Published in the Bulletin of Experimental Treatments for AIDS April 1999 issue, by the San Francisco AIDS Foundation.

April 1999 Table of Contents

Main Page

beta@sfaf.org

HAART Attack: Metabolic Disorders During Long-Term Antiretroviral Therapy

By Ethan David

HIV infection and changes in body metabolism have always gone hand in hand. The most evident sign of this relationship has been AIDS-related wasting syndrome, in which lean body tissues (muscle) are scavenged for energy resources. Cellular protein is broken down while fat reserves are underutilized.

Abnormalities in blood lipid (fat) levels begin early in HIV disease. Researchers noticed in the first years of the epidemic that triglyceride levels in the blood increased over the course of the disease whereas various types of cholesterol decreased. Triglycerides, a union of three fatty acid chains, are the standard form of energy storage in the body. Cholesterol is a ringed compound produced by the liver that acts as the building block for cell membranes and certain sex, anti-stress, and metabolic hormones.

Scientists attributed the raised lipid levels to the increases of interferon-alpha (IFN-alpha) production occurring during HIV infection. IFN-alpha is an antiviral signaling molecule that renders cells less hospitable to viral infection. Chronic increases in other immune system activators during HIV infection, such as tumor necrosis factor alpha (TNF-alpha), also may play a role in these metabolic changes. The interplay between these inflammatory molecules and alterations in lipid profiles is not unique to HIV. It is a classic response to chronic infection.

With the introduction of highly active antiretroviral therapy (HAART) four years ago, one would assume that these changes would reverse themselves as immune stimulants and inflammatory agents decreased along with HIV viral load. Instead, a mysterious new syndrome has arisen which is characterized by a loss of subcutaneous fat layers throughout the body, most noticeably in the limbs and cheeks. These signs of wasting are frequently accompanied by an abnormal increase in fat deposits around the abdominal organs (visceral fat), between the shoulder blades, and in women's breasts. (Less frequently, men develop gynecomastia, or fatty tissue breast development.)

A number of metabolic changes parallel these physical changes and may trigger them. Triglyceride levels in the blood may increase still further during HAART, accompanied by increases in cholesterol, especially the low density lipoprotein (LDL) component. LDL is commonly referred to as "bad" cholesterol because high levels of LDL are associated with increased risk of heart attack. This generalized rise in blood lipids is known as hyperlipidemia. Hyperglycemia (excess blood levels of the simple sugar glucose) sometimes appears as well. For an in-depth article on HAART and body changes, see "Body Fat Changes: More than Lipodystrophy" in the January 1999 issue of BETA.

Current Data

The frequency and extent of HAART-related metabolic changes are open to controversy because studies have been subject to numerous qualifications. Most studies were small "look-backs" at a number of case reports, and information on triglyceride and cholesterol levels prior to the initiation of HAART were usually unavailable.

In addition, the blood samples available for testing had been taken at random points during the day. Fasting samples (typically taken in the morning) are highly preferable, as recent meals will greatly affect the results and preclude obtaining data that reflect the drugs' possible influences.

Until recently, Andrew Carr, MD, David Cooper, MD, and colleagues at St. Vincent's Hospital in Sydney, Australia, had published the most rigorous study, in the May 1998 issue of AIDS. This study, which utilized fasting blood specimens, compared 113 men and two women on protease inhibitors to 32 HIV positive men who had not received protease inhibitors and 47 HIV negative men.

Viral loads were a little lower in the protease inhibitor group (1,300 copies/mL) than in the non-protease inhibitor group (4,000 copies/mL), but total blood cholesterol was one-third higher and triglyceride levels were twice as high. Lipid levels did not differ significantly between the non-protease inhibitor and HIV negative groups, and glucose measurements were about the same for all three groups. But the level of the hormone insulin produced by the persons on protease inhibitors was 80% higher than that produced by the HIV negative men.

In other words, the persons on protease inhibitors had to produce 80% more insulin in order to achieve normal glucose levels, an observation that provides evidence of insulin resistance. The study further found lipodystrophy (defined simply as loss of fat deposits in the face, arms, or legs) in 64% of the people receiving protease inhibitors and in only 3% of the other HIV positive participants. This figure resulted from participants' subjective reports of their own physical changes after starting protease inhibitors, and has been highly controversial because of its unexpectedly high magnitude.

The quality of metabolic data continues to improve. Researchers recently described several of the larger studies at the 6th Conference on Retroviruses and Opportunistic Infections (CROI) held in Chicago in late January/early February. The Center for Disease Control's Adult and Adolescent Spectrum of Disease Project was able to sift through enough data to present the experiences of 745 of its 2,892 enrollees. Instances of high triglycerides were defined as being the first time blood serum levels rose above 500 mg/dL. (This extremely high value was set to account for the use of nonfasting samples.)

The rates of the triglyceride abnormality were 15.1% in those taking a protease inhibitor plus nucleoside analogs; 6.1% in those on a non-nucleoside reverse transcriptase inhibitor (NNRTI) and nucleoside analogs regimen; 6.5% in those taking only nucleoside analogs; and only 3.0% in those receiving no treatment at all. All of these differences were statistically significant (except the difference between nucleoside analogs only and no treatment).

J. Falutz, a researcher from Montreal, reported at CROI on a study that analyzed data from fasting blood samples. Falutz checked the body composition and lipid levels of 86 men on stable HAART regimens. About half the men had viral loads below 400 copies/mL. Cholesterol levels were above the normal range in 33% of the group without lipodystrophy and in 52% of the group with only fat depletion. Seventy-five percent of the group with both fat depletion and abnormal fat accumulations had high cholesterol. Excessive triglyceride levels were found in 26% of those with no lipodystrophy, 65% of those with fat depletion only, and 69% of those with fat depletion plus accumulation.

The differences in average lipid levels between the two lipodystrophy subgroups were not clear-cut. The existence of fat deposits in the fat depletion plus accumulation group might reflect successful, if distorted, uptake of blood lipids by fat cells in the truncal and dorsal bulges. Without those deposits, blood lipid levels might have been much higher.

A similar small study in 33 women taking protease inhibitors found that waist-to-hip ratios correlated with lipid and glucose levels in fasting blood samples. As with Falutz's report, L. Bausserman reported that average elevations were only moderately above normal. An association of this nature, observed after the fact at a single point in time, is not proof of causality: the women with high pretreatment lipid and glucose levels may have been more susceptible to truncal gains and peripheral shrinkage. Since baseline measurements are unavailable, it is hard to say what role protease inhibitors play in upsetting metabolic balance.

Also at CROI, Merck Research Laboratories reported on the 24-week effects of their protease inhibitor indinavir (Crixivan). The company noted modest increases in blood cholesterol but no significant changes in triglycerides, in nonfasting samples taken from 151 volunteers in early indinavir trials. Again, the absence of cholesterol elevation is not proof that indinavir does not cause lipodystrophy. The Sydney participants studied by Carr who were taking indinavir also had a fairly high incidence of lipodystrophy, although at a lower prevalence rate than in those in the cohort who were taking the ritonavir (Norvir)/saquinavir (Fortovase) combination.

Ongoing Studies

More thorough investigations are now taking place. One of these is a substudy by the AIDS Clinical Trials Group (ACTG), ACTG 384, which is comparing d4T/3TC or AZT/3TC plus either the protease inhibitor nelfinavir (Viracept), the NNRTI efavirenz (Sustiva), or both. The substudy will prospectively follow fasting glucose and lipid levels as well as body composition changes in 354 of the 800 treatment-naive persons to be in enrolled in the overall study. Results are not expected until 2001.

The Terry Beirn Community Program for Clinical Research on AIDS (CPCRA) is administering CPCRA 058, which will include 1,000 participants without prior treatment. They will be assigned to receive either a regimen containing a protease inhibitor, an NNRTI, or both, with the drugs individually selected by their doctors. The protocol includes measurements of glucose and lipid levels and body composition during the first year of treatment. Accrual will take three years, however, and the results will not become available until much later.

The ACTG is also drawing on most of its ongoing trials to enroll a 3,000-person, five-year study of participants' experiences with antiretroviral therapy. Among other data, metabolic information will be collected every four months and correlated with participants' treatment history. In the meantime, Carr, Cooper, and their associates are now conducting a four-month, Australia-wide study of 2,000 persons to gauge the current prevalence of metabolic and physical abnormalities and their relation to present and past use of antiretroviral drugs.

Additionally, Serono Laboratories, which makes human growth hormone (HGH) for AIDS-related wasting, is now conducting a quick survey of these phenomena. Persons with metabolic and/or physical alterations and their doctors fill out an extensive questionnaire about their medical history, lab tests, and symptoms. Each of these persons will be matched with a normal (i.e., with normal blood glucose and lipid levels and without body fat redistribution) HIV positive and an HIV negative person of similar basic characteristics. They will also fill out questionnaires to create a comparative case-control study. The data from these last two inquiries should be available within a year.

Ongoing Theories

The lack of available information has not stopped some people from formulating plausible explanations for HAART-associated imbalances in lipid and sugar metabolism. Much of the speculation concerns the liver because this organ has primary control over amino acid, glucose, and lipid levels. The liver can use amino acids and glucose to create triglycerides or use triglycerides and amino acids to create glucose as needed during, after, or between meals.

Regulation of cholesterol levels also is centered in the liver, which either breaks down or synthesizes new cholesterol according to digestive intake. At the 6th CROI, French investigators noted that protease inhibitor use seemed to be associated with lower levels of several important enzymes, in particular triglyceride lipase in the liver, which eliminates the triglyceride remnants not absorbed by fat cells. Activity of a more universal cellular enzyme, lipoprotein lipase, which enables cells to absorb triglycerides, also seemed reduced in persons taking a protease inhibitor.

Carr, Cooper, and collaborators were the first to propose a general hypothesis explaining the relationship between protease inhibitors and abnormal fat utilization and storage. Protease inhibitors generally have been considered highly specific in binding only to the HIV protease enzyme while not interacting with other cellular enzymes.

They are not specific enough, say the Australian researchers, because the protease enzyme's active site, to which the protease inhibitors bind, is approximately 60% similar in structure to active regions in two major lipid-regulating compounds, lipoprotein receptor-related protein (LPR) and cytoplasmic retinoic-acid binding protein type 1 (CRABP-1).

LPR exists on the surfaces of liver cells as well as the cells lining blood vessel walls. It helps these cells absorb and break down triglycerides. More importantly, CRABP-1 works with the liver's cytochrome P450 3A enzymes, which themselves are known to be affected by protease inhibitors, to transform "all-trans" retinoic acid (a vitamin A derivative) to 9-cis retinoic acid. This molecule promotes the maintenance, proliferation, and maturation of fat cells, but has less influence in fat tissues around the abdominal organs and between the shoulder blades.

One can easily conclude that a 9-cis retinoic acid deficiency leads to high blood lipid levels and fat accumulation in the less affected sites. Disturbances in sugar metabolism would be a secondary consequence (see below).

When published a year ago, this hypothesis touched off a storm of opposition largely because it was formulated without supporting data. Nevertheless, Glaxo Wellcome, which has a large diabetes/obesity research unit, did take the hypothesis seriously.

James Lenhart, PhD, a Glaxo Wellcome research project leader, refers to, however, animal data that show that giving mice supplemental 9-cis retinoic acid increases blood triglycerides whereas using drugs to block its signaling pathway decreases triglyceride levels. (These data were presented at the 6th CROI and are awaiting publication.) This is the opposite result of what the Carr-Cooper hypothesis assumes.

Still, there is an alternative pathway through other receptors in which all-trans retinoic acid does lead to the release of extra lipids into the blood. In cell culture experiments, indinavir, but not the other protease inhibitors, seems to promote such signaling. Further experiments now in progress will determine if indinavir does this by blocking CRABP-1 and thus shunting the excess all-trans retinoic acid onto these other cell receptors.

As for ritonavir, nelfinavir, and saquinavir, the Glaxo scientists found that these protease inhibitors, but not indinavir, block lipid accumulation and enhance lipid degradation in an experimental fat cell line, as Lenhart also reported at CROI. These protease inhibitors also inhibit maturation of new fat cells.

The mechanism by which they produce these effects remains undetermined. The Glaxo experiments ruled out blockage of 9-cis retinoic acid production. The drug-related inhibition is more likely due to some disruption in 9-cis retinoic acid signaling. For example, ritonavir, nelfinavir, and saquinavir may interfere with the binding of this compound to its receptors.

Ritonavir had less inhibitory effect than nelfinavir and saquinavir in these experiments, although it is probably the greatest inducer of hyperlipidemia in the human body of the three. Ritonavir may act by yet another mechanism. The fact that different protease inhibitors have different effects on fat metabolism raises the hope that it is possible to develop a protease inhibitor without lipid-related side effects.

Glaxo Wellcome's research would be more credible if it were not tied to the publicity surrounding the launch of the company's new protease inhibitor amprenavir (Agenerase), which received FDA approval on April 15. Presently, Glaxo Wellcome's promotional materials claim that amprenavir will in fact not cause such side effects. As proof, the company points to the mere four reports of lipodystrophy during the amprenavir trials, which included 550 persons treated for at least six months.

It is important to be aware that these were spontaneous reports that doctors filed on their own as Glaxo Wellcome made no systematic attempt to collect data on lipodystrophy. A continuing French trial reported by P.-A. Bart, MD, of 41 persons taking amprenavir plus abacavir (Ziagen) has noted that triglyceride and cholesterol levels seemed to double in the course of a year of treatment, although only nonfasting blood samples were available for analysis.

Beyond Protease Inhibitors

The increases in lipid levels seen with amprenavir as well as the other protease inhibitors may or may not be due to their properties as protease inhibitors. Changes in lipid metabolism also occur with protease inhibitor-sparing regimens. As reported by Schlomo Staszewski, MD, at the 4th International Conference on Drug Therapy in HIV Infection in November 1998 in Glasgow, DuPont's pivotal 006 trial that included 154 persons on AZT/3TC/efavirenz observed 25% increases in nonfasting cholesterol on this nucleoside analog/NNRTI regimen.

T. Saint-Marc, MD, and a group of French physicians at the 6th CROI reported on a large group of people that received only dual nucleoside analogs and experienced a loss of subcutaneous fat along with increased fat deposits around the central organs. These changes occurred in 17 of 27 persons (63%) on d4T plus either ddI or 3TC but in only 3 of 16 (19%) on AZT plus either ddI, d4T, or 3TC.

Viral loads were the same (averaging only about 1,000 copies/mL) whether on AZT or d4T, and CD4 cell counts averaged an almost normal 540 cells/mm³. Metabolic differences among the two treatment groups and an HIV positive control group not on medication were subtler than in those taking protease inhibitors and generally did not reach statistical significance (see "Conference Coverage" for more detail).

It is possible that several mechanisms are at work, some of which are not related to protease inhibitors themselves. Donald Kotler, MD, of St. Luke's-Roosevelt Hospital in New York points out that a similar syndrome, including body composition changes and disturbances in glucose and fat metabolism, occurs in other chronic infections. The condition seems to be a result of chronic stress and includes elevated cortisol secretion as well as reduced sensitivity to insulin.

Kotler believes that protease inhibitors merely exacerbate this phenomenon by drastically reducing viral load and permitting renewed immune system function. In an analysis that Kotler conducted in 77 men and 19 women, CD4 cell count and type of therapy did not appear to influence fat distribution when other factors were taken into account. Low viral load and female gender were the only independent correlations that emerged.

The Sugar Connection

Regulation of lipid levels in the blood is intimately connected to sugar processing, so it would not be surprising if signs of deranged glucose metabolism accompanied the lipid problems. The hormone insulin causes cells to absorb more glucose; in fat cells, the excess glucose is converted to triglycerides for storage. Impaired fat cell function and the consequent hyperlipidemia may lead to decreased use of glucose for either immediate energy production or energy storage. More glucose may remain in the blood, resulting in a compensatory increase in insulin.

Persons on protease inhibitor regimens have glucose levels that are, on average, similar to or only moderately higher than those of healthy HIV positive persons, but the original Carr-Cooper study in the May 7, 1998 AIDS found that their mean insulin levels can be nearly two times greater. The problem with this form of compensation is that high insulin levels lead to a lower number of insulin receptors on cell walls, a form of "insulin resistance." Barring the presence of HIV, insulin resistance is the hallmark of "glucose intolerance," seen with adult-onset diabetes or treatment with glucocorticoid steroids like prednisone.

In a recent German investigation by R. Walli and others, 67 people on protease inhibitors were checked for insulin resistance and blood lipids. The investigators compared these 67 to untreated people with HIV and to HIV negative controls. Investigators also checked 24 of the 67 for proper processing of oral glucose intake.

Although there was a large variation in individual results, protease inhibitor therapy correlated overall with insulin resistance, poor glucose tolerance, and hyperlipidemia. Eleven of the 24 who were checked for glucose processing showed serious impairments, with blood glucose levels rising sharply for prolonged periods after the ingestion of a standard glucose dose.

Although the number of frank diabetes cases attributed to protease inhibitors so far has been small, the FDA issued a public warning a year ago concerning the association between protease inhibitors, diabetes, and hyperglycemia.

The original Carr-Cooper study also noted that leptin levels in people on protease inhibitors were one-third the levels in similar HIV positive persons not on protease inhibitors. Leptin is a peptide hormone secreted by fat cells and is thought to be involved in the regulation of lipid and glucose metabolism, therefore body weight regulation.

Fat cells unable to accumulate normal amounts of lipids reduce their production of leptin and other hormones to impel the body to increase glucose as well as triglyceride levels in the blood, primarily by eating more. Normally, this hormonal signaling eventually leads to restored fat stores, but in this case it would be futile, further driving up insulin levels to reverse the superfluous glucose release that leptin activates.

The Long-Term Effects of Hyperlipidemia

It may be only a matter of time before the prevalence of diabetes increases. Similarly, high lipid levels are a major risk factor for coronary heart disease. This condition begins as a buildup of fatty plaque in the arteries. The plaque forms lesions on the walls of arteries that may ultimately rupture, leading to blood clotting and further blockage of the artery. A heart attack may ensue if the oxygen supply to a section of heart muscle declines to a critical point.

Similarly, blockages in neck arteries can lead to strokes (in the brain). Cholesterol complexed with proteins into LDL particles for transport through the blood stream is considered the major source of this plaque.

In contrast, high-density lipoprotein cholesterol (HDL, which transports excess lipids back to the liver for degradation) is protective and prevents LDLs from collecting. The increased risk of heart attack stemming from high amounts of blood triglycerides is not yet well defined.

People taking protease inhibitors generally have raised levels of LDL cholesterol and triglycerides while HDL cholesterol remains constant, resulting in a decreased HDL-to-LDL ratio. This ratio is the most commonly used standard to gauge heart attack risk. The DuPont 006 trial did note a 15% increase in HDL for the AZT/3TC/efavirenz combination along with a 25% increase in total cholesterol (blood samples were nonfasting). Here, too, there was a decrease in the HDL/LDL ratio.

A year ago, Keith Henry, MD, and associates at Regions Hospital in St. Paul, Minnesota, published a description of two people with severe blockage of heart arteries, and many other anecdotal descriptions have since appeared. Today, the incidence of heart attack remains low. Carl Grunfeld, MD, of the San Francisco Veterans Administration Medical Center estimates that the lipid changes seen in people on protease inhibitors at the VA clinic would lead to 1.41 additional cases of coronary artery disease per 100 persons over 10 years.

Still, this increased risk of coronary artery disease and related death does not begin to offset the drastically reduced rate and risk of death due to AIDS, achieved in large part, in fact, through HAART.

This calculation of increased risk from protease inhibitors did not take into account the other common risk factors for heart disease that persons with HIV may have. Risk factors are more than additive; they combine synergistically to multiply an individual's risk of heart disease. HIV infection by itself may lead to decreased HDL and to heart muscle damage. Cardiac disease was the primary cause of death in 9.1% of persons with HIV in the two years prior to the introduction of protease inhibitors, according to a report by R.C. Patel and W.H. Frishman published in 1996 in Medical Clinics of North America. Heart problems will likely increase in prevalence as more people with HIV advance into their fifties.

Besides HIV and aging, other risk factors may affect such persons, including hyperglycemia and diabetes, family history of heart disease, sedentary lifestyle, smoking, high blood pressure, and use of testosterone, anabolic steroids, and certain other medications that increase blood clotting while simultaneously multiplying an individual's chance of developing heart disease. The rate of treatment failure also may well increase given HIV's rapid ability to evolve resistance to present drugs. In future, the risk/benefit assessment for individuals may change considerably.

For the time being, it would behoove all people living with HIV, especially those taking HAART, to work closely with their health-care providers to receive regular checkups that include evaluations of blood fats and sugar, and to discuss strategies to maintain optimal levels. It is also important to be regularly assessed for heart-blood vessel disease and to evaluate and implement--as well as continue to monitor--interventions as necessary.

Page last updated 1 June 1999