Lipodystrophy and metabolic disturbances

Graeme Moyle for

Metabolic and morphologic changes in persons receiving therapy are a key obstacle to the initiation and continuation of therapy, are stigmatising, and potentially place individuals at risk of future vascular morbidity.

Whilst probable contributors to the aetiology of this condition have been described, the mechanisms by which these changes occur are not fully understood and no reliably effective therapy has been established.

Lipodystrophy plenary

Issues with regard to lipodystrophy aetiology were summarised by David Cooper. [Abstract PL-11] In particular, approved protease inhibitors (PIs) appear to impact lipids and insulin resistance (including in healthy volunteers), affect adipocyte differentiation and hepatocyte lipid handling and release in vitro and interfere with the insulin-stimulated glucose uptake by GLUT 4 in adipocytes.

Data are less complete as to how nucleoside analogues (NAs) contribute although they have synergistic impact on adipogenesis (fat cell creation) and lipolysis (fat release) with PIs in adipocyte cultures in vitro. Diminished adipocyte mitochondrial DNA is difficult to interpret as it may reflect impending apoptosis rather than NA toxicity. Clearly, however, cohort studies implicate a role for NAs. Cross-sectional point prevalence studies suggest patients treated with NA + PI regimens have the highest rates of lipodystrophy with combined estimates of greater than 50% with these regimens.

Non-drug factors such as age, sex, disease severity, pre-therapy fat mass and exercise may also play a role in risk or severity. He also noted that a minority of studies have found associations with viral load and CD4+ cell count and that while immune reconstitution in transplant patients was associated with dyslipdaemia, insulin resistance and visceral obesity lipoatrophy was not described.

Mitochondrial dysfunction

An in vitro study of NAs and PIs in adipocyte cultures looked at the impact of these drugs on fat cell metabolism and gene expression. Both mature and differentiating cells were evaluated over up to 19 days of culture. Both stavudine and zidovudine had no effects at physiologic concentrations (<10?M) but at exposure above 100?M adipogenesis (new fat cell formation) and triglyceride accumulation were suppressed.

Mitochondrial function did not alter significantly and total mitochondrial DNA did not alter with stavudine or zidovudine at exposures of 100-300?M up to 8 days. Changes in gene expression suggested alterations in genes involved in adipogenesis and lipogenesis but not mitochondria. [Abstract 521] This suggests that these agents are not exerting effects in fat cells via mitochondrial mechanisms.

Fat biopsies from patients with lipoatrophy, however, indicate declines in mitochondrial DNA and increased mitochondrial protein whereas these changes were not observed in untreated patients. These types of changes are seen in cells from persons with congenital mitochondrial diseases but may also be a pre-apoptotic event or a response to increased cell energy needs (such as seen in cold adaptation). The finding is suggestive of mitochondrial DNA problems in fat cells but not confirmatory. [Abstract 522]

A second study of fat biopsies from lipoatrophic areas in 10 HIV infected patients treated with HAART for 6 to 21 months was reported. In all 10 cases, adipocytes showed progressive disruption of cell membranes, fragmented cytoplasmic rims, irregular cell outlines, and eventually large fat droplets lying free in the connective tissue, consistent with apoptosis. In addition, many adipocytes showed variable compartmentalization of fat droplets, with decrease in cell size, and large, mitochondria-rich cytoplasm. These authors suggested that lipoatrophy is characterized by apoptosis, defective lipogenesis, and also an increased metabolic activity in many of the fat cells. [Abstract 494]

Lipids and body shape changes

Considerable data with regard to both lipids and morphological changes were described in a large poster session on Tuesday afternoon, although limited new conclusions could be drawn. One key problem with assessing the aetiology of morphological changes has been that cross-sectional studies have tended to enrol individuals who have received multiple treatment regimens, thus compounding the potential for multi-factorial outcomes that are easily misinterpreted.

Effects of PI-based HAART

A study of 6 men with lipoatrophy receiving PI-based HAART were compared with six healthy male volunteers matched to their age and body mass index (BMI). Individuals were studied for insulin sensitivity using a hyperinsulinaemic euglycaemic clamp. Basal insulin concentrations were substantially higher in the lipoatrophic patients and mean endogenous glucose production 38% higher, basal lipolysis 78% higher in these individuals.

During the clamp procedure, endogenous glucose production was suppressed by only 50% in the lipoatrophic patients but by 85% in the controls. Whilst total glucose disposal increased in both groups, it only increased a modest 4% in the lipodystrophic patients compared with 23% in the controls. Lipolysis was also less efficiently depressed. Thus, patients with lipoatrophy appear to have both peripheral and hepatic insulin resistance and elevated glucose turnover. [Abstract 495]

ARV therapy and lipodystrophy

A large cross-sectional study of 672 individuals on initial therapy was reported from a collaborative effort of 19 Spanish centres. Evaluating patients on first line therapy helps avoid some confounders or biases in cross-sectional studies. Patients were required to be on initial HAART regimens for greater than six months (mean 20 months) with body fat changes being evaluated clinically and categorised as facial lipoatrophy, abdominal adiposity or both.

Unfortunately, evaluation was not reported to be blinded to therapy. The study included 75% male and 25% female patients with a mean CD4+ cell count of 528 cells/mm3 and mean viral load of 2.2 log10 copies/mL. Patients were categorised by PI and NA regimen. Overall, 209 of 672 individuals were classified as having moderate to severe lipodystrophy. Overall, the rates of facial lipoatrophy (FA), lipoatrophy at other sites, abdominal adiposity (AA) and all lipodystrophy (LDS – atrophy and adiposity) did not differ across groups confirming that this is a class effect (Table 1).

Table 1

Regimen N All LD Facial Atrophy Abdominal Adiposity
AZT+3TC+IDV 148 26% 8% 18%
AZT+3TC+NFV 67 25% 7% 8%
d4T+3TC+IDV 208 33% 13% 22%
d4T+3TC+NFV 121 29% 10% 15%
d4T+ddI+IDV 67 36% 15% 20%
d4T+ddI+NFV 40 24% 6% 12%

Observed lipoatrophy was as low as 6% (d4T, ddI plus Nelfinavir) and as high as 15% (d4T, ddI, plus Indinavir) suggesting an influence of PI choice rather than NA backbone over lipoatrophy. Combined lipoatrophy plus abdominal adiposity rates were lowest in individuals receiving d4T, ddI and Nelfinavir (24%) and highest in those receiving d4T, ddI, and Indinavir (36%). Time on antiretrovirals, the presence of hypertriglyceridaemia, and waste-hip ratio were amongst the independent factors associated with lipodystrophy in a multi-variate model, but no individual drug was associated with increased risk. [Abstract 486]

The Atlantic study provided data with regard to both fat redistribution and metabolic disturbances. These were ARV therapy naive individuals given stavudine plus didanosine with either nevirapine, lamivudine or indinavir as the third agent. 298 individuals were randomised equally to the 3 regimens. Although no baseline evaluations were available, 150 questionnaires were completed by physicians with regard to the appearance of individuals within the study (indinavir 46, nevirapine 48, lamivudine 56).

The reasons and methods for this sub-selection were not reported. Patients were assessed for loss of fat in arms, legs, face and buttocks and fat gain in abdomen, neck or breast area. An additional 37 individuals underwent abdominal CT scan, a median of 121 weeks after enrolment. One of the problems with such an analysis lies in the fact that 238 individuals were randomised to the study and only 150 individuals have been evaluated and it remains unclear whether these individuals are necessarily representative of the overall population.

Fat accumulation, fat atrophy or both were reported in 16%, 35% and 15% respectively for individuals surveyed. No significant differences were observed between treatment groups and in the sub-set of individuals assessed by single slice CT scan, no differences in sub-cutaneous to total adipose tissue or visceral to total adipose tissues were observed. [Abstract 488] Lipid changes favoured nevirapine with rises in HDL-cholesterol being significantly greater with this agent (49%) relative to lamivudine (10%) or indinavir (16%) arms. Total:HDL-cholesterol ratio also significantly improved with nevirapine and lamivudine but not indinavir. Differences in triglycerides were not observed. [Abstract 496]

ARVs and lipid disturbances

The Combine study is another study of naive patients randomised to either PI containing or sparing regimens (in this case on backbone of zidovudine and lamivudine). A cross-sectional study of individuals after 12 months of therapy was reported. This sub-study included 43 patients, 20 receiving nelfinavir, 23 receiving nevirapine. Consistent with the Atlantic study, HDL levels improved substantially with nevirapine but less so with nelfinavir. In both groups total cholesterol rose modestly. LDL rose significantly in nelfinavir but not nevirapine treated patients. Consistent with the Atlantic data the authors concluded regardless of NA backbone, PIs tend to lead to a more atherogenic lipid profile where as the use of nevirapine seems to lead to more lipid protected results. [Abstract 506]

Switching therapy

With regards to treatment, the possibility of switching away from PIs or NAs has been raised as a means to managing this problem. Limited new information was presented with regard to switching away from PIs where available evidence suggest modest improvements in lipid parameters and generally good virologic control. Improvements in fat atrophy have not been consistently reported and no adequately powered studies have compared approaches.

A small pilot randomised trial evaluated thymidine analogue withdrawal in lipoatrophic patients, virologically controlled on PI-sparing therapy. Patients with lipoatrophy and viral load <400 copies/ml, who had replaced PI therapy with quadruple RT inhibitor therapy 6-12 months earlier either continued therapy or ceased either stavudine or zidovudine. Regional changes in body fat were assessed by DEXA scans at weeks 0, 12 and 24. Nineteen patients were enrolled, 10 to continue and 9 to cease stavudine or zidovudine. At baseline all 19 were on abacavir, nevirapine or efavirenz (18 / 1), stavudine or zidovudine (17 / 2) and lamivudine or didanosine (18 / 1).

At week 24, limb fat had increased modestly but significantly in those in the discontinue arm. Total body weight, central fat and lean body mass did not differ between groups. However, 5 of the 9 stop patients had viral rebounds whereas there were no virological failures that occurred in the continue-therapy patients. The authors concluded that this approach appears too risky for loss of virologic control, as with de-intensification studies in the past, hence different strategies to reduce NA burden need to be investigated. [Abstract 96]

A second small study evaluated the potential to replace stavudine with abacavir in 16 individuals who were followed for six months after therapy change. The only modification of therapy was stavudine to abacavir. All patients had lipoatrophy on entry to the study with completely virological suppression. The switch was associated with a significant decline in cholesterol but with no significant improvement in lactate, triglycerides or morphological changes. Two individuals (12.5%) experienced a virological rebound. [Abstract 491]


The impression from the lipodystrophy presentations was that we have some understanding of how PIs may cause metabolic disturbances but it remains unclear as to why fat cells are dying. The highest risk syndrome appear to be those involving PI plus NA but increasingly, evidence from evaluations of persons on first line therapy suggests that these problems are class effects with limited difference between drug choice.


Unless otherwise noted, all abstract references in the text are to the 1st International IAS Conference on HIV Pathogenesis and Treatment. July 8-11, 2001. Buenos Aires, Argentina.


Copyright: United States and international copyright laws protect the entire design and contents of HIV and Information posted on the site may be used for personal, scientific or informational purposes, but NOT for commercial reproduction

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