Conclusion
In community-living elders, total and HMW adiponectin showed similar U-shaped relationships with CVD. The inverse relation in the lower range, but not the direct association at the higher end, disappeared after inclusion of putative intermediates, suggesting that high levels may reflect adverse processes separate from adiponectin's beneficial glycometabolic properties.
Objective
The aim of the study was to test the hypothesis that total adiponectin is directly associated, but HMW adiponectin is inversely related, with CVD in older adults. Design, setting, and participants: We evaluated 3290 participants free of prevalent CVD in a longitudinal cohort study of U.S. adults aged 65 yr and older. Main outcome measures: We measured incident CVD (n = 1291), comprising coronary heart disease and ischemic stroke.
Results
Total and HMW adiponectin were tightly correlated (r = 0.94). Cubic splines adjusted for potential confounders revealed that the associations of total and HMW adiponectin with CVD were U-shaped, with inflection points of 20 and 10 mg/liter, respectively. After controlling for potential confounding, levels of total and HMW adiponectin below these cutpoints tended to be inversely associated with incident CVD, driven by their significant or near-significant relations with coronary heart disease [hazard ratio (HR), 0.85 per sd increase; 95% confidence interval (CI), 0.75-96; and HR, 0.87; 95% CI, 0.75-1.01, respectively]. These associations were abrogated by additional inclusion of putative metabolic intermediates. Above these cutpoints, however, both total and HMW adiponectin were significantly directly associated with CVD after adjustment for confounders and, particularly, mediators (HR, 1.20 per sd increase; 95% CI, 1.06-1.35; and HR, 1.12; 95% CI, 1.02-1.24, respectively).