Lipodystrophy

Body fat changes and metabolic changes known as lipodystrophy, which occur when a person is taking highly active antiretroviral therapy (HAART), are no longer regarded as a direct consequence of protease inhibitor treatment alone. The causes of body fat changes may be multi-factorial and more complex than first thought.

Metabolic changes, including lipid elevations and disturbances in glucose (sugar) metabolism, are discussed in more detail in Metabolic changes on antiretroviral therapy in Anti-HIV therapy: Body fat and metabolic changes whilst on treatment. The metabolic changes may have more serious long-term consequences than body fat changes, because they may increase the risk of heart disease and of diabetes.

Increasingly, experts argue the two key physical features of lipodystrophy, fat accumulation and fat loss, need to be understood as separate problems. This is because these features may have different causes and treatments. The extent to which these features of the 'lipodystrophy syndrome' are interrelated is not understood.

A number of factors have been associated with body fat changes:

• Treatment with protease inhibitors and nucleoside reverse transcriptase inhibitors (NRTIs) together.

• Specific drugs.

• Type and duration of anti-HIV therapy.

• Duration of HIV infection or low CD4 cell count.

• Gender.

• Age.

• Race.

• Higher body mass and fat prior to treatment.

These factors are discussed in more detail below.

Background

Lipodystrophy was first noticed during 1997 among people who had been taking protease inhibitors for several months. Originally named 'Crix belly', the syndrome was reported as abdominal weight gain among people taking indinavir (Crixivan). However, the link with indinavir was partly a consequence of the fact that the overwhelming majority of people receiving highly active anti-HIV treatment up to that point had been prescribed indinavir. Subsequently, the syndrome was linked to the protease inhibitors as a class of drugs, with anecdotal reports that people taking combinations that did not include protease inhibitors were also developing the syndrome.

Abnormal body fat changes were occurring over and above those that might be expected in people recovering from serious illness, even taking into account the reduced resting energy expenditure requirements of people who no longer had high levels of HIV replication (HAART typically reduces such requirements by about 10% once undetectable viral load is reached). If resting energy expenditure declines without a reduction in food intake, the result is weight gain in the form of fat (Pernerstorfer-Schoen 1999).

What are the changes?

The following signs have been described as part of the syndrome in reports since 1997:

• Increased waist size.

• Increased breast size.

• 'Buffalo hump' (fat accumulation around the neck and upper back).

• Fat accumulation around the neck and jaw ('moon face').

• Fat deposits in other locations.

• Facial wasting, especially of the cheeks.

• Loss of subcutaneous fat in all parts of the body, most visibly in the limbs.

• Wasting of the buttocks.

• Prominent leg veins.

Metabolic changes have been also been widely reported. The connection between body fat changes and metabolic changes remains unclear. The relationship between blood fat (lipid) levels, blood sugar and insulin abnormalities, and other laboratory data, and body fat changes is not presently understood. These changes are discussed in detail in Metabolic changes on antiretroviral therapy.

The Marrakech workshop in 2000 classified the fat changes into four types:

• Type 1: fat loss only.

• Type 2: fat accumulation only.

• Type 3: a combination of fat loss and fat accumulation.

• Type 4: metabolic changes without changes in body fat distribution.

The European Medicinal Evaluation Agency established a Committee for the Evaluation of the Metabolic Complications of HAART in 1999 to conduct four research studies into the side effects of antiretroviral therapy. The first study was aimed at developing a case definition of lipodystrophy and reported its results in 2002. The study involved 1081 adults with lipodystrophy. It attempted to find identifying features and measurements that clearly distinguished lipodystrophy, and developed a definition that could be used to detect around 80% of cases (Carr 2002).

The study established a case definition for lipodystrophy based on the following criteria:

• Female gender.

• Age over 40 years.

• HIV duration longer than four years.

• CDC category C.

• Increased waist to hip ratio.

• Decreased HDL cholesterol.

• Increased anion gap.

• Increased ratio of visceral adipose tissue to subcutaneous adipose tissue (greater than 1.59).

• Increased trunk to limb fat ratio.

• Decreased leg fat percentage (below 8.8%).

A scoring system allows clinicians to assess how completely a patient meets this definition.

The problem with this definition is that it requires DEXA scans in order to establish most of the ratios, and this equipment is not available in many hospitals, nor is this form of testing quick to carry out where it is available. A definition pruned of body composition measurements has also been published, and the scoring system is available online at http://www.ti3m.com/hiv/default_ld.htm

Disagreements over the definition of lipodystrophy

Fat accumulation that occurs in the abdominal area of individuals with lipodystrophy is known to be visceral rather than subcutaneous fat. This means the fat accumulates around the organs rather than under the skin. Although increased abdominal fat is regarded as a core feature of lipodystrophy, the Fat Redistribution and Metabolism (FRAM) study presented at the International AIDS Conference in 2002 suggested that HIV-infected men are not more prone to abdominal fat and visceral fat accumulation than age-matched men in the general population (FRAM 2005). This points to the ongoing disagreement about what constitutes HAART-related lipodystrophy.

The FRAM study went about identifying the common features of the syndrome in a different way from any previous research, and its methods are considered to be epidemiologically sound. 412 individuals with a median age of 40 and a median CD4 count of 394 cells/mm³ were recruited at random from treatment centres in the US, and compared with a control group of 153 HIV-negative men from the CARDIA cohort, a long-term cardiovascular risk cohort. This was a cross-sectional study - in other words, measurements were carried out once, and people had variable durations of HIV infection and treatment. The study started with no pre-conceptions about linkages between the body fat and metabolic changes reported in people with HIV. The two populations were comparable in terms of age and height, but the HIV-positive group weighed less. Whether this introduced a bias or not is unclear since the data have not been presented in detail.

The study found that people with HIV had less subcutaneous fat, and those with lipoatrophy (defined by patient self-report and doctor's examination) had least subcutaneous fat (defined by DEXA scan). Subcutaneous fat loss was greatest from the legs, then from the arms and then from the lower and upper trunk. These findings are unsurprising (Gripshover 2003; Saag 2003).

In contrast, the findings relating to visceral fat accumulation have been met with surprise. People with HIV had less visceral adipose tissue than the control group, and whilst people with lipoatrophy displayed a higher ratio of waist to hip measurements characteristic of the metabolic syndrome seen in HIV-negative people, this increased ratio was due to a reduced hip measurement, not a bigger waistline. The reduced hip measurement was most likely a consequence of the loss of buttock fat. There was no relationship between the degree of subcutaneous fat lost and changes in visceral adipose tissue, suggesting that the changes are not reciprocal. Buffalo hump occurred with the same frequency in people with HIV and the control group, although the size of the humps was larger among men with HIV. HIV-positive men with buffalo hump also had significantly less lower trunk subcutaneous fat than HIV-negative men (Zolopa 2003).

The researchers concluded that the only change specific to HIV-positive people was lipoatrophy, and that central fat accumulation cannot be used as a defining characteristic of the syndrome. However, the design of this study makes it impossible to rule out fat accumulation; since people with HIV weighed less, it is also arguable that they weighed less before becoming infected with HIV, so any changes in total central fat would start from a lower baseline. On the one hand this means that relatively small changes (similar to those experienced in the HIV-negative population) might become visible more easily, but it also means that relatively small reductions in subcutaneous fat over time will be more apparent in the HIV-positive group. The finding that patients with lipoatrophy had the lowest volume of central fat tends to argue against the former explanation.

The more robust finding of this study is the lack of relationship between fat loss and fat gain in different parts of the body, implying that central accumulation is not a redistribution of fat from the periphery, but a separate process with distinct causes.

Further analyses of this study, which has recruited over 1000 patients, will compare body fat distribution in men and women, and look at relationships between metabolic changes and body fat changes. Relationships between drug regimens and body distribution will also be analysed (although these data will be of limited usefulness given the criticisms of previous cross-sectional studies).

How common are fat changes?

There are still widely differing estimates of the incidence of body fat changes among people on HAART. Some studies have reported that between 5%-30% of people on protease inhibitors develop lipodystrophy. However, the final results of an Australian study found a very high incidence: 83% of the PI group experienced some symptoms of lipodystrophy after 21 months of therapy, while 11% experienced severe body fat changes (Carr 2001). A review of 624 French patients who had been taking at least one PI for an average of 18 months found that 85% had experienced at least one physical change during that time. The most common changes were increased abdominal wall thickness, enlargement of veins in the legs and arms, increased waist size, wasting of the lower limbs and/or buttocks (each reported by 38-48% of the patients).

The French APROCO cohort found that after 12-20 months treatment the prevalence of fat loss in the face and limbs was 21%; 17% had fat accumulation and 24% had a mix of fat loss and fat accumulation (Saves 2002). A cross-sectional French study of 685 HIV-infected people on treatment in early 1999 found 59% had lipodystrophy including 63 who had never received a PI (Boufassa). A cross-sectional analysis of the Aquitaine Cohort conducted in January 1999 found 38% of 581 people on treatment had lipodystrophy (Thiebaut 2000).

The definition of lipodystrophy used in these studies was fairly broad, and an HIV-negative control group was not used. In comparison the FRAM study (see above) found that 40% of men with HIV had developed lipoatrophy, and that this was the only distinctive feature of the syndrome.

A carefully controlled study in Germany has shown that approximately 50% of people on HIV treatment developed body fat changes within two years of starting treatment (Bogner 2001). Another German cohort of 221 HIV-infected people followed for three years found 34% developed lipodystrophy (Mauss 2002).

However, a retrospective review of over 1,600 HIV-infected children on protease inhibitors found that only 1% had visibly recognisable changes in their body shape and composition (Babi 1999). But detecting body fat changes in growing children is quite complicated. Two recent studies have reported body fat changes in a larger proportion of children on HAART. Increased truncal fat was reported in 22% and 28% of the children on HAART (Arpadi 2001; Jacquet 2000) while five of 37 also had signs of peripheral wasting. Interestingly, lipids were not significantly higher among children with body fat changes compared to children with no body fat changes but insulin resistance was apparent (Jacquet). Another French study found that 24% of children had at least one symptom of lipodystrophy, but that African children were significantly less likely to experience body fat changes (Beregszaszi 2005).

An Australian survey of 724 patients conducted in 2000 found that the most frequently affected body parts were the face (45%), the legs (43%), the abdomen (39%), the arms (39%) and the buttocks (36%). When the group looked at 1348 consecutive patients seen at Australian HIV clinics during the first five months of 2000, they found that 54% had body fat changes. Of these, 52% had both peripheral and central fat changes, 37% had fat wasting only, and 11% had fat accumulation only. Of particular note is the fact that 21% of the antiretroviral naive patients had body fat changes, suggesting that some body fat changes may be confused with changes due to ageing or HIV infection itself (Miller 2000).

Similar rates of body fat changes were seen amongst the Multicenter AIDS Cohort Study (MACS) which, like the Australian study, included a high number of long-term survivors of HIV and a control group of HIV-negative men. Of the HIV-infected men, 25 to 35% had lost fat in the face, arms and legs compared to 2% of the HIV-negative group. While fat accumulation was equally common in both groups, the combination syndrome of a fat belly plus thinning of the arms and/or legs occurred in 20% of the HIV-infected men and only 1% of the negative men (Kingsley 2001).

An Indian study from an HIV clinic in Pune has shown that lipodystrophy and elevated lipid levels are just as common in patients there as in the United States. Patients in the study had been receiving the World Health Organization's recommended first line regimens of d4T or AZT with 3TC and nevirapine for over one year (median 18 months), had been over 95% adherent and had not switched drugs due to intolerance. Lipoatrophy was seen in 26% of the patients taking d4T and 10% of those taking AZT. Patients on d4T were three times more likely than HIV-negative controls to have developed fat loss. Fat accumulation was found in 10% of the d4T-treated patients and 6% of the AZT-treated patients (Pujari 2005).

The role of protease inhibitors

Lipodystrophy was initially blamed on protease inhibitor treatment, due to its recognition in the HIV-positive population within eighteen months of the widespread adoption of PI treatment.

Although an association between protease inhibitor treatment and body fat changes has been identified repeatedly in studies, this does not necessarily mean that protease inhibitors are responsible for the syndrome, or that they are its sole cause.

The problem with much of the research is that it is retrospective, so it may simply identify an association, rather than a causal relationship, between a particular drug and fat changes. That is, analysis may simply reflect which drugs are more commonly taken. Such a study cannot control for the reasons why particular drugs are prescribed. For example, d4T and indinavir have been associated with the syndrome(s) more frequently than other drugs, but this may be a consequence of the more frequent prescription of indinavir and d4T to treatment-experienced patients with advanced HIV disease after 1996. Prospective studies are needed to investigate the relationship between particular drugs and body fat changes in an unbiased manner.

The type and duration of antiretroviral therapy may be an important factor in determining what type of body fat and metabolic disorders an individual may experience. For example, various studies have found ritonavir/saquinavir, ritonavir, indinavir and saquinavir are associated with lipodystrophy. Indinavir has been particularly linked to abdominal fat accumulation (Boubaker 2002; Lichtenstein 2002).

The role of nucleoside reverse transcriptase inhibitors

Protease inhibitor therapy is not the sole factor associated with body fat changes. The Sixth Retroviruses Conference in 1999 heard of over 60 cases of lipodystrophy in people taking anti-HIV therapy that did not include protease inhibitors. Subsequent data confirmed earlier reports linking nucleoside analogues to lipodystrophy, lipoatrophy and metabolic irregularities.

Key studies have demonstrated body fat changes in people who have never taken protease inhibitors:

• A Spanish prospective study of 295 patients who started non-PI containing triple regimens found that 15% had developed lipodystrophy after an average follow-up of 19 months, with d4T treatment and greater than 12 months antiretroviral treatment associated with an increased risk of lipodystrophy (Martinez 2003).

• A Spanish study reported alterations in fat tissue in 14% of 335 people who had been on long-term NRTI therapy. Nine people developed fat accumulation, 12 fat loss, and 25 combined fat loss and accumulation with fat loss taking longer to develop. Women were more likely to experience body fat changes (Galli 2002).

NRTI-related fat loss has been linked to alterations in mitochondrial DNA, high levels of lactate in the blood and liver dysfunction (Brinkman 1999; Carr 2000).

A number of studies have also shown that peripheral subcutaneous fat levels increase, or at least cease to decline, after individuals switch from d4T to abacavir, or switch from a nucleoside analogue-containing regimen to a PI and NNRTI-based regimen. These studies are discussed in Treating body fat and metabolic changes - switching drugs in Anti-HIV therapy: Body fat and metabolic changes whilst on treatment.

Despite the substantial body of evidence linking the NRTIs to lipodystrophy, there is little evidence linking NNRTIs to lipodystrophy, although efavirenz has been associated with elevated lipid levels. Combinations including an NNRTI but not a protease inhibitor are associated with less risk of lipodystrophy, according to a recent German study (Mauss 2002).

The role of d4T compared to other nucleoside reverse transcriptase inhibitors

As some of the above studies indicate, d4T is more strongly associated with body fat changes than other NRTIs, specifically fat loss (lipoatrophy).

Two major studies presented in 2002 showed that d4T and d4T plus ddI respectively are associated with greater fat loss than AZT or AZT and 3TC when patients are followed over periods of up to two years. Another study, presented in 2003, showed that patients randomised to d4T had significantly lower limb fat content than people randomised to tenofovir after 96 weeks. The studies have convinced many clinicians that d4T does have a causative role in fat wasting, and that the drug should not be used where other treatment options exist, especially for first-line treatment.

The first study, conducted in Western Australia, tracked 53 male treatment-naive patients who received DEXA scans at baseline and during their treatment course. This was not a randomised study; it prospectively tracked individuals who initiated treatment with regimens that contained AZT or d4T. The vast majority of patients also received 3TC. Patients were followed for at least 24 months, and the study only analysed patients who took the drug throughout the 24 months follow-up (an ontreatment analysis).

At baseline, participants had approximately 22% fat content in their legs. During the first year of treatment, individuals tended to gain weight. But as time went on, leg fat declined in both treatment groups. After 24 months, this had fallen to 13% in the d4T group and 19% in the AZT group (P<0.005). No difference could be detected when fat loss was analysed according to concomitant use of protease inhibitors or non-nucleoside reverse transcriptase inhibitors. Although individuals with AIDS, and older individuals, had lower leg fat levels at baseline, these individuals did not lose fat faster or more profoundly, suggesting that these host factors do not dictate fat loss.

The study also evaluated the rate of fat loss in the legs compared to overall changes in body fat, and found that leg fat as a proportion of total body fat declined more steeply in the d4T group.

The researchers also analysed fat loss in 59 patients who had received prior treatment with AZT before commencing a triple HAART regimen. Those who switched to d4T after prior AZT treatment had greater fat loss and a greater decline in leg fat as a proportion of total fat than patients who continued on AZT for 30 months. This finding goes some way to refuting the view that fat loss associated with d4T in large cohort studies is attributable to prior AZT treatment or duration of nucleoside analogue exposure.

The researchers suggested that future studies should look in more detail at changes in fat over time rather than looking for the presence or absence of a diagnosis of lipoatrophy (Nolan 2002).

In the second study (ACTG 384) the combination of d4T and ddI was associated with a significantly greater loss of fat in the arms and legs amongst participants after 48 and 60 weeks on treatment, when compared with AZT/3TC.

One hundred and fifty six patients were randomly selected for a sub-study in which DEXA scans (a form of X-ray designed to assess body composition) were carried out at baseline and during the 80 week follow-up of the larger study. Participants in ACTG 384 were randomised to one of two nucleoside analogue backbones, AZT/3TC or d4T/ddI, and also randomised to add either nelfinavir, efavirenz or both to that backbone. If the first combination failed, individuals in the triple therapy arms switched to the remaining drugs (e.g from ddI/d4T/nelfinavir to AZT/3TC/efavirenz).

The fat loss analysis is the first time that a randomised comparison of nucleoside analogue backbones has used a relatively objective measurement of lipoatrophy to compare the effect of different drugs on fat wasting. Previous studies have used physician report, and have not reported on trends over time to the same extent.

In the ACTG 384 sub-study, individuals randomised to ddI/d4T had lost 7.5% of their baseline limb fat by week 48, despite an initial increase in fat levels during the first 16-32 weeks of treatment. In comparison, the AZT/3TC group still had median fat levels above baseline (+4.7%). This difference was statistically significant (P=0.027). This difference became more pronounced at week 64 (-13% vs +2%). Despite a higher proportion of Caucasians in the ddI/d4T group, there was no difference when the results were analysed by racial group.

The analysis also looked at trunk fat, but did not distinguish between subcutaneous and visceral fat. Levels of trunk fat remained above baseline in both groups at week 64, but trunk fat levels were significantly higher in the AZT/3TC group.

The sub-study was not powered to look at differences between triple regimens in effects on limb fat. At the time the study was designed (1998), the suggestion that protease inhibitors and nucleoside analogues might have additive influences on the development of lipodystrophy had not surfaced. Instead, the sub-study was designed to compare the effects of the most commonly prescribed protease inhibitor of the day (nelfinavir) with the new drug in town (efavirenz).

Loss of limb fat only became apparent at week 60 in the nelfinavir group and at week 80 in the efavirenz group. At week 80, limb fat had declined by 18% in the nelfinavir group and 10.7% in the efavirenz group (p=0.019) (Dubé ²002).

It should be noted that significant fat loss was also observed in the AZT-containing arms in the two studies cited above that used DEXA scans rather than physician observation and anthropometric measurement to determine rates of fat loss.

Several other prospective studies have linked d4T and fat wasting.

For example, Gilead 903 compared tenofovir with d4T against a background of efavirenz and 3TC. At week 96, limb fat data were available on 132 d4T-treated patients and 126 tenofovir-treated patients, although baseline characteristics were not stated (Staszewski 2003).

Twelve percent of patients in the d4T arm had developed investigator-defined lipodystrophy by week 96 of the study, compared to 1% of patients in the tenofovir arm. Analysis of total limb fat by DEXA scan, designed to measure fat loss from the limbs, showed that tenofovir-treated patients had significantly greater fat content at week 96 (an average difference of 6lbs) (p<0.001).

The investigators did not report on changes in limb fat from baseline, clearly an important factor in assessing long-term toxicity. However, patients in the tenofovir-treated group were reported to have gained more weight than the d4T group after 96 weeks (mean change +6lbs versus +1lb, p=0.02), which might support the view that fat loss had occurred among patients who received d4T to a greater extent than amongst those who received tenofovir, given that weight gain due to reduced resting energy expenditure is a well recognised consequence of initiating HAART.

Another large randomised study, CPCRA 058, reported that when compared with abacavir/3TC treated patients, d4T/ddI-treated patients lost significantly more subcutaneous fat after a median of 33 months. CPCRA 058 randomised treatment-naive patients to PI-containing, NNRTI-containing or PI and NNRTI-containing regimens, and included a sub-study comparing two nucleoside analogue backbones. A metabolic substudy recruited 96 patients to undergo measurements of body fat including skin fold thickness, waist and hip measurements, and total body fat using bioimpedance assay (BIA). By every measure, the d4T/ddI treated patients had lost fat after 33 months, whilst the abacavir/3TC group had gained fat (Raghavan 2003).

However, not all studies have confirmed a particular role for d4T in fat wasting. For example, one study which followed 354 HIV-infected people for a year found duration of AZT and duration of d4T equally associated with lipodystrophy, while fat loss was more significantly associated with AZT than d4T (Laurenroth-Mai 2002). The French NOVAVIR study found little difference in fat loss between drugs (Joly 2001). This randomised study compared people switching to d4T/3TC or AZT/3TC after approximately two years of exposure to AZT, ddI or ddC. After 30 months follow-up, the only body fat measurement that differed between the two groups was the thigh skinfold fat measurement - d4T recipients had less thigh fat. There was a significant difference in body fat by clinician assessment; 70% of d4T recipients had fat loss compared with 44% of the AZT recipients. It should be noted that the reliability of these clinical assessments has been challenged because doctors knew which drug the patient was taking, which may have influenced their assessments.

Another study to implicate d4T in fat wasting was the ALBI study, in which patients were randomised to receive either AZT/3TC, d4T/ddI or to alternate these combinations. The ALBI study showed a 6.7-fold higher risk of lipodystrophy in patients who received d4T/ddI (largely lipoatrophy). Adding a protease inhibitor after six months did not increase or reduce the risk. This study also found that alternating d4T/ddI with AZT/3TC resulted in an increased risk if the first regimen was d4T/ddI, but not if the first regimen was AZT/3TC. The study also found that patients with lipodystrophy were more likely to have viral load below 50 copies at month 30 (indicating the long-term success of treatment). Only modest changes in metabolism were seen on nucleoside analogue therapy.

The drawback of the ALBI study, as with NOVAVIR, is that clinicians were not blinded to the treatment regimen of their patients, and so may have been influenced by reports that d4T was associated with lipodystrophy. Also, over 40% of cases were described as `mild`(Chene 2002).

The mechanism by which nucleoside analogues might cause fat loss is unclear. One suggestion is that nucleoside analogues may damage the DNA of mitochondria in fat cells (adipocytes) that store fat in the limbs. These issues are discussed in more detail in Possible causes of body fat and metabolic changes in Anti-HIV therapy: Body fat and metabolic changes whilst on treatment.

A number of other observational studies have also found a link between d4T and fat wasting. In some instances, other NRTIs are also implicated. Key observational findings are summarised below.

• Length of time on d4T increased the risk of fat wasting in the limbs and face irrespective of prior treatment experience, CD4 count or viral load in a study of 43 people taking dual nucleoside combinations at a French clinic (Saint-Marc 1998). This team subsequently reported that stopping d4T led to increased subcutaneous fat.

• An Australian study of 227 individuals found that duration of exposure to dual nucleoside analogue therapy prior to commencing protease inhibitor therapy, and length of time on d4T, were associated with subcutaneous fat loss. In individuals who started treatment with PI-based HAART, fat loss was significantly more likely in those who received d4T, suggesting a direct relationship between d4T and fat wasting. Even after adjusting for the duration of prior nucleoside analogue exposure and any other confounding factors, d4T treatment was still associated with a greater risk of developing fat loss (Mallal 2000).

• Fat wasting also occurred in those taking AZT-based therapy: after 24 months over 50% of AZT recipients had evidence of fat wasting, compared with 80% of d4T recipients. Furthermore, those who added a PI to dual nucleoside therapy developed signs of fat loss more quickly than those who took dual nucleoside therapy, suggesting that PIs accelerate any underlying effect of nucleoside analogues on fat loss (Mallal 2000).

• A much stronger association was reported between d4T/ddI (89%) or d4T/3TC (52%) and body fat changes when compared with AZT/3TC (4%) in 156 individuals (30% women) mainly receiving indinavir as their protease inhibitor. Forty-one percent of the patients developed body fat changes after less than nine months on HAART (Polo 2000).

• A prospective study of 40 men followed for at least 96 weeks found that commencing HAART with a d4T-containing regimen was associated with fat loss throughout 96 weeks, regardless of whether d4T treatment was sustained throughout that period (Mallon 2003).

• Long term 3TC treatment (3 years plus) was most strongly associated with fat redistribution in women, although there was also a strong association with d4T treatment. Fat redistribution was less likely in women taking AZT/3TC compared to d4T/3TC (Gervasoni 1999).

• An analysis of 494 consecutive patients commencing HAART in a Barcelona clinic found that duration of HAART was associated with an increased risk of any form of lipodystrophy, as was increasing age, but duration of d4T treatment further increased the risk of fat loss compared to other drugs (relative risk 1.3 per 6 months of d4T treatment) (Martinez 2001).

• A German study followed 221 patients with identical duration of therapy who started HAART as antiretroviral naive patients in mid-1996. 34% had lipodystrophy by 2000, and the multivariate analysis showed that prolonged exposure to d4T doubled the risk of fat changes. People who took NNRTI-based HAART had a significantly lower risk of lipodystrophy (Mauss 2002). This study did not use DEXA and was an unrandomised cohort analysis.

• However another German study looked at people who had taken one HAART regimen only, and who had never switched between d4T and AZT. Patients had taken HAART for an average of 23 months. No difference in the incidence of fat loss or fat accumulation was found; only duration of HAART was significantly associated with the risk of developing body fat changes (Bogner 2001).

Multiple risk factors for fat loss

d4T may be an innocent bystander in the view of some researchers, blamed because it is associated with the factors that may actually be causing the problem of fat loss.

An analysis of patients in the Hospital Outpatients Study (HOPS) found no association between particular drugs and fat loss, and instead found that the following factors were associated with fat loss:

• CD4 cell count below 100 cells/mm³ after two years follow-up.

• CD4 cell count increase of less than 50 cells/mm³ after nearly two years of observation on treatment, regardless of drugs taken or total duration of treatment.

• Low to average body mass (below 24kg/m², another marker of advanced HIV disease.

• White race (a possible marker of better access to care).

The cohort followed 337 patients for an average of 21 months, during which time 44 patients developed moderate to severe lipoatrophy (Lichtenstein 2003).

Patients with low CD4 cell counts at the time of the second study visit were found to be more likely to have had low CD4 cell counts earlier in the course of their HIV infection. Indeed, a lowest ever (nadir) CD4 cell count of less than 100 cells/mm³ at any time as early as 1994 was predictive of the emergence of lipoatrophy.

Lipoatrophy was unusual in the absence of these non-drug-related factors.

The study results were also controlled for diseases causing fat loss, including MAI, pneumonia and depression. Such illnesses occurred in 18.2% of the group with lipoatrophy, but in over 40% of patients with no fat wasting.

These findings suggest that later initiation of therapy may increase the risk of lipoatrophy.

However, some of these factors were included in the multivariate analysis carried out in the Western Australia study (see The role of d4T compared to other NRTIs above), and the researchers discovered that whilst older age and a prior AIDS diagnosis predisposed towards lower leg fat levels at baseline, they did not affect the degree of fat loss on treatment (Nolan 2002). Age over 40 years and elevated triglycerides at baseline have been linked to lipoatrophy (Noor 2004).

Furthermore, research into the phosphorylation of nucleoside analogues in cells has shown that higher levels of AZT triphosphate are present in the cells of people with CD4 cell counts below 100, suggesting a mechanism for the interaction between immune suppression and body mass seen in the HOPS study.

Another large prospective analysis, of the Italian Cohort of Antiretroviral-Naﶥ Patients, found that sex, exposure route and age all affected the risk of developing lipodystrophy, whilst individuals who added a PI to a dual NRTI combination and PIs were significantly more likely to develop lipodystrophy than those who took dual NRTI combinations (Galli 2002). After controlling for other risk factors, d4T was not found to be associated with lipodystrophy.

The researchers also found that HCV coinfection (present in 36% of the cohort) was a significant risk factor for development of lipoatrophy, as was older age. However, exposure to HIV by injecting drug use was associated with a lower risk of lipodystrophy, leading the authors to speculate that poorer adherence due to lower educational and income level may have affected the degree of drug exposure experienced by this group of patients. This finding tends to reinforce the view that drug-related factors play a substantial part in the development of the syndrome.

On the other hand, a gender difference not linked to treatment history also emerged. Women were significantly more likely to develop fat accumulation or a mixed syndrome of fat gain and fat loss.

Nucleoside reverse transcriptase inhibitors and protease inhibitors together

A study presented to the 2001 International AIDS Society meeting in Buenos Aires showed that the risk of developing lipodystrophy is vastly increased when nucleoside analogues are combined with protease inhibitors.

A long-term study of ritonavir/saquinavir treatment was analysed to determine whether patients who intensified their treatment with two nucleoside analogues after 48 weeks (or due to viral rebound) had higher rates of lipodystrophy. Body composition was examined after three years on treatment, and at year 4, and showed that whilst 23% of those who had intensified treatment had developed three or more body fat abnormalities, none of those who received treatment solely with protease inhibitors had developed severe lipodystrophy. Less severe lipodystrophy, classified as one change, was seen in 18% of the PI-only group, and 44% of the PI + NRTI group. However, rates of metabolic abnormalities were similar in the two groups. Intensification with NRTIs resulted in a 8.4-fold increase in the risk of body fat changes after adjusting for age, baseline CD4 cell count, duration of prior NRTI treatment and baseline viral load. A further analysis of this study showed that patients who added an NRTI to the protease inhibitors were eight times more likely to experience facial fat loss and six times more likely to lose buttock fat (Cohen 2002).

It was not possible to analyse ACTG 384 to determine whether a particular combination of the protease inhibitor (nelfinavir) and NRTIs led to greater fat loss or fat accumulation. However, an analysis of the Western Australia study found no difference in the rate of fat loss when controlling for use of either NNRTIs or protease inhibitors.

Duration of treatment and evolution of fat changes

The emergence of physical changes has been associated with the length of time on protease inhibitors and NRTIs in some studies (Duncombe 2000; Carr 2001; Lichtenstein 2001). However, the prevalence of body fat changes stabilised in the large MACS study after two years of treatment, which may suggest that some individuals are not at risk of lipodystrophy no matter how long they stay on treatments (Kingsley 2001). Prolonged treatment with an NNRTI-based combination has been associated with reduced risk of lipodystrophy when compared with prolonged PI treatment (Mauss 2002).

A US study of 1077 patients in three large clinical practices found that new cases of lipodystrophy were still being diagnosed in people who had been taking antiretroviral therapy for at least three years, with a 12.5% incidence of new cases during years 3 to 5 of antiretroviral therapy, compared with an incidence of 4% per annum up to year 3. However, this study may underestimate the incidence rate at all stages because it was carried out in the last quarter of 1998, when protease inhibitors had been standard therapy for less than two years in the USA (Lichtenstein 2001; 2002).

Over the course of one year, another study found, 5% of men and 2% of women developed fat loss in a cohort of 433 men and women, in which 18% of men and 7% of women already had signs of fat loss at the beginning of the follow-up period (Jacobson 2002).

A prospective study of the development of body fat changes in patients commencing PI-containing regimens found that after 24 months, changes in body fat distribution were modest; only reductions in leg fat were found to be significant (Walmesley 2002).

An Australian natural history study which followed 40 men who started PI or NNRTI-containing HAART, using DEXA scans at baseline, at weeks 12, 24, 48 and then every 48 weeks for at least 96 weeks found that:

• Individuals gained limb fat until week 24 (as observed in the ACTG 384 study)

• Individuals then began to lose limb fat; they lost a median of 13% of limb fat per year (Half of the group were followed for at least three years).

• Central fat accumulation was detectable in PI treated but not NNRTI-treated patients by week 24 (Mallon 2003).

No evidence of the reversion of body fat to a normal distribution has been reported, except in certain studies of treatment switching. See Treating body fat and metabolic changes - switching drugs in Anti-HIV therapy: Body fat and metabolic changes whilst on treatment for further details.

Australian and French doctors have reported the appearance of lipodystrophy in people treated during primary infection. Of a group of 41 patients who began treatment during primary infection and were studied over a six month period, where the average duration of therapy was 11.7 months, 22 self-identified fat redistribution, and this was confirmed by their doctor in six cases (Miller 2000). In France, 14 out of 84 patients prescribed HAART during primary infection developed body fat changes after a mean of 18.9 months, and among those who had been followed for two years, the incidence was 26% (Goujard 2001).

A German doctor reported the development of substantial abdominal fat accumulation within six weeks of a four week course of post-exposure prophylaxis with a d4T and efavirenz-containing regimen. A previous course had been given ten weeks earlier. The fat accumulation persisted for at least three months (at the time of report it had not gone away). The patient did not develop any metabolic abnormalities (Mauss 2002b).

Pre-treatment weight

Another trend that has emerged from recent research into lipodystrophy is that fat accumulation is more common among people who are overweight, or have a high body mass index, before starting therapy. People with a lower body mass index are more likely to experience fat depletion (Muurahainen 2000).

An Italian study of 704 people followed for a mean of 96 weeks after starting antiretroviral therapy found that higher body weight increased the risk of fat accumulation, but did not increase the risk of a mixed syndrome of fat accumulation and peripheral fat wasting (Galli 2000). Another study reported that those with high body mass index at baseline were more likely to develop hypertrophy (increase in cell size) or a mix of hypertrophy and lipoatrophy (Raffi 2000).

An Australian prospective study of 40 men found that higher lean body mass at baseline was associated with the development of lipodystrophy (Mallon 2003).

Age

The Australian Lipodystrophy Prevalence Survey found that the risk of body fat changes increased with age. Individuals aged 40-49 had a two fold increased risk compared with those below 40, and the risk increased four fold in those over 50 years old (Miller 1999).

This pattern has been reflected in most other studies of risk factors for lipodystrophy.

Gender

Gender affects body fat changes. Several studies have found that women are more likely to experience fat accumulation, and to experience more severe accumulation of fat, whereas men are more likely to have severe fat depletion. Men also seem more likely to experience metabolic abnormalities. The reasons for these gender differences are not understood.

For example, the Self-Ascertained Lipodystrophy Syndrome Assessment (SALSA) study found that 97% of women and 84% men reported fat accumulation. In contrast, 77% of men and 61% of women reported fat loss of the face, limbs and/or buttocks. Men were significantly more likely than women to develop high lipid levels. Only 32% of the women in this study had high triglycerides and high cholesterol (Muurahainen 2000; Falutz 1999).

A prevalence study conducted in the US has also found that women were more likely to experience fat accumulation than fat loss (63% and 11%) compared with men (36% and 19%). However, race cut across this gendered effect; African American men and women were less likely to experience fat accumulation (Wanke 2000)

An Italian study of 704 individuals followed for an average of 96 weeks after starting antiretroviral therapy found no association between gender and fat accumulation only, but women were significantly more likely to experience a mixed syndrome of fat accumulation and peripheral fat wasting than men (Galli 2000).

However a DEXA analysis of 183 HIV-positive women in the Womens Interagency HIV Study and a control group of HIV-negative women found that whilst HIV-positive women receiving antiretroviral treatment had significantly lower levels of leg fat than either untreated HIV-positive women or HIV-negative women, truncal fat was also lower in HIV-positive women than in HIV-negative women. These results showed a similar pattern to the findings of the FRAM study in men (Mulligan 2004).

Men and women also appear to have different risk factors associated with body fat changes. A review of 526 patients with lipodystrophy in New York and Montreal found that whilst fat accumulation in men was associated with higher body mass index and viral load below 500 copies, these factors were not associated with fat accumulation in women. In men, fat wasting was significantly associated with d4T treatment, whereas this association was not evident in women. Instead, fat wasting was significantly associated with a body mass index below 28.5, an association that was not evident in men (Muurahainen 2000).

An Italian study of 2258 individuals (29.8% women) found that women were significantly more likely to develop body fat changes, but men were significantly more likely to develop lipoatrophy, despite no significant difference in d4T exposure between the two groups (Galli 2003).

Other possible causes - response to treatment

Some researchers have suggested that factors other than antiretroviral drugs are causing body fat changes and high blood lipids. Dr Donald Kotler has conducted a retrospective study of body composition among people with HIV, compared to HIV-negative controls, and found that body composition did not differ among the HIV-positive group according to previous therapy. However, Kotler reported that body composition changes were associated with the intensity of HIV infection.

There is some evidence to support this theory. One study (Keruly 1998) found that high lipid levels were associated with greater rises in CD4 count and greater declines in viral load in comparison to people who did not develop high lipids, although two studies contradicted this finding.

A review of 155 patients, the majority male, from the Tufts University Nutrition for Healthy Living Cohort found that higher baseline CD4 count was associated with greater trunk fat increase, but HAART was not predictive of fat changes. Higher baseline viral load was associated with greater fat loss in both the trunk and the limbs (-3.2% per year per log₁₀ HIV RNA), but fat loss was not predicted by baseline CD4 count. The maximum follow-up was 24 months (McDermott 2003).

Unlike some other studies, the Western Australian team cited previously found no association between the severity of lipodystrophy and current CD4 count or viral load. However, CD4 count prior to PI treatment was a predictor of current total fat mass, as was duration of time on protease inhibitors and weight prior to treatment (Carr 2001).

Although viral load response to therapy has not been correlated with body fat changes in any study, a small French study found that people who experienced larger CD4 increases after commencing HAART were more likely to experience body fat changes. In 18 people, the average CD4 count among people with body fat changes was 351, compared with an average CD4 count of 203 among people who did not develop fat changes (Bani-Sadr 1999).

In contrast, a review of the HOPS study found that poor CD4 cell reconstitution was associated with fat loss; those with CD4 cell increases of less than 100 cells/mm³ after two years of follow-up were more likely to have lipoatrophy (Lichtenstein 2002).

Lowest CD4 count ever, degree of CD4 cell reconstitution on treatment and CD4 nadir below 200 have all been associated with severity of fat redistribution (Lichtenstein 2001; Mauss 2002), as have viral load below 50 copies/ml after 30 months of treatment (Chene 2002).

Race

Race may impact on lipodystrophy. Two studies have found that African Americans are less likely to experience body fat changes than other races (Wanke 2000; Lichtenstein 2001). An Australian study found that white race was associated with a greater prevalence of lipodystrophy compared with Asian and African races (Mallal 1999). Absence of lipodystrophy has been reported amongst Koreans (Chang 2002).

However, a prospective study of 12 black (mainly African) and 11 white men in London found no difference in metabolic and fat changes based on race (Day 2000).

Diet

Studies have not looked at the effect of dietary changes on body fat redistribution, but an Australian study of fat intake in people with and without lipodystrophy found no significant difference in the proportion of total calories derived from saturated fats. They pointed out that in the HIV-negative population, people with body changes indicating a high risk of heart disease (such as visceral fat accumulation) do derive a higher proportion of their calorie intake from saturated fats than the rest of the population, and can modify their risk of heart disease and reduce fat accumulation by adjusting their diet. In contrast, dietary changes probably have little impact on the mechanisms causing body fat changes and elevated lipids in HIV-related lipodystrophy, the authors argued (Batterham 2000).

However, a cross-sectional American study found that high intakes of polyunsaturated fats and alcohol, and a low fibre intake were strongly associated with metabolic abnormalities and lipodystrophy (Hadigan 2001). Some dietary patterns observed may have been a consequence of a changing need for calories due to body fat changes. For example, those experiencing fat loss had a higher daily calorie intake than those with either abnormal fat accumulation or a mixture of fat accumulation and wasting. The authors suggested that a prospective study (which follows people before they develop metabolic or body fat changes) may help to establish the extent to which pre-existing dietary patterns influence metabolic changes.

Body fat changes in children

Body fat changes have also been reported in children receiving HAART. French researchers carried out a cross-sectional study in 39 children receiving HAART, and found body fat changes in one third of the group. The combined syndrome of fat wasting and central fat accumulation tended to be more frequent and severe in adolescents, consistent with the observation that inherited forms of lipodystrophy tend to develop or worsen after the onset of puberty. Changes in glucose tolerance were also less pronounced in pre-pubescent children.

Twenty-three per cent of children without body fat changes in the cohort had lipid abnormalities, suggesting that even in the absence of physical changes, an underlying metabolic distrurbance exists. The long-term implications of elevated lipid levels in children are also a matter of concern, because these changes may predict an increased risk of early onset heart disease in adult life (Jaquet 2000).

A European review of 374 children found that 28% of children had at least one sign of lipodystrophy, and that girls were twice as likely to show signs of lipodystrophy than boys. Children aged 13 or over were more likely to show signs of central fat accumulation (Vigano 2003).

Another European study, of 477 HIV-positive children and adolescents, found that lipodystrophy was associated with an AIDS diagnosis, female sex, and the use of a protease inhibitor or d4T. 26% of the children and adolescents showed signs of body fat redistribution. Of these 46 (37%) had a combination of both fat loss and fat gain, 42 (34%) had fat accumulation, and 36 (29%) had fat loss (European Paediatric Lipodystrophy Group 2004).

Four other cross-sectional studies have also reported on body fat changes. Seven of 35 children receiving HAART (mean age 9 years) had body fat changes in one study, while a further 19 had metabolic but not physical changes (Amaya 2002). In a second group of 29 children (median age 6 years), 8 showed signs of fat redistribution and 17 had metabolic changes (Meneilly). Neither group showed any relationship between duration of PI treatment and body fat or metabolic changes, and no relationship with age was reported. A third study found that about 20% of 85 children had body fat changes, most commonly increased abdominal fat, while nearly half had some sort of lipid abnormality (Tellol 2002). Of those with body fat changes, all but one were taking protease inhibitors. Finally, an Italian group compared body fat changes in 37 HIV-positive children and adolescents receiving HAART matched with 52 age-matched controls. The majority (26) entered puberty during the study period, and the researchers found that the stage of puberty and the duration of HAART exposure were the only significant risk factors for development of lipodystrophy. Girls were protected from fat loss, as were children with a higher body mass index (Vigano 2003).

Connection with metabolic disorders

People with lipodystrophy commonly have elevated levels of fats (called lipids) in their blood (high triglycerides and high cholesterol) and insulin resistance. However, the connection between high lipids and other metabolic abnormalities is not fully understood.

Whilst some studies show that elevated triglyceride levels predict the development of lipodystrophy (Galli 2002), others fail to show a relationship. Once again, the use of cross-sectional rather than prospective data from large randomised studies make it difficult to draw conclusions.

Somewhat unreliable data has also come from a post-hoc analysis of two randomised studies comparing standard and slow-release d4T. Lipoatrophy was associated with the use of d4T, elevated baseline triglycerides (over 200mg/dl or 2.25mmol/l) and age over 40 years (Noor 2004).

See Metabolic changes on antiretroviral therapy in Anti-HIV therapy: Body fat and metabolic changes whilst on treatment for a full discussion.

Treatment of body fat changes

For a discussion of treatment of body fat changes, see Treating body fat and metabolic changes.