Editor's note: this article was originally published on September 27, but an html coding error caused six paragraphs midway through the story to be suppressed. This is the updated version.
When are lipid levels cause for concern in people on antiretroviral treatment? At last year’s Lipodystrophy Workshop Dr Stefan Mauss set the cat among the pigeons by presenting data showing that the cholesterol increases seen in people on HAART might not be as atherogenic as previously assumed, due to the high proportion of VLDL cholesterol in their blood. One UK medic was later heard to remark that Dr Mauss shouldn’t be allowed to present these data in public, so profoundly did they upset assumptions about the atherogenicity of some HAART regimens – and the company marketing strategies also based on these assumptions.
At this year’s workshop Jim Neaton, cardiovascular epidemiology advisor to the CPCRA and one of the epidemiologists who analysed the huge MR FIT study of cardiovascular risk factors in the US population, attempted to shed some light on how we should be judging the atherogenicity of different HAART regimens.
His first striking observation was to highlight the very low levels of cholesterol seen in treatment-naïve people recruiting to a new CPCRA study called FIRST. In this population, median baseline cholesterol is 160mg/dL (4.14mmol/L) versus 200mg/dL (5.18mmol/L) in 36 to 39 year olds joining the SMART study, another CPCRA study that is recruiting treatment-naïve and treatment-experienced people (for randomisation to stop treatment until their CD4 cell count falls below 250, or to continue treatment with the aim of maximal viral suppression).
Neaton pointed out that the SMART population had comparable cholesterol levels to the general population, matched for age, as captured in the NHANES study (another large US study of cardiovascular risk factors). Cholesterol levels in the MR FIT study also tracked the levels found in each band in the SMART population in baseline.
Of course, the findings from the SMART study disguise the fact that people who have been on HAART for some time may have higher cholesterol levels, as with any population averaging. What do other parameters tell us?
FIRST participants had lower levels of LDL cholesterol (-34mg/dL, or 0.88mol/L) compared to the NHANES cohort, but they also had lower levels of HDL cholesterol (-13mg/dL, or 0.15mmol/L) and higher triglyceride levels (+13 mg/dL, or 0.15mmol/l). HDL cholesterol levels do not tend to improve with protease inhibitor therapy (although increases have been seen in NNRTI-treated patients), but triglyceride levels do increase on many HAART regimens, further widening the gap between people with HIV and the general population in this respect. A 1mmol/L increase in fasting or non-fasting triglyceride levels is associated with a 15% increase in the risk of coronary heart disease in the Framingham Study, the most widely referred-to guide to coronary risk used by clinicians. As an example of the severity of the problem created by triglyceride increases, Jim Neaton cited data published by Andrew Carr and his Australian colleagues in 1998. They compared metabolic parameters in treatment-naïve people with HIV, HIV-negative controls and HIV-positive people taking protease inhibitor-containing HAART (usually with d4T). They found that people on PIs had triglyceride levels averaging 292mg/dL (3.3mmol/L), compared to an age-matched average of 115mg/dL (1.3mmol/L) in HIV-negative controls.
Thus, choice of regimens that don’t cause large trigylceride increases is one important consideration. In a bad conference for d4T (see also reports on fat wasting in ACTG 384 and when compared with AZT), an Italian group reported a significantly increased risk of triglyceride increases in people who commenced a d4T-containing regimen when compared with AZT/3TC (n=322, relative risk 1.74; relative risk of hypertriglyceridemia with d4T/ddI: 4.61) (Galli). Further data confirming this connection will be presented at the 42nd ICAAC on Friday September 27, when the 48 week results of a comparison of d4T and tenofovir in first line regimens will be reported.
What are these triglycerides saying? One of their messages is that there is a lot of VLDL cholesterol around, carrying triglycerides about the body. In a stimulating presentation that was afforded little discussion, Gary Simon of George Washington University, Washington DC, explained that not all VLDL cholesterol is equally harmless, as Dr Stefan Mauss and colleagues suggested last year. In fact, some species of VLDL cholesterol are a very bad thing. VLDL particles can be divided into 6 sub-groups, and groups V5 and V6 (the largest particles) carry the highest risk of coronary heart disease. A low level of HDL cholesterol coupled with high levels of V5 and V6 VLDL increases the risk of CHD substantially, Simon told the meeting.
Simon’s group analysed VLDL sub-groups and other lipid profiles in treatment-naïve patients and in 15 people taking nelfinavir and 17 taking efavirenz after 16 weeks of therapy. Most individuals were taking AZT/3TC, although a few of the nelfinavir group were taking ddI/d4T, making these data a little muddy.
The group found that whilst triglyceride levels rose significantly in both groups (26.5% in the NFV group and 38.3% in the EFV group, non-significant difference), the change in VLDL levels was modest, and largely occurred in the less atherogenic subset of medium-sized VLDL particles.
However, a clear difference did emerge in HDL cholesterol levels. Although HDL cholesterol is considered to be `good’ cholesterol, increases in levels of smaller HDL particles confer an increase in CHD risk. The rise in small HDL particle levels was significantly greater in the nelfinavir group when compared to the efavirenz group (44.8% vs 25.2%), whereas the efavirenz group experienced a significantly greater increase in large HDL cholesterol particles (8.2% vs 40.9%, p<0.05).
LDL cholesterol levels also increased more in the nelfinavir group than the efavirenz group (45% vs 11.7%, p<0.05).
Given that participants were taking different nucleoside backbones, these data are best treated as proof of principle rather than clinical gospel. What they suggest is the value of measuring cholesterol sub-set changes in licensing studies and large strategic trials like Initio and ACTG 384.
But what does a clinician do with all these parameters when assessing whether to switch patients or individualise regimens based on cardiovascular risk? Peter Reiss of the Academic Medical Centre, Amsterdam, encouraged clinicians to use a website developed to assess cardiovascular risk, http://www.CHD-taskforce.com, which calculates the risk of cardiovascular events based on current data. It is intended to spare clinicians from poring over Framingham tables and other data they may not be familiar with. What it doesn’t do is spell out what to do with the HIV-positive patient, for whom the short-term benefits of HAART will usually outweigh the short term risks of cardiovascular events. What is the long-term risk, and when we will know that we have reached the long-term?
In his presentation Jim Neaton highlighted that we do not know the incubation period of any cardiovascular disease that might result from metabolic changes now being seen on HAART. Nor can we assess the extent to which changes that confer or reduce risk when viewed through the lens of Framingham tables and other data from cardiovascular epidemiology in the general population will be mirrored in people with HIV. As an illustration Jim Neaton highlighted the results of a recently published randomised study of oestrogen and progesterone replacement therapy in 16,608 healthy post-menopausal women. At year one, HDL cholesterol levels had increased by 7.3%. But at year five, when the study was stopped, the risk of cardiovascular events in the treated group exceeded the placebo group by 22%.
Large-scale studies are thus needed, but new techniques that might deliver quicker answers also need to be considered. Coronary calcification is one measure suggested by Jim Neaton; the development of calcified plaques in the walls of blood vessels can occur quickly and is a reliable marker of cardiovascular risk. Intima media thickness, markers of coagulation and fibronolysis and coronary calcification will be measured in 500 participants in the SMART study, along with another sub-study that will carry out serial ECGs to check for silent myocardial infarctions, a strong predictor of future coronary events that would be missed if studies rely solely on the reporting of identified events.
For further discussion of previous evidence on cardiovascular disease and metabolic changes on HAART, see Heart disease and HAART elsewhere on this website.
Galli M et al. Factors associated with the development of high serum triglyceride levels in patients receiving two NRTI combinations. Fourth International Workshop on Adverse Drug Reactions and Lipodystrophy in HIV, San Diego, abstract 40, 2002.
Simon G et al. Antiretroviral therapy-induced changes in lipoprotein subclass phenotype: comparison of protease inhibitor and non-nucleoside reverse transcriptase inhibitor regimens. Fourth International Workshop on Adverse Drug Reactions and Lipodystrophy in HIV, San Diego, abstract 14, 2002.