HAART to HEART: cardiovascular risk in HIV

Judith Aberg MD, for

The debate at the CROI over whether HIV and its therapy are associated with the development of coronary heart disease (CHD) was as prominent as it was at last year’s conference.

Again we heard conflicting data as the studies are either retrospective with many confounding factors or prospective with short duration thus making it impossible to know how much, if any, HIV and its therapies are contributing to the development of CHD. In fact the New England Journal of Medicine (20 February 2003) contained an editorial by Drs Dan Kuritzkes and Judith Currier debating this same issue as they reviewed the results of a retrospective study of the VA database by Dr S. Bozzette and colleagues. [1, 2] I agree with their conclusions that longer follow-up of the VA cohort and other cohorts is needed before cardiac risk can be ascertained.

Dr Friis-Møller presented the results of the prospective observational D:A:D study of 23,468 subjects from 11 cohorts in three continents [3]. Subjects were enrolled from July 1999-April 2001 with final data collection/analysis August 2002. Although the subjects were since followed prospectively starting July 1999, the investigators entered data based on years subjects were started on HAART (PI or NNRTI), which varied from none to greater than six years. It is unclear if they truly have source documentation of all 23,468 subjects’ first date of HAART or whether they derived this data from the subjects’ history as documented in a chart. Data on HIV disease, risk factors for myocardial infarction (MI) and incident MI were collected. The median age of this cohort is 39 years, 24% female, 60% smokers and 30% had elevated triglycerides (TG). During 36,479 person years, a total of 126 subjects had a MI – a subgroup comprised of 90% men and median age 48 years. Thirty-six (25%) of the subjects had a fatal MI. Traditional risk factors such as age, male sex, previous history of CHD and smoking remained independent predictors of developing a MI. Other important risks included diabetes and hypertension. The relative risk per year of exposure to HAART was 1.26. Interestingly, lipodystrophy was a protective risk (RR 0.6). However this term was defined by subjective measures and needs to be further studied before any conclusions can be drawn. The investigators concluded that HAART use was associated with a 26% relative increase in the rate of MI per year of exposure over the first seven years. As Dr. Friis-Møller noted, the risk of MI at this point is low and the obvious benefits of HAART outweigh concerns over potential CHD. Nevertheless, it remains important to consider all aspects of health care for HIV-positive individuals including modification of traditional CHD risk factors if possible as well as treating the complications of HAART as they arise.

This database has many of the same limitations that other databases have that collect information retrospectively and then follow subjects prospectively. Hopefully, over time as they follow this cohort, particularly the many subjects who entered the study naive to ART or on their initial regimens, we can begin to sort out those factors contributing to CHD from HIV and its therapies versus traditional risk factors. The investigators are to be commended for developing such a large cohort and collecting data from various international centres. This has taken enormous effort and will be a very valuable cohort to follow long-term.

Dr Lucas presented the results from the Johns Hopkins database, another large cohort that has been followed longitudinally since 1990, regarding the incidence of CHD and cerebrovascular disease (CVD) among their HIV-positive patients [4]. A nested case control study was designed to assess factors associated with CHD and CVD. Subjects without CHD and CVD risks were randomly selected as controls. Five controls per case were identified and matched on cohort enrolment date and duration of follow-up. Of 2,671 patients followed for 7,330 person-years (PY) after 1 January 1996, there were 43 CHD and 37 CVD events for an incidence rate of 5.9 events/1000 PY and 5.0 events/1000 PY, respectively. Factors associated (p < 0.05) with having a CHD or CVD event included age (mean = 46 years-cases, 41 years-controls), cholesterol (mean 186 g/dl cases, 156 g/dl controls), prior diabetes (15% cases, 7% controls), prior hypertension (43% cases, 17% controls), CD4 (mean 351 cells/mm3 cases, 251 cells/mm3 controls). There was no difference between cases and controls in race, injecting drug use, or HIV-1 RNA. Cases were significantly more likely than controls to receive protease inhibitors (PI) (59% vs 43%) and D4T/3TC (63% cases, 43% controls); however, no differences were found for other nucleoside RTIs, NNRTIs, or any individual PI. The risk factors were similar for CHD and CVD when assessed separately. The investigators did not adjust for smoking or lipodystrophy. Also given the limitations of this database, they could not account for nadir CD4 count or duration of HIV infection. The incidence of CHD and CVD is significantly higher than one would have expected for the same age-sex-race population rates of 2/1000 PY for CHD and 3/1000 PY for CVD as reported from the National Health and Nutrition Examination Survey (NHANES). However, it is difficult to compare historic controls to the current database plus given the many confounding factors and lack of reliable information (smoking, metabolic syndrome, HIV duration, CD4 nadir), it remains unclear if HIV and its therapies are independent risks for the development of CHD. In contrast to these two large population studies, Dr Bozzette reported no change or a slight decrease in incidence of CHD among an even larger cohort of 39,766 HIV-positive persons followed at Veterans Affairs facilities in the USA. The answer to these question will take time and further study.

Carotid artery intima thickness test in HIV-positive individuals on HAART, first study

Dr Judith Currier, University of California, Los Angeles presented the baseline results of the trial evaluating ‘Carotid artery intima thickness (IMT) in HIV-positive and HIV-negative adults’. [5]. The AIDS Clinical Trails Group 5078 team carefully designed a study limiting the high-risk confounding factors and having matched HIV seronegative controls to really focus on the contribution of HIV and its therapies unlike other trials we have seen.

Carotid IMT has been shown to be predictive of clinical cardiac events in individuals with and without cardiac symptoms. All seven sites were trained by a standard protocol, performed in duplicate and sent to a central reader at the Cleveland Clinic. All subjects were evaluated at baseline and at weeks 24, 48, 72 and 96. Each site enrolled a triad at a time consisting of one HIV positive individual on a protease inhibitor (PI) for at least two years, one HIV positive individual not on a PI, and one HIV seronegative control. All three subjects were matched for age within five years, race/ethnicity, sex, blood pressure status, smoking status and menopausal status for women. Subjects with high-risk CHD risk factors such as diabetes, history of CHD either by the subject or a first degree relative and uncontrolled hypertension were excluded from the study to limit the number of confounding factors. A total of 45 triads (patient groupings) were enrolled and one subject in the PI group of one triad discontinued prematurely. The median duration of PI use was 216 weeks. Each of the triads was well matched and the HIV groups were equally matched for CD4 and HIV viral load. The median CD4 was 530 in the PI group and 482 in the no PI group. The nadir CD4 count was unavailable as was duration of HIV seropositivity. The cohort was predominantly male (90%) and white (76%) with a median age of 42 years and 56% non-smokers. All subjects were normotensive. The ACTG 5078 team defined smokers as per the guidelines for the American Cancer Society, which is used for most cardiology studies. The median value of the labs were similar between the groups except for triglycerides (TG) and total cholesterol (TC) were slightly higher in the PI group at 219 and 192 mg/dL respectively compared with 142, 107 mg/dL TG and 179, 187 mg/dL TC in the no PI and HIV-negative groups. There was also an increased weight to hip ratio among the PI group compared with the other two, however there was no difference in respect to the body mass index or waist circumference.

The 5078 team reported that there was no significant difference in the measurements of the carotid IMT among the three groups. When one controls for the known CHD risk factors, the PI group did not have increased carotid IMT measurements compared with the no PI or HIV seronegative matched controls. Factors associated with an increased carotid IMT were low HDL, elevated TG (more pronounced when HDL is low), older age and increased BMI. This by itself is very interesting as low HDL is associated with HIV itself and was among the first lipid abnormalities described prior to the introduction of HAART. HAART, particularly ritonavir containing regimens, have been implicated as a cause of hypertriglyceridemia. Although the 5078 team did not find a significant difference among the groups now, it may be too early to detect any differences. CHD may take decades to develop and it will be critical to follow this extremely well matched cohort for years, in fact longer than the team has initially planned.

Carotid artery intima thickness test in HIV-positive, second study

On behalf of investigators from San Francisco General Hospital (SFGH), Dr P Hsue presented longitudinal results of measuring carotid IMT in 106 HIV-positive subjects on ART. [6] Similar to the ACTG 5078 study reported above, this group of investigators sought to determine the predictors of carotid IMT in HIV-positive subjects and to follow IMT progression over time. The mean age of the 106 subjects was 45 years and 88 (83%) were male. The duration of HIV infection was 11 + 5 years, median CD4+T-cell count was 354 cells/mm3 and the median duration of PI use was four years prior to enrollment. Compared to historic controls, the baseline mean carotid IMT was 0.90 + 0.27 mm was thicker than expected. Multi-variable predictors of increased IMT at baseline were age, LDL, hypertension and nadir CD4 <200. C-reactive protein, fibrinogen, HDL, triglycerides, lipodystrophy and duration of HIV were not predictive of increased baseline carotid IMT. The investigators followed the first 22 subjects enrolled over a year. The mean rate of carotid IMT progression was 0.1 + 0.1 mm/yr, which the investigators noted was increased compared to that reported in previous published studies of the general population. Age and duration of PI use were associated with carotid IMT progression. Interestingly, 41% of the 22 subjects had hypertension.

In contrast to the findings by the 5078 study, the investigators of this study concluded that HIV and its therapy in addition to traditional risk factors may contribute to the development of CHD. So, why the difference between the two studies? There was some debate over the methodology used by the ACTG 5078 team in comparison to one done by investigators at SFGH. The 5078 team chose to measure the carotid IMT at a specific site on the far wall on the internal carotid artery where there is laminar flow and this method has been reported to be a very precise and reproductive measurement. The SFGH chose to measure six sites near the bifurcation on each internal carotid artery for a total of 12 measurements and used the mean. I am unaware of any head to head comparisons of these two methodologies but I think the most important aspect will be following these cohorts long term using the same method to determine the longitudinal differences within each cohort.

The difficulty in interpreting the SFGH results is that there is no HIV seronegative control group. The investigators compared their results to historic controls who may not be representative of our current population, plus the equipment may be different with results being very operator dependent. Just this past year, there were national headlines stating that more than 20% of Americans met the definition for the metabolic syndrome compared with <10% ten years ago. And even more disturbing is that over 6% of 20 year olds met the criteria for the metabolic syndrome. Given the marked improvement of technology plus the changing demographics, I think it is imperative that these trials have HIV seronegative matched controls. I encourage the investigators at SFGH to increase their sample size and enrol controls as they continue their studies. Let’s see what next year’s results tell us.


Of the two IMT studies, the SFGH is methodologically by far the inferior study: it is cross sectional with a small number of patients and no control of the traditional risk factors is made in the subjects who were followed longitudinally. The increase in IMT is a log greater than associated with any other cardiovascular risk factor. The low number of patients creates the problem of multiple testing, i. e. to have a statistical positive results by chance with no real meaning. The ACTG study balances the risk factors by a valid design and is excluding only patients with existing cardiovascular disease who will have a rapid progression, probably due to high risk constellations in the past most probably before ART.

D:A:D like all the other cohort studies has the critical problem that it does not have good data on the duration of HIV-infection which may be a risk factor by itself. The low number of events makes establishing high-risk profiles difficult at the time being including the use of different antiretroviral agents, but most identified variables point in the direction of traditional cardiovascular risk factors. However it would be interesting to know if different lipid changes result in different risks. A hint in this study that different lipid profiles due to ART may translate into different risks is the observation, that triglycerides are not associated with an increased risk in myocardial infarction.

Neither the D:A:D nor the Hopkins study can control for the duration of HIV-infection. Antiretroviral treatment may simply be a surrogate marker for being HIV-positive for a longer time. In addition the replication rate of HIV may matter as a proatherogenic factor. Again a higher replication rate of HIV will lead to immunosuppression and earlier treatment with ART. This dilemma is impossible to solve in both studies due to their design. This should not mean that hypercholesterolemia may not be a risk factor, but it may be more complicated than it seems. In both studies HIV-negative controls are missing.


Unless stated otherwise, all references are to the Programme and Abstracts of the 10th Conference on Retroviruses and Opportunisitc Infections (CROI), 10–14 February 2003, Boston.

  1. Kuritzkes DR, Currier J. Cardiovascular risk factors and antiretroviral therapy. N Engl J Med 348:679-680
  2. Bozzette S, Ake C, Tam HK et al, Cardiovascular and cerebrovascular events in patients treated for human immunodeficiency virus infection. N Engl J Med. 348:702-710
  3. Friis-Møller N, Weber R, Monforte AD et al. Exposure to HAART is associated with an increased risk of myocardial infarction: the D:A:D study. Abstract 130
  4. Moore RD, Keruly JC, Lucas G. Increasing incidence of cardiovascular disease in HIV-infected persons in care. Abstract 132
  5. Currier J, Kendall M, Henry K et al. Carotid imtima-media thickness in HIV-infected and uninfected adults: ACTG 5078. Abstract 131
  6. Hsue P, Lo J, Franklin A et al. Increased atherosclerotic progression in patients with HIV: the role of traditional and immunologic risk factors. Abstract 139lb.


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