Abstract
Context: Adiponectin is a hormone secreted by adipocytes that acts as an antidiabetic adipokine. Adiponectin exists as multimers in plasma, and high molecular weight (HMW) adiponectin is particularly thought to be the active form of the protein.
Objective: The aim of the study was to assess whether decreased total and HMW adiponectin are independent risk factors for the development of type 2 diabetes.
Design: Study subjects were Japanese-Americans enrolled in the Hawaii-Los Angeles-Hiroshima study between 1992 and 2002. Duration of follow-up was an average of 5.4 yr.
Participants: We investigated 321 men and 445 women who were nondiabetic Japanese-Americans. Glucose tolerance was evaluated according to 1997 American Diabetes Association criteria, and 112 subjects developed type 2 diabetes during the follow-up period.
Main Outcome Measure: The influence of baseline total and HMW adiponectin on the development of type 2 diabetes was the main outcome measure.
Results: Subjects who developed type 2 diabetes had significantly decreased plasma total and HMW adiponectin compared with those who did not develop the disease (P < 0.001, respectively). In a Cox proportional hazards model, both decreased total and HMW adiponectin levels were independent risk factors for the progression to type 2 diabetes after adjusting for sex, age, body mass index, waist-to-hip ratio, homeostasis model assessment, and classification of 75-g glucose tolerance test (hazards ratio: total, 0.600, P = 0.018; HMW, 0.614, P = 0.001, respectively). Dividing tertiles of adiponectin, hazards ratios in the lowest vs. highest tertile were total, 1.787 (95% confidence interval, 1.006–3.173); and HMW, 2.493 (95% confidence interval, 1.342–4.632), after similar adjustments.
Conclusions: Decreased total adiponectin is an independent risk factor for the progression to type 2 diabetes in Japanese-Americans. Moreover, HMW adiponectin more closely associates with the progression to type 2 diabetes when compared with total adiponectin.
INSULIN RESISTANCE, which is induced by a high-fat diet and is associated with obesity, is a major risk factor for type 2 diabetes. Adipose tissue does not simply act as a storage organ of excess energy, but also secretes a variety of proteins and peptides into circulating blood known as adipokines, which include leptin, TNF-α, plasminogen-activator inhibitor type 1, resistin, and so on (1–5). Adiponectin is also one of the adipokines, and mRNA expression of adiponectin and its plasma level is significantly reduced in obese/diabetic mice and humans (6, 7). In a longitudinal study of monkeys, adiponectin decreased before the onset of diabetes in parallel with a decrease in insulin sensitivity (8). Furthermore, in studies of adiponectin knockout mice, glucose tolerance was significantly attenuated after an oral glucose challenge compared with wild-type mice on a high-fat diet, although differences in glucose tolerance were marginal on a standard chow diet (9–11). Hence, adiponectin-deficient mice were more susceptible to diet-induced insulin resistance. Therefore, in humans, it is thought that one who has decreased serum adiponectin levels is liable to develop type 2 diabetes on a high-fat diet but not on a normal diet.
Recently, it was demonstrated that adiponectin circulates as a trimer called low molecular weight adiponectin, as a hexamer called the medium molecular weight form of adiponectin, and as a high molecular weight (HMW) adiponectin comprising 12–18 protomers (12, 13). The HMW complex is the most active form of adiponectin in depressing blood glucose levels in mice (14). In clinical data, patients with type 2 diabetes and coronary heart disease have a selective reduction in HMW adiponectin (14, 15). Moreover, weight reduction or treatment with the insulin-sensitizing drug known as thiazolidinedione preferentially elevates the HMW adiponectin compared with the other two oligomeric complexes (14, 16, 17). The aforementioned studies suggest that the oligomeric complex distribution of adiponectin is important for the antidiabetic and antiatherogenic activity of this hormone. Therefore, the influence of HMW adiponectin as a risk for type 2 diabetes is thought to be more important than total adiponectin.
Japanese-Americans who are identical to their Japanese progenitors have lived Westernized lifestyles for decades and consumed high-fat and high simple carbohydrate diets (18, 19). We have reported that Japanese-Americans are hyperinsulinemic and insulin resistant compared with native Japanese, and their prevalence of diabetes is two to three times higher than their Japanese counterparts (18). For Japanese-Americans with Westernized lifestyles, the influence of adiponectin on the development of type 2 diabetes is expected to be more apparent than in Japanese. We investigated the influence of decreased HMW adiponectin on the development of type 2 diabetes as well as total adiponectin in Japanese-Americans.
Subjects and Methods
Subjects
We enrolled Japanese-Americans in medical surveys in Hawaii and Los Angeles, California. Initiated in 1970, this survey (Hawaii-Los Angeles-Hiroshima study) is part of a long-term epidemiological study to detect the risk factors for diabetes and cardiovascular disease; the subjects are limited to a population in which both parents were Japanese. This survey was conducted in accordance with the Declaration of Helsinki and was approved by the ethics committee of Hiroshima University School of Medicine. All study participants provided written informed consent.
Study subjects were Japanese-Americans enrolled two or more times in the Hawaii-Los Angeles-Hiroshima study between 1992 and 2002. The study population consisted of 321 men and 445 women who did not have diabetes, as ascertained by a 75-g glucose tolerance test (75gGTT) at baseline. Subjects who were not pure Japanese and had a medical history of gastric resection were excluded. During each follow-up examination, participants who were free of diabetes at baseline had a 75gGTT performed after an overnight fast every 3.44 ± 0.04 yr, individually. Incident diabetes was diagnosed according to 1997 American Diabetes Association criteria [fasting glucose ≥ 7.0 mmol/liter or 2-h glucose ≥ 11.1 mmol/liter after a glucose load of 75 g] (20).
Measurements
Blood was centrifuged, and the obtained serum was immediately frozen. Serum samples were subsequently brought back to Japan. Total adiponectin was measured using an ELISA method (Otsuka Pharmaceuticals, Tokushima, Japan). HMW adiponectin was also measured by ELISA methods, which were available as a kit from Fujirebio Co. (Tokyo, Japan). Nakano et al. (21) described that HMW adiponectin could be measured effectively and specifically by this kit. Insulin [immunoreactive insulin (IRI)] was measured by double-antibody RIA method (Shionogi, Osaka, Japan). Insulin resistance was evaluated with a homeostasis model assessment (HOMA-R) (22). The HMW/total adiponectin ratio was calculated from HMW adiponectin/total adiponectin.
Statistical analyses
Data were expressed as means ± se. Categorized data were analyzed by χ2 test, and unpaired Student’s t tests were used for group comparisons (developed diabetes vs. not developed). A Cox proportional hazards model was used to test the significance of adiponectin levels in predicting the incidence of type 2 diabetes as a dependent variable (1 = developed diabetes, 0 = not developed). Independent variables included sex, age, body mass index (BMI), waist-to-hip ratio (WHR), HOMA-R, and classification of 75gGTT (1 = impaired glucose tolerance, 0 = normal glucose tolerance). In addition, total or HMW adiponectin was included as an independent variable. Total adiponectin concentrations were divided into tertiles based on population. Tertile-specific hazards ratios were estimated through a Cox proportional hazards model. Tertile values of total adiponectin were as follows: tertile 1, 12.32–41.39 μg/ml; tertile 2, 7.61–12.31 μg/ml; and tertile 3, 1.98–7.60 μg/ml. The number of subjects was 255, 256, and 255, respectively. HMW adiponectin instead of total adiponectin was also used in a Cox proportional hazards model. Tertile values of HMW adiponectin were as follows: tertile 1, 8.15–27.89 μg/ml; tertile 2, 4.35–8.14 μg/ml; and tertile 3, 0.42–4.33 μg/ml. The number of subjects was 255, 255, and 256, respectively. For all analyses, SAS package version 8.2 (SAS, Cary, NC) was used.
Results
Fifty-nine of 321 men and 53 of 445 women developed type 2 diabetes during the follow-up period (5.44 ± 0.09 yr). Baseline characteristics of subjects who developed diabetes, or who did not, are shown in Table 1. The proportion of men to women, BMI, WHR, the prevalence of impaired glucose tolerance, fasting glucose, 2-h glucose, fasting IRI, 2-h IRI, HOMA-R, and triglycerides were significantly higher in the group that developed diabetes than the group that did not. The high-density lipoprotein cholesterol level was lower in the group that developed diabetes. Age and total cholesterol were not significantly different among the groups. Serum levels of total adiponectin, HMW adiponectin, and HMW/total adiponectin ratios were significantly lower in the group that developed diabetes. In addition, a similar result was revealed when subjects were divided by gender. Total adiponectin, HMW adiponectin, and HMW/total adiponectin ratio were lower in men than women (data not shown).
Baseline characteristics of the subjects who developed or did not develop diabetes during follow-up examination
| Not developed | Developed | P | |
|---|---|---|---|
| Men/women | 262/392 | 59/53 | 0.012 |
| Age (yr) | 60.9 ± 0.6 | 63.5 ± 1.1 | 0.067 |
| BMI (kg/m2) | 23.3 ± 0.1 | 24.4 ± 0.4 | 0.002 |
| WHR | 0.84 ± 0.01 | 0.87 ± 0.01 | <0.001 |
| NGT/IGT | 561/93 | 55/57 | <0.001 |
| Fasting glucose (mmol/liter) | 4.77 ± 0.02 | 5.17 ± 0.06 | <0.001 |
| 2-h glucose (mmol/liter) | 5.94 ± 0.06 | 7.66 ± 0.18 | <0.001 |
| Fasting insulin (pmol/liter)a | 43.0 ± 1.1 | 57.4 ± 3.6 | <0.001 |
| 2-h insulin (pmol/liter)a | 330.7 ± 11.4 | 472.8 ± 33.6 | <0.001 |
| HOMA-Ra | 1.55 ± 0.04 | 2.23 ± 0.15 | <0.001 |
| Total cholesterol (mmol/liter) | 5.87 ± 0.04 | 5.88 ± 0.09 | 0.881 |
| Triglyceride (mmol/liter)a | 1.65 ± 0.05 | 2.14 ± 0.16 | <0.001 |
| HDL cholesterol (mmol/liter) | 1.37 ± 0.01 | 1.26 ± 0.03 | 0.003 |
| Total adiponectin (μg/ml)a | 11.69 ± 0.25 | 9.47 ± 0.48 | <0.001 |
| HMW adiponectin (μg/ml)a | 7.60 ± 0.20 | 5.38 ± 0.35 | <0.001 |
| HMW/total adiponectin ratioa | 0.61 ± 0.01 | 0.53 ± 0.01 | <0.001 |
| Not developed | Developed | P | |
|---|---|---|---|
| Men/women | 262/392 | 59/53 | 0.012 |
| Age (yr) | 60.9 ± 0.6 | 63.5 ± 1.1 | 0.067 |
| BMI (kg/m2) | 23.3 ± 0.1 | 24.4 ± 0.4 | 0.002 |
| WHR | 0.84 ± 0.01 | 0.87 ± 0.01 | <0.001 |
| NGT/IGT | 561/93 | 55/57 | <0.001 |
| Fasting glucose (mmol/liter) | 4.77 ± 0.02 | 5.17 ± 0.06 | <0.001 |
| 2-h glucose (mmol/liter) | 5.94 ± 0.06 | 7.66 ± 0.18 | <0.001 |
| Fasting insulin (pmol/liter)a | 43.0 ± 1.1 | 57.4 ± 3.6 | <0.001 |
| 2-h insulin (pmol/liter)a | 330.7 ± 11.4 | 472.8 ± 33.6 | <0.001 |
| HOMA-Ra | 1.55 ± 0.04 | 2.23 ± 0.15 | <0.001 |
| Total cholesterol (mmol/liter) | 5.87 ± 0.04 | 5.88 ± 0.09 | 0.881 |
| Triglyceride (mmol/liter)a | 1.65 ± 0.05 | 2.14 ± 0.16 | <0.001 |
| HDL cholesterol (mmol/liter) | 1.37 ± 0.01 | 1.26 ± 0.03 | 0.003 |
| Total adiponectin (μg/ml)a | 11.69 ± 0.25 | 9.47 ± 0.48 | <0.001 |
| HMW adiponectin (μg/ml)a | 7.60 ± 0.20 | 5.38 ± 0.35 | <0.001 |
| HMW/total adiponectin ratioa | 0.61 ± 0.01 | 0.53 ± 0.01 | <0.001 |
Data are expressed as means ± se. NGT, Number of normal glucose tolerance subjects at baseline; IGT, number of impaired glucose tolerance subjects at baseline; HDL, high-density lipoprotein. P values are determined by unpaired t test comparing not developed group to developed. Categorized data are analyzed by χb test.
Parameters are transformed logarithmically before analysis.
Baseline characteristics of the subjects who developed or did not develop diabetes during follow-up examination
| Not developed | Developed | P | |
|---|---|---|---|
| Men/women | 262/392 | 59/53 | 0.012 |
| Age (yr) | 60.9 ± 0.6 | 63.5 ± 1.1 | 0.067 |
| BMI (kg/m2) | 23.3 ± 0.1 | 24.4 ± 0.4 | 0.002 |
| WHR | 0.84 ± 0.01 | 0.87 ± 0.01 | <0.001 |
| NGT/IGT | 561/93 | 55/57 | <0.001 |
| Fasting glucose (mmol/liter) | 4.77 ± 0.02 | 5.17 ± 0.06 | <0.001 |
| 2-h glucose (mmol/liter) | 5.94 ± 0.06 | 7.66 ± 0.18 | <0.001 |
| Fasting insulin (pmol/liter)a | 43.0 ± 1.1 | 57.4 ± 3.6 | <0.001 |
| 2-h insulin (pmol/liter)a | 330.7 ± 11.4 | 472.8 ± 33.6 | <0.001 |
| HOMA-Ra | 1.55 ± 0.04 | 2.23 ± 0.15 | <0.001 |
| Total cholesterol (mmol/liter) | 5.87 ± 0.04 | 5.88 ± 0.09 | 0.881 |
| Triglyceride (mmol/liter)a | 1.65 ± 0.05 | 2.14 ± 0.16 | <0.001 |
| HDL cholesterol (mmol/liter) | 1.37 ± 0.01 | 1.26 ± 0.03 | 0.003 |
| Total adiponectin (μg/ml)a | 11.69 ± 0.25 | 9.47 ± 0.48 | <0.001 |
| HMW adiponectin (μg/ml)a | 7.60 ± 0.20 | 5.38 ± 0.35 | <0.001 |
| HMW/total adiponectin ratioa | 0.61 ± 0.01 | 0.53 ± 0.01 | <0.001 |
| Not developed | Developed | P | |
|---|---|---|---|
| Men/women | 262/392 | 59/53 | 0.012 |
| Age (yr) | 60.9 ± 0.6 | 63.5 ± 1.1 | 0.067 |
| BMI (kg/m2) | 23.3 ± 0.1 | 24.4 ± 0.4 | 0.002 |
| WHR | 0.84 ± 0.01 | 0.87 ± 0.01 | <0.001 |
| NGT/IGT | 561/93 | 55/57 | <0.001 |
| Fasting glucose (mmol/liter) | 4.77 ± 0.02 | 5.17 ± 0.06 | <0.001 |
| 2-h glucose (mmol/liter) | 5.94 ± 0.06 | 7.66 ± 0.18 | <0.001 |
| Fasting insulin (pmol/liter)a | 43.0 ± 1.1 | 57.4 ± 3.6 | <0.001 |
| 2-h insulin (pmol/liter)a | 330.7 ± 11.4 | 472.8 ± 33.6 | <0.001 |
| HOMA-Ra | 1.55 ± 0.04 | 2.23 ± 0.15 | <0.001 |
| Total cholesterol (mmol/liter) | 5.87 ± 0.04 | 5.88 ± 0.09 | 0.881 |
| Triglyceride (mmol/liter)a | 1.65 ± 0.05 | 2.14 ± 0.16 | <0.001 |
| HDL cholesterol (mmol/liter) | 1.37 ± 0.01 | 1.26 ± 0.03 | 0.003 |
| Total adiponectin (μg/ml)a | 11.69 ± 0.25 | 9.47 ± 0.48 | <0.001 |
| HMW adiponectin (μg/ml)a | 7.60 ± 0.20 | 5.38 ± 0.35 | <0.001 |
| HMW/total adiponectin ratioa | 0.61 ± 0.01 | 0.53 ± 0.01 | <0.001 |
Data are expressed as means ± se. NGT, Number of normal glucose tolerance subjects at baseline; IGT, number of impaired glucose tolerance subjects at baseline; HDL, high-density lipoprotein. P values are determined by unpaired t test comparing not developed group to developed. Categorized data are analyzed by χb test.
Parameters are transformed logarithmically before analysis.
A Cox proportional hazards model was used to examine the association of serum levels of total adiponectin with the progression to type 2 diabetes. Higher levels of total adiponectin significantly reduced the risk for progression to type 2 diabetes before adjustment [per μg/ml: hazards ratio, 0.918 (95% confidence interval [CI], 0.882–0.956), P < 0.001; per log μg/ml: hazards ratio, 0.429 (95% CI 0.303–0.607), P < 0.001, respectively]. As shown in Fig. 1A, higher concentrations of total adiponectin (per log μg/ml) remained protective against type 2 diabetes after adjustments for sex, age, BMI, WHR, HOMA-R, and classification of 75gGTT [0.600 (0.393–0.916); P = 0.018]. The baseline classification of 75gGTT was also an independent risk factor for the development of diabetes, although sex, age, BMI, WHR, and HOMA-R were not.
Risk factors for the progression to type 2 diabetes by follow-up examinations. Results from the Cox proportional hazards model are shown. Hazards ratios are shown above the vertical bars. Horizontal bars are 95% CIs. As independent variables, total adiponectin levels (A) and HMW adiponectin levels (B) are included.
The association of adiponectin with the progression to diabetes was analyzed in another way, and subjects were divided into tertiles based on population. In a Cox proportional hazards model including tertiles of total adiponectin, hazards ratios for developing diabetes by decreasing tertiles of adiponectin were 1.000, 1.548 (0.891–2.689), and 1.787 (1.006–3.173) (P = 0.054 for trend) adjusted for sex, age, BMI, WHR, HOMA-R, and classification of 75gGTT (Fig. 2A).
A, Adjusted hazards ratios for developing type 2 diabetes from a Cox proportional hazards model according to total adiponectin concentrations. Data are expressed after adjusting for sex, age, BMI, WHR, HOMA-R, and classification of GTT. Error bars indicate 95% CIs. The trend analysis is P = 0.054. *, P < 0.05 compared with the highest tertile. B, Adjusted hazards ratios for developing type 2 diabetes from a Cox proportional hazards model according to HMW adiponectin concentrations. Data are expressed after adjusting for sex, age, BMI, WHR, HOMA-R, and classification of GTT. Error bars indicate 95% CIs. The trend analysis is P = 0.005. *, P < 0.05 compared with the highest tertile.
Next, a Cox proportional hazards analysis was performed using HMW adiponectin levels instead of total adiponectin as a variable. Higher levels of HMW adiponectin also significantly reduced the risk for the progression to type 2 diabetes before adjustment [per μg/ml: hazards ratio, 0.879 (95% CI, 0.834–0.928), P < 0.001; per log μg/ml: hazards ratio, 0.500 (95% CI, 0.394–0.635), P < 0.001, respectively]. Higher concentrations of HMW adiponectin remained protective against type 2 diabetes after adjusting for the same variables [0.614 (0.456–0.828), P = 0.001; Fig. 1B). The baseline glycemia was also an independent risk factor for type 2 diabetes.
In a Cox proportional hazards model including tertiles of HMW adiponectin, hazards ratios for developing diabetes by decreasing tertiles of HMW adiponectin were 1.000, 2.098 (1.167–3.772), and 2.493 (1.342–4.632) (P = 0.005 for trend) adjusted for sex, age, BMI, WHR, HOMA-R, and classification of 75gGTT (Fig. 2B).
In addition, for assessing HMW/total adiponectin ratio, higher levels of HMW/total adiponectin ratio also remained protective against developing type 2 diabetes after adjustments for sex, age, BMI, WHR, HOMA-R, and classification of 75gGTT [hazards ratio, 0.121 (95% CI, 0.033–0.447); P = 0.002].
Discussion
The present study clearly reveals that decreased total adiponectin is an independent risk factor for the progression to type 2 diabetes in Japanese-Americans who demonstrate a strong insulin resistance and a high incidence of type 2 diabetes. Moreover, this is the first report about the association of decreased HMW adiponectin levels and the development for type 2 diabetes. In addition, our results indicate that the association might be more closely related to the development of type 2 diabetes than total adiponectin levels.
Japanese-Americans are reported to ingest more fat, more protein of animal origin, and more simple carbohydrates than their native Japanese counterparts (23). In a previous study, although adiponectin knockout mice did not develop diabetes on a normal diet, those with a high-fat/high-sucrose diet developed severe insulin resistance and diabetes (10). The aforementioned data support our results that decreased adiponectin is a determinate marker for the progression to type 2 diabetes in Japanese-Americans, whose lifestyles are strongly Westernized compared with native Japanese.
It was previously reported that decreased total adiponectin is a risk factor for the progression to type 2 diabetes in Pima Indians, white Europeans, Asian Indians, and African-Americans (24–27). In the Japanese population, Daimon et al. (28) reported that decreased serum levels of adiponectin were risk factors for the progression to type 2 diabetes; the report examined the Funagata study, which was a representative population-based study of Japan. In the Japanese population, type 2 diabetes patients were relatively small in number, whereas in our study, the incidence of diabetes was higher than in the native Japanese (28). Furthermore, in our study, hazards ratios for developing diabetes in decreased total adiponectin remained significantly higher after adjustments were made for sex, age, BMI, WHR, HOMA-R, and classification of 75gGTT. Therefore, the influence of low circulating adiponectin concentrations on the development of type 2 diabetes is more evident than in the Japanese report (28).
In the present study, decreased HMW adiponectin levels and HMW/total adiponectin ratios are also independent risk factors for the development of type 2 diabetes. The hazards ratios for developing the disease by decreasing tertiles of HMW adiponectin were obviously higher than total adiponectin. Furthermore, although trend analysis of total adiponectin was not significant, that of HMW adiponectin was statistically significant. HMW adiponectin is thought to be the most active form of adiponectin (14). The amount of the HMW adiponectin complex, but not the total amount of adiponectin, was reported to correlate with a thiazolidinedione-mediated improvement in insulin sensitivity (14). Fisher et al. (29) recently reported that HMW adiponectin ratios inversely correlate more closely with 2-h glucose levels after oral GTT than total adiponectin levels. Our findings support the emerging hypothesis that HMW adiponectin is the active form of the protein, and HMW adiponectin levels, which are thought to be important markers of the development of type 2 diabetes, will play a prominent role in the future.
In this study, we provided evidence that in Japanese-Americans whose lifestyles were Westernized, decreased total adiponectin is an independent risk factor for the development of type 2 diabetes after adjusting for several other risk factors. Furthermore, decreased HMW adiponectin concentrations may be a more useful predictor than total adiponectin for assessing the risk of type 2 diabetes.
Acknowledgments
Disclosure Statement: The authors have nothing to disclose
Abbreviations:
- BMI,
Body mass index;
- CI,
confidence interval;
- 75gGTT,
75-g glucose tolerance test;
- HMW,
high molecular weight;
- HOMA-R,
homeostasis model assessment of insulin resistance;
- IRI,
immunoreactive insulin;
- WHR,
waist-to-hip ratio.
