Abstract 3632: Increases in HDL-C in the CHICAGO Study Explain the Benefits of Pioglitazone in Reducing CIMT Progression in Patients With Type 2 Diabetes
Background/Objective: The Carotid Intima-Media Thickness in Atherosclerosis Using Pioglitazone (CHICAGO) study showed that pioglitazone (PIO) vs glimepiride (GLM) slowed progression of atherosclerosis, measured as carotid intima-media thickness (CIMT) in patients with type 2 diabetes mellitus (T2D). Here we evaluate baseline and on-treatment predictors of benefit.
Methods: CHICAGO was a randomized double-blind, controlled study in 462 patients with T2D treated to A1c target with PIO (15– 45 mg) or GLM (1– 4 mg) for 72 weeks. The treatment effect on changes in CIMT after 18 months was estimated using ANCOVA. We adjusted for baseline CIMT, site effects and treatment effect and found the mean 18-month treatment difference to be 0.0134 mm (p =0.02). We next adjusted for a variety of individual predictors of change in CIMT by adding them one at a time to the initial model in order to determine if any would result in the treatment effect becoming non-significant. We considered baseline and 6-month changes for the following parameters: blood pressure, standard lipids, free-fatty acids, ApoB, ApoA1, A1c, fasting plasma glucose, insulin, pro-insulin, weight, BMI, waist/hip ratio, hs-CRP, duration of diabetes, number of metabolic syndrome components, statin use, baseline age and smoking status.
Results: Inclusion of baseline values for CV risk parameters did not affect the significance of the PIO effect on CIMT. At 6 months, PIO significantly increased (p<0.01) HDL (6.3 mg/dl) and decreased triglycerides (−22.6 mg/dl), A1c (0.3%), fasting glucose (15 mg/dl), insulin (48.3 pmol/L), pro-insulin (23.3 pmol/L), and hs-CRP (0.6 mg/L) relative to baseline. Only the inclusion of changes in HDL and insulin levels resulted in the loss of significance for the treatment effect on CIMT. Adjustment for the change in HDL resulted in a 30% decrease in the regression co-efficient and changed the p-value from 0.02 to 0.12 for the effect of PIO on CIMT. The adjustment for the insulin levels decreased the regression co-efficient by 19.8% and the p-value to 0.08.
Conclusion: Of all the parameters evaluated, only the changes in HDL and insulin with PIO compared to GLM treatment explained a major component of the treatment effect benefit on CIMT progression.