Additive effects of adiponectin and C-reactive protein in predicting the risk of type 2 diabetes: a 10-year prospective study

Abstract

Background and Aims: Both serum adiponectin, an adipokine with insulinsensitizing and anti-inflammatory action, and high sensitivity C-reactive protein (hsCRP), a marker of chronic low-grade systemic inflammation, have been reported to predict the development of type 2 diabetes. Here we examined, in a 10-year prospective study, whether these two biomarkers had an additive effect on predicting the long-term risk of type 2 diabetes. Materials and Methods: Serum adiponectin and hpCRP were measured in 417 non-diabetic Chinese subjects from the Hong Kong Cardiovascular Risk Factor Prevalence Study, who had returned for their 10-year follow-up in 2005/06. Of these, 76 had developed DM, according to the 1998 WHO diagnostic criteria. Adiponectin was measured using an in-house ELISA assay and hsCRP was measured using a particle-enhanced immunoturbidimetric assay. The association of adiponectin and hsCRP, alone or in combination, with the 10-year risk of diabetes was investigated. Results: Compared to the 341 subjects who remained non-diabetic at year- 10, subjects who had developed diabetes consisted of more men (p<0.003). They had higher baseline BMI, waist circumference, mean arterial blood pressure, plasma glucose (fasting and 2-hour post-OGTT), fasting insulin, HOMA-insulin resistance index (HOMA-IR), triglycerides, LDL-cholesterol and hsCRP levels (all p<0.001 except for LDL-cholesterol with p<0.05). They also had lower baseline serum adiponectin and HDL-cholesterol levels (both p<0.001). In a stepwise multiple logistic regression analysis model including sex, age, BMI, fasting insulin, presence of metabolic syndrome according to NCEP criteria, adiponectin and hsCRP, only baseline adiponectin (adjusted OR 0.44; 95% CI 0.27–0.74; p=0.002), hsCRP (adjusted OR 1.43; 95% CI 1.10–1.84; p=0.007) and metabolic syndrome (adjusted OR 2.72; 95% CI 1.55–4.78; p=0.001) were independently predictive of diabetes at 10 years. The addition of baseline adiponectin or hsCRP to a model consisting of sex, age and BMI significantly improved the prediction of diabetes risk, as reflected by the likelihood ratios in logistic regression analysis (p=0.001 for adiponectin and p=0.019 for hsCRP). Furthermore, the combined addition of adiponectin and hsCRP to the model provided significantly greater improvement than the addition of adiponectin alone (p=0.048) or the addition of hsCRP alone (p=0.003). Conclusion: Serum adiponectin and hsCRP were both independent predictors of the 10-year risk of diabetes in this Chinese cohort. Their usefulness as biomarkers for predicting the development of type 2 diabetes were over and above those of conventional risk factors including sex, age and BMI, and appeared to be additive when these two new biomarkers were used together. Supported by the STR seeding fund on Healthy Aging of the University of Hong Kong and the Innovation & Technology Fund