Individuals receiving antiretroviral therapy for the treatment of HIV-1 infection have experienced peripheral lipoatrophy , gain in visceral fat , hyperlipidaemia and insulin resistance, all of which have been grouped under the common heading of HIV-associated lipodystrophy (HIVLD).
The pathogenesis of HIVLD is likely to involve numerous mechanisms, such as the inhibition of DNA polymerase gamma in mitochondria by nucleoside reverse transcriptase inhibitors (NRTIs) leading to mitochondrial dysfunction in fat [3,4], abnormalities of fat differentiation resulting from the effects of protease inhibitors on transcription factors such as sterol regulatory element-binding protein-1 (SREBP-1) [5,6] and increased apoptosis of fat cells secondary to these many insults . Changes in immune status resulting from therapy, and the effects of secondary cytokine activation on fat tissue, could explain some changes seen in HIVLD  although evidence to support this view is lacking.
Cross-sectional cohorts have shown an association between HIVLD and older age, lower baseline CD4 cell counts, the use of and duration of use of protease inhibitors (PI) and of NRTIs [9–14] particularly when PIs and NRTIs are used in combination . Although the syndrome is well described, a lack of objective, prospective data prevents identification of possible primary and secondary defects. In one prospective study, any form of patient- or physician-reported lipodystrophy was seen in 17% of individuals at 18 months , but the study only included individuals taking PI-containing regimens.
We hypothesised that lipoatrophy occurs as a result of treatment with, and duration of, antiretroviral therapy, rather than being purely an immune phenomenon. To explore this hypothesis, we studied a cohort of antiretroviral-naive, HIV-infected men starting a variety of antiretroviral regimens, to determine changes in body composition resulting from therapy, and factors associated with these changes.
HIV-infected men, referred by their treating physicians, were sequentially recruited from the HIV Clinics of St Vincent's Hospital, Sydney, Australia. All patients had documented HIV-1 infection, were at least 18 years old, had no active AIDS-defining illness, no previous antiretroviral exposure and were at a stage of their illness where their treating physicians considered antiretroviral therapy appropriate. As this was a non-randomized study, the treating physician made decisions regarding the components of baseline antiretroviral regimens, and any subsequent treatment changes. All patients provided written, informed consent after approval of the study by the Human Research Ethics Committee.
Baseline demographics were collected. The following assessments were performed at baseline, weeks 12, 24 and then every 24 weeks thereafter; CD4+ lymphocyte counts, plasma HIV RNA, fasting total cholesterol, estimated low-density lipoprotein (LDL) cholesterol, high-density lipoprotein (HDL) cholesterol, triglyceride, glucose and insulin [1,15].
Total and regional body composition was quantified at screening, and at weeks 12, 24, 48 and every 48 weeks thereafter by dual-energy X-ray absorptiometry (DEXA) (Lunar DPXL; Madison, Wisconsin, USA) at a single reading site. Estimates of central abdominal fat (CAF) and spinal bone mineral density (represented by ‘T’ score) were determined from a central window measurement [1,15].
All analyses were by intention-to-treat and included follow-up data on all subjects who started their treatment regardless of subsequent treatment changes. The primary endpoint was change in limb fat measured by DEXA with secondary endpoints of changes in CAF, total cholesterol and estimated LDL cholesterol, triglycerides, glucose, insulin and ‘T’ score. Data were collected on a specifically designed database (Logicielle, Sydney, Australia) and non-parametric analyses were applied. Paired comparisons were analysed using the Wilcoxon signed rank test, unpaired comparisons using the Mann–Whitney test and correlations between sets of variables using Spearman rank correlation coefficient tests. The Kruskal–Wallis test was used to compare three group samples. Simple and multiple regression analyses were performed to examine the relationship between variables. P values of less than 0.05 were considered significant.
Forty patients were recruited between July 1997 and May 2000 with analysis performed in December 2001. At the time of analysis, all patients were alive, 90% of the cohort had completed 96 weeks of therapy and 50% had completed 144 weeks of therapy.
All median baseline metabolic parameters were within normal limits (Table 1), as were body mass index (BMI) and bone mineral density. Almost half the cohort received a regimen containing a PI, with indinavir being the most common PI used. Five subjects received a regimen containing both PIs and non-nucleoside reverse transcriptase inhibitors (NNRTI). Roughly equal numbers were prescribed one of the three most common dual NRTI backbone regimens at that time: didanosine and stavudine, stavudine and lamivudine, and zidovudine and lamivudine.
Over the course of the study, 16 (40%) subjects changed therapy. Four (10%) subjects changed class from a PI to a NNRTI-containing regimen an average of 23 (range 12–36) months into the study. One subject changed from a PI to a NNRTI-containing regimen due to virological failure 2 years into the study. All remaining changes were secondary to side effects. The most common drugs involved were stavudine with 11 changes after a mean of 7 [3–12] months of treatment (six secondary to neuropathy, one secondary to symptomatic lactic acidaemia, one secondary to adherence problems and three for side effects not otherwise specified), didanosine (ddI) with five changes after a mean of 10 (3–24) months (three secondary to neuropathy, one secondary to nausea and one for other side effects), and indinavir with four changes after a mean of 8 (3–12) months (two secondary to renal calculi, one secondary to paronychia and one secondary to other side effects). No subject changed therapy as a result of changes in body composition.
Immunological and virological responses to therapy
Seventy percent of the cohort achieved viral load < 400 copies/ml by week 12, with no significant change seen between weeks 12 and 144 (Fig. 1b). At the time of analysis, 82% of the cohort achieved an undetectable viral load (HIV RNA < 400 copies/ml). Median baseline CD4 cell count was 246 × 106 cells/l (range 0 to 836 × 106 cells/l). The largest rise in median CD4 lymphocyte count occurred by week 24 (median increase of 126 × 106 cells/l, P < 0.0001), with a further significant rise between weeks 24 and 144 (increase of 56 × 106 cells/l, P < 0.05).
There was a significant rise in BMI (22.5 to 24, P < 0.05) and lean mass (3% rise from baseline, P < 0.01) by week 12, which was sustained to week 48. At week 96 BMI had decreased to a level not significantly different from baseline. Total body fat increased by median 20% to week 24 (P < 0.001) with no significant changes from week 24 onwards.
There was an increase in limb fat (median 5.6 versus 6.9 kg), CAF (1.0 versus 1.3 kg) and lean mass (52.5 versus 54.7 kg) to week 24 coinciding with the largest changes in CD4 and HIV RNA (Figs 1a and 2a). From week 24 onwards there was a selective, progressive loss of limb fat with median limb fat mass at weeks 96 and 144 being significantly less than baseline (median loss of 1.6 and 3.0 kg, respectively). During the initial period, 57% patients experienced greater than 10% gain in limb fat mass. Subsequently, however, 46% patients lost more than 10% limb fat mass and 48% patients lost at least 1 kg between baseline and week 96. From week 24, there was a median 13.6% [interquartile range (IQR), 0.9–26.3] loss of limb fat per year of treatment. This loss occurred after the period of greatest change in CD4 and HIV RNA (Fig. 1).
Older age correlated with a lower baseline limb fat. Higher baseline CAF and truncal fat correlated with higher baseline limb fat. The biggest gains in limb fat from baseline to week 24 were seen in those with higher baseline CAF and truncal fat, higher baseline HIV RNA and higher baseline ‘T’ scores. When baseline and week 24 patient, treatment, immune, morphological and metabolic factors were analysed, subjects treated with stavudine, those with higher baseline HIV RNA, higher baseline ‘T’ score and higher week 24 lean mass had a significantly greater rate of loss of limb fat from week 24 (percentage loss per year) (Tables 2–4). Multivariate analysis revealed use of stavudine to be associated with a significantly increased rate of loss of limb fat (P = 0.05).
Central abdominal fat
In contrast to limb fat, after an initial increase to week 24, CAF was maintained after week 24, remaining significantly greater than baseline out to week 144 (Fig. 2b). Baseline factors predicting larger mass of CAF at week 24 included higher baseline total fat, limb fat, truncal fat and CAF, higher insulin concentrations and higher BMI (Table 2). Factors associated with a greater percentage change in CAF between baseline and week 24 included lower baseline cholesterol and CD4 cell count and higher baseline HIV RNA. Subjects with more CAF at week 24 also had significantly higher week 24 truncal, limb and total fat, higher cholesterol, triglycerides, insulin, and lower HDL cholesterol and a higher BMI (Table 3). A greater change in CD4 cell count from baseline correlated with changes in limb fat, truncal fat, CAF and total fat. Those taking a PI-containing regimen had significantly higher and those taking a NNRTI regimen had significantly lower CAF at week 24.
Bone mineral density
The ‘T’ score rose slightly to week 24 and then fell significantly to week 48 remaining significantly lower than baseline out to week 144. The percentage of the cohort with ‘T’ scores less than −1 (consistent with osteopenia) rose from 13% at baseline to 22% at week 144. There was little change in the number of the subjects with ‘T’ scores less than −2.5 consistent with osteoporosis (1 at baseline, 2 at week 144).
Changes in metabolic parameters are illustrated in Figure 3. Fasting total cholesterol and estimated LDL cholesterol rose significantly early into treatment with the largest rise occurring by week 12 (Fig. 3a). Both remained significantly higher than baseline out to week 144. In contrast triglyceride concentrations did not rise significantly until week 96 and remained elevated out to week 144 (Fig. 3b). Similarly, insulin concentrations rose late with significantly higher insulin concentrations occurring at week 144 (Fig. 3b). Changes in fasting insulin were not significantly different between those prescribed PI-containing regimens [median, 1.2 mmol/l per year (IQR, −0.87 to 3.27)] and those prescribed PI-sparing regimens [0.65 mmol/l per year (IQR −0.7 to 2)].
The change in total and LDL cholesterol per year of treatment was 0.44 and 0.24 mmol/l, respectively. In a multivariate analysis of baseline variables, treatment with PIs (P = 0.006), lower baseline cholesterol (P = 0.02), and lower baseline lean mass (P = 0.04) were independently associated with greater increases in total cholesterol. Similarly, use of PIs (P = 0.002) and lower baseline LDL (P = 0.04) were associated with the largest rises in LDL cholesterol.
The change in triglycerides concentrations was 0.14 mmol/l per year of treatment. No baseline factor determined change in triglyceride, although those with greater change in cholesterol per year also had significantly larger rises in triglycerides per year (P = 0.01).
HDL cholesterol rose by 0.1 mmol/l at week 24 (P < 0.01) but the rise was not sustained past week 96. Fasting glucose concentrations did not change (4.7 mmol/l at baseline versus 4.85 mmol/l at week 144).
The complexity of the mechanisms underlying HIVLD is reflected in the results of this exploratory study. Most of the components of HIVLD, such as central fat accumulation, peripheral lipoatrophy and hyperlipidaemia were observed. Contributions by drug classes and specific drugs, in addition to immunological and virological responses to therapy, were all shown to have some correlation with changes in body composition. From these results, a sequence of events begins to emerge, providing an insight into possible primary and secondary events in the development of HIVLD. Of importance, these results reveal trends that run counter to some currently held theories about the possible mechanisms underlying HIVLD.
The gains in both central abdominal fat and limb fat seen during the first 24 weeks of treatment could be explained by general improvements in health and nutrition associated with treatment of HIV viraemia and reversal of the associated catabolic state. The simultaneous increase in lean mass also seen during this period support this viewpoint, suggesting that these changes result from general improvements in nutrition rather than a fat-specific process.
The gain in CAF during the initial 24 weeks of therapy may reflect the visceral fat accumulation reported in HIVLD . From week 24 onwards, there was an obvious selective, progressive loss of limb fat whereas both CAF and lean mass were maintained (Figs 1 and 2). This likely represents the lipoatrophy that is characteristic of HIVLD. Maintained levels of CAF in the presence of limb fat loss (Fig. 1) could result in the central fat becoming more pronounced clinically to patients or physicians as limb fat declined.
Of interest is the correlation noted between CAF and limb fat. Although by 3 years both gain in central fat and loss of limb fat are present, the two processes do not appear to be occurring simultaneously and may be part of separate aetiologies – the increase in CAF probably in response to nutritional improvement (and therefore a secondary effect of drug treatment), whereas the loss of limb fat was probably a result of long-term antiretroviral drug use. However, excess fat accumulation resulting from changes in nutritional requirements alone would be expected to decrease over time as the body gains muscle mass. The fact that this fails to occur in this cohort may indicate that the persistent accumulation of abdominal fat seen after week 24 is a reflection of a pathological process. Whether this is related to immune restoration, continued drug exposure or other process is unclear from this study.
Loss of limb fat, however, developed in the absence of significant changes in HIVVL, and after the largest change in CD4 cell count had already occurred (Fig. 1a and 1b). This runs contrary to the hypothesis that HIVLD is part of a cytokine-driven ‘immune reconstitution’ phenomenon [8,16]. Indeed, during the period of most intense immune recovery, there was actually a gain, rather than a loss, of both limb fat and CAF.
Complex relationships between morphological and metabolic changes are apparent from this study. Those with higher baseline cholesterol values gained less limb fat and CAF to week 24 (Table 2). A first impression may be that these subjects already have high baseline body fat deposits, thus explaining the higher cholesterol values and lower relative fat gains. However this was not the case, as no significant correlations existed between baseline cholesterol levels and mass of total or compartmental body fat (data not shown).
The importance of the cholesterol values is reflected in their strong association with subsequent loss of limb fat (Table 3). Such an association has previously been reported . There are marked inter-individual variations in the function of adipose tissue, with evidence supporting a genetic basis for this variation . If higher cholesterol levels reflect poorly functioning adipose tissue, these individuals may be at higher risk of further adipose tissue dysfunction, resulting from the additional and continuing insults from antiretrovirals on adipose tissue (e.g. SREBP-1c) [5,6] or mitochondrial function . This could explain the changes seen in this cohort.
Higher insulin concentrations associated with high CAF mass, seen in this study, is an association that has been well described in HIV-infected  and uninfected populations . Other PI- induced mechanisms of insulin resistance, such as those involving dysfunction of the GLUT-4 glucose transporter [21,22], result in an acute insulin resistance. Although in vitro evidence exists to support such a mechanism, such an acute process should result in high insulin concentrations early in the course of treatment. This was not seen in this study, with hyperinsulinaemia occurring later into treatment.
Although the most detailed prospective study to date, the results from this study should be viewed in the context of study limitations. It could be argued that the fact that this was a non-randomized study, with no untreated controls, using various antiretroviral regimens, in a cohort of male only patients is a limitation. In addition, the use of total body DEXA to estimate CAF and bone mineral density is imperfect. Although significant associations were detected, these should be viewed in the context of the number of subjects involved in the study. It could be argued that the large number of comparisons studied and the multiple data points used leaves the results prone to false positives, although non-parametric analyses were employed in an attempt to limit this.
The number of treatment changes may or may not be viewed as a limitation. Treatment changes happen frequently in routine practice and as such, this study may provide a more accurate description of the development of HIVLD in clinical practice. In addition, development of HIVLD was not used as a reason for changing treatment. The impact of treatment changes on the overall analysis was minimal. For example, there were no significant differences in the rate of limb fat loss in those who stopped stavudine, nor in the accumulation of CAF in those who stopped indinavir (data not shown).
Objective changes in body composition correlating with HIV associated lipodystrophy [1,2] only became readily apparent after 2 years of antiretroviral treatment in this cohort. This has consequences with respect to published studies alluding to the metabolically friendly profile of certain drugs or drug combinations , which have a relatively short follow-up period. Importantly, this study shows how the development of lipoatrophy is relatively independent of changes in immune function, being more a product of continued exposure to antiretroviral drugs.
The authors wish to thank the patients for their participation in the study. We also wish to acknowledge the help with the design of the statistical analysis provided by Dr Matthew Law from the National Centre in HIV Epidemiology and Clinical Research, Sydney, Australia.
Sponsorship: The National Centre in HIV Epidemiology and Clinical Research is funded by the Commonwealth Department of Health and Aged Care through the Australian National Council on AIDS, Hepatitis C and Related Diseases and its Research Advisory Committee.
P.W.G.M. is supported by funds from the National Heart Lung and Blood Institute of the National Institutes of Health, grant number RO1 HL65953-01.
1.Carr A, Samaras K, Burton S, Law M, Freund J, Chisholm DJ, et al
. A syndrome of peripheral lipodystrophy, hyperlipidaemia and insulin resistance in patients receiving HIV protease inhibitors.AIDS
2.Miller KD, Jones E, Yanovski JA, Shankar R, Feuerstein I, Falloon J. Visceral abdominal-fat accumulation associated with the use of indinavir.Lancet
3.Shikuma CM, Hu N, Milne C, Yost F, Waslien C, Shimizu S, et al
. Mitochondrial DNA decrease in subcutaneous adipose tissue of HIV infected individuals with peripheral lipoatrophy.AIDS
4.Brinkman K, Smeitink JA, Romijn HA, Reiss P. Mitochondrial toxicity induced by nucleoside-analogue reverse-transcriptase inhibitors is a key factor in the pathogenesis of antiretroviral-related lipodystrophy.Lancet
5.Caron M, Auclair M, Vigouroux C, Glorian M, Forest C, Capeau J. The HIV protease inhibitor indinavir impairs sterol regulatory element-binding protein-1 intranuclear localisation, inhibits preadipocyte differentiation, and induces insulin resistance.Diabetes
6.Dowell P, Flexner C, Kwiterovich P, Lane MD. Suppression of preadipocyte differentiation and promotion of adipocyte death by HIV protease inhibitors.J Biol Chem
7.Domingo P, Matias-Guiu X, Pujol R P, Francia E, Lagarda E, Sambeat M A, et al
. Subcutaneous adipocyte apoptosis in HIV-1 protease inhibitor-associated lipodystrophy.AIDS
8.Ledru E, Christeff N, Patey O, de Truchis P, Melchior J-C, Gougeon M-L. Alteration of tumor necrosis factor-( T-cell homeostasis following potent antiretroviral therapy: contribution to the development of human immunodeficiency virus-associated lipodystrophy syndrome.Blood
9.van der Valk M, Gisolf EH, Wit FWNM, Japour A, Weverling GJ, Danner SA on behalf of the Promethus study group. Increased risk of lipodystrophy when nucleoside analogue reverse transcriptase inhibitors are included with protease inhibitors in the treatment of HIV-1 infection.AIDS
10.Martinez E, Mocroft A, Garcia-Viejo MA, Perez-Cuevas JB, Blanco JL, Mallolas J, et al
. Risk of lipodystrophy in HIV-1-infected patients treated with protease inhibitors: a prospective cohort study.Lancet
11.Thiebaut R, Daucourt V, Mercie P, Ekouevi DK, Malvy D, Morlat P, et al. Lipodystrophy, metabolic disorders, and human immunodeficiency virus infection: Aquitaine Cohort, France, 1999.Clin Infect Dis
12.Lichtenstein KA, Ward DJ, Moorman AC, Delaney KM, Young B, Palella FJ Jr, et al
. Clinical assessment of HIV-associated lipodystrophy in an ambulatory population.AIDS
13.Carr A, Miller J, Law M, Cooper DA. A syndrome of lipoatrophy, lactic acidaemia and liver dysfunction associated with HIV nucleoside analogue therapy: contribution to protease inhibitor-related lipodystrophy syndrome.AIDS
14.Boufassa F, Dulioust A, Lascaux AS, Meyer L, Boue F, Delfraissy JF, et al
. Lipodytrophy in 685 HIV-1-treated patients: influence of antiretroviral treatment and immunovirological response.HIV Clin Trials
15.Carr A, Samaras K, Thorisdottir A, Kaufmann G, Chisholm DJ, Cooper DA. Diagnosis, prediction and natural course of HIV protease inhibitor-associated lipodystrophy, hyperlipidaemia and diabetes mellitus.Lancet
16.Mynarcik D C, McNurlan M A, Steigbigel R T, Fuhrer J, Gelaot MC. Association of severe insulin resistance with both loss of limb fat and elevated serum tumor necrosis factor receptor levels in HIV lipodystrophy.J Acquir Immune Defic Syndr
17.Bogner JR, Vielhauer V, Beckmann RA, Michl G, Wille L, Salzberger B, et al
. Stavudine versus zidovudine and the development of lipodystrophy.J Acquir Immune Defic Syndr
18.Miserez AR, Muller PY, Barella L, Schwietert M, Erb P, Vernazza PL, et al
. A single-nucleotide polymorphism in the sterol-regulatory element-binding protein 1c gene is predictive of HIV-related hyperlipoproteinaemia.AIDS
19.Gan S K, Samaras K, Thompson C, Kraegen E, Carr A, Cooper D, et al
. Correlations between intramyocellular lipid, visceral fat and insulin sensitivity: a study of HIV positive subjects with and without peripheral lipodystrophy.Diabetes
20.Carey DG, Jenkins AB, Campbell LV, Freund J, Chisholm DJ. Abdominal fat and insulin resistance in normal and overweight women.Diabetes
21.Noor MA, Lo JC, Mulligan K, Schwarz JM, Halvorsen RA, Schambelan M, et al
. Metabolic effects of indinavir in healthy HIV-seronegative subjects.AIDS
22.Murata H, Hruz P W, Mueckler M. The mechanism of insulin resistance caused by HIV protease inhibitor therapy.J Biol Chem
23.Dube MP, Qian D, Edmondson-Melancon H, Sattler FR, Goodwin D, Martinez C, et al
. Prospective, intensive study of metabolic changes associated with 48 weeks of amprenavir-based antiretroviral therapy.Clin Infect Dis