Study results shed light on cardiovascular risks: role of antiretrovirals and metabolism of lipids and sugars

Simon Collins, HIV i-Base

It is clear that current antiretroviral strategies result in changes in lipid and sugar metabolism to some extent in nearly all patients and to a serious degree in a significant number.

That these symptoms overlap with risk factors for Cardiovascular Disease (CVD) are such a growing concern that several large prospective data collection studies have already been established and are posting early results. [1]

A more precise and practical estimate of the implications for these risk factors was proposed by Matthias Egger at the 2nd Workshop on Lipodystrophy in Toronto (Oct 2000) and was widely reported [2, 3]. It is unlikely and unfortunate that his recommendations to consider HAART as an additional risk factor for CVD in the decision to initiate therapy have not been fully integrated in clinical practice, although references to CVD risk are now included in US treatment guidelines. Even simple web-based resources for calculating CVD risk would benefit from being used. [4]

Several presentations at this conference looked to provide more detailed information about the rates of CVD in HIV patients and the risk factors associated with specific antiretroviral treatments. While follow up periods are impressive for HIV disease, the general feeling is that these are still too short to pick up CVD events which usually result from 30 years of cumulative risk factors.

Cardiovascular and cerebrovascular (CV-CV) events recorded on the databases of the Veterans Association from both post and pre-HAART era were reported by Sam Bozzette as a late breaker oral presentation. [5] Data was presented biannually from 1993-2001 for almost 40,000 veterans (98% male, 41% white, 7% prior vascular disease) covering 122,000 patient years. Antiretroviral (ARV) use from the >1.4 million prescriptions was included in the analysis).

There were 1764 hospital admissions and 521 attributable deaths to CV-CV, but these decreased despite a six-fold increase in use of antiretrovirals from all classes. Approximately 15,000 total all cause deaths occurred over this period.

All cause mortality reduced by 75% over this period, but as CV-CV events dropped less dramatically by 10-20%, this meant CV-CV events accounted for a greater proportion of all deaths. However, even when looking at those people at higher risk, the incidence per 100 patient years fell by approximately 50% (from 14 to 7 in the 2428 patients with a previous history of vascular disease; and from seven down to three in the 4051 patients >55 years old) in the pre- and post-HAART years respectively. (see Figure 1)

Fig 1. CV-CV events and ARV exposure recorded in VA databases

Year exposure/100 patient yrs Events /100 patient yrs
1993 1995 1997 1999 2001 1993 1995 1997 1999 2001
Anu ARV 23.1 23.1 56.1 57.9 60.1 CV-CV related admission 1.6 1.7 11.3 1.3 0.9
RTI 23.1 23.1 53.9 54.1 56.0 CV-CV admission or death 2.0 2.2 1.8 1.8
PI 33.0 40.1 35.9 All cause mortality 18.0 21.3 9.3 6.9 5.0
NNRTI 2.9 16.5 22.9

Fig 2: Prevalence of established risk factors by study group

CHD risk factors
HIV +ve cases
HIV -ve controls
US Nat Data
(HIV +ve vs HIV -ve) (general population)
Diabetes mellitus
Current smoker

Seven year follow-up is a good basis for continuing to collect data from subsequent years, but this may still be too early to see an increase in incidence of events. Additional concern was indicated that patients would be missed from the database if, as often happens with CV-CV events, healthcare was accessed by VA patients at other clinics. The results, while interesting, should not in themselves lower overall concern for the overlap of metabolic changes and risk factors for cardiovascular disease.

A similar trend was reported from another large US cohort following 55,000 patients from 1992-2000 and reporting 7,188 deaths (68% of pre-HAART). Compared with the pre-HAART period, proportions of deaths decreased for TB, non-TB mycobacterial infections, PCP and toxoplasmosis. Proportions of deaths increased for liver, non-Hodgkins lymphoma, cachexia/wasting, kidney disease and sepsis. Although not reaching statistical significance, the trend for ischemic heart disease (OR 1.9; CI 0.99-3.62) was suggestive of an increase in proportion of deaths.

Whilst a similar rate of incidence and lack of association with specific drugs or ARV drugs in general was reported by Klein and colleagues from the Kaiser Permanente programme in California, this team were more cautious about the implications. [7] This group collected results for up to 5.5 years from men aged 34-65. The study reported a significant 6.5 against 3.8 CHD events/1000 patient years, adjusted for age in over 4100 HIV positive patients compared to 40,000 HIV negative controls (p=0.003), and a trend towards increased risk of myocardial infarction (4.3 vs 2.9, p=0.07). So whilst an increase in CHD could not be matched to ARV use, the risk of heart disease remained higher in HIV-positive patients.

Risk factors for CHD were common among HIV positive patients hospitalised for CHD: hypertension (27%), diabetes mellitus (16%), hypercholesterolemia (72%) and current smoker (29%). However differences between the HIV positive and the HIV negative control group, and indeed differences between the control group and incidence in US national data were not explained by traditional risk factors. (see Figure 2)

Holmberg and colleagues from the HOPS cohort reported higher incidence of myocardial infarction (but not of angina or cerebrovascular events) among patients using protease inhibitors. [8] The number of MI were small but the difference between the groups was significant; 13 in 3013 patients on PI compared to 2 in 2663 non-PI-using patients (p=0.012). This association remained in multivariate models controlling for hypertension, smoking, diabetes, age, gender, and evidence of dyslipidemia although the confidence intervals were relatively wide.

An interesting and potentially optimistic study looking at increased lipid levels was presented in a poster by Stefan Mauss. [9] This study looked more carefully at increases in cholesterol in 187 consecutively seen patients at a clinic in Düsseldorf, 45% of whom had hypercholesterolemia >200mg/dl. Hypercholesterolemia in these patients was due to increases in LDL in 14% cases, to high LDL and VLDL in 16% cases, to VLDL in 66% cases and in 4% cases due to increases in HDL alone. Further analysis of the VLDL composition revealed large triglyceride-rich particles associated with low risk familial hypertriglyceridaemia. The possibility that many HIV positive patients with high triglycerides may be at a lower risk than previously thought, perhaps supports the lower incidence seen in the cohorts reported above, and lipid profiles in lipodystrophic patients should be included in larger studies.


  1. Studies underway on metabolic side effects of antiretrovirals. HTB Vol2No2.
  2. Egger M – Metabolic complications of HAART – Second workshop on lipodystrophy and adverse events. Toronto, Oct 2000. Oral presentation 023.
  3. Egger M – HAART and the heart: lipodystrophy & cardiovascular risk: need for perspective. PRN Notebook Vol6 No2.
  4. Web based resources to calculate CVD risk:
  5. Bozzette S et al – Cardio- and Cerebrovascular Outcomes with Changing Process of Anti-HIV Therapy in 36,766 US Veterans. 9th CROI. Late Breaker LB9
  6. Wolfe M et al – Deaths from Non-AIDS-Related Diseases Have Increased as a Proportion of Deaths of HIV-Infected Persons since the Advent of HAART. 9th CROI. Oral abstract 14
  7. Klein D et al – Hospitalisations for Coronary Heart Disease and Myocardial Infarction among HIV+ Patients in the HAART Era, 9th CROI. Abstract 696.
  8. Holmberg S – Protease Inhibitor Use and Adverse Cardiovascular Outcomes in Ambulatory HIV Patients. 9th CROI. Abstract 698 – P
  9. Mauss S – Differentiating Hypercholesterolemia Associated with Antiretroviral Therapy (ART). 9th CROI. Abstract 689. P

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