Adipokines and Risk of Type 2 Diabetes in Older Men

RESEARCH DESIGN AND METHODS—

The British Regional Heart Study is a prospective study of cardiovascular disease involving 7,735 men aged 40–59 years who were selected from the age-sex registers of one general practice in each of 24 British towns and who were screened between 1978 and 1980 (20). In 1998–2000, all surviving men, now aged 60–79 years, were invited for a 20th-year follow-up examination. Ethics approval was provided by all relevant local research ethics committees. All men provided informed written consent to the investigation, which was carried out in accordance with the Declaration of Helsinki, and completed a questionnaire (Q20) that included questions on medical history and lifestyle behavior. The men were asked to fast for a minimum of 6 h, during which time they were instructed to drink only water and to attend for measurement at a prespecified time between 0800 and 1800 h. They then provided a blood sample, collected using the Sarstedt Monovette system. The samples were stored at −20°C on the day of collection and transferred in batches for storage at −70°C until analysis, carried out after no more than one freeze-thaw cycle. A total of 4,252 men (77% of survivors) attended for examination; 4,086 men had at least one measurement of the adipocyte variables (leptin, adiponectin, or IL-6). We excluded 487 men with a doctor's diagnosis of diabetes or those with a fasting glucose ≥7 mmol/l (World Health Organization criteria) who were considered to have prevalent diabetes. A total of 3,599 men were then available for analysis.

Adipokine measurements

Plasma leptin was measured by an in-house radioimmunoassay validated thoroughly against the commercially available Linco assay, as previously described (21). The intra- and interassay coefficients of variation (CVs) were <7 and <10%, respectively, over the sample concentration range. The detection limit of the assay was 0.5 ng/ml. Plasma adiponectin concentrations were determined using enzyme-linked immunosorbent assay (ELISA) (R&D systems, Oxford, U.K.), and the intra- and interassay CVs were each <7.5%. IL-6 was assayed using a high-sensitivity ELISA (R&D Systems). The intra- and interassay CVs were 7.5 and 8.9%, respectively. There is no evidence that the adipokines measured in the present study are influenced by prolonged storage or repeat free-thawing of samples. Of the 3,599 men with at least one measure of adipokines, 32 men had missing data on adiponectin, 33 on IL-6, and 192 on leptin. More men had missing leptin levels since this assay required a larger sample volume.

Cardiovascular risk factors

Weight, height, and waist circumference were measured. BMI (weight divided by the square of height in meters [kg/m²]) was calculated for each man at reexamination. Details of questionnaire assessment and classification of smoking status, physical activity, social class, alcohol intake (20,22), and the measurement of blood pressure and blood lipids in this cohort have been described elsewhere (20,23,24). Men were asked to recall a doctor diagnosis of coronary heart disease (myocardial infarction or angina), stroke, and diabetes. Plasma glucose was measured by a glucose oxidase method using a Falcor 600 automated analyzer. Serum insulin was measured using an ELISA that does not cross-react with proinsulin (25). Triglycerides, blood glucose, and insulin concentrations were adjusted for the effects of fasting duration and time of day (24). Insulin resistance was estimated according to the homeostasis model assessment of insulin resistance (HOMA-IR) (the product of fasting glucose [mmol/l] and insulin [units/ml] divided by the constant 22.5) (26). C-reactive protein (CRP) was assayed by ultra-sensitive nephelometry (Dade Behring, Milton Keynes, U.K.).

Follow-up

All men have been followed up for all-cause mortality, cardiovascular morbidity, and development of a diagnosis of type 2 diabetes from initial examination to June 2004 (27), and follow-up has been achieved for 99% of the cohort. This analysis is based on follow-up from rescreening in 1998–2000, with a mean follow-up period of 5 years (4–6 years). Information on death was collected through the established tagging procedures provided by the National Health Service central registers. Information on new cases of diabetes was obtained by regular biennial reviews of the patients’ notes (including hospital and clinic correspondence) through to the end of the study period and from repeated personal questionnaires to surviving subjects after initial examination. Cases are based on self-reported diagnoses confirmed by primary care records, an approach that has been validated in the present study (28)

Statistical methods

The distributions of adiponectin, leptin, and IL-6 were skewed, log transformation was used, and geometric means and interquartile ranges were presented. The men were divided into three equal thirds on the basis of adiponectin, leptin, and IL-6 distributions. Cox proportional hazards model was used to assess the multivariate-adjusted relative risk for each third compared with the reference group (lowest third). Person-years was used to calculate the incidence of diabetes, with men censored at time of death. In the adjustment, smoking (categorized as never, long-term ex-smokers [>15 years], recent ex-smokers [<15 years], and current smokers), social class (manual workers, nonmanual workers, or Armed Forces), physical activity (four groups), alcohol intake (five groups), preexisting coronary heart disease (CHD) (yes/no), stroke (yes/no), use of statins (yes/no), and treatment for hypertension (yes/no) were fitted as categorical variables. Inactive men included those who reported no physical activity or who were only occasionally active (22). BMI, HOMA-IR, systolic blood pressure, HDL cholesterol, and CRP were fitted as continuous variables. Tests for interaction were carried out by adding an interaction term (obesity × adipokine variable) to the regression model with the adipokine groups fitted continuously (1–3). All analyses were carried out using SAS (version 8.2; SAS Institute, Cary, NC).

RESULTS

Baseline characteristics of incident case and control subjects

During the mean follow-up period of 5 years, there were 108 incident diabetes cases (rate 6.0/1,000 person-years). Table 1 shows the baseline characteristics in the men who developed diabetes and in those who remained free of diabetes. Men who developed diabetes had higher BMI and waist circumference measurements and were more likely to be physically inactive and to have a higher prevalence of CHD. They had significantly higher mean levels of metabolic risk factors, CRP, leptin, and IL-6 and lower levels of adiponectin.

Association of adipokines with incident diabetes

Table 2 shows the correlations between the adipokine measures and age, BMI, waist circumference, metabolic risk factors, and CRP. The incidence rates and adjusted relative risks (RRs) of type 2 diabetes by tertiles of the adipocyte markers, using those in the lowest third as the reference group, are shown in Table 3. Adiponectin (inversely) and leptin and IL-6 (positively) were significantly predictive of type 2 diabetes even after adjustment for age, social class, physical activity, smoking status, alcohol intake, preexisting CHD, stroke, use of statins, treatment of hypertension, systolic blood pressure, and BMI. We repeated the analyses in Table 3, adjusting for waist circumference instead of BMI; similar results were obtained.

Further adjustment for HOMA-IR attenuated the relationship between adiponectin and diabetes and abolished the association between leptin and diabetes but made little difference in the association between IL-6 and diabetes. The relationship between IL-6 and incident diabetes remained significant even after further adjustment for HDL cholesterol and CRP (Table 3). Since adiponectin and IL-6 were not correlated, further adjustment for each other made little difference to the associations seen (adjusted RR 2.01 [95% CI 1.06–3.81] for IL6 and 0.63 [0.35–1.11] for adiponectin [top vs. bottom third]). The IL-6–to–adiponectin ratio showed similar magnitude of association, as seen for IL-6 (2.20 [1.22–3.97]) after adjustment for HOMA-IR.

Adipokines, obesity, and diabetes

We examined the relationships between the adipokine variables and risk of diabetes separately in obese men (waist circumference >102 cm or BMI ≥30 kg/m²) and nonobese men (Table 4). High adiponectin was associated with significantly decreased risk of diabetes in obese men, but the benefit was less apparent in nonobese men (test for interaction P = 0.04). This decreased risk in obese men was seen even after adjustment for HOMA-IR; no association was seen in nonobese men. By contrast, no interaction was seen between obesity, IL-6, and risk of diabetes. Elevated IL-6 was associated with increased risk of diabetes in both groups of men even after adjustment for HOMA-IR (Table 4). Further adjustment for CRP and HDL cholesterol made little difference on the associations seen in Table 4. The associations seen for adiponectin in obese men persisted even after further adjustment for IL-6. Leptin showed no association after adjustment for insulin in either group.

CONCLUSIONS—

In this large prospective study of men aged 60–79 years at baseline, we have shown that low adiponectin and high leptin levels associate with a higher risk of incident type 2 diabetes independent of age, obesity, and a comprehensive range of other potential confounders or explanatory factors. However, further adjustment for HOMA-IR attenuated these relationships, especially for leptin. By contrast, elevated IL-6 levels remain significantly and independently associated with incident diabetes following additional adjustment for HOMA-IR. Perhaps more importantly, our results suggest that the relationship between low adiponectin and incident diabetes is potentially dependent on baseline adiposity levels.

These results have several potential implications. First, the novel observation of a significant impact of obesity on the relationship between adiponectin and subsequent diabetes may have clinical relevance. Recent data from the DREAM (Diabetes Reduction Assessment with Ramipril and Rosiglitazone Medication) Study (19) indicate that rosiglitazone reduces the risk of progression to diabetes most in the at-risk subjects who had a higher baseline BMI. The mechanism of action of glitazones is not fully elucidated but may include enhancing both β-cell function and adiponectin synthesis. Adiponectin promotes hepatic fatty acid oxidation and reduces hepatic fat. As found in the DREAM Study, a significant reduction in alanine aminotransferase concentrations implies a reduction in liver fat (19). One may therefore speculate that a stronger link of low adiponectin to subsequent diabetes in obese subjects could partly explain a greater RR reduction in diabetes seen with glitazones in obese subjects (19). Interestingly, rimonabant, a selective cannabinoid-1 receptor (CB1) blocker, also increases adiponectin (29) and improves glycemic control (30).

As regards leptin's link with diabetes, recent nested case-control work from the ARIC (Atherosclerosis Risk in Communities) Study (including both men and women and different ethnicities) suggested that following adjustment for age, sex, ethnicity, obesity indexes, fasting insulin, inflammation score, hypertension, triglycerides, and adiponectin, higher leptin was significantly associated with a 40% lower risk of subsequent diabetes (12). The authors suggested that their findings were commensurate with a possible protective effect of leptin against diabetes. By contrast, higher leptin levels predicted higher risk of diabetes in Japanese men but not women (11), although the number of cases of diabetes in that study were small (23 men and 17 women), and other studies showed no independent association (10). Our results, generated in a more homogenous population of predominantly white men but adjusted for similar range of confounders, did not confirm a protective effect of leptin on diabetes risk. Rather, much of the association between leptin and subsequent diabetes could be accounted for by obesity and insulin resistance (31). Further prospective studies are needed to disentangle the relationship between leptin and subsequent diabetes.

Interestingly, high IL-6 was independently associated with subsequent diabetes, even with further adjustment for insulin resistance or CRP. Although high IL-6 has been previously associated with subsequent diabetes (13–17), few studies have assessed the role of insulin resistance as a potential confounder, and, in the one other study that has, IL-6 has also been shown to be related to diabetes independent of insulin (15). This is important, since low-grade systemic inflammation could mediate higher diabetes risk via insulin resistance (32), but our observation suggests that IL-6 may associate with diabetes via alternative mechanisms; at present, there is no evidence for an independent role of IL-6 in impaired β-cell function/apoptosis. As such, our work adds to the literature on the potential links between IL-6 and diabetes. A recent comprehensive review that took account of a range of in vitro, in vivo, and genetic studies, together with studies on diverse tissues such as liver, muscle, adipose tissue, and pancreas, concluded that chronically elevated IL-6 may indeed contribute to development of type 2 diabetes via mechanisms including altered insulin signaling in hepatocytes/adipocytes and effects on the central nervous system to impair energy regulation (33). High IL-6 may also drive hepatic fatty acid synthesis and cause endothelial dysfunction (34). Indeed, high IL-6 is consistently linked to a range of metabolic abnormalities typical of an insulin-resistant state (34). Finally, that IL-6 was linked to incident diabetes independently of CRP is noteworthy, particularly since CRP may not be causally linked to insulin resistance and related features (35). Such observations direct attention toward upstream cytokines, such as IL-6. In terms of causality, an IL-6 receptor antagonist (Toclizumab) (36) may offer the potential to directly block this pathway and determine metabolic effects in obese individuals.

Our study is not without some limitations. Our study was carried out in an older, predominantly white, male population, and we cannot generalize our findings to women, younger men, or other ethnic groups. Moreover, the numbers developing diabetes were modest and greater follow-up time would enhance power and narrow CIs. We also acknowledge that adiponectin association with other risk factors appeared weaker than in other studies, but we feel this may be related to the more advanced age of our cohort since adiponectin rises with age. It is also relevant that we measured total adiponectin; the high–molecular weight adiponectin fraction, which is as yet not easily measured in such large numbers, may be more strongly associated with incident diabetes, but this requires further study. Finally, we acknowledge that we did not measure other cytokines, such as tumor necrosis factor-α, that may have relevance to diabetes pathogenesis and thus prediction.

On the basis of our prospective cohort study, the findings indicate that 1) the association between leptin and diabetes is largely explained by obesity and insulin resistance, 2) the association between IL-6 and diabetes remains independent of obesity and insulin resistance, and 3) the adiponectin-diabetes association is mediated to some extent by insulin resistance and appears to be stronger in obese individuals.