Reports on lipodystrophy and metabolic complications

Andrew Carr, MD, for

[Note: Unless otherwise noted, all references are to Program and Abstracts of the Fifth Congress on Drug Therapy in HIV Infection, October 22-26, 2000, Glasgow, Scotland.]



Morrie Schambelan from UCSF reviewed the possible factors that contribute to lipodystrophy and metabolic disturbances. He presented a thorough summary of much of this year’s published or presented studies (particularly the Lipodystrophy Workshop) and drew 2 main conclusions:

  • each physical and metabolic feature probably has a multifactorial aetiology;
  • different protease inhibitors appear to exert different metabolic effects – ritonavir causes hyperlipidaemia within 2 weeks, whereas indinavir and amprenavir both induce insulin resistance but not hyperlipidaemia (at least over a 4 week period).

The implication from the latter conclusion is that it may be possible to switch within a class for a given metabolic disturbance.

Dawn Averitt presented her perspective as an HIV-infected woman with lipodystrophy. Apart from graphically reporting many patients’ issues of self-worth and stigmatisation, she also convincingly showed what has been anecdotally reported elsewhere, that patients can recover from lipodystrophy without change in antiretroviral regimen. How common this is remains unclear (although one suspects this is very unusual), but there may be a lot to learn about the pathophysiology of lipodystrophy and its potential therapy from such individuals.

The only new pathogenesis study evaluated tumour necrosis factor-alpha, a major inflammatory protein and stimulator of adipocyte lysis (abstract P137A). In 10 patients ceasing protease inhibitor therapy, levels of the soluble receptors for tumour necrosis factor-alpha (TNF-1r, and TNF-2r) both declined, although TNF-alpha (which binds to TNF-r) was undetectable in plasma at all times. Whether this is a causal relationship remains to be determined, but lends credence to the notion that cytokine dysregulation, either secondary to HIV or to imperfect immunological recovery in the presence of antiretroviral therapy, is somehow linked to lipodystrophy.


The consensus is that the only way to diagnose lipodystrophy is by physical examination, preferably with patient agreement; in other words, a very subjective way. An objective, sensitive, specific, reproducible, simple and cheap method of diagnosis would greatly assist industry, regulatory bodies and clinicians. So far, no single physical or laboratory parameter has been identified that will distinguish lipodystrophy from normal, because “normal” body composition varies greatly, and because the lipid and glycaemic parameters common with lipodystrophy also are common in the general population.

The Lipodystrophy Case Definition study has been established in order to generate such a diagnostic tool. The rationale and design of the study were reviewed in a session devoted to antiretroviral toxicity. The study is following the model established by the American Rheumatology Association over the last 30 years for the diagnosis of over 50 autoimmune diseases, such as rheumatoid arthritis, systemic lupus erythematosis, and numerous forms of vasculitis. For example, the diagnosis of SLE is made when a patient is found to have any 4 clinical or laboratory features out of a total of 11 such parameters. This model seems appropriate for study of lipodystrophy diagnosis because of the numerous parallels between many autoimmune diseases and lipodystrophy: both have a complex aetiology, no “gold-standard” diagnostic test, and can have very varied clinical and laboratory features.

The Lipodystrophy Case Definition study is ongoing in 36 sites globally, and is recruiting 400 cases (patients with at least 1 moderate or severe feature of lipodystrophy on physical examination and patient report) and 400 controls (patients who on physical exam and patient report have no lipodystrophy). A host of parameters are being collected: lipodystrophy-specific physical examination and questionnaire, demographics, HIV disease parameters, details about all antiretroviral therapy ever received, about all current metabolic and dietary therapies, metabolic (lipid, glycaemic, lactate) measures, and body composition (total and regional body fat and lean tissue by simple anthropometry, DEXA scan and abdominal CT scan).

About half the patient data will used to generate models independent of the questionnaire, physical exam and antiretroviral data. These models will be tested in the remaining patients. It is possible that diagnosis may require presence of a small number of parameters, or the presence of a number of parameters from a longer list. Subdefinitions for men and women, for patients with pure lipoatrophy or fat accumulation, and for patients receiving various types of antiretroviral therapy, will be explored but it is possible that the study will be underpowered to address all such definitions.

Prevalence and Risk Factors

Seven posters, mostly representing new European cohorts, reported on factors associated with lipodystrophy, lipoatrophy or central fat accumulation (P130, P133 to P137, and P150). The reports were remarkably consistent with each other, and with cohort data presented over the last 15 months at numerous meetings. In general, current stavudine therapy, increasing age, greater antiretroviral or nucleoside analogue duration, and lower weight prior to therapy were associated with lipoatrophy, and female sex and protease inhibitor therapy with central fat accumulation (although some studies no differences in risk factors for either body phenotype). Lipid and glucose disturbances were more likely in those receiving protease inhibitor therapy or with any lipodystrophy. As might be expected, at least one study found that metabolic disturbance in the absence of lipodystrophy was more a function of age and male sex than of antiretroviral therapy or duration, confirming these disturbances are multifactorial and not all HIV-related.

Lipids and Glucose

The phase 3 study of lopinavir (LPV) reported grade 3-4 elevations in triglyceride or cholesterol in 30% of patients (abstract PL 6.5). Although the metabolic data were not presented in detail, the lipid increases appeared greater with LPV than with NFV; given the otherwise good tolerability of LPV reported to date, it appears that these lipid effects may be one Achilles heel for the drug. Unfortunately insulin sensitivity was either not studied or not reported.

Phase 2 data of the once-daily protease inhibitor BMS-232632 found no increases in fasting total cholesterol or triglyceride, in contrast to nelfinavir (abstract PL 6.6); again, no insulin sensitivity data were presented.

The increasingly popular bid regimen of ritonavir/indinavir (RTV/IDV), was found to induce greater changes in fasting lipids than did indinavir tid (abstract PL 6.7). Whether this effect is due to the addition of RTV or to greater overall levels of IDV is not clear. The bid regimen was also associated with greater risk of nephrolithiasis.

Lastly, nelfinavir (NFV) was associated with relatively rapid (within 4 weeks) increases in fasting lipids when compared to nevirapine (NVP) in naive patients initiating either drug with ZDV/3TC (and was also less virologically effective, including in patients with plasma HIV RNA greater than 100,000 copies/ml).

A feature of the reporting of the metabolic details at this meeting (and in other presentations this year) is the marked lack of consistency in how metabolic parameters are reported. There are several effective ways to under-report metabolic abnormalities. Favourite methods appear to be to report only the rates of grade 4 changes, to only measure fasting glucose (which is a very poor measure of insulin resistance with antiretroviral therapy) or to not do the tests. A better way undoubtedly would be to report mean changes from baseline of all the key parameters (total, LDL and HDL cholesterol, triglycerides, glucose, insulin, C-peptide). I am unaware of any study (either from academia or industry) that has done this. It is high time that metabolic parameters were reported in a consistent fashion, so that studies can be compared in even a limited fashion.

Mitochondrial Toxicity and Lactic Acidemia

There was little new here. David Cooper summarized the available data on all antiretroviral classes, as well as individual drugs. These data were recently reviewed (Lancet, October 24, 2000). He did report on unpublished data from a colleague, Bruce Brew, that patients with d4T-induced peripheral neuropathy had higher lactate levels (mean 3.2 mmol/l) than patients with clinically similar neuropathy thought to be due to HIV disease (1.8 mmol/l), as well as in patients receiving d4T who did not have neuropathy (1.7 mmol/l). This extends our possible indications for measuring lactate, and after a decade suggests there may be a simple test for distinguishing HIV neuropathy from nucleoside analogue neuropathy.

A case presentation demonstrated that lactic acidemia with acute hepatitis may in fact have long-term consequences (abstract P165A). A patient who presented with a lactate of 7 mmol/l and acute hepatic failure (but no other risk factor for liver disease) was reported to clinically recover fully after stopping dual nucleoside analogue therapy for 12 months, only to represent with hepatic synthetic failure (hypoalbuminemia and prolonged prothrombin time), portal hypertension (bleeding oesophageal varices and encephalopathy), and a liver biopsy that showed steatosis, marked depletion of mitochondrial complex type I, but no cirrhosis. It was speculated that long-term nucleoside therapy associated with lower level lactic academia (which occurs in 15 to 20% of all nucleoside analogue recipients) might also occur some risk, and that monitoring lactate might be appropriate in those with underlying liver disease or with unexplained liver dysfunction.


Frank Goebel reviewed studies of both the physical and metabolic complications of lipodystrophy. Unfortunately, he felt obliged to begin his lecture with the statement that there are very few prospective, randomised studies, and that only one had been published (Hadigan et al, JAMA 2000). The published study had shown that metformin was effective as an insulin-sensitising agent for centrally obese, insulin-resistant adults receiving protease inhibitors, and also lead to reduced visceral and subcutaneous fat. Therefore, metformin may be less than ideal in patients with peripheral lipoatrophy and central adiposity (the most common scenario). Growth hormone has been found to reduce both intra-abdominal fat accumulation and buffalo humps, but the effect is transient if the drug is withdrawn, and is achieved with at the cost of well-described side effects and at great expense. Even worse is the situation for lipoatrophy, where no agent has shown been well studied, and for which protease inhibitor substitution has been unhelpful. He chastised the pharmaceutical industry for not sponsoring studies of the thiazolidinediones (eg. rosiglitazone, piaglitazone). In non-HIV lipodystrophy, troglitazone was recently reported (Arioglu, Ann Intern Med, 2000) to lead to significant increases in peripheral fat mass, as well as reductions in insulin resistance (lipid changes were not reported in this study, whether because they were not studied or there was no benefit is unknown). Protease switch studies have been somewhat helpful for lipids, particularly elevated total cholesterol and triglyceride, but variable effects on insulin sensitivity have been noted, with little effect on body composition. Importantly, few studies have been randomised, and have rarely included comprehensive metabolic analysis and objective measures of body composition. Switch studies were reviewed in detail by Rob Murphy at another session, with similar conclusions reached by Bill Powderly at ICAAC.

Avascular Necrosis

Only 1 study of only 5 patients was reported. Three of the 5 patients required bilateral hip replacement, and 2 patients had disease in the knee or ankle. One interesting finding, and in contrast to reports in Toronto, was that none of the 5 patients had lipodystrophy, a parameter previously weakly linked to avascular necrosis. Also, there was a significant delay in the time to diagnosis in 4 of the patients, emphasising that good patient care requires clinical acumen. (for further details see report in this issue of HTB)


Four studies evaluated adverse reactions to nevirapine (NVP). A Spanish multicentre, open-label randomised study evaluated whether 1 of 3 strategies could prevent NVP hypersensitivity in 562 patients (abstract PL 9.2):

Rash (%) Discontinuation (%)
1. Escalation (100mg increase/week for the first 4 weeks 8.8 5.3
2. Oral prednisolone 50mg every 2 days for 2 weeks 8.6 4.3
3. Loratadine 10mg bid for 2 week 11.6 4.2
4. Both escalation and prednisolone 7.7 3.9
5. No intervention 18.7 8.5

Interventions 1, 2 and 4 resulted in significantly less rash rates than did no intervention, but it is interesting that the effects of dose escalation and prednisolone were not additive. However, no intervention lead to reduced rates of NVP cessation, which is the clinically relevant question, however the study may not have powered to determine this adequately. Lastly, the escalation regimen was apparently not associated with significantly lower NVP concentrations, although these data were not shown. Overall, the results are at odds of a Boerhinger-Ingelheim-sponsored placebo-controlled study presented by Julio Montaner in the last 6 months showing that prednisolone 20 mg/day for 2 weeks was associated with no reduction in rash rate, and with a greater rate of severe rash.

Data from a South African randomised study comparing stavudine, plus either 3TC (lamivudine) or FTC (emtricitabine), plus NVP or efavirenz (EFV), in 468 patients found that NVP was associated with a far greater risk of grade 4 hepatotoxicity (9.4% vs 0%), with almost all cases occurring in the first 4 weeks. The rate was significantly higher in women than in men (12% and 6%, respectively; p=0.05) but there was no difference between blacks and non-blacks, and only 3 patients were hepatitis B or C seropositive (a known risk factor for NVP hepatitis).

These results are consistent with data from previous NVP studies, including in HIV-uninfected women, that NVP is more hepatotoxic in women than in men. These data suggest that the recent review by the EMEA recommending more frequent estimation of liver enzymes in the first 6 weeks of NVP therapy is warranted, particularly in women and hepatitis carriers.

Data from 2 other studies perhaps offer a little perspective. BI1090, the phase 3 NVP clinical endpoint study (abstract PL 8.6), found that clinical hepatitis occurred in 2.8% of patients randomised to NVP versus 1.4% of those who received placebo, suggesting that most cases are asymptomatic (or unrecognised). Therefore, most grade 3/4 liver enzymes with nevirapine are presumably asymptomatic. A cohort study from Spain (abstract PL 8.5) came to similar conclusions, with discontinuation rates of 2% for hepatitis and 9% for rash.

At the laboratory level, a cross-sectional study found that abacavir (ABC) hypersensitivity was associated with greater proportion of CD4+ and CD8+ T lymphocytes that secreted interleukin-4 (IL-4) in response to in vitro stimulation (abstract P168). This increase was in turn associated with a greater frequency in the blood of Th0 cells, (the cells that are capable of maturing into T cells that secrete different profiles of cytokines, depending upon the maturing stimulus received). An interesting study, and deserving of prospective evaluation including with other drugs associated and not associated with hypersensitivity.

Other Studies

One cohort study of 1,481 (1,067 male, mean age 37 years) patients followed for a mean 19 months reported on the rates of grade 3-4 adverse events seen with various nucleoside analogues, after adjustment for protease inhibitor type (abstract P170). Rates per 100 person-years were as follows:

ddI 36% AZT/ddI 46%
d4T 26% d4T/ddI 35%
3TC 23% d4T/3TC 23%
ddC 21% AZT/3TC 26%
AZT 21% AZT/ddC 13%

The study did not separate the patients by whether this was their first nucleoside regimen or not, and did not describe the prevalence of milder adverse events, which are perhaps more likely to result in virological failure (because of continued therapy but with imperfect adherence). Nevertheless, this study provides the first population data of such toxicity, and clearly some assistance in guiding the choice of therapy.

Copyright 2000 by HIV and Hepatitis Treatment Advocates and HIV and All Rights Reserved.

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