Background:  Intermittent use of ART (DC group) was associated with an increased risk of serious cardiovascular disease (CVD) compared to continuous ART (VS group) in the SMART Study. We previously reported that high-density lipoprotein cholesterol (HDL-c) declined significantly more in DC than VS patients at 12 months and this unfavorable lipid change could offer a possible explanation for the increased risk of CVD. To explore this finding further, lipoprotein particle concentrations were measured in a random sample of DC and VS patients and for CVD cases and matched controls. Studies in the general population have shown that low density particles (LDL-p), particularly small LDL-p, and HDL-p predict CVD.

Methods:  Lipoprotein particle concentrations were measured in stored plasma from a random sample of 451 patients by nuclear magnetic resonance at baseline and 1 month (218 DC and 233 VS). In addition, lipoprotein particles at baseline were measured in 249 cases of CVD through study closure on July 11, 2007 and in 494 controls matched for age, sex, country, and date of randomization odds ratios (OR) (upper vs lowest quartile) of HDL-p, LDL-p, and VLDL-p were estimated using conditional logistic regression with and without adjustment for baseline age, race, ART, HIV RNA, CD4⁺ count, smoking, prior CVD, diabetes, blood pressure and lipid-lowering drugs, body mass index, hepatitis B or C co-infection, ECG abnormalities, LDL-c (or HDL-c when studying LDL-p), and triglycerides.

Results:  Average DC-VS differences for change after 1 month µmol/L were –2.3 (p <0.0001) for total HDL-p, –0.27 for large HDL-p (p = 0.19), –1.1 (p = 0.003) for medium HDL-p, and –1.0 (p = 0.02) for small HDL-p. Unadjusted OR of CVD were 0.40 (95%CI 0.26 to 0.63, p = <0.0001) for total HDL-p (lower values associated with increased risk), 0.66 (0.44 to 1.05, p = 0.08) for large HDL-p, 0.91 (0.60 to 1.38, p = 0.66) for medium HDL-p, and 0.53 (0.34 to 0.83, p = 0.005) for small HDL-p. Adjustment did not alter the strength of the HDL-p associations. VLDL-p and LDL-p concentrations were not significantly associated with CVD. Unadjusted and adjusted OR for small LDL-p were 1.38 (0.87 to 2.19, p = 0.17) and 1.09 (0.56 to 2.13, p = 0.79).

Conclusions:  Treatment interruption has a rapid unfavorable effect on HDL particle concentrations. HDL-p in contrast to LDL-p concentrations were significantly related to risk of CVD independent of other risk factors.