Controlling for apolipoprotein A-I concentrations changes the inverse direction of the relationship between high HDL-C concentration and a measure of pre-clinical atherosclerosis

- Inclusion of Apo A-I in regression model had a marked impact on the association between HDL-C and CAC score.
- The direction of the association between HDL-C and CAC score varies by levels of Apo A-I.
- If Apo A-I was included in regression models there was a positive association between HDL-C and CAC score.
- It appears that high HDL-C levels can not be assumed to confer protection against CVD risk.

BackgroundThe independent effect of high density lipoprotein cholesterol (HDL-C) concentration to confer cardiovascular disease protection has been questioned. We investigated whether the inverse association between HDL-C concentration and a measure of preclinical atherosclerosis was modified by other risk factors.MethodsCross-sectional data were analysed from an occupational cohort of 12,031 men who had measurements of cardiovascular risk factors and a cardiac computed tomography (CT) estimation of coronary artery calcium (CAC) score, a measure of pre-clinical atherosclerosis. Logistic regression was used to describe associations between both HDL-C and Apo-A-I concentrations and their ratio as exposures, and CAC scores >0, ≥20 and ≥100, as outcomes.Results1351 (11.2%), 665 (5.5%) and 230 (1.9%) of participants had a CAC score >0, ≥20 and ≥100, respectively. Adjusting for age, glucose, triglyceride, LDL-C, systolic blood pressure, waist circumference, prior cerebrovascular accident, prior coronary artery disease, prior hypertension, alcohol consumption, smoking status and exercise, a negative association existed between HDL-C and CAC score. (E.g. odds ratio (OR) for top compared to bottom HDL-C quartile for CAC >0 = 0.78 [95%CI 0.64, 0.94], p = 0.01). Further adjustment for Apo A-I changed the direction of the association between HDL-C and CAC score >0 (OR for top compared to bottom quartiles 1.61 [95%CI 1.18, 2.21], p = 0.003). Sensitivity analyses showed that point estimates for ORs were very similar regardless of CAC score threshold (CAC >0, ≥20 and ≥100).ConclusionControlling for Apo A-I concentrations changes the inverse direction of relationship between high HDL-C concentration and a measure of pre-clinical atherosclerosis.