The confusion surrounding the etiology and treatment of HIV-related metabolic complications has been maddening for both patients and their clinicians. Different studies -- typically small and powered for endpoints other than changes in body fat -- have produced contradictory results. The few interventions now available that have proved to be effective generally produce only marginal gains and are often difficult to obtain and/or are expensive. Further frustration stems from the inability of experts in the field to agree on the definition of body shape changes or whether certain changes even exist. In the absence of conclusions, anecdotes become mistaken for evidence and belief systems are entrenched as dogma.

Given this data-vacuum, it was satisfying that the results from a series of studies of body shape changes presented at the 2006 Conference on Retroviruses and Opportunistic Infections (CROI) were complementary -- together painting a more focused picture of the role of the insulin-sensitizing thiazolidinediones (also called glitazones) and metformin in reversing changes in regional adipose tissue volume. In addition, long-awaited data on the use of testosterone on visceral adipose tissue were presented.

All That Glitters: Glitazones for Changes in Fat Mass in HIV-Infected Patients

The incidence of fat wasting at the hands of stavudine (d4T, Zerit) has decreased with the abandonment of this nucleoside reverse transcriptase inhibitor (NRTI) in the United States, Europe and Australia. However, veterans of this therapy still confront disfiguring lipoatrophy and demand continued study of interventions to reverse these changes. Additionally, stavudine is not alone in its ability to reduce fat. Evidence suggests some patients receiving zidovudine (AZT, Retrovir), still commonly used as a first-line treatment for HIV infection, can also experience significant loss of limb fat and subcutaneous adipose tissue of the abdomen.

A lingering issue in the field of metabolic disorders has been the role of glitazones in the management of lipoatrophy. These drugs are attractive candidates as agents to reverse lipoatrophy given conceptual models of the etiology of changes in fat depots during HIV infection. The largest study of a glitazone used to treat HIV-associated lipoatrophy was the Australian ROSEY study,1 which did not find a statistically significant difference in limb fat gain between rosiglitazone (Avandia) (4 mg twice a day) and placebo in lipoatrophic patients. When presented and published, these results signaled the end was near, if not already here, for any application of glitazones for fat wasting.

However, a handful of smaller studies, including a 39-subject trial comparing rosiglitazone to metformin, suggested that there was an effect of rosiglitazone in increasing subcutaneous adipose volume in patients with lipoatrophy.2-5 Further, some aspects of the ROSEY study raised questions about whether the results were as clear-cut as presented. For example, the control group experienced an increase in limb fat that reduced the difference between the two study arms. In addition, a surprisingly high proportion of the ROSEY participants were on abacavir (ABC, Ziagen), a drug that the investigators had used in a previous study, MITOX,6 which switched patients from stavudine or zidovudine to abacavir, suggesting some of the ROSEY study participants had also participated in MITOX. Rosiglitazone for lipoatrophy refused to die.

Kathleen Mulligan of University California-San Francisco presented the results of AIDS Clinical Trials Group (ACTG) study 5082, a study of rosiglitazone (4 mg daily -- as opposed to the 4 mg twice a day of the ROSEY study and several other trials of this drug) versus metformin (Glucophage, Fortamet) (1,000 mg twice daily) versus rosiglitazone + metformin versus double placebo.7

Unlike the ROSEY study, which enrolled those with lipoatrophy, ACTG 5082 targeted patients with increased abdominal girth and, importantly, evidence of insulin resistance or glucose intolerance based on fasting insulin levels and glucose tolerance testing. The median trunk fat at study entry was 12.5 kg, which by way of comparison was 5 kg greater than the median trunk fat of participants who were antiretroviral therapy naive upon entering another ACTG study, A5005s. However, although designed as a study of visceral hypertrophy, the median leg fat of the cohort was low at 3.8 kg. This was 1 kg less than the median leg fat of A5005s participants at baseline.

The study was relatively small, with about 25 patients in each of the four study arms. Although powered primarily for glycemic outcomes, it is the body shape changes that are of greatest interest. After 16 weeks, there was no difference between the treatment arms and double-placebo in visceral or subcutaneous abdominal fat as assessed by dual energy x-ray absorptiometry (DEXA) scanning. However, leg fat increased significantly compared to double-placebo in the rosiglitazone study arm (P = .034).

The change in fat in the rosiglitazone-only group was about 200 g (a 5% increase). While rosiglitazone increased leg fat, it was not observed to improve arm fat compared to double-placebo. In contrast to the ROSEY study, in which differences between rosiglitazone and placebo were obfuscated by the increase in limb fat of the control subjects, in A5082 the continued fat wasting of the double-placebo-assigned subjects helped to drive the statistical difference between study groups.

Interestingly, metformin had no effect on limb fat compared to double-placebo and seemed to negate the effect of rosiglitazone when the two were combined. Further, metformin was poorly tolerated, leading to excess treatment discontinuation among those taking this drug. Of the 26 patients who were assigned to metformin alone, eight developed treatment-limiting toxicity, including increases in lactate (n = 4), diarrhea (n = 3) and an unspecified toxicity (n = 1).

Neither metformin nor rosiglitazone led to major changes in triglyceride levels. However, there was a trend toward an increase in low-density lipoprotein (LDL) cholesterol and low high-density lipoprotein (HDL) cholesterol among patients assigned rosiglitazone. An opposite trend toward lower LDL cholesterol and higher HDL cholesterol was observed in the metformin group. As expected, both drugs improved insulin sensitivity. In addition, there were increases in adiponectin -- a fat-derived protein inversely correlated to visceral fat and insulin resistance -- seen with both agents.

When contacted by e-mail after the conference, Dr. Mulligan suggested that one explanation for the lack of an effect of metformin on visceral adipose tissue in A5082, despite the opposite finding by Hadigan and colleagues,8 was that a substantial proportion of the metformin-assigned participants in A5082 required dose reductions and/or treatment discontinuation secondary to stringent toxicity management guidelines, limiting the ability to detect significant changes in this group. She also noted that continued NRTI use could have muted the effects of both agents as was clearly demonstrated in a study of pioglitazone (Actos) described below. Unfortunately, a breakdown of the NRTIs taken during the study was not presented.

Consistent results are what lead to changes in standard of care and it was a relief when the results of two other studies presented at CROI generally supported the findings of ACTG 5082. Willy Rozenbaum of the French Agency for AIDS Research (ARNS) took the podium (do French physicians wear all black even when seeing patients?) and reported that pioglitazone produced a small but statistically significant increase in limb fat, as measured by DEXA scanning in a randomized, placebo-controlled study of 110 patients with self-reported and clinician-confirmed lipoatrophy.9

No evidence of insulin resistance was required for entry into this study and 28% of the patients were receiving stavudine at baseline. Of the 64 patients who were assigned to treatment with pioglitazone (30 mg daily) for 48 weeks, there was a 380 g increase in limb fat compared to a negligible 50 g increase in the placebo arm (P = .051).

However, those assigned pioglitazone who were receiving stavudine had only a 170 g increase in limb fat mass. Analysis restricted to those not taking stavudine found a 450 g gain in limb fat mass with pioglitazone -- a statistically significant difference from placebo (P = .013). There was no significant change in abdominal visceral or subcutaneous fat in either study arm.

Change in Limb Fat From Baseline at 48 Weeks
No d4Td4T
450 g40 g.013170 g70 g.4

As has been seen with almost every interventional study that has found a statistically significant increase in fat in lipoatrophic patients -- such as antiretroviral switch studies -- the gain was small, considering the severity of the fat loss among the patients, and was not well-appreciated by the trial participants using a visual analog scale to rate their limb fat.

However, someone with a glass-half-full perspective would appreciate that pioglitazone seemed to prevent further fat loss and that longer-term administration may yield further fat gain. Additionally, pioglitazone was well tolerated.

As opposed to the results from A5082, lipid levels did not substantially change during therapy with this glitazone -- which has not been as well studied as its rosier sister drug in HIV-infected patients. In fact, HDL cholesterol increased during the study among those in the intervention arm -- although it was pointed out that this was a study of French patients and not Americans, who are more prone to the super-size-it-syndrome. A major finding of this study was that pioglitazone did not lead to significant changes in limb fat among the participants who remained on stavudine.

The absence of a positive effect of metformin on fat depots described in A5082 was also reported by Rakhi Kohli from Tufts University.10 In this double-blind, placebo-controlled trial, 48 patients received metformin 1,500 mg daily. In contrast to other studies, including A5082, in which trial participants with increased abdominal girth were required to have insulin resistance or glucose intolerance, the participants in this trial had to have a high waist-to-hip ratio but a fasting insulin level less than 18 uU/mL. Despite this, the cohort tended to have mild to moderate glucose intolerance on glucose tolerance testing. At study entry, limb fat mass was 9.8 kg and 12.3 kg in the metformin and placebo arms, respectively. This is in stark contrast to the lipoatrophy seen in the French study in which baseline limb fat was approximately 3 kg.

After 24 weeks, the metformin-receiving participants did not have a statistically significant change in visceral fat, as measured by computed tomography (CT) scan, even after adjusting for age, height and baseline visceral adipose tissue volume (-10.1% in metformin group versus -3.2% in placebo group, P = .58).

In the unadjusted analysis, there was a significant decrease in limb fat among those assigned metformin. Metformin resulted in a 686 g drop in limb fat volume as measured by DEXA compared to a 161 g increase in limb fat mass among the control subjects (P = .03).

In the adjusted analysis, however, the difference between the two groups became statistically insignificant -- mostly due to a reduced increase in limb fat among the controls (-614 g for metformin versus +95 g for placebo, P = .12). There was no significant change in lipids during metformin therapy in this small study. Interestingly, metformin did not produce improvements in glycemic or insulin parameters. Body mass index did drop with metformin -- a well-described effect. Safety data were not reported.

This study supports the finding of A5082 in which there was a trend toward reduced leg fat in the metformin-assigned participants and in which metformin negated rosiglitazone-associated increases in limb fat when the two were combined. The authors correctly warn that a consequence of metformin usage is a loss of limb fat and that metformin should not be used in patients without marked insulin resistance and ample limb fat.

Collectively, these three studies tell us several things. First, glitazones as a treatment for lipoatrophy are not off the table. In both studies that looked at these agents, there were significant increases in limb fat -- findings that are consistent with several other small studies2-5 but not with the one large randomized controlled study of rosiglitazone,1 during which increases in limb fat were seen in the treatment and control arms, or another 30-person randomized study.11

The finding that rosiglitazone can increase LDL cholesterol in some studies calls for the careful monitoring of lipids when glitazones are prescribed to HIV-infected patients. Second, the results suggest that insulin resistance need not be a prerequisite for glitazone therapy for lipoatrophy. Rather, removal of the offending NRTI appears to be critical to success.

Third, these studies continue to demonstrate that the changes being observed -- as was the case for antiretroviral switch studies -- are positive but small. Little if any improvement was noted by patients, but the studies were generally short in duration. Unfortunately, facial fat was not assessed in any of these studies. Lastly, metformin did not look good. More fat wasting, limited effect on visceral fat (particularly in those without insulin resistance) and toxicity all mean that metformin is not ready for prime time, and, if used, should be reserved for those with clear insulin resistance and no peripheral lipoatrophy.

Another large study of glitazones for lipoatrophy would have been helpful to settle the issue. Such a study was planned by the ACTG several years ago but was scuttled when it was learned that the ROSEY study was underway. In the absence of such a study, clinicians need to examine the data that are available and consider whether the magnitude of the effects reported, the toxicity of therapy and other costs support the use of glitazones to treat fat wasting. What about fat accumulation? Read on.

Testosterone Supplementation for Visceral Fat

Androgens are lipolytic. Their ability to reduce fat cells is evident in the chiseled bodies of the doping weight lifter and the lean-limbed HIV-infected guy who takes androgens to look good in T-shirts. Cecilia Shikuma presented the results of ACTG 5079, a randomized, controlled trial of testosterone supplementation in 88 HIV-infected, HAART (highly active antiretroviral therapy)-treated men with abdominal obesity, as assessed by waist-to-hip ratio, and mild to moderately reduced testosterone levels (total testosterone level 124-400 ng/dL or, if greater than 400 ng/dL, a level of bioavailable testosterone that was below normal).12

Trial participants received 10 g of testosterone gel to apply topically or placebo during the first 24 weeks of the study. This was followed by an open-label phase that lasted an additional 24 weeks. At baseline, participants' median trunk fat -- as assessed by CT scanning -- was 11.5 kg and their median limb fat by DEXA scanning was between 4.8 and 5.7 kg.

At 24 weeks, visceral adipose tissue volume rose slightly in both study arms, but the change was not significantly different from baseline. Subcutaneous adipose tissue of the abdomen decreased significantly in the testosterone group by 7.2% and rose in the placebo group by 8.1% (difference between arms P = .001). Total abdominal fat (a composite of visceral and subcutaneous abdominal fat) was significantly different between the groups (P = .04). Not surprisingly, testosterone also reduced extremity fat. Limb fat fell by almost 0.5 kg in the active treatment group compared to a slight increase in limb fat in the placebo-assigned participants. Testosterone was found to be well tolerated, but lipid and insulin parameters have yet to be analyzed.

In this study, testosterone did not reduce visceral fat but led to decreases in subcutaneous fat that improved the waist-to-hip ratio and increased lean tissue mass. Limb fat was not spared, however, and was significantly reduced during therapy. As mentioned previously, androgens are lipolytic (except when it comes to visceral fat in HIV-infected men).


Given the results of these and prior studies, it is evident that the effects of insulin sensitizers on limb fat are fairly modest. For many patients any gain in extremity (or facial) fat is worth pursuing and one cannot fault a patient or clinician for trying rosiglitazone or pioglitazone to halt or reverse lipoatrophy.

As mentioned above, careful attention to lipids is needed. Regarding visceral fat accumulation, there were disappointing results from studies of both metformin and testosterone gel. Based on prior studies, it is possible that patients who have insulin resistance may have better results than suggested by A5082 where toxicity may have limited the evaluation of this agent.

Clearly, we need better answers. Other studies are exploring novel approaches to body fat problems associated with HIV infection. One of the most exciting is the use of uridine for lipoatrophy. But, as we saw here at CROI, these studies take time and anything can happen.


  1. Carr A, Workman C, Carey D, et al, for the Rosey investigators. No effect of rosiglitazone for treatment of HIV-1 lipoatrophy: randomised, double-blind, placebo-controlled trial. Lancet. February 7, 2004;363(9407):429-438.

  2. Hadigan C, Yawetz S, Thomas A, Havers F, Sax PE, Grinspoon S. Metabolic effects of rosiglitazone in HIV lipodystrophy: a randomized, controlled trial. Ann Intern Med. May 18, 2004;140(10):786-794.

  3. Gelato MC, Mynarcik DC, Quick JL, et al. Improved insulin sensitivity and body fat distribution in HIV-infected patients treated with rosiglitazone: a pilot study. J Acquir Immune Defic Syndr. October 1, 2002;31(2):163-170.

  4. van Wijk JPH, de Koning EJP, Cabezas MC, et al. Comparison of rosiglitazone and metformin for treating HIV lipodystrophy: a randomized trial. Ann Intern Med. September 6, 2005;143(5):337-346.

  5. Calmy A, Hirschel B, Hans D, Karsegard VL, Meier CA. Glitazones in lipodystrophy syndrome induced by highly active antiretroviral therapy. AIDS. March 28, 2003;17(5):770-772.

  6. Carr A, Workman C, Smith DE, et al, for the Mitochondrial Toxicity (MITOX) Study Group. Abacavir substitution for nucleoside analogs in patients with HIV lipoatrophy: a randomized trial. JAMA. July 10, 2002;288(2):207-215.

  7. Mulligan K, Yang Y, Koletar S, et al, and ACTG Protocol 5082 Team. Effects of metformin and rosiglitazone on body composition in HIV-infected patients with hyperinsulinemia and elevated waist/hip ratio: a randomized, placebo-controlled trial. In: Program and abstracts of the 13th Conference on Retroviruses and Opportunistic Infections; February 5-8, 2006; Denver, Colo. Abstract 147.
    View slides: Download PDF

  8. Hadigan C, Corcoran C, Basgoz N, Davis B, Sax P, Grinspoon S. Metformin in the treatment of HIV lipodystrophy syndrome: a randomized controlled trial. JAMA. July 26, 2000;284(4):472-477.

  9. Slama L, Lanoy E, Valentin MA, et al. Effect of pioglitazone on HIV-1 related lipoatrophy: a randomized double-blind placebo-controlled trial (ANRS 113) with 130 patients. In: Program and abstracts of the 13th Conference on Retroviruses and Opportunistic Infections; February 5-8, 2006; Denver, Colo. Abstract 151LB.

  10. Kohli R, Wanke C, Gorbach S, Shevitz A. A randomized placebo-controlled trial of metformin for the treatment of HIV lipodystrophy. In: Program and abstracts of the 13th Conference on Retroviruses and Opportunistic Infections; February 5-8, 2006; Denver, Colo. Abstract 148.
    View slides: Download PowerPoint

  11. Sutinen J, Hakkinen AM, Westerbacka J, et al. Rosiglitazone in the treatment of HAART-associated lipodystrophy -- a randomized double-blind placebo-controlled study. Antivir Ther. June 2003;8(3):199-207.

  12. Shikuma C, Parker R, Sattler F, et al, and AIDS Clin Trials Group Protocol A5079 Study Team. Effects of physiologic testosterone supplementation on fat mass and distribution in HIV-infected men with abdominal obesity: ACTG 5079. In: Program and abstracts of the 13th Conference on Retroviruses and Opportunistic Infections; February 5-8, 2006; Denver, Colo. Abstract 149.
    View slides: Download PowerPoint