HDL particle concentration predicts cardiovascular disease in SMART

Nathan Geffen, TAC

Daniel Duprez presented a nested case-control study from the SMART study showing that intermittent HAART was associated with a decrease in high-density lipoprotein (HDL) particle concentration in comparison with continuous HAART and that lower total HDL particles and especially the concentration of small HDL particles predicted cardiovascular (CVD) events. [1]

SMART found that CD4-guided treatment interuptions caused an excess risk of CVD compared to continuous treatment (HR:1.57; 95%CI 1.00-2.46; p=0.05). Stopping treatment was associated with a decline in both HDL and low-density lipoprotein (LDL) cholesterol. However, the net change in terms of the total cholesterol TC:HDL ratio was unfavourable. The positive effect of a reduction in LDL cholesterol was not sufficient to make up for the decline in HDL on the interruption arm. [2]

HDL is one of five major lipoprotein groups and HDL particles transport cholesterol and triglycerides in the blood. While HDL particles generally removes cholesterol from artery walls and are considered “good” cholesterol, LDL particles directly contribute to plaque formation and are considered “bad” cholesterol. However, HDL can be further grouped into large, medium and small particles.

Duprez explained that subjects with the same HDL-cholesterol might have different concentrations of HDL-particles. To illustrate this he described two 52-year old men who both had HDL-cholesterol of 36mg/dL. However, concentrations of large and small HDL particles respectively were 8 and 23µmol/L in one compared to 3 and 28µmol/L in the other.

The aim of the case-control study was to describe the relationship between lipoprotein particle size and concentration with CVD. About 240 subjects with CVD events prior to the closure of SMART were compared with two controls, matched on country, age, gender and date of randomisation.

There were statistical differences in baseline characteristics for prior AIDS (37% v. 25%, p=0.0005), current smoker (52% v. 40%, p=0.001), diabetes (17% v. 8%, p=0.0007), on blood pressure lowering drugs (45% v. 31%, p<0.0001) and prior CVD (13.3% v. 5.2%, p=0.0004) between the CVD and control cases respectively. There were, however, no statistically significant differences in age, gender, race, CD4 count viral load count and on anti-cholesterol drugs.

At baseline, there were no differences in total cholesterol, LDL particles, very low density lipoprotein (VLDL) particles, LDL cholesterol and triglycerides. but HDL cholesterol (38 v. 42, p=0.03), TC: HDL ratio (5.2 v. 4.7, p=0.05) and HDL (28.4 v. 30.2, p=0.0001) were significantly different.

Odds ratios (with subjects in lowest quartile used as reference) were adjusted for age, race, viral load, CD4 count, BMI, smoking, diabetes, HBV or HCV, use of anti-cholesterol medications, prior CVD and major baseline ECG abnormalities.

The CVD events were as follows:

  • 124 cases of non-fatal coronary heart disease (CHD),
  • 62 cases of non-fatal atherosclerotic non-CHD (strokes, peripheral arterial disease),
  • 26 cases of non-fatal congestive heart failure (CHF) and
  • 36 fatal cardiovascular cases.

Concentrations of VLDL particles, LDL particles and HDL particles were all significantly associated with non-fatal CHD, but only HDL particle concentration was associated with non-fatal atherosclerotic CHD.

HDL particle concentration approached statistical significance for predicting fatal CVD (unadjusted OR:0.3 95%CI 0.1-1.1, p=0.08).

Total HDL particle concentration was statistically associated with CVD (adjusted OR: 0.41, p=0.001).

When grouped by size, small – but not large or medium – HDL-particles were significantly associated with CVD (Adjusted OR: 0.55, p=0.03). LDL particles did not predict CVD events overall.

In the intermittent HAART arm, there was a significant decline in total (2.2, p<0.0001), medium (1.1, p=0.002) and small (p=0.03) HDL particles concentrations at one month relative to the continuous HAART arm.


This study improves our understanding of the role of various lipid particles in CVD events in HIV-positive people, and further research on the relationships between lipoproteins, HIV, ARVs and CVD is warranted.

Factors at baseline other than HIV, such as smoking, diabetes and hypertension, were all statistically correlated with CVD events in SMART.


  1. Duprez D et al. High-density lipoprotein particles but not low-density lipoprotein particles predict cardiovascular disease events in HIV patients: Strategies for Management of ART study. Oral abstract 149.
  2. Phillips AN et al. Interruption of antiretroviral therapy and risk of cardiovascular disease in persons with HIV-1 infection: exploratory analyses from the SMART trial. Antivir Ther. 2008 13(2):177-87.

Links to other websites are current at date of posting but not maintained.