Introduction

Diabetic microangiopathy, specifically retinopathy and nephropathy, is a hallmark of type 1 diabetes. Diabetic retinopathy is the leading cause of sight-threatening diseases in patients of working age [1], while diabetic nephropathy is the leading cause of renal failure in France and other industrialised countries [2, 3]. Nephropathy dramatically increases the risk of premature death [4] and cardiovascular morbidity [5] and mortality [4] associated with type 1 diabetes.

Prospective studies have recently shown that insulin resistance is significantly associated with the risk of microvascular complications in type 1 diabetic patients. Furthermore, parental insulin resistance is associated with diabetic nephropathy, which strongly suggests that insulin resistance plays a role in the development of microangiopathy in type 1 diabetes patients [6, 7].

Adiponectin is a hormone that is secreted by adipocytes to regulate glucose and lipid metabolism [8]. The finding that adiponectin infusion reverses insulin resistance in rodent models suggests that adiponectin plays a role in this condition [9]. Adiponectin concentrations are abnormally low in patients with insulin resistance, such as those with present or incident cases of type 2 diabetes [10], and in patients with macrovascular disease [11].

Based on these findings, the aim of the present study was to investigate the relationships between adiponectin and renal and retinal diabetic complications in patients with type 1 diabetes.

Subjects, materials and methods

Patient selection and follow-up

We studied 126 normoalbuminuric patients, aged 25 years or over, who were among the 310 type 1 diabetic subjects included in the Survival Genetic Nephropathy (SURGENE) study, a prospective observational study [12] that was designed to analyse the determinants of diabetic nephropathy. Type 1 diabetes was diagnosed on the basis of age at disease onset <40 years and requirement for permanent insulin therapy within 1 year. All patients had a duration of diabetes of over 3 years and had been followed-up for over 2 years. None of the patients displayed any evidence of macrovascular disease at baseline.

The diabetic patients were compared with non-diabetic age- and sex-matched subjects who were selected from the general population to be included in the DESIR (Donnees Epidemiologiques sur le Syndrome d’Insulino-Resistance) study [13]. Each patient was compared with two non-diabetic control subjects. All patients gave written informed consent and the study protocol was approved by local ethics committees.

At baseline, retinopathy was classified by a trained ophthalmologist according to a clinical classification system, using direct fundoscopy after pupillary dilatation and/or fluorescenic angiography. Retinopathy was classified as severe (pre-proliferative retinopathy requiring laser pan-photocoagulation, or past or present proliferative retinopathy identified by retinal new vessels), mild (background retinopathy, defined as micro-aneurysms and/or exudates and/or micro-hemorrhages) or absent (no evidence of retinopathy).

Patients were examined every 4–6 months for a median period of 6 years (range 2–9 years). Blood pressure was recorded using an automatic device, and HbA1c and urinary albumin concentrations were determined. Incident persistent microalbuminuria was diagnosed when the urinary albumin concentration of two or more of three consecutive random sterile samples exceeded 20 mg/l, as validated previously [14].

Laboratory measurements

Urinary albumin concentration was determined using a nephelometric method [15]. Fasting HDL cholesterol was determined after phosphotungstic precipitation. HbA1c was determined by HPLC (Biorad Diamat; Biorad, Ivry Sur Seine, France). Serum creatinine was determined using a modification of Jaffes’s method; this was used to estimate creatinine clearance according to the Cockcroft formula [16].

Plasma adiponectin was determined in duplicate with a RIA (Linco, St Charles, MO, USA) that uses 125I-labelled murine adiponectin and multispecies adiponectin rabbit antiserum to determine plasma adiponectin levels by the double-antibody/polyethylene glycol technique. Adiponectin standards were prepared using recombinant human adiponectin. The sensitivity of this assay is 1 ng/ml (100 μl sample size) and the intra-assay and interassay CVs are 4.4 and 9.9%, respectively. The limit of linearity of the assay is 200 ng/ml (100 μl sample size). Plasma samples were diluted 1:500 in assay buffer prior to use. Samples were run in duplicate, and the means of the duplicate values were considered for statistical analyses. According to manufacturer, this RIA kit measures the trimer, hexamer, and high-molecular-weight forms, as well as some cross-reactivity to the globular domain. It does not recognize the monomer.

Statistical methods

All data were stored and analysed using the STATVIEW V program (SAS Institute, Cary, NC, USA). Data are presented as means±SD, or as medians (ranges) if the distributions were skewed. Groups were compared using the chi square test for categorical variables, and parametric (if normally distributed; ANOVA or Student’s t-test) or non-parametric tests (if not normally distributed; Mann–Whitney U-test) for continuous variables. The non-parametric Spearman’s rank test was used to analyse the relationship between continuous variables. Multivariate logistic regression analysis was used to investigate the relationship between several candidate variables and the outcome variable (severity of retinopathy or incident microalbuminuria). An additional survival analysis using a Cox model was used to examine the contribution of adiponectin levels, with time to development of microalbuminuria as the outcome, taking potential confounding variables into account. Results are expressed as adjusted hazard ratios (HRs) with 95% confidence intervals. A p value less than 0.05 was considered significant.

Results

Baseline characteristics

The baseline characteristics of the type 1 diabetic patients and matched control subjects are shown in Table 1. Plasma adiponectin concentrations were significantly higher in women than in men, both in the diabetic and non-diabetic groups (p<0.0001). Adiponectin concentration was correlated with age among the type 1 diabetic patients (rho=0.238, p=0.0079), but not among the control subjects (rho=0.07, NS). Conversely, adiponectin concentration was inversely correlated with BMI among the control subjects (rho=−0.165, p=0.0023), but not among the type 1 diabetes patients (rho=−0.07, NS). We found no correlation between adiponectin and HbA1c concentrations in diabetic patients (rho=0.049, NS). Plasma adiponectin concentration was positively correlated with HDL cholesterol concentration among the type 1 diabetic patients (rho=0.222, p=0.0148) and the non-diabetic control subjects (rho=0.355, p<0.0001). Adiponectin concentration was correlated with estimated creatinine clearance among the control subjects (rho=0.132, p=0.0409), but not among the type 1 diabetic patients (rho=0.006, NS).

Table 1 Comparison of type 1 diabetic patients and control subjects
Full size table

The clinical and biological characteristics of the patients stratified according to diabetic retinopathy stage are summarised in Table 2. Adiponectin concentration was significantly higher in patients with severe diabetic retinopathy than in those with mild or no diabetic retinopathy (p=0.0005). Logistic regression analysis showed that adiponectin concentration (p=0.0032) and diabetes duration (p=0.0001) were independently associated with severe retinopathy, whereas sex, BMI and systolic blood pressure were not.

Table 2 Characteristics of type 1 diabetic patients according to diabetic retinopathy
Full size table

Follow-up study

Eighteen patients developed persistent microalbuminuria during the follow-up period (mean 6 years, range 2–9 years). The patients who developed microalbuminuria excreted more albumin in their urine and had higher adiponectin concentrations than the other patients, although these differences were statistically significant only in men (Table 3). We also found that microalbuminuric patients were more likely to be men. Logistic regression analysis showed that adiponectin (p=0.0158), sex (p=0.0054) and baseline urinary albumin concentration (p=0.004) were independently associated with incident nephropathy, whereas diabetes duration, BMI, mean HbA1c during follow-up and systolic blood pressure were not. A Cox model survival analysis revealed that time to development of microalbuminuria was independently and significantly determined by adiponectin concentration (adjusted HR 1.06, 95% CI 1.02–1.10, p=0.0031 for each increase of 1 μg/ml). Other contributing variables were sex (adjusted HR 12.05, 95% CI 2.45–62.5, p=0.0023 for men) and urinary albumin excretion (adjusted HR 1.19, 95% CI 1.06–1.33, p=0.0022 for each increase of 1 mg/l). Diabetes duration (p=0.660), BMI (p=0.856), HbA1c (p=0.305), and systolic BP (p=0.394) were not associated with time to development of microalbuminuria.

Table 3 Characteristics of type 1 diabetic patients according to the occurrence of microalbuminuria during follow-up
Full size table

A second plasma sample for the duplicate determination of adiponectin concentration was available for 38 patients (not different from the whole group) over a period of 3.2 years (−1 to 8 years). The adiponectin concentrations measured in the duplicate samples (−2.4±35.1%) were not different from baseline values; this was also the case in the nine patients who developed microalbuminuria (0.9±45.1%) and the 29 who did not (−2.9±31.9%).

Discussion

We found that the type 1 diabetic patients had higher plasma adiponectin concentrations than the sex- and age-matched healthy control subjects. According to our baseline data, plasma adiponectin concentrations were independently higher in type 1 diabetic patients with severe retinopathy than in those with no or mild retinopathy. The results of our prospective study confirmed that elevated adiponectin was independently associated with the development of microalbuminuria in type 1 diabetic patients. This is the first report to show that an elevated concentration of adiponectin is associated with diabetic microvascular disease.

Our study was associated with certain limitations. First, it was primarily designed to study renal complications, and our data on retinal disease could not be examined using a follow-up design. Second, this study was carried out in a single centre and is of limited size, though the risk factors for diabetic retinopathy (mainly diabetes duration) and incident microalbuminuria (baseline urinary albumin excretion) were consistent with data from large-scale studies [17, 18].

Our results confirm recent reports that plasma adiponectin levels are higher in type 1 diabetic patients than in healthy control subjects [17–19]. In our study this difference was not due to differences in age or sex, as patients and control subjects were carefully matched. It is unlikely to be due to differences in body composition, as it persisted after adjustment for BMI.

Plasma adiponectin levels have repeatedly been shown to be lower in patients with insulin resistance or macrovascular complications than in control subjects [8, 10, 11]. We found a positive association between HDL cholesterol and plasma adiponectin levels. This reflects their association with insulin resistance in type 1 diabetic patients and in non-diabetic control subjects, and is consistent with many other observations [19]. As insulin resistance has been shown to be related to diabetic retinopathy and nephropathy in prospective [20, 21] and family-based studies [6, 7], we expected patients with severe retinopathy and incident microalbuminuria to have relatively lower levels of adiponectin. Furthermore, recent reports suggest a role for low-grade inflammation in diabetic microangiopathy, particularly that caused by IL-6 [22, 23]. As IL-6 levels are inversely correlated with adiponectin levels [24], we expected to find an association between low plasma adiponectin concentrations and microvascular disease; however, our results revealed the opposite. This finding is probably not due to differences in body composition, since subjects with severe retinopathy and those with mild or no retinopathy had similar BMIs, as did patients with and without incident microalbuminuria. It is also unlikely to be due to differences in metabolic control, since we found no correlation between HbA1c and plasma adiponectin concentrations.

Our follow-up observational study showed that higher adiponectin concentrations are associated with an increased risk of incident microvascular complications. This finding is opposite to that reported for the association between adiponectin and macrovascular complications [11]. It is therefore likely that adiponectin is regulated differently in type 1 diabetes than it is in type 2 diabetes.

The relationship between diabetic microangiopathy and adiponectin concentration has yet to be fully established, though it may be related to chronic low-grade inflammation. Unfortunately, neither highly sensitive C-reactive protein levels nor IL-6 levels were assessed in this study. Inflammation markers (possibly involved in the development of nephropathy) have previously been shown to be associated with low adiponectin concentrations in non-diabetic and type 2 diabetic subjects [25]; however, these results are at complete variance with those of the present study. Interestingly, adiponectin can stimulate angiogenesis both in vitro and in vivo [26, 27]. Thus, the increase in angiogenesis caused by the increase in adiponectin could lead to proliferative retinopathy, which would explain our results. However, we have no data to support this hypothesis in terms of the renal circulation.

The link between adiponectin and the development of microalbuminuria is currently unknown. As adiponectin levels were recently shown to be correlated with CD146 levels in healthy volunteers, it can be proposed that the increase in adiponectin promotes alterations in the endothelial junctions [28], possibly leading to leakage of albumin into the urine. Another (and not exclusive) mechanism that has been proposed is based on the fact that adiponectin induces endothelial nitric oxide production in vitro [29]. The haemodynamic hypothesis of diabetic microangiopathy emphasises the importance of increases in capillary pressure and flow as key factors for the development of complications [30]. Enhanced endothelial nitric oxide synthase (eNOS) activity has been suggested to mediate renal hyperfiltration [31]. Indeed, chronic NOS inhibition abolishes glomerular hyperfiltration [32], and the afferent arterioles of streptozotocin diabetic rats show high levels of eNOS expression [33]. This could be particularly relevant to our study, as we used the development of microalbuminuria as the study endpoint. In this regard, renal haemodynamic studies could help to determine whether this hypothesis is relevant to the association we observed between adiponectin and incident microalbuminuria. Large-scale studies are needed to answer this question.