A Longitudinal Study of the Changes in Body Fat and Metabolic Parameters in a South African Population of HIV-Positive Patients Receiving an Antiretroviral Therapeutic Regimen Containing Stavudine

Abstract

The aim of this study was to determine the patterns of change in body fat and metabolic parameters in a South African cohort on a first line ART regimen containing stavudine. Fasting lipogram, blood glucose and insulin levels, CD4 cell count, viral load, BMI, waist-to-hip ratio (WHR), and skinfold thickness at the triceps, scapula, and iliac crest were measured before starting ART in 42 (27 female) subjects. Repeat measurements were performed at four monthly intervals for 2 years. Lipodystrophy was diagnosed using patient perception and assessment by a physician. At baseline, subjects who went on to develop lipodystrophy (LD group) were fatter and had higher skinfold thickness at all three sites and higher insulin levels than subjects who never developed lipodystrophy (NLD group). The WHR increased to a greater extent while hip circumference and tricep skinfolds fell more significantly in the LD than NLD group. Triglyceride and cholesterol levels increased significantly in both groups while lactate and glucose levels increased more and insulin levels increased less in the LD than the NLD group. Neither viral load nor CD4 count differed between the groups during the study. Viral load correlated positively with insulin levels at baseline. Thus, lipodystrophy in the South African population is characterized by a higher BMI before initiation of ART and lipoatrophy of the arms and hips, lipohypertrophy of the waist, and increased lactate production. When compared to the NLD group, the LD subjects display attenuated insulin secretory output in response to a higher weight gain.

Introduction

M orphologic and metabolic complications associated with long-term suppressive antiretroviral (ARV) therapy (ART) include abnormalities in fat distribution (lipodystrophy), dyslipidemia, insulin resistance, hyperlactatemia, and osteopenia.^1,2 These complications can impact future morbidity and mortality³ and are increasingly being recognized in countries with previously poor access to ART.^4,5 Furthermore, ART drug toxicity is associated with reduced quality of life and poor compliance.⁴ The South African government began a national antiretroviral program in 2004 and by the end of 2007 there were 428,951 patients on ART.⁶ Initial treatment combines stavudine (d4T), lamivudine (3TC), and efavirenz (EFV), with EFV replaced by nevirapine in women wishing to become pregnant.⁷ This choice was based on WHO recommendations for ART in resource-poor settings.⁸ Estimates from various large South African state clinics suggest that over 90% of patients start on d4T, 3TC, and EFV, and that most toxicity is related to d4T, usually managed by replacement with zidovudine (AZT).⁴ Stavudine has been removed as the preferred first choice for HIV therapy in most developed countries; however, due largely to cost d4T remains a preferred first-line therapy in many developing countries. Thus, recent surveys in low- and middle-income countries with a high prevalence of HIV infection have shown that over 60% of such countries use a first-line ARV regimen containing d4T.^9,10 Little is known about the metabolic and morphologic side effects of ART in southern Africa and no longitudinal toxicologic studies have been conducted in African populations receiving d4T therapy. The aim of the present study was therefore to determine the prevalence of side effects to ART in a highly characterized, prospective South African cohort on a first-line ART regimen containing d4T.

Materials and Methods

Subjects

Fifty HIV-infected participants referred to the clinic for ART were recruited by convenience sampling from April 2004 to December 2005 and followed up at four monthly intervals for 24 months. Participants were eligible for this study if they intended to begin ART, were ART naive, between the ages of 18 and 55 years, and were not pregnant at the time of enrollment. The criteria for initiating ART was a CD4 count of less than 250 or the presence of an AIDS-defining illness. The human Ethics Committee of the University of the Witwatersrand approved the study with written informed consent obtained at enrollment.

Body composition

All participants had weight, height, and waist and hip circumference recorded, with subsequent calculation of body mass index (BMI; kg/m²) and waist-to-hip ratio. The participants were identified as having “lipodystrophy” in this study if the patient and the health care worker both identified noticeable lipoatrophy or lipohypertophy at more than one of the following body sites: face, dorsocervical region, breasts, arms, abdomen, buttocks, and legs.¹¹ Specifically, the patients were asked if they had noticed thinning of the face, arms, buttocks, hips, and legs. They were also asked if they had noted any increase in fat deposition, particularly over the abdomen, breasts, or on the back. These changes must have occurred after initiation of ART. The patients were seen by the same physician (J.G.) at every visit.

Body composition was assessed at baseline and at four-monthly visits using standardized procedures by the same two trained nurses. Subjects were weighed after removing shoes and outdoor clothing. Waist circumference was measured at the narrowest part of the torso between the lower rib and the iliac crest, and hip circumference at the level of greatest gluteal protuberance. Skin fold thicknesses were assessed from the mean of two repeated skinfold thickness measurements made at three sites: the triceps, subscapular area, and over the abdomen as previously described.¹²

Metabolic measurements

Blood was collected after an overnight fast for glucose, lactate, lipids (Roche Modular analyser, Roche Diagnostics, Mannheim, Germany), and insulin measurements (Immulite 1000 analyzer Diagnostics Corp., Los Angeles, CA). Low-density lipoprotein (LDL)-cholesterol levels were calculated using the Friedewald formula¹³ and insulin resistance assessed using the homeostasis model assessment (HOMA) method.¹⁴ Other measurements taken were CD4 counts (Beckman Coulter flow cytometry, Miami, FL) and viral load (Roche Cobas Amplicor, Indianapollis, IN). Venous blood samples were collected with the use of a tourniquet and samples were centrifuged and assayed immediately. All metabolic and immunological measurements were repeated every 4 months. Impaired fasting glucose (≥5.6 mmol/liter), hypercholesterolemia (≥5.0 mmol/liter), hypertriglyceridemia (≥1.7 mmol/liter), hyper-LDL-cholesterolemia (≥3.0 mmol/liter), and hypo-HDL-cholesterolemia (<1.0 mmol/liter in males and <1.3 mmol/liter in females) were defined using the guidelines from the National Cholesterol Education Programme, Adult Treatment Panel III (NCEP-ATPIII).¹⁵

Statistical analysis

Data that were not normally distributed were log transformed to normality. Metabolic and anthropometric variables at different time points were compared using the Student's paired t-test or the Wilcoxon matched pairs test. Differences between the lipodystrophic and nonlipodystrophic groups were compared using the Student's nonpaired t-test or the Mann–Whitney U test. The prevalence of dyslipidemia or impaired fasting glucose (IFG) was compared between groups using the chi-squared test.

The principal determinants of fasting insulin and lipid levels prior to commencement of ART were investigated using univariate (Pearson correlation) analysis. The variables studied for correlation with insulin levels were gender, BMI, weight, waist and hip circumference, waist-to-hip ratio (WHR) and tricep, scapular, and iliac crest skinfolds, viral load, and glucose levels. The same variables with the addition of insulin and HOMA were investigated for their correlation with lipid levels. The variables that gave significant correlations in the univariate analysis were then included as independent variables in stepwise, backward multiple regression analysis. Data are expressed in the tables and figures as mean ± SD or median (interquartile range).

Statistical analyses were performed using the software package Statistica version 7.1 (StatSoft, Tulsa, OK).

Results

Patient characteristics

Study entry characteristics are summarized in Table 1. Fifty patients were recruited into the study of which 42 completed 24 months of follow-up and eight did not complete the study for the following reasons: two moved away from the locality, one became pregnant, two died, and three withdrew from the study. These events occurred early in the study and therefore data collected for these participants were not included in the analyses. All participants were started on the standard first line antiretroviral therapy of d4T, 3TC, and EFV. d4T was dosed according to body weight: 16 patients were on 30 mg twice daily because their weight was below 60 kg and 26 patients were on 40 mg twice daily for weights above 60 kg. No differences were observed for changes in metabolic or anthropometric variables over the course of the study when the 30-mg and 40-mg d4T dose groups were compared.

Table 1.

Baseline Characteristics for All Subjects a

Variable	All subjects
BMI (kg/m²)	22.7 (2.55)
Weight (kg)	60.5 (13.81)
Waist (cm)	74.5 (9.29)
Hips (cm)	91.8 (7.88)
Waist–hip ratio	0.68 (0.1)
Triceps (mm)	13.3 (7.7)
Scapular (mm)	11.15 (6.43)
Iliac crest (mm)	9.9 (7.2)
Glucose (mmol/liter)	4.40 (0.90)
Insulin (μIU/ml)	5.93 (7.27)
HOMA	1.19 (1.62)
Total cholesterol (mmol/liter)	3.62 ± 0.92
Triglycerides (mmol/liter)	0.90 (0.50)
HDL (mmol/liter)	0.91 ± 0.30
LDL (mmol/liter)	2.20 ± 0.74
Lactate (mmol/liter)	1.28 (1.08)
Viral load	213 000 (478 333)
CD4 count	93 (107)

Data expressed as mean ± SD or median (interquartile range).

Data are reported only from those 42 patients who completed the study. All patients were black African, and 65% were female. Median CD4 counts and viral loads were 93 cells/mm³ (range: 6.0–315.0) and 248,000 HIV RNA copies/ml (range: 6800–1,400,000), respectively, at initiation of ART and changed to 312 (range: 114–745; p < 0.0005) and 0.0 (range: 0.0–15,000), respectively, after 2 years of therapy. At baseline the 42 participants who completed the study had a mean age of 34.4 years (range: 23–51 years) and a BMI of 22.7 (range: 20.5–41.1).

Only 16 subjects remained on the original ART regimen after 2 years. d4T-associated toxicity accounted for most drug changes (2 for peripheral neuropathy, 4 for lactic acidemia, and 18 for lipoatrophy). Two participants changed therapy due to the development of resistance to EFV. The overall prevalence of lipodystrophy was 42.9% (18 of 42 subjects) after 2 years of treatment. Median time to diagnosis of lipodystrophy was 78 weeks (range: 53–82 weeks); of those who developed lipodystrophy a mixed pattern was reported in the majority (n = 14); two subjects each displayed either lipoatrophy alone or lipohypertrophy alone. The prevalence of lipodystrophy was higher in females (57.1%) than males (12.5%; p < 0.05). At the time of diagnosis of lipodystrophy patients had d4T replaced with AZT and at the end of the study period these subject had received AZT for a median period of 36 weeks (range: 14–77 weeks).

Baseline anthropometric and metabolic parameters that predicted lipodystrophy

Subjects who developed lipodystrophy tended to have a greater BMI prior to initiation of therapy (p < 0.05; Table 2) and greater skinfold thickness at the triceps (p < 0.05), scapular (p < 0.005), and the iliac crest (p < 0.0005) prior to commencing therapy than those who never developed lipodystrophy (Table 2).

Table 2.

Body Composition and Metabolic Variables over Time and Percentage Changes from Baseline at 111 Weeks in Subjects with (LD) and without (NLD) Lipodystrophy a

	Baseline		59 weeks		111 weeks		111 weeks (% change from baseline)
Variables	LD	NLD	LD	NLD	LD	NLD	LD (%)	NLD (%)
BMI	23.6 (3.5)	22.0 (2.2)*	25.2 (5.3)	23.0 (3.3)*	24.6 (3.4)	23.5 (3.7)	5.4 (11.5)	3.1 (15.7)
Weight (kg)	61.6 (11.6)	59.4 (11.1)	69.4 (11.7)	62.6 (13.7)	66.8 (9.7)	65 (16.3)	4.9 (12.1)	3.3 (11.1)
Waist (cm)	75.5 (11.0)	73.0 (8.6)	81.5 (11.9)	77.5 (12.0)	82.3 (11.0)	80.0 (10.5)	7.4 (12.3)	6.2 (9.7)
Hips (cm)	93.5 (7.6)	90.0 (5.1)*	97.3 (11.0)	92.5 (8.0)	91.2 (12.0)	92.0 (5.0)	−3.2 (7.7)	3.8 (10.3)***
Waist–hip ratio	0.78 (0.07)	0.82 (0.08)	0.86 (0.12)	0.85 (0.12)	0.89 (0.07)	0.88 (0.13)	11.7 (7.4)	5.2 (14.5)**
Triceps (mm)	16.1 (6.0)	11.4 (8.4)*	15.6 (9.1)	10.5 (6.9)	12.0 (7.7)	9.9 (6.8)	−16.8 (44.2)	−5.1 (41.9)*
Scapula (mm)	13.5 (5.3)	8.8 (5.1)**	22.2 (11.3)	11.3 (7.6)*	24.2 (11.2)	12.6 (10.2)*	58.7 (88.2)	39.0 (91.2)
Iliac crest (mm)	14.0 (4.2)	7.3 (3.9)***	18.2 (10.9)	11.1 (6.0)*	19.4 (10.8)	10.4 (10.4)*	45.2 (89.8)	58.6 (48.0)
Glucose (mmol/liter)	4.2 (0.50)	4.5 (0.95)	4.7 (0.68)	4.8 (0.60)	4.6 (0.50)	4.6 (0.55)	10.0 (10.0)	4.1 (15.9)*
Insulin (μIU/ml)	7.4 (8.0)	4.3 (4.5)**	10.4 (5.8)	5.1 (2.7)*	9.4 (5.0)	7.4 (8.3)*	7.5 (189)	42.3 (260)
HOMA	1.3 (1.9)	0.90 (1.6)	2.3 (1.2)	1.2 (0.8)**	1.8 (0.9)	1.4 (1.8)	16.0 (172)	19.4 (207)
Lactate (mmol/liter)	1.3 (0.50)	1.3 (1.2)	3.0 (1.7)	1.5 (0.60)**	1.9 (1.4)	1.6 (0.93)	75.5 (89.4)	−2.3 (101)

Data expressed as median (interquartile range); ^* p < 0.05, ^** p < 0.005, ^*** p < 0.0005 vs. the LD group at the same time point.

Insulin levels were higher at baseline in subjects who later developed lipodystrophy (p < 0.05; Table 2), but there was no difference in insulin resistance as determined by HOMA.

Principal determinants of insulin and lipid levels prior to commencement of ART

Fasting insulin at baseline correlated significantly (p < 0.05) with mass, hip circumference, scapular and iliac crest skinfold thickness, and viral load in univariate analyses. When these factors were included as independent variables in a stepwise, backward regression model only iliac crest skinfold measures (beta = 0.53, p < 0.0005) and viral load (beta = 0.44, p < 0.005) demonstrated statistically significant relationships with insulin. The r-value for this model was 0.63 (p < 0.0001). Viral load explained 18% of the variance in insulin levels. A similar analysis showed that HOMA also correlated with viral load (beta = 0.40, p < 0.05) and iliac crest skinfold measures (beta = 0.36, p < 0.05) in a multiple regression model (r = 0.48, p < 0.01).

Significant, negative correlations were observed for fasting LDL cholesterol with insulin, HOMA, and viral load in univariate analyses. Due to collinearity HOMA and insulin could not be included together in a multiple regression model and therefore were analyzed with viral load in two separate models. In both models only viral load correlated significantly with LDL cholesterol (beta = –0.40, p < 0.05 in both models) explaining 19% of the variance in LDL concentrations.

Fasting triglyceride levels demonstrated weak positive correlations in univariate analyses with age (p = 0.07), viral load (p = 0.07), and HOMA (p = 0.09). When these variables were included in a multiple regression model, viral load was found not to be a significant determinant of triglyceride levels (beta = 0.16, p = 0.30), but age (beta = 0.39, p < 0.05) and HOMA (beta = 0.37, p < 0.05) were. The r-value for this model was 0.53 (p < 0.05).

Changes in anthropometry during ART

Subjects who developed lipodystrophy had a significantly higher BMI at 59 weeks when compared to the no lipodystrophy (NLD) group (p < 0.05; see Table 2), and the mean percentage change in mass from the baseline level was greater in the LD subjects (Fig. 1A). All patients showed a significant increase in waist circumference at the end of the study (p < 0.001 for baseline versus week 111 for the NLD and p < 0.005 for the LD group); however, the percentage increase above baseline values was greater in LD than NLD subjects, although this trend did not reach statistical significance (Fig. 1B). Percentage change from baseline values for hip circumference was negative in the LD group and significantly different from that for NLD subjects at 59, 77, 94, and 111 weeks (Fig. 1C). By week 111 patients who developed lipodystrophy had significantly increased in WHR from baseline values compared to the NLD subjects (p < 0.005; see Table 2).

FIG. 1.

Percentage change from baseline values for weight (A), waist (B), and hip circumference (C) and tricep skinfold thickness (D). (▪) Subjects with lipodystrophy; (▴) subjects without lipodystrophy; ^* p < 0.05, ^** p < 0.005, ^*** p < 0.0005 versus value at week 0; ⁺ p < 0.05, ⁺⁺ p < 0.005, ⁺⁺⁺ p < 0.0005 versus subjects with lipodystrophy. Values are medians (interquartile ranges were not displayed to aid clarity).

Subjects who developed lipodystrophy experienced a greater increase in triceps skinfold thickness at 41 weeks than did the NLD group, after which triceps skinfold fell markedly (Fig. 1D). Skinfold thickness of the scapular and iliac crest showed a steady increase in both groups but remained significantly higher at 59 and 111 weeks in the LD group (Table 2; p < 0.05 for all)

Metabolic and immunologic changes with therapy

All subjects showed a steady increase in triglycerides, total cholesterol, and HDL and LDL cholesterol over the course of the study with no significant differences noted between subjects who did or did not develop lipodystrophy. There was a marked increase in the percentage of participants who developed abnormal lipid profiles based on NCEP-ATPIII guidelines by the end of the study period (see Table 3). Despite this only one study subject met the NCEP-ATPIII criteria for the initiation of lipid-lowering therapy by the end of the study, with no subjects meeting these criteria at the start of the study.

Table 3.

Changes in Glucose and Lipid Levels over Time and Percentage of All Subjects Not Meeting NCEP III Guidelines for Lipids and Glucose at Start and End of Study a

Variable	Baseline blood levels (mmol/liter)	Blood levels at 111 weeks (mmol/liter)	Percentage of subjects at baseline with lipid or glucose levels above NCEP cut-off values	Percentage of subjects at 111 weeks with lipid or glucose levels above NCEP cut-off values
Glucose	4.40 (0.90)	4.60 (0.56)*	2.40	2.50
Total cholesterol	3.62 ± 0.92	4.75 ± 1.24*	2.34	35.0***
Triglycerides	0.90 (0.50)	1.20 (1.18)*	4.76	30.8**
HDL	0.91 ± 0.30	1.41 ± 0.42*	25.0	70.0***
LDL	2.20 ± 0.74	2.65 ± 0.97*	19.0	29.0

Data expressed as mean ± SD or median (interquartile range) or percentage; ^* p < 0.05, ^** p < 0.005, ^*** p < 0.0005 vs. baseline. NCEP III, National Cholesterol Education Programme, Adult Treatment Panel III.

The HOMA level did not increase significantly between baseline and week 111 in the LD subjects (p = 0.62) but in the NLD group there was an increase in HOMA that just failed to reach statistical significance (p = 0.076; Table 2). In the LD group insulin did not increase from baseline levels as significantly as for the NLD subjects (Fig. 2A), with fasting insulin being significantly higher at the end of the study compared to baseline in NLD (p < 0.05) but not LD subjects (p = 0.40; Table 2).

FIG. 2.

Percentage change from baseline values for insulin (A), glucose (B), and lactate levels (C). (▪) Subjects with lipodystrophy; (▴) subjects without lipodystrophy; ^* p < 0.05, ^** p < 0.005 versus value at week 0; ⁺ p < 0.05, ⁺⁺ p < 0.005 versus subjects with lipodystrophy. Values are medians (interquartile ranges were not displayed to aid clarity).

A steady rise in glucose was observed in all subjects with marked differences from baseline values noted at 111 weeks of therapy (Table 3). Although there were no differences in absolute glucose between LD or NLD subjects (Table 2), those patients who developed lipodystrophy experienced a greater percentage increase in plasma glucose at weeks 94 and 111 than the NLD patients (p < 0.05 for both; Fig. 2B) and fasting glucose levels at the end of the study were significantly higher than the baseline levels for LD (p < 0.005) but not NLD (p = 0.60) subjects. The prevalence of impaired fasting glucose was 2.40% at the start of the study and this did not change significantly by the end of the study (Table 3).

Plasma lactate increased in all patients. After 59 weeks of therapy both absolute lactate levels (p < 0.005; Table 2) and percentage change in levels from baseline (p < 0.005; Fig. 2C) were significantly higher in the LD than NLD subjects. After 111 weeks there was no difference in mean lactate between these two groups (see Table 2 and Fig. 2C) because lactate dropped when therapy in the LD patients was changed from d4T to AZT.

No differences in viral load or CD4 cell counts were observed between the LD and NLD subjects at baseline or after 2 years of therapy (data not shown).

Effects of replacing d4T with AZT in subjects with lipodystrophy

All subjects who developed lipodystrophy were transferred from d4T to AZT therapy. Table 4 shows anthropometric and metabolic variables measured at initiation of ART at time of replacement of d4T with AZT and at the end of the study period. It can be seen that during d4T therapy, BMI and waist both increased significantly (p < 0.05 and p < 0.005, respectively) while tricep skinfold thickness fell (p = 0.07). During the time of AZT therapy BMI and waist tended to fall, but both remained significantly higher than at baseline. Tricep skinfold thickness rose slightly over this time period but remained significantly lower than the baseline value. Glucose, lactate, total cholesterol, and triglyceride blood levels all rose significantly with d4T therapy. Glucose levels remained static with AZT therapy while triglyceride levels rose slightly and both remained significantly higher than baseline levels. Lactate and total cholesterol levels fell with AZT therapy, the latter significantly (p < 0.05), but both remained at levels significantly higher than at baseline.

Table 4.

Effect of Replacing d4T with AZT on Anthropometric and Metabolic Variables in Subjects with Lipodystrophy a

Variable	Baseline visit at which d4T therapy began	Visit where d4T was replaced with AZT	Final visit
BMI	23.6 (3.5)	24.9 (3.5)*	24.6 (4.6)*
Waist (cm)	75.5 (11.0)	84.0 (13.0)**	82.2 (12.0)**
Triceps (mm)	16.1 (6.0)	11.2 (7.7)	12.1 (6.3)*
Glucose (mmol/liter)	4.2 (0.5)	4.6 (0.9)*	4.6 (0.5)**
Lactate (mmol/liter)	1.28 (0.50)	2.80 (2.02)**	1.95 (1.33)*
Cholesterol (mmol/liter)	3.37 ± 0.83	4.78 ± 1.26***	14.57 ± 1.45**
Triglyceride (mmol/liter)	0.89 (0.60)	1.25 (0.60)*	1.50 (1.00)*

Data expressed as mean ± SD or median (interquartile range); ^* p < 0.05, ^** p < 0.005, ^*** p < 0.0005 vs. baseline; ^† p < 0.05 vs. visit at which d4T was replaced with AZT.

Increases in viral load during ART

In 11 subjects (4 from 18 with lipodystrophy and 7 from 24 without lipodystrophy) viral loads increased from undetectable levels during the course of ART. These rises in viral load could be explained in seven of the patients: four (one subject with lipodystrophy) were due to nonadherence to therapy and three (two with lipodystrophy) were due to temporary removal from ART due to lactic acidosis. Therefore, in four subjects (one with lipodystrophy) the rise in viral load could not be explained. If these are assumed to be due to noncompliance then the total prevalence for noncompliance to ART would be 11.1% (2 of 18) in those with lipodystrophy and 25.0% (6 of 24) in subjects without lipodystrophy (p = 0.26).

Discussion

This study is very relevant because within developing countries the most common first-line ARV regimens for treating HIV infection make use of d4T.^9,10 A number of studies performed in the developed world have shown that the use of d4T leads to detrimental metabolic and morphologic side effects.^11,16
–18 However, despite the prevalent use of d4T in African countries, no longitudinal studies have been performed to monitor the development of such side effects in these population groups. One large study has been performed within a Rwandan population, but this was a cross-sectional analysis.¹⁹

We have demonstrated a high prevalence of lipodystrophy among subjects taking a widely used ART regimen in a black South African population, with several baseline clinical and metabolic factors predicting the development of this syndrome. Subjects who developed lipodystrophy had greater adiposity before receiving ART than subjects who did not. It is possible that lipoatrophy is easier to identify in such subjects; however, there was considerable overlap in BMI and hip and skinfold measurements between the LD and NLD groups and the NLD group displayed small reductions in hip circumference and triceps skinfolds compared to the LD group.

The prevalence of lipodystrophy in our study is similar to that in studies from developed countries^11,20,21 and from a large African study.¹⁹ We described a mixed pattern of lipoatrophy and lipohypertrophy as the most common feature of lipodystrophy, which is similar to the findings from a large Rwandan cohort¹⁹ and agrees with data generated from many studies of lipodystrophy in European and American populations.^2,3 However, it must be noted that studies have reported that lipohypertrophy is not a feature of ART-associated lipodystrophy when HIV-positive subjects are compared with a control group of HIV-negative subjects.^22,23 However, both these studies included patients using a variety of different ART regimens and the control populations had higher body weights than the HIV-positive subjects.^22,23 Furthermore, in females HAART was actually associated with greater visceral fat mass.²³

The most notable feature of lipodystrophy in the current study was the fall in body mass, hip circumference, and triceps skinfold after an initial rise in all three measurements. Waist circumference also tended to rise more dramatically in the LD subjects and WHR was therefore higher in those with lipodystrophy after 111 weeks on ART. Thus, this study supports the many previous investigations demonstrating that lipoatrophy is the prime feature of the lipodystrophic syndrome in patients receiving ART.²

All subjects developed significant increases in serum lipid levels confirming data from other studies.^19,21,24 Although some studies have shown that ART is associated with decreased glucose tolerance,^19,25 the prevalence of impaired fasting glucose did not increase in this study. However, both fasting glucose and insulin were significantly higher by week 111 compared to baseline measurements for the total study cohort. On further analysis it was observed that the LD group showed significant increases in fasting glucose but not fasting insulin from baseline to week 111, whereas the reverse was observed for the NLD subjects. This trend was observed throughout the study period with the percentage change from baseline values for insulin being greater in the NLD than the LD group at all but one time point, whereas the percentage change for glucose was higher in the LD than the NLD group at most time points. Furthermore, BMI, weight, and waist measurements all rose more significantly in LD than NLD subjects. These data suggest that in LD subjects, fasting insulin secretion does not rise sufficiently to compensate for the rise in insulin resistance produced by increased abdominal fat deposition; this leads to poorer glucose tolerance compared to NLD subjects. The reason for the lower increase in fasting insulin levels from baseline in the LD group is not known. However, it is noteworthy that before ART commenced the LD group had higher BMIs and had higher fasting insulin than the NLD subjects. It is therefore possible that the LD subjects, even before ART was initiated, have experienced a period of high beta cell insulin secretory output and that the added burden of increased fat deposition after the commencement of ART has reduced beta cell insulin secretory capacity. The use of NRTIs, particularly d4T treatment, has been shown to increase fasting insulin levels in a longitudinal study¹⁸ and thus d4T in the present study may also contribute to increased insulin secretory demands. Only fasting insulin levels were measured in the current study and therefore we are not able to comment on the effect of body fat changes or ART on insulin secretory responsiveness.

The description of greater adiposity in LD compared to NLD subjects prior to commencement of ART is a novel finding and suggests that fat levels influence the development of the syndrome. The molecular mechanism underlying this relationship is not known; however, it does demonstrate that adipose tissue is a key target organ in ART-associated lipodystrophy. The present study also demonstrated that in comparison to the NLD group the LD subjects tended to accumulate more adipose tissue immediately after initiation of ART at anatomical sites at which lipoatrophy subsequently occurred. Thus, lipoatrophy seems to be preceded by relative lipohypertrophy. These data are intriguing since in vitro data show that NRTIs can block adipogenesis.²⁶ Therefore, lipohypertrophy in the presence of NRTI therapy must be a consequence of adipocyte hypertrophy rather than adipocyte hyperplasia. This suggests that lipoatrophy is a result of the inability of adipose tissue depots to replace fully matured adipocytes undergoing apoptosis with new adipocytes recruited from the preadipocyte population. The reason for the early increased fat deposition observed in LD subjects is not known but may be a consequence of their higher insulin levels or a genetic predisposition to obesity. The reason why lipoatrophy does not occur at the abdominal fat depot is unclear. Visceral adipocytes are metabolically distinct from subcutaneous adipocytes²⁷ and this metabolic difference may also extend to the sensitivity of visceral adipocytes to the inhibitory activity of NRTI drugs on adipogenesis. Studies have also shown that hyperplasia is more prominent in subcutaneous than visceral fat depots^28,29 suggesting that inhibition of adipogenesis may have a greater effect on lipid accumulation in subcutaneous adipose tissue.

Lipodystrophy was associated with a dramatic rise in fasting lactate levels up to 59 weeks of therapy. A previous investigation has also shown that lipodystrophy is a risk factor for hyperlactatemia³⁰ and a number of studies have demonstrated that d4T use is strongly related to elevated lactate levels^30,31 and that discontinuation of d4T leads to a fall in serum lactate concentrations.³¹ The association between hyperlactatemia and lipodystrophy observed in this study further supports the hypothesis that the adverse effects of NRTI therapy are due to mitochondrial toxicity.³²

The replacement of d4T with AZT in the subjects who developed lipodystrophy led to slight improvements in the body fat changes observed after initiation of ART (see Fig. 1 and Table 4). The AZT therapy was monitored for only a median time period of 36 weeks and therefore it is possible that greater improvements may have been observed over a longer period of observation. Previous studies have shown that replacing d4T with AZT does lead to a reversal of lipoatrophy.^33,34 Metabolic variables also improved after AZT therapy was initiated, most particularly lactate and cholesterol levels (see Table 4). This is in agreement with data showing that lactic acidosis does not reoccur after replacing d4T with AZT.^35,36

The lipodystrophy and changes in metabolic variables observed in the present study seem to be the result of ART, as also described in many previous studies.^{11,16

–19,21,24,25} Stavudine is well known to have effects on body fat distribution and metabolic profiles^11,16
–18; however, EFV has also been shown to cause lipoatrophy³⁷ and lead to increases in lipid levels.³⁸ It is therefore possible that the metabolic and anthropometric changes observed in the present study may be due to the combined effects of d4T and EFV.

The positive relationship observed between viral load and HOMA levels in the current study is a novel finding. It is well recognized that insulin resistance and hyperglycemia are features of infection.³⁹ Furthermore, an earlier investigation, although not observing any correlation between insulin resistance and viral load, did report a negative relationship between CD4 counts and insulin resistance in a large group of ART-naive patients.⁴⁰ However, it must also be noted that two previous studies conducted on a small number of ART-naive HIV-infected subjects observed improved insulin sensitivity in the HIV-positive compared to the healthy, control group.^41,42 The data from the current study therefore suggest that the hyperinsulinemia observed in subjects receiving ART may not be a consequence of the therapy alone but may also involve the virus itself.

The negative relationship between LDL cholesterol and viral load observed in this study has also been demonstrated in previous studies^40,43 as has the positive correlation of viral load with triglyceride concentrations.⁴⁰ However, a unique observation has been that the relationship between triglyceride levels and viral load is a consequence of insulin resistance and is therefore not a direct effect of the virus on triglyceride metabolism.

There was no formal assessment of adherence to ART in the current study. All subjects were questioned about adherence at each clinic visit and four did admit to short breaks in therapy, which were mirrored by a rise and fall in viral loads. Unexplained rises and falls in viral load were observed in a further four subjects and this was used as a proxy indicator of noncompliance. Thus, 25% of subjects without and 11.1% of subjects with lipodystrophy were ART noncompliant. A previous study has shown that lipodystrophy is associated with greater adherence to ART.⁴⁴

Limitations of this study include the small sample size. Despite this the study was sufficiently powered to demonstrate statistically significant differences in anthropometric and metabolic variables between LD and NLD subjects and to observe significant relationships of viral load with lipid and insulin levels. Furthermore many of the observations reported in the current study are supported by data from previous investigations.^{1,5,21,24,30,31,40,41} The clinical stage of the disease was not recorded at baseline in the present study. This may be a limitation of the study because it has been shown that the prevalence of metabolic and anthropometric side effects to ART is high in subjects with late-stage HIV infection.⁴⁵ A further drawback of the current study was the method used for the diagnosis of lipodystrophy, which was based on self-reporting by the patient, and confirmation of body fat changes by the physician. This method, however, has been used in many previous studies^{11,19,22,23,46,47} and is reported to be the best procedure to use if scanning techniques for quantifying body fat redistribution, such as DEXA, are not available.⁴⁸ Furthermore, the method used in the present study for diagnosing lipodystrophy is possibly superior to that used in other studies because all the patients were monitored and diagnosed by the same clinician, thus overcoming any problems of interobserver variability and unfamiliarity with the patient's previous body habitus.

A control group of HIV-negative subjects was not included in the present study and therefore we were unable to compare metabolic variables between such subjects and our HIV-positive patients. However, data from South Africa have demonstrated a less atherogenic fasting lipid profile in African than European or Indian population groups^49,50 and this mirrors the low prevalence of hypercholesterolemia and hypertriglyceridemia at baseline in our study. Furthermore, at baseline no subjects met the NCEP-ATPIII criteria for the initiation of lipid-lowering therapy¹⁵ and at the end of the study period only one subject met these criteria. Investigations have also shown that in African subjects insulin resistance is higher than in European subjects but similar to Indian subjects,^51,52 while the prevalence of type 2 diabetes is higher in the Indian than African group with the lowest prevalence observed in the European population.⁵³

The longitudinal nature of the current study allowed changes in metabolic and anthropometric variables to be monitored for up to 2 years from the start of ART. This enabled us to demonstrate that subjects who went on to develop lipodystrophy had greater BMI and higher insulin levels at baseline than subjects who did not develop lipodystrophy and that with initiation of ART insulin levels increased from baseline levels less in the LD than the NLD group. Thus, LD is associated with beta cell insulin secretory dysfunction in combination with insulin resistance and this novel finding is supported by studies demonstrating reduced proinsulin-processing efficiency⁵⁴ and an attenuated insulin secretory response to insulin resistance⁵⁵ in subjects with ART-associated lipodystrophy. Furthermore, profound hyperlactatemia was noted in the LD group, a finding observed in only one previous study.³⁰

The majority of African ART roll out programs continue to use d4T, largely due to cost, availability, and ease of use. In South Africa, which has the largest ART program in the world, d4T is currently still the first-line therapy for the majority of patients. Lipodystrophy and metabolic dysfunction are common using this first-line regimen,^5,19 and it is therefore essential that guidelines for early detection of this highly stigmatizing condition is built into any ART program using d4T as first-line therapy. Nurse and lay-worker-driven ART programs are common in Africa and detection of lipodystrophy with rapid substitution of d4T should be the norm in this setting.

Footnotes

Acknowledgments

We thank the South African Department of Health staff and the clinic patients. The Carnegie foundation and the National Health Laboratory Service funded the study. Dr. Venter is supported by PEPFAR.

Disclosure Statement

No competing financial interests exist.

References

Carr

, Samaras

, Burton

et al. A syndrome of peripheral lipodystrophy, hyperlipidemia and insulin resistance in patients receiving HIV protease inhibitors. AIDS, 1998; 12:F51–F58.

Grinspoon

, Carr

. Cardiovascular risk and body-fat abnormalities in HIV-infected adults. N Engl J Med, 2005; 352:48–62.

Currier

, Havlir

. Complications of HIV disease and therapy. Top HIV Med, 2008; 16:23–30.

Boulle

, Orrel

, Kaplan

et al. Substitutions due to antiretroviral toxicity or contraindication in the first 3 years of antiretroviral therapy in a large South African cohort. Antivir Ther, 2007; 12:753–760.

van Griensven

, De Naeyer

, Mushi

et al. High prevalence of lipoatrophy among patients on stavudine-containing first-line antiretroviral therapy regimens in Rwanda. Trans R Soc Trop Med Hyg, 2007; 101:793–798.

Towards universal access: scaling up priority HIV/AIDS interventions in the health sector. Progress report. Geneva: WHO, 2008. http://www.who.int/hiv/mediacentre/2008progressreport/en/index.html(accessed on December 11th, 2008).

South African national antiretroviral treatment guidelines 1st ed.. Pretoria: National Department of Health, 2004. http://www.doh.gov.za/docs/factsheets/guidelines/artguidelines04/intro.pdf(accessed on December 11th, 2008).

Antiretroviral therapy for HIV infection in adults and adolescents in resource-limited settings: Towards universal access. Geneva: WHO, 2006. http://www.who.int/hiv/pub/guidelines/adult/en/index.html(accessed on December 11th, 2008).

Beck

, Vitoria

, Mandalia

, Crowley

, Gilks

, Souteyrand

. National adult antiretroviral therapy guidelines in resource-limited countries: Concordance with 2003 WHO guidelines? AIDS, 2006; 20:1471–1479.

10.

Renaud-Théry

, Nguimfack

, Vitoria

et al. Use of antiretroviral therapy in resource-limited countries in 2006: Distribution and uptake of first- and second-line regimens. AIDS, 2007; 21,Suppl 4:S89–S95.

11.

Lichtenstein

, Ward

, Moorman

et al. Clinical assessment of HIV-associated lipodystrophy in an ambulatory population. AIDS, 2001; 15:1389–1398.

12.

Wang

, Thornton

, Kolesnik

, Pierson

Jr . Anthropometry in body composition. An overview. Ann NY Acad Sci, 2000; 904:317–326.

13.

Friedewald

, Levy

, Fredrickson

. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem, 1972; 18:499–502.

14.

Matthews

, Hosker

, Rudenski

, Naylor

, Treacher

, Turner

. Homeostasis model assessment: Insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia, 1985; 28:412–419.

15.

Executive summary of the third report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). J Am Med Assoc, 2001; 285:2486–2497.

16.

Heath

, Hogg

, Singer

, Chan

, O'Shaughnessy

, Montaner

. Antiretroviral treatment patterns and incident HIV-associated morphologic and lipid abnormalities in a population-based cohort. J Acquir Immune Defic Syndr., 2002; 30:440–447.

17.

Jones

, Sawleshwarkar

, Michailidis

et al. Impact of antiretroviral choice on hypercholesterolaemia events: The role of the nucleoside reverse transcriptase inhibitor backbone. HIV Med, 2005; 6:396–402.

18.

Brown

, Li

, Cole

et al. Cumulative exposure to nucleoside analogue reverse transcriptase inhibitors is associated with insulin resistance markers in the Multicenter AIDS Cohort Study. AIDS, 2005; 19:1375–1383.

19.

Mutimura

, Stewart

, Rheeder

, Crowther

. Metabolic function and the prevalence of lipodystrophy in a population of HIV-infected African subjects receiving highly active antiretroviral therapy. J Acquir Immune Defic Syndr, 2007; 46:451–455.

20.

Miller

, Carr

, Smith

et al. Lipodystrophy following antiretroviral therapy of primary HIV infection. AIDS, 2000; 14:2406–2407.

21.

Heath

, Hogg

, Chan

et al. Lipodystrophy-associated morphological, cholesterol and triglyceride abnormalities in a population-based HIV/AIDS treatment database. AIDS, 2001; 15:231–239.

22.

Bacchetti

, Gripshover

, Grunfeld

et al. Study of fat redistribution and metabolic change in HIV infection (FRAM). Fat distribution in men with HIV infection. J Acquir Immune Defic Syndr, 2005; 40:121–131.

23.

From the Study of Fat Redistribution and Metabolic Change in HIV Infection (FRAM). Fat distribution in women with HIV infection. J Acquir Immune Defic Syndr, 2006; 42:562–571.

24.

Pujari

, Dravid

, Naik

et al. Lipodystrophy and dyslipidemia among patients taking first-line, World Health Organization-recommended highly active antiretroviral therapy regimens in Western India. J Acquir Immune Defic Syndr, 2005; 39:199–202.

25.

Carr

, Samaras

, Thorisdottir

, Kaufmann

, Chisholm

, Cooper

. Diagnosis, prediction, and natural course of HIV-1 protease-inhibitor-associated lipodystrophy, hyperlipidaemia, and diabetes mellitus: A cohort study. Lancet, 1999; 353:2093–2099.

26.

Pace

, Martin

, Hammond

, Mamotte

, Nolan

, Mallal

. Mitochondrial proliferation, DNA depletion and adipocyte differentiation in subcutaneous adipose tissue of HIV-positive HAART recipients. Antivir Ther, 2003; 8:323–331.

27.

Hamdy

, Porramatikul

, Al-Ozairi

. Metabolic obesity: The paradox between visceral and subcutaneous fat. Curr Diabetes Rev, 2006; 2:367–373.

28.

Tchkonia

, Giorgadze

, Pirtskhalava

et al. Fat depot origin affects adipogenesis in primary cultured and cloned human preadipocytes. Am J Physiol Regul Integr Comp Physiol, 2002; 282:R1286–1296.

29.

Drolet

, Richard

, Sniderman

et al. Hypertrophy and hyperplasia of abdominal adipose tissues in women. Int J Obes (Lond), 2008; 32:283–291.

30.

Bonnet

, Balestre

, Bernardin

, Pellegrin

, Neau

, Dabis

. Risk factors for hyperlactataemia in HIV-infected patients, Aquitaine Cohort, 1999–2003. Antivir Chem Chemother, 2005; 16:63–67.

31.

Hocqueloux

, Alberti

, Feugeas

et al. Prevalence, risk factors and outcome of hyperlactataemia in HIV-infected patients. HIV Med, 2003; 4:18–23.

32.

Brinkman

, ter Hofstede

, Burger

, Smeitink

, Koopmans

. Adverse effects of reverse transcriptase inhibitors: mitochondrial toxicity as common pathway. AIDS, 1998; 12:1735–1744.

33.

McComsey

, Paulsen

, Lonergan

et al. Improvements in lipoatrophy, mitochondrial DNA levels and fat apoptosis after replacing stavudine with abacavir or zidovudine. AIDS, 2005; 19:15–23.

34.

Cherry

, Lal

, Thompson

et al. Increased adipocyte apoptosis in lipoatrophy improves within 48 weeks of switching patient therapy from stavudine to abacavir or zidovudine. J Acquir Immune Defic Syndr, 2005; 38:263–267.

35.

Lonergan

, Barber

, Mathews

. Safety and efficacy of switching to alternative nucleoside analogues following symptomatic hyperlactatemia and lactic acidosis. AIDS, 2003; 17:2495–2499.

36.

Bolhaar

, Karstaedt

. A high incidence of lactic acidosis and symptomatic hyperlactatemia in women receiving highly active antiretroviral therapy in Soweto, South Africa. Clin Infect Dis, 2007; 45:254–260.

37.

Cameron

, da Silva

, Arribas

et al. A 96-week comparison of lopinavir-ritonavir combination therapy followed by lopinavir-ritonavir monotherapy versus efavirenz combination therapy. J Infect Dis, 2008; 198:234–240.

38.

Tashima

, Bausserman

, Alt

, Aznar

, Flanigan

. Lipid changes in patients initiating efavirenz- and indinavir-based antiretroviral regimens. HIV Clin Trials, 2003; 4:29–36.

39.

McGuinness

. Defective glucose homeostasis during infection. Annu Rev Nutr, 2005; 25:9–35.

40.

El-Sadr

, Mullin

, Carr

et al. Effects of HIV disease on lipid, glucose and insulin levels: Results from a large antiretroviral-naive cohort. HIV Med, 2005; 6:114–121.

41.

Hommes

, Romijn

, Endert

, Eeftinck Schattenkerk

, Sauerwein

. Insulin sensitivity and insulin clearance in human immunodeficiency virus-infected men. Metabolism, 1991; 40:651–656.

42.

Heyligenberg

, Romijn

, Hommes

, Endert

, Eeftinck Schattenkerk

, Sauerwein

. Non-insulin-mediated glucose uptake in human immunodeficiency virus-infected men. Clin Sci (Lond), 1993; 84:209–216.

43.

Rose

, Woolley

, Hoy

et al. HIV infection and high-density lipoprotein: The effect of the disease vs the effect of treatment. Metabolism, 2006; 55:90–95.

44.

Guaraldi

, Murri

, Orlando

et al. Morphologic alterations in HIV-infected people with lipodystrophy are associated with good adherence to HAART. HIV Clin Trials, 2003; 4:99–106.

45.

Santos

, Palacios

, Ruiz

, González

, Márquez

. Study of patients diagnosed with advanced HIV in the HAART era–OMEGA Cohort. Int J STD AIDS, 2005; 16:252–255.

46.

Carr

, Emery

, Law

, Puls

, Lundgren

, Powderly

. An objective case definition of lipodystrophy in HIV-infected adults: A case-control study. Lancet, 2003; 361:726–735.

47.

Asensi

, Martín-Roces

, Collazos

et al. Association between physical and echographic fat thickness assessments and a lipodystrophy grading scale in lipodystrophic HIV patients: Their practical implications. AIDS Res Hum Retroviruses, 2006; 22:830–836

48.

Tien

, Grunfeld

. What is HIV-associated lipodystrophy? Defining fat distribution changes in HIV infection. Curr Opin Infect Dis, 2004; 17:27–32.

49.

Seftel

, Asvat

, Joffe

et al. Selected risk factors for coronary heart disease in male scholars from the major South African population groups. S Afr Med J, 1993; 83:891–897.

50.

Naran

, Chetty

, Crowther

. The influence of metabolic syndrome components on plasma PAI-1 concentrations is modified by the PAI-1 4G/5G genotype and ethnicity. Atherosclerosis, 2008; 196:155–163.

51.

van der Merwe

, Crowther

, Schlaphoff

, Gray

, Joffe

, Lönnroth

. Evidence for insulin resistance in black women from South Africa. Int J Obes Relat Metab Disord, 2000; 24:1340–1346.

52.

Ferris

, Naran

, Crowther

, Rheeder

, van der Merwe

, Chetty

. The relationship between insulin sensitivity and serum adiponectin levels in 3 population groups. Horm Metab Res, 2005; 37:695–701.

53.

Motala

, Omar

, Pirie

. Diabetes in Africa. Epidemiology of type 1 and type 2 diabetes in Africa. J Cardiovasc Risk., 2003; 10:77–83.

54.

Haugaard

, Andersen

, Halsall

, Iversen

, Hales

, Madsbad

. Impaired proinsulin secretion before and during oral glucose stimulation in HIV-infected patients who display fat redistribution. Metabolism, 2007; 56:939–946.

55.

Haugaard

, Andersen

, Volund

et al. Beta-cell dysfunction and low insulin clearance in insulin-resistant human immunodeficiency virus (HIV)-infected patients with lipodystrophy. Clin Endocrinol (Oxf), 2005; 62:354–361.