NATAP at the  Durban World AIDS Conference 2000
Durban, South Africa, July 9-14

Metabolic and Lipodystropy issues: Confusion continues to reign Prevalence, Mytochondrial Toxicity, Growth Hormone, Lipid Lowering Agents:
Reported By
Dr Graeme Moyle MD, MBBS, Chelsea & Westminster Hospital, London UK


Longer-term (> 1 year) antiretroviral therapy has been associated with lipid elevations, insulin resistance sometimes accompanied by new-onset diabetes mellitus, and abnormal body fat distribution. This may involve peripheral fat loss, localized or truncal fat gain or both. It has not been clearly established if all these phenomena are inter-related although unifying theories have been proposed. Both PIs and nucleoside analogs (NAs) have been suggested to play a role in the pathogenesis of these changes. Hypertriglyceridemia and reduced high density lipoprotein (HDL) cholesterol, but not diabetes mellitus or fat redistribution were recognized as complications of HIV-1 infection, and most typically wasting, prior to the availability of PI.

Elevations in cholesterol and triglycerides have been reported in HIV-negative healthy volunteers receiving ritonavir monotherapy over a 2-week period, confirming that PIs agents have a direct effect on lipid handling.

A definition for the syndrome(s) to enable consistent case definition in studies does not currently exist. Reported clinical manifestations of the fat redistribution syndrome are heterogeneous and range from central or localised adiposity alone to peripheral fat wasting or combinations of both. Central adiposity, at least in males, appears mainly secondary to visceral fat accumulation and may be best assessed by CT or MRI scanning A range of approaches have been considered for management of fat redistribution and lipid abnormalities. These include:

        Diet modification and exercise programs

        Food supplements particularly acetyl-L-carnitine, Ubiquinone (co-enzyme Q-10) and riboflavin

        Hormonal therapies including growth hormone and anabolic steroids

        Lipid lowering agents such as statins, fibrates and omega-10 fish oils

        Insulin sensitizing agents such as metformin and glitazones

        Cosmetic surgery

        Switching to alternative antiretrovirals with a perceived lower risk of the problems (see separate review)

        Structured treatment interruptions (see also separate review)

Nucleoside analogues: Do available data support the mitochondrial toxicity hypothesis?

Nucleoside analogues have been occasionally associated with lactic acidosis and hepatic steatosis, which is thought to relate to mitochondrial dysfunction caused by these agents inhibiting mitochondrial DNA polymerase gamma. This problem has been reported with all the available nucleoside analogue combinations. Inhibition of this enzyme has been suggested to also be important in the pathogenesis of fat redistribution syndrome. A difference in relative incidence of lactic acidosis or hyperlactatemia between combinations has not been established. In a cross sectional survey of 211 asymptomatic patients, 161 (76%) of whom were on nucleoside analogue therapy, mild hyperlactataemia (defined as 2.1-5mmol/l) was present in 23% of treated and 8% of untreated patients (Chi2 p=0.3). Serious hyperlactataemia (>5mmol/l) was observed in only one patient, which normalized without alteration of therapy. Of the patients with hyperlactataemia, 19% of ZDV recipients therapy, and 28% of d4T recipients had hyperlactatemia (1). The presence of elevated lactate in the absence of therapy is intriguing but may reflect laboratory issues, sampling variation (for example use of cuffs or not), use of other mitochondrial toxins (such as alcohol), familial mitochondrial disease, recent exercise or the possibility that HIV infection alone may impact mitochondrial function. A similar survey of 70 treated patients found lactate levels were >2.1 in 36% of patients and the anion gap widened in 19%. Three of 4 patients with lactate >3mmol/l and an anion gap of >12 had symptoms suggestive of drug toxicity such as fatigue, weight loss or myopathy (2). In 8 patients with clinical lipodystrophy who were exercised to assess oxidative and glycolytic capacity in skeletal muscle tissue using ergometer cycling, oxygen consumption (VO2) and blood lactate (L) were measured. Before and after exercise, muscle biopsies were obtained to measure activity of citrate synthase (CS) and hydroxyacyl-CoA  dehydrogenase (HD) to assess mitochondrial oxidative  capacity. The HAART treated HIV patients performed less work than healthy controls and had significantly higher baseline and post exercise lactates. Muscle biopsy data did not differ from controls. The absence of muscle abnormalities suggest that lactate elevations may have related to diminished hepatic clearance of lactate rather than excess production (3).

The physiology of the lipid metabolism was studied measuring lipid absorption, gastrointestinal lipid handling and lipid oxidation in 6 HIV positive individuals  with established 2 years lipidaemia on treatment with a protease  inhibitor (PI), 6 age matched healthy HIV negative controls. Additionally, seven HIV positive men were studied prior to treatment (study 1) after one month (study 2) on a PI containing  regimen and again after three months (study 3). All patients were given a standard test meal and recovery of tracer in the breath as 13CO2 was determined hourly for six hours  after the meal. The rate of lipid absorption and oxidation were not significantly different between the pretherapy patient group and controls. However, the rate of lipid oxidation had increased by 3 months when compared with pretherapy. This rate was lower than in those who had established dyslipidaemia. The authors concluded that prior to commencing PI lipid metabolism in HIV positive patients is normal. However, once on a PI containing regimen a progressive abnormality of lipid handling occurs in all patients with decreased peripheral uptake of lipid and enhanced oxidation. As the mitochondria are responsible for fat oxidation, these data suggest that mitochondrial function is normal or may even be increased to compensate for the increases in circulating fat (4).

A small biopsy study of persons (n=4) with fat redistribution and health controls, however, reported histological differences in the appearance of adipose tissue. Whilst mitochondria were abundant many abnormal forms were present with alterations of mitochondrial cristae and with lipid droplet accumulation in the cytoplasm (5). These data are more consistent with some mitochondrial toxin at work although the need for more control data, particularly in older individuals where mitochondrial changes have previously been reported (the mitochondria deteriorate with age).

Prevalence of metabolic abnormalities: Too many confounders for clear conclusions

The problem with most available data on prevalence is that it is either retrospective or cross sectional and doesnít adequately correct for past treatment history. This may be particularly relevant if the appearance of fat redistribution is a delayed consequence of a particular drugs (as was seen with fialuridine and mitochondrial toxicity). A German group evaluated 250 patients who had commence ART in 1996 (mean time since start of ART 36 months) using a standardized physician and patient questionnaire and visual analogue scales. Patients were mostly (80%) male with a mean age of 39 years. Physicians diagnosed lipodystrophy in 36-37% of the cohort (6). Risk factors were analysed by univariate testing followed by multivariate  logistic regression models of demographics,  clinical history, CD4+ cells, viral load (VL), and treatment  history. Variables significantly associated with lipodystrophy in logistic models were: CD4+ nadir < 200/ml (OR 2.2, p > 0.05), treatment  with d4T > 12 months (OR 2.2, p > 0.005), treatment with  NNRTI > 12 months (OR 0.2, p > 0.05). Treatment with PI (OR 2.0, p = 0.03) and male gender (OR 0.4, p = 0.02). were significant  in univariate but not in multivariate analyses (7).

Similar risk factors were evaluated in the Swiss cohort study

Data on abnormal body fat distribution reported by the patient, or by the visiting physician, were collected in the cohort. Serum levels of total cholesterol, HDL-cholesterol and triglycerides were measured at the time of each visit. Out of 1379 patients treated with antiretroviral drugs, 585 (42.4%) developed signs of fat redistribution. Peripheral fat loss was observed in 386 (28.0%) patients, whereas 416 (30.2%) had signs of fat accumulation. In the analysis by cross tabulation, AZT, 3TC and the combination AZT + 3TC as part of HAART had a diminished rate of fat loss (Odds ratio (OR) of 0.5 (0.4-0.6),  0.8 (0.6-1.0), and 0.1 (0.03-0.5), respectively) relative to D4T and the combination DDI + D4T (OR of 2.1 (1.6-2.7),  and 1.5 (0.4-6.5), respectively). In regard to fat accumulation, a risk increase of 1.4 fold (1.1-1.8) was observed with indinavir, but not with other protease inhibitors. Lipid values were frequently  abnormal: hypercholesterolemia (> 6.2 mmol/l) was detected  in 28.4% and hypertriglyceridemia (> 2.3 mmol/l) in 37% of patients, although no specific asscoiations were  reported (8).

However, other cohorts did not find these significant associations. For example, in 118 HIV-infected patients on HAART including stavudine (n =  95) or zidovudine (n = 23) fat wasting was assessed by physical  examination, regional fat distribution was estimated using four sites anthropomitry and central adiposity  was assessed by measurement of waist-hip ratio. Both groups were well balanced with respect to age,  sex, duration of HIV infection, risks factors for acquiring HIV  infection, prior AIDS defining conditions, duration of HAART,  daily caloric intake, CD4 cell counts, viral load and percent  with undetectacble plasma viral load. The proportion of antiretroviral-naive patients was significantly greater in the stavudine-treated  group (26.7% vs. 0%, p = 0.01) but the mean time of exposure to  NRTIs prior to HAART was not significantly different (46.9 ±  30.7 vs. 45.0 ± 32.1 months, p = .81). The lean body mass and fat parameters of body composition were similar among the groups either when expressed in absolute numbers ore when expressed as percentage of body weight. There were no statistically significant differences between the skinfold thickness measured at four different sites, nor there were between arm and leg or metabolic parameters. Visceral adiposity estimated through the waist-hip ratio was not statistically different in both groups (9).

Data examining the association of personal and/or family  history of diabetes mellitus, cardiovascular diseases or  obesity  with the development of body habitus and metabolic  complications from a US cohort of 175 patients (85% African American, 73% on PIs). Body habitis change was assessed by self report and measurement of waist-hip ratio plus unfasted lipids and glucose values. In general, strong associations were observed with high cholesterol and family history of high cholesterol, high glucose and famialial diabetes and history of obesity and raised BMI or waist-hip ratio suggesting that therapy may unmask fimilaial conditions or accelerate their onset in susceptible individuals (10).


Switching studies have been discussed in a previous report from this conference. In general, improvements in insulin resistance, modest but incomplete improvements in lipids but not much change in peripheral lipoatrophy have been reported from changing to PI sparing regimens.

One study evaluated the impact of a planned treatment interruption on metabolic parameters. 26 men with viral loads <500 copies/ml for at least  12 months while on PI based HAART were studied. Parameters measured before HAART cessation and immediately prior to its reinstitution were: fasted cholesterol, LDL, HDL, triglycerides, oral glucose tolerance  test with insulin levels, and anthropometrics (BMI, abdominal  circumference, waist-to-hip ratio, sum of 4 skinfolds). HAART was interrupted for a median 5.9 weeks (range 4.0-13.1). 13/17 had increased visceral abdominal fat at baseline. There was a significant decrease in levels (mean±SD) of total cholesterol (194±47.3 vs. 160±29.4; p>0.0001),  LDL (114± 32.6 vs. 95 ± 25.8; p = 0.0008), and triglycerides  (261 ± 244.3 vs. 216 ± 267.3; p = 0.011) after the period  of HAART interruption. However, there were no significant changes in glucose, insulin levels, or anthropometrics (11). Thus, whilst some metabolic changes were rapidly reverse clinical abnormalities, if reversible may require longer periods of treatment interruption than 6 weeks.

Growth hormone has been suggested as a potential therapy for particularly fat increase manifetsations of the syndrome. Several posters describing small series of patients supported this potential.

One study compared changes in body composition after 12 and 24 weeks of therapy and after therapy had been discontinued for at least 12 weeks. 14 subjects received open-label 6 mg/day recombinant human growth hormone (the wasting dose of growth hormone). Measurements include DEXA scan for regional and total lean and fat and MRI of the abdomen. Weight was similar at all time points. GH led to a gain in lean mass and a relative loss of fat. The loss of fat was greater in the trunk than in the arms or legs. Reduction in the intra-abdominal fat mass was observed There were no additional benefits in >12 weeks therapy. Unfortunately, benefits substantially regresses towards baseline after stopping therapy suggesting if benefits are to me maintained a maintenance dose of growth hormone needs to be established (12). CD4 cell count and viral load benefits appear to be maintained and adverse effects of lipids or glucose handling are reported but infrequent and resolve after treatment withdrawal (12, 13).

Lipid lowering agents

The elevation of cholesterol with HAART is principally in the LDL and VLDL fractions and in health volunteers has been suggested to be related to increased hepatic production of VLDL (14). Inhibition of HMG-co-reductase in the liver diminishes de novo synthesis of cholesterol, the mechanism responsible for production of >50%of total body cholesterol. Inhibition of HMG-coA activates increased synthesis of hepatic LDL-receptors leading to increased clearance of circulating LDL (15). The role of statins in drug induced hyperlipidemia is not established. Pravastation may represent the best choice of agent in the circumstance of protease inhibitor use as, unlike other statins it is not substantially metabolised by cytochromes p450 and has the lowest binding to plasma proteins of the statin agents (16). Significant drug interactions with protease inhibitors have not been observed. ACTG A5047, reported the effects of ritonavir (RTV) +  saquinavir (SQV) on statins levels: modest reductions in pravastatin levels were seen whereas atorvastatin levels rose 4-5 fold and simvistatin around 27-fold.  Nelfinavir, or either ritonavir or saquinavir levels were not affected by pravastatin. Atrovastin also did not effect ritonavir or saquinavir levels but trend towards modest reductions in saquinavir levels were observed with simvistatin (17). Thus simvistatin appears contraindicated with PIs. Statins are generally well tolerated agents with mild gastrointestinal effects being most commonly reported. The most important adverse event is myotoxicity which may manifest as isolate creatinine kinase elevation, myalgia, myosotis myopathy or,  most importantly, rhabdomyolysis (18).

In a randomized, open-label comparative trial of dietary advice alone or with 40mg od pravastatin (PS) in persons established on PI-based regimens with viral load (VL) <500cps/ml and cholesterol >6.5mmol/l (240mg/dl), all patients remained with VL <500cps/ml.  Mean fasting cholesterol fell by 4% and 17% for dietary advice and pravastatin groups respectively. The fall in total cholesterol in each group was accounted for entirely by reduction in LDL as HDL rose non-significantly by 0.6mmol/l in both groups. The reduction in LDL at week 24 was 1.24mmol/l (19%) with pravastatin and 5.5% with dietary advice alone. Weight, fasting glucose or triglycerides did not significantly change in either group. No significant clinical or laboratory events occurred (19).


1.       Vrouenraets S, Treskes M, Regez RM, et al. The occurrence of hyperlactataemia in HIV-infected patients on NRTI treatment. XIII International AIDS Conference, Durban, July 9-14, 2000: abstract TuPpB1234

2.       Harris M, Tesiorowski A, Montaner JSG. Screening for nucleoside-associated lactic acidois XIII International AIDS Conference, Durban, July 9-14, 2000: abstract TuPpB1233

3.       Roge BT, Calbet J, M¸ller K, et al.  Mitochondrial function and exercise capacity in HIV-infected patients with lipodystrophy. XIII International AIDS Conference, Durban, July 9-14, 2000: abstract TuPpB1232.

4.       Morlese J, Ware L, Kruger M, et al. A prospective study of lipid metabolism in HIV-1 seropositive patients treated with HAART. ThOrB759

5.       Mallal S, Nolan D. Light and electron microscopy findings in subcutaneous  fat in antiretroviral treatment and naÔve HIV-infected patients. LbPeB7054

6.       Goetzenich A, Corzillius M, Mauss S, et al.  The impact of lipodystrophy on quality of life. WePpB1381

7.       Mauss S, Corzillius M, Wolf E, et al. Risk factors for the HIV-associated lipodystrophy syndrome in patients with homogeneous duration of ART. ThPpB1486

8.       Bernasconi E, Boubaker K, Sudre P, et al. Metabolic side effects of antiretroviral therapy in the swiss cohort study ThOrB703

9.       Domingo P,  Francia E, Torres OH, et al.  Differential effects of zidovudine and stavudine on fat distribution and its associated the metabolic abnormalities in HIV-infected patients on highly active antiretroviral therapy ThOrB705

10.   Raghavan SS, Stanton M, Lester K, et al.  The association of personal and or family history of diabetes, cardiovascular complications and obesity with the development of anti-retroviral associated body habitus changes and metabolic complications WePeB4276

11.   Hatano H, Miller KD, Yoder CP, Metabolic and anthropometric consequences of interruption of highly active antiretroviral therapy (HAART): improvement in serum triglyceride and cholesterol levels. WePeB4221

12.   Engelson ES, Glesby M,  Sheikan J, Body composition changes during and after growth hormone therapy for lipodystrophy with truncal adiposity ThPpB1437

13.   Greiger-Zanlungo P, Blick G, Garton T, Finkelstein F. Follow-up data on the effects of r-hGH (Serostim) on glucose levels and hepatic function in AIDS wasting patients switched to protease sparing treatment regimens. ThPeB5061

14.   Purnell JQ, Zambon A, Knopp RH, et al. Effect of ritonavir on lipids and post-heparin lipase activities in normal subjects. AIDS 2000;14:51-57.

15.   Grundy SM Cholesterol metabolism in man. West J Med 1978;128;13-25

16.   Pentikainen PJ, Saraheimo M, Schwartz JI, et al. Comparative pharmacokinetics of lovastatin, simvistatin and pravastatin in humans. J Clin Pharmacol 1992;32:136-140.

17.   Fichtenbaum C, Blaschke T, Rosenkranz S  et al  Pharmacokinetic interactions between protease inhibitors and selected HMG-CoA reductase inhibitors  WeOrB544

18.   Ucar M, Mjorndal T, Dahlqvist R. HMG-CoA reductase inhibitors and myotoxicity. Drug Safety 2000;22:441-457.

19.   Moyle G, Lloyd M, Reynolds B, Baldwin C. A randomised open label comparative trial of dietary advice with and without pravastatin for the management of protease inhibitor (PI)-associated hypercholesterolaemia. ThPpB1438