjects. The clinical characteristics of 227 patients are shown in Table 1. The 2 groups were well matched with regard to gender distribution, age at diagnosis of diabetes, disease duration after diagnosis of diabetes, HbA_1C at diagnosis of diabetes, and BMI at diagnosis of diabetes.

Table 2 shows Bgl II genotype distribution in patients and normal subjects. The genotype distribution in each group shown in Table 2 was in Hardy-Weinberg equilibrium. The frequency of the Bgl II (+/+, +/−) genotype in group B plus C patients (69.7%) was significantly higher than that in group A patients (55.6%, OR [95% CI] = 1.84 [1.07-3.16], P = .0270). The frequency of Bgl II (+/+, +/−) genotype in group C patients (71.4%) was significantly higher than in group A plus B patients (58.0%, OR = 1.81 [1.02-3.22], P = .0437), or in group A patients (55.6%, OR = 2.00 [1.10-3.65], P = .0242). When the analysis was confined to those patients with a disease duration of 10 years or more, the frequency of Bgl II (+/+, +/−) genotype was 72.4% for group B plus C patients, 55.1% for group A patients, 73.9% for group C patients, and 58.1% for group A plus B patients (OR = 2.14 [1.20-3.82], P = .0102, group B plus C versus group A patients; OR = 2.04 [1.06-3.94], P = .0262, group C versus group A plus B patients; OR = 2.31 [1.20-4.44],P = .0120, group C versus group A patients). These observations suggest that patients with the Bgl II (+/+, +/−) genotype have an increased risk for diabetic retinopathy or nephropathy. For normal subjects, the frequency of the Bgl II (+/+, +/−) genotype was 65.7% and the frequency of the Bgl II (−/−) genotype was 34.3%.

Next, the dose effect of the Bgl II (+) allele on the risk for retinopathy or nephropathy was analyzed in diabetic patients with a disease duration of 10 years or more. It was shown that the greater the number of the Bgl II (+) allele, the greater the risk for diabetic retinopathy or nephropathy. For retinopathy, the frequencies of Bgl II (+/+) genotype were 23.5% and 11.2% for group B plus C and group A patients, respectively, and the frequencies of Bgl II (+/−) genotype were 48.9% and 43.9% for group B plus C and group A patients, respectively. The ORs (95% CI) to Bgl II (−/−) were 1.82 (0.98-3.38, P = .0580) and 3.41 (1.49-7.78, P = .0036) for Bgl II (+/−) and Bgl II (+/+) genotypes, respectively (see Table 2). For nephropathy, the frequencies of Bgl II (+/+) genotype were 21.7% and 14.7% for group C and group A plus B patients, respectively, and the frequencies of Bgl II (+/−) genotype were 52.2% and 43.4% for group C and group A plus B patients, respectively. The ORs (95% CI) to Bgl II (−/−) were 1.93 (0.99-3.76, P = .0533) and 2.38 1.02 –5.54, P = .0444) for Bgl II (+/−) and Bgl II (+/+) genotypes, respectively (see Table 2).

Increased frequency of Bgl II (+)-containing allele was observed in all stages of diabetic microangiopathies, that is, genotype distributions did not differ significantly among different stages of retinopathy and nephropathy (data not shown).

As shown in Table 3, a multiple logistic regression model with a dependent variable (presence or absence of retinopathy) and several independent variables (Bgl II genotype, sex, age at diagnosis of diabetes, disease duration after diagnosis of diabetes, and BMI at diagnosis of diabetes) had an adjusted OR of 1.515 (95% CI, 1.020-2.251, P = .0397) for the relation between retinopathy and the presence of the Bgl (+) allele, suggesting that the Bgl II genotype is an independent predictor of retinopathy. To examine whether the relation of the Bgl II genotype to nephropathy shown in Table 2 depended on the relationship between the genotype and retinopathy, we analyzed the genotype distribution only in patients with retinopathy, comparing those with and without nephropathy. The genotype frequency for Bgl II (+/+, +/−) was 71.4% and 65.7% for group C and group B, respectively, which were not significantly different (P > .05). Thus, the Bgl II genotype is associated with the prevalence of retinopathy independent of the other variables, but it is not independently associated with the prevalence of nephropathy.

Current disease status of patients in relation to the Bgl II genotype is shown in Table 4. No significant difference was observed among 3 Bgl II genotypes with regard to HbA_1C, prevalence of antihypertensive drug treatment, or hyperlipidemia, either in group A patients or group B plus C patients although the mean of HbA_1C and the prevalence of antihypertensive drug treatment in group B plus C patients were higher than those in group A patients.

We next analyzed the genotype distribution of the PLA1/A2 polymorphism among 154 diabetic patients (82 with retinopathy, 72 without retinopathy). However, we found no subject with the PLA1/A2 or PLA2/A2 genotypes, that is, all patients had the PLA1/A1 genotype. This result is consistent with previous reports that the PLA2 allele is rare in Japan.28 29 

Several epidemiologic and experimental studies indicate that disease duration of diabetes and glycemic control are major determinants for the development of diabetic retinopathy or nephropathy.30-32 In the presence of prolonged hyperglycemia, alteration in retinal or renal blood flow, metabolic changes, hemostatic abnormality, or nonenzymatic glycosylation of long-lived tissue proteins are observed.14,33-36 These changes are associated with vascular dysfunction in microcirculation,13 36-40 which are thought to contribute to the occurrence or progression of diabetic retinopathy and nephropathy.

Platelets might be involved in the occurrence of diabetic retinopathy and nephropathy. Hyperreactive platelets in diabetic patients would be more likely to interact with an exposed subendothelium of damaged vessels and enhance microthrombus formation or small vessel occlusion in vivo. This, in turn, might alter retinal or renal blood flow. Moreover, evidence indicating the beneficial effect of antiplatelet therapy18-21 on retinopathy and nephropathy suggests the involvement of platelets in the pathogenesis of microangiopathy. The focus in the present study was on platelet adhesion, which is a first critical step for primary thrombus formation and leads to intracellular activation processes.

This study shows that the Bgl II polymorphism is associated with the prevalence of retinopathy among patients with type II diabetes mellitus. The Bgl II (+/+, +/−) genotypes increased the risk of retinopathy and nephropathy, but the association between the Bgl II genotype and nephropathy was not independent of retinopathy. However, because the majority of diabetic patients with retinopathy suffer from concomitant nephropathy, the Bgl II genotype could be also regarded as a predictor of nephropathy. The Bgl II polymorphism has been reportedly associated with platelet α2β1 density, the extent of platelet adhesion to collagen, and the prevalence of myocardial infarction or stroke.7,8 10-12 This study is the first to demonstrate an association between the Bgl II polymorphism and diabetic retinopathy.

The amount of nonenzymatically glycosylated collagen, which more easily interacted with platelets, was higher in diabetic patients than in nondiabetic controls.33 41 It is possible that Bgl II (+)-containing platelets can more easily interact with nonenzymatically glycosylated collagen and accelerate the occurrence of retinopathy. In addition, our findings might affect antiplatelet drug treatment for diabetic retinopathy. For instance, patients with the Bgl II (+/−, +/+) genotype might benefit more from antiplatelet therapies. Although the role for α2β1 integrin in the development of diabetic retinopathy is not fully understood, our data suggest a contribution of platelets in the development of diabetic retinopathy.

The PLA1/A2 polymorphism has recently been highlighted because of its association with thrombotic disorders.42 The PLA2 allele, however, was reported to be extremely rare in Japanese populations.28 29 In agreement with these reports, we did not observe the genotypes with the PLA2 allele among 154 diabetic patients. Therefore, it remains unclear whether the PLA1/A2 polymorphism might affect the susceptibility to microangiopathy in patients with type II diabetes mellitus.

In conclusion, Bgl II polymorphism of the α subunit of α2β1 integrin is associated with the prevalence of retinopathy in patients with type II diabetes mellitus. The present study will aid in early identification of risk or possibly prevention and will direct studies on more beneficial uses of antiplatelet drugs for diabetic microangiopathy.

We thank Dr Thomas J. Kunicki, Scripps Research Institute, CA, for providing the detailed protocol on Bgl II genotyping. We also thank Dr Koichi Kawano and Dr Nobuo Aoki, Kyorin Universtiy, Tokyo, Japan, for providing samples from healthy subjects.