Abstract
Context: The expression of adipogenic genes in sc adipose tissue has been reported to be lower among patients with HIV-associated lipoatrophy than HIV-uninfected controls. It is unclear whether this is a result or cause of lipoatrophy.
Objective: The objective of the study was to investigate the temporal relationships among changes in adipogenic gene expression in sc adipose tissue and changes in body fat distribution and metabolic complications in HIV-infected subjects on antiretroviral therapy.
Design: This was a prospective longitudinal study.
Setting: The study was conducted at HIV clinics in Seattle, Washington.
Participants: The study population included 31 HIV-infected and 12 control subjects.
Interventions: Subjects were followed up for 12 months after they initiated or modified their existing antiretroviral regimen.
Main Outcome Measures: Changes in body composition, plasma lipids, insulin sensitivity, and gene expression in sc abdominal and thigh adipose tissue.
Results: Subjects who developed lipoatrophy (n = 10) had elevated fasting triglycerides [3.16 (sd 2.79) mmol/liter] and reduced insulin sensitivity as measured by frequently sampled iv glucose tolerance test [1.89 (sd 1.27) × 10−4 min−1/μU·ml] after 12 months, whereas those without lipoatrophy (n = 21) did not show any metabolic complications [triglycerides 1.32 (sd 0.58) mmol/liter, P = 0.01 vs. lipoatrophy; insulin sensitivity 3.52 (sd 1.91) × 10−4 min−1/μU·ml, P = 0.01 vs. lipoatrophy]. In subjects developing lipoatrophy, the expression of genes involved in adipocyte differentiation, lipid uptake, and local cortisol production in thigh adipose tissue was significantly reduced already at the 2-month visit, several months before any loss of extremity fat mass was evident.
Conclusions: In HIV-infected subjects, lipoatrophy is associated with elevated fasting triglycerides and insulin resistance and might be caused by a direct or indirect effect of antiretroviral drugs on sc adipocyte differentiation.
Reduced expression of adipogenic genes in thigh, but not abdominal, subcutaneous adipose tissue precedes the development of lipoatrophy in HIV-infected subjects on highly-active antiretroviral therapy.
The last decade has seen remarkable advances in the treatment of HIV infection. Highly active antiretroviral therapy (HAART) has proven very effective in preventing disease progression (1). However, a significant number of patients treated with HAART develop lipodystrophy syndrome characterized by changes in body fat distribution, dyslipidemia, and insulin resistance (2) that is associated with drug noncompliance and an increased risk of cardiovascular disease (3).
Lipodystrophy is characterized by a loss of peripheral fat mass (4,5). Among the mechanisms being considered for this peripheral lipoatrophy are effects of antiretroviral drugs on adipocyte differentiation and apoptosis (2,6). Several protease inhibitors have been shown to inhibit adipocyte differentiation in vitro (7,8,9,10). Bastard et al. (11) reported that the expression of genes involved in adipocyte differentiation, including sterol regulatory element binding protein (SREBP)-1, were reduced in sc fat tissue of patients suffering from lipoatrophy. From this study, however, it remained unclear whether the change in adipose tissue gene expression preceded or followed peripheral lipoatrophy.
We performed the present study to investigate the temporal relationships among changes in adipogenic gene expression in abdominal and thigh sc adipose tissue and changes in body fat distribution, plasma lipid concentrations, and insulin sensitivity in HIV-infected subjects starting HAART or changing their regimen. Whereas most previous studies used a cross-sectional approach, we followed up subjects for 12 months with repeated measurements of body fat mass and distribution as well as frequently sampled iv glucose tolerance tests to assess insulin sensitivity. We performed sc adipose tissue biopsies at baseline and after 2 and 12 months of HAART initiation or modification to analyze the expression of adipogenic genes in three categories: adipocyte differentiation, cellular lipid uptake, and adipocyte conversion of cortisone to cortisol. Changes in body fat distribution of HIV-infected subjects were interpreted relative to changes observed in HIV-uninfected control subjects.
Subjects and Methods
Study design and subjects
We enrolled 57 HIV-infected patients and 14 HIV-uninfected control subjects into this prospective longitudinal study. Patients received routine clinical care from their providers at their regular clinics, all in Seattle, WA, throughout the study. Of the enrolled HIV-infected subjects, 17 did not appear for their 2-month visit because they either had not started HAART as planned (n = 4) or stopped taking their medication within the first few weeks (n = 5), because of preexisting lipodystrophy (n = 1) or death (n = 1), or because they were unable or unwilling to complete all required follow-up visits (n = 6). Another five subjects dropped out from the study at a later time point because of medication noncompliance (n = 4) or death (n = 1), whereas four were excluded from analysis because of preexisting lipodystrophy (n = 1), prolonged illness (n = 1), cocaine and heroin abuse (n = 1), and cross-sex hormone treatment (n = 1). Other exclusion criteria included active opportunistic infection or tumor; fasting plasma triglyceride levels of greater than 11.3 mmol/liter or fasting plasma glucose levels greater than 7 mmol/liter at baseline; and previous or current treatment with anabolic drugs, anticytokine agents, glucocorticoids, or inhibitors of glucocorticoid production. Of the remaining 31 subjects (30 men, 1 woman), 24 were HAART naive at the time of enrollment. The mean duration since the subjects’ diagnosis before starting the study was 70 (sd 66) months. We also enrolled 14 HIV-uninfected men as control subjects. Two control subjects did not complete all study visits and were therefore excluded from analyses.
All HIV-infected subjects were admitted to the University of Washington General Clinical Research Center at baseline. The baseline visit occurred no more than 2 wk before starting HAART in naive subjects or switching at least two drugs in the regimen of subjects already receiving HAART. Subjects then came in for 2-, 6-, and 12-month follow-up visits. During all visits, we assessed medication history, vital signs, height and weight, and waist and hip circumference. A whole-body dual-energy x-ray absorptiometry (DEXA) scan and a computed tomography (CT) scan at the umbilicus were performed at baseline, 6 months, and 12 months. At baseline, 2 months, and 12 months, we drew fasting blood samples for measurement of plasma lipids and performed a frequently sampled iv glucose tolerance test and sc abdominal and thigh adipose tissue biopsies. All study procedures were in accordance with The Declaration of Helsinki and were approved by the University of Washington Human Subjects Committee. All subjects gave written consent.
Procedures
Whole-body DEXA scans were performed on a Hologic QDR 1500 (Hologic Inc., Bedford, MA) or a GE Lunar Prodigy (General Electric Healthcare, Waukesha, WI) scanner. Regression analysis was used to convert measurements performed on the Hologic scanner to those performed on the Lunar scanner, using data from 11 men who were scanned on both machines (for trunk fat mass: r2 = 0.93; for extremity fat mass: r2 = 0.98). Abdominal adipose tissue distribution was assessed by CT scan on a GE 8800 scanner (General Electric Medical Systems Americas, Milwaukee, WI) as described in detail previously (12).
A modification of the 4-h frequently sampled iv glucose tolerance test described by Beard et al. (13) was used to measure insulin sensitivity. The modification consisted of substituting a 10-min iv insulin infusion (0.025 U/kg beginning 20 min after the glucose bolus) for an iv tolbutamide infusion. The minimal model of glucose kinetics of Bergman et al. (14) was used to compute the insulin sensitivity index.
Blood samples were placed on ice immediately after withdrawal, and spun within 2 h at 3000 × g for 15 min, and EDTA plasma was frozen at −70 C until analysis. Plasma for the measurement of free fatty acids contained tetrahydrolipstatin at a final concentration of 1 mg/liter. Total and high-density lipoprotein (HDL)-cholesterol as well as triglycerides were measured in plasma samples by the Northwest Lipid Research Laboratory, Seattle, WA, a reference laboratory within the National Reference System for cholesterol. Low-density lipoprotein-cholesterol was calculated using the Friedewald equation. Plasma free fatty acids were measured by an enzymatic colorimetric assay kit (Wako Chemicals USA, Richmond, VA).
Fat tissue was collected to analyze the expression of SREBP-1; peroxisome proliferative activated receptor (PPAR)-γ; the CCAAT/enhancer binding proteins (C/EBP)-α, -β, and -δ; lipoprotein lipase (LPL); hydroxysteroid 11-β dehydrogenase 1 (11β-HSD1); and glyceraldehyde-3-phosphate dehydrogenase (GAPDH). Subcutaneous adipose tissue needle biopsies were performed at the abdomen just lateral to the umbilicus, and at the lateral mid-thigh at baseline and after 2 and 12 months of HAART initiation or modification. About 0.5 g of adipose tissue was removed, immediately frozen on dry ice, and stored at −70 C until analysis. Total RNA was isolated using Absolutely RNA miniprep kits (Stratagene, La Jolla, CA) with on-column DNase treatment to remove trace genomic DNA. Quantitative PCR was performed on an Mx4000 Multiplex QPCR system (Stratagene) with samples loaded in triplicate using approximately 10–50 ng of total RNA per reaction. Quantitative PCR was run in a 25-μl reaction using Stratagene Brilliant single-step quantitative RT-PCR kit. Pooled total RNA from patient samples was used for standard curves as 1:3 serial dilutions. The standard curves showed efficiencies between 75 and 83%, and r2 values of 0.99 or higher for all genes. Ct values for each gene from patient samples were converted to nanograms based on the standard curve run with each plate and normalized to the actual values of total RNA added to each reaction (15). Total RNA was quantified on the Mx4000 Multiplex QPCR system in duplicate wells with the RiboGreen RNA quantitation kit (Molecular Probes, Eugene, OR) using standards supplied by the manufacturer. Probes and primers were from either Integrated DNA Technologies, Inc. (Coralville, IA) or Operon (Huntsville, AL) and designed using Primer Express 2.0 software (Applied Biosystems Inc., Foster City, CA). Sequences of probes and primers are available upon request.
Statistical analysis
Analyses were performed using SPSS, version 11.5 (SPSS Inc., Chicago, IL). Distribution of variables was analyzed by checking histograms and normal plots of the data, and normality was tested by means of Kolmogorov-Smirnov and Shapiro-Wilk tests. The association between changes in trunk and extremity fat mass was tested by Pearson’s correlation test. Anthropometric and metabolic data at baseline as well as the change from baseline to 12 months were analyzed by means of ANOVA, using the unmodified data as the dependent variable for normally distributed variables and the ranks for nonnormally distributed variables. If the ANOVA indicated a significant difference, we performed post hoc least significant difference tests. Changes in metabolic variables from baseline to 6 and 12 months were analyzed by Friedman tests, and differences between the lipoatrophy and nonlipoatrophy subjects at different time points were tested by Mann-Whitney U tests, adjusted for multiple testing. CD4 cell count and viral load were analyzed by repeated-measures ANOVA, with the degrees of freedom adjusted according to Greenhouse and Geisser where appropriate. Gene expression data were compared by Wilcoxon signed rank tests to compare expression levels at later time points with baseline levels and by Mann-Whitney U tests to compare independent groups.
Results
In the 12 months of the study, the HIV-uninfected control subjects gained 3.9% of trunk fat mass (sd 19.9%) and 0.02% (sd 18.7%) of extremity fat mass. We used these data as objective figures against which body fat changes in HIV-infected subjects could be gauged. We defined central lipohypertrophy as a gain in trunk fat mass of more than 23.8% (mean change in the control group +1 sd) and peripheral lipoatrophy as a loss of extremity fat mass of more than 18.7% (mean change in the control group −1 sd, Fig. 1). In our study, changes in trunk fat mass were positively correlated with changes in extremity fat mass (r2 = 0.63; P < 0.001). Only one subject developed both central lipohypertrophy and peripheral lipoatrophy; all others with peripheral lipoatrophy also experienced a loss of central fat, whereas all others with central lipohypertrophy gained peripheral fat mass as well. We therefore compared the subjects who experienced a general loss of fat mass (lipoatrophy, n = 9) and the subjects who gained fat mass overall (lipohypertrophy, n = 11) with those who had unchanged fat distribution (no lipodystrophy, n = 10) after 1 yr; the single subject who developed both central lipohypertrophy and peripheral lipoatrophy was excluded from analyses comparing lipoatrophy and lipohypertrophy.
Figure 1.
Changes in trunk and extremity fat mass in HIV-infected subjects on HAART (▪) and HIV-uninfected controls (○) over the 12 months of the study. The two lines indicate the mean change ± 1 sd in trunk fat mass (+23.8%) as well as the mean change −1 sd loss in extremity fat mass (−18.7%) observed in controls. HIV-infected subjects who gained more than 23.8% of trunk fat mass were defined as having developed central lipohypertrophy; HIV-infected subjects who lost more than 18.7% of extremity fat mass were defined as having developed peripheral lipoatrophy.
Subjects who went on to develop lipoatrophy had higher total fat mass at baseline [25.3 (sd 10.0) kg] than subjects developing lipohypertrophy [14.2 (sd 6.6) kg; P < 0.05 vs. lipoatrophy], subjects with no lipodystrophy [15.1 (sd 7.5) kg; P < 0.05 vs. lipoatrophy], or control subjects [15.5 (sd 6.8) kg; P < 0.05 vs. lipoatrophy]. Their trunk fat mass was approximately 70% greater than that of any other group (P < 0.05 for lipoatrophy vs. each of the other three groups), and they had about twice as much intraabdominal fat as subjects not developing lipoatrophy [118 (sd 46) cm2 vs. 59 (sd 28), 59 (sd 42), and 62 (sd 37) cm2 for lipohypertrophy, no lipodystrophy, and control groups; P < 0.05 for lipoatrophy vs. each of the other three groups; see supplemental Table 1, published as supplemental data on The Endocrine Society’s Journals Online Web site at http://jcem.endojournals.org]. Both the no-lipodystrophy group and the controls had very little body composition change over 12 months (Table 1). In contrast, subjects in the lipohypertrophy group gained a large amount of fat mass, by all DEXA and CT scan measures. Subjects in the lipoatrophy group lost fat mass overall, which was due to a loss in trunk fat mass and a more pronounced loss in extremity fat mass. Abdominal sc fat area as measured by CT scan decreased, whereas intraabdominal fat area increased in this group despite the overall loss of body fat. The trunk to extremity fat ratio increased comparably in both the lipohypertrophy and lipoatrophy groups, although the fat distribution changes contributing to this increase differed markedly in the two groups. Lean body mass increased in all HIV-infected subjects by a mean of 2.8 kg (sd 3.9 kg) during the 12 months of the study, without any significant differences between subjects developing lipoatrophy, lipohypertrophy, or no lipodystrophy. It is noteworthy that the loss of trunk and extremity fat mass in the lipoatrophy group began only after the sixth month of study (Fig. 2). In contrast, gains in trunk and extremity fat mass were already evident in the lipohypertrophy group at 6 months.
Table 1.
Changes in anthropometric variables in HIV-infected subjects during the first year of initiating or modifying HAART and HIV-uninfected controls (in % of baseline)
| Lipoatrophy (n = 9) | Lipohypertrophy (n = 11) | No lipodystrophy (n = 10)a | Controls (n = 12) | |
|---|---|---|---|---|
| Group-defining characteristics based on DEXA scans | ||||
| Total fat mass | −22.7 (17.3) | +49.7 (25.6) | +4.7 (17.0) | +2.5 (18.4) |
| Trunk fat mass | −13.0 (18.1) | +52.0 (17.0) | +3.2 (19.7) | +3.9 (19.9) |
| Extremity fat mass | −36.5 (16.1) | +49.7 (45.2) | +7.9 (20.8) | +0 (18.7) |
| Trunk to extremity fat ratio | +27.8 (23.1) | +19.5 (23.3) | +0.3 (15.4) | +5.7 (14.6) |
| Other anthropometric measures | ||||
| Lean body mass | +5.8 (4.6) | +6.5 (10.0) | +6.5 (11.0) | ±0.0 (4.5) |
| Body weight | −4.8 (6.0)a | +6.3 (10.0)b | +0.5 (4.0)a | +0.6 (3.7)a |
| Intraabdominal fat | +21 (33)a | +103 (61)b | +5 (32)a | +20 (61)a |
| Abdominal sc fat | −27 (28)a | +47 (46)b | −4 (25)a,c | −2 (19)c |
| Waist to hip ratio | +0.7 (2.1)a | +6.0 (6.5)b | +0.3 (2.3)a | +0.1 (2.1)a |
Data are expressed as means (sd). Values with different superscript lettersdiffer from each other at P < 0.05 in post hoc least significant difference tests. No statistical tests were performed on group-defining characteristics.
, n = 9 for intraabdominal fat and abdominal sc fat.
Figure 2.
Changes in trunk (A) and extremity (B) fat mass in HIV-infected subjects developing lipohypertrophy (• and solid lines, n = 11), lipoatrophy (▴ and dashed lines, n = 9), and no lipodystrophy (▪ and broken lines, n = 10) during 12 months of antiretroviral therapy. Because the group definitions were based on these variables, no statistical tests were performed. Data represent means and se values.
Despite their marked gain in intraabdominal fat, the lipohypertrophy subjects resembled the no lipodystrophy subjects in having normal fasting triglyceride and insulin levels as well as normal insulin sensitivity at 12 months (Table 2). Despite the fact that they had lost a large amount of fat mass, the lipoatrophy subjects showed higher fasting triglyceride and insulin levels and lower insulin sensitivity by 12 months than subjects without lipoatrophy. Because these metabolic complications were confined to the lipoatrophy group and because this form of lipodystrophy seems to be the most characteristic and stigmatizing change in fat distribution associated with HIV infection (5), we combined the lipohypertrophy and no-lipodystrophy groups for the analyses of changes in metabolic variables and adipose tissue gene expression over time. The time course of the changes in triglyceride and HDL-cholesterol levels as well as insulin sensitivity is shown in Fig. 3 for those subjects who did and did not develop lipoatrophy. Fasting triglycerides had increased already at the 2-month time point in the group developing lipoatrophy, before any loss of peripheral fat mass had occurred, and remained elevated throughout. By contrast, there were only minor changes in plasma triglycerides in subjects not developing lipoatrophy. There were few differences in HDL-cholesterol or changes in HDL-cholesterol between these two groups. It was notable that HDL-cholesterol increased slightly but significantly over time in both groups. Insulin sensitivity tended to be lower at baseline (P = 0.10) in subjects developing lipoatrophy and decreased further in the study period, albeit not to a statistically significant extent. The use of specific antiretroviral medications did not seem to distinguish subjects who developed lipoatrophy at 12 months from those who did not (supplemental Table 2).
Table 2.
Lipid and glucose metabolism at 12 months in HIV-infected subjects who developed lipoatrophy, lipohypertrophy, or no lipodystrophy within 12 months of initiating or modifying HAART
| Lipoatrophy (n = 9) | Lipohypertrophy (n = 11)a | No lipodystrophy (n = 10) | |
|---|---|---|---|
| Triglycerides (mmol/liter) | 3.27 (2.94)a | 1.38 (0.64)b | 1.25 (0.52)b |
| Cholesterol (mmol/liter) | 4.51 (1.19) | 3.96 (0.85) | 3.81 (0.73) |
| LDL-cholesterol (mmol/liter) | 2.15 (0.65) | 2.41 (0.67) | 2.33 (0.52) |
| HDL-cholesterol (mmol/liter) | 0.83 (0.23) | 0.93 (0.52) | 0.91 (0.34) |
| Free fatty acids (mmol/liter) | 0.64 (0.19) | 0.58 (0.24) | 0.57 (0.26) |
| Insulin sensitivity (× 10−4 min−1/μU·ml) | 1.56 (0.79)a | 3.90 (2.69)b | 3.67 (1.74)b |
| Fasting insulin (pmol/liter) | 144 (60)a | 90 (40)b | 76 (47)b |
| Fasting glucose (mmol/liter) | 5.26 (0.37) | 5.20 (0.39) | 5.37 (0.38) |
| Adiponectin (μg/ml) | 3.9 (2.0)a | 5.3 (2.3)a | 10.4 (7.1)b |
Data are expressed as means (sd). Values with different superscript lettersdiffer from each other at P < 0.05 in post hocleast significant difference tests (P = 0.059 for post hoc comparison of triglycerides and lipoatrophy vs. lipohypertrophy). LDL, Low-density lipoprotein.
, n = 10 for insulin sensitivity, fasting insulin, and fasting glucose.
Figure 3.
Changes in plasma triglycerides (A), HDL-cholesterol (B), and insulin sensitivity (C) over the 12 months of the study in HIV-infected subjects developing lipoatrophy (▴ and dashed lines, n = 10) or not developing lipoatrophy (• and solid lines, n = 21 except for insulin sensitivity, where n = 19). *, P < 0.05 for comparison of the two groups at that time point. †, P < 0.05 for change within group over time (Friedman test). Data represent means and se values.
The only changes in adipogenic gene expression in subjects not developing lipoatrophy were an increase in 11β-HSD1 expression in both abdominal and thigh adipose tissue (P = 0.04 and P = 0.03, respectively) and an increase in the expression of C/EBPβ in thigh adipose tissue (P = 0.03) between baseline and the 12-month visit (Fig. 4, A and B). In the subjects developing lipoatrophy, we saw reductions in thigh adipose tissue expression of SREBP-1 (P = 0.03), PPARγ (P = 0.02), LPL (P = 0.01), and C/EBPα (P = 0.02) between baseline and 12 months. Because extremity fat mass was already reduced at this time point, these changes could have been the result of the process leading to lipoatrophy.
Figure 4.
Absolute changes in mRNA concentrations of genes involved in adipocyte differentiation, lipid uptake, and local cortisol production as well as GAPDH in the 12-month study period in abdominal (A) and thigh (B) sc adipose tissue as well as in the first 2 months of the study in abdominal (C) and thigh (D) sc adipose tissue. Data are mRNA concentrations normalized for total RNA content (in nanograms of standard per nanograms total RNA in reaction). Error bars represent se values. ▪, No lipoatrophy group [n = 16 for abdominal and n = 15 for thigh adipose tissue data (A and B); n = 20 for abdominal and n = 21 for thigh adipose tissue (C and D)]; □, lipoatrophy group [n = 9 for abdominal and n = 8 for thigh adipose tissue (A and B); n = 10 for abdominal and n = 9 for thigh adipose tissue (C and D)]. *, P < 0.05 for comparison of mRNA concentrations at baseline and 12-month visit (A and B) and baseline and 2-month visit (C and D), respectively, within each group. **, P < 0.05; ***, P < 0.01 for comparison of change between the two groups.
However, thigh adipose tissue expression of almost all adipogenic genes was reduced already at the 2-month time point in subjects who would later develop lipoatrophy, well before any loss of extremity fat mass was detected (Fig. 4, C and D). From baseline, the mRNA concentrations of SREBP-1, PPARγ, LPL, C/EBPα, C/EBPβ, C/EBPδ, and 11β-HSD1 were reduced by an average of 31–44% in the lipoatrophy group, whereas the expression of none of these genes had changed significantly in the subjects not developing lipoatrophy. Accordingly, these changes between baseline and 2 months were significantly different between those subjects who developed and those who did not develop lipoatrophy for these seven genes that play key roles in adipocyte differentiation, lipid uptake, and intracellular conversion of cortisone to cortisol. In abdominal sc adipose tissue, by contrast, the expression of all genes tended to increase, most notably for C/EBPα and C/EBPβ. The expression of GAPDH did not change to a statistically significant extent in this period and did not differ between subjects developing and not developing lipoatrophy. To ensure that antiretroviral drug treatment before baseline did not affect our findings, we repeated all analyses with the 23 subjects who were HAART naïve at baseline only; all results were very similar (data not shown).
Discussion
The primary findings of our prospective longitudinal study were that lipoatrophy, but not lipohypertrophy, was associated with the metabolic complications of elevated fasting triglyceride concentrations and insulin resistance and that lipoatrophy was preceded by a reduction in thigh sc adipose tissue expression of genes involved in adipocyte differentiation, lipid uptake, and local cortisol production. It is important to note that insulin sensitivity tended to be lower already at baseline in subjects who would later develop lipoatrophy, suggesting that insulin resistance or hyperinsulinemia might contribute to the effects of HAART on gene expression and peripheral fat loss.
Whereas it was initially believed that peripheral lipoatrophy was associated with central lipohypertrophy in the majority of affected patients, our data confirm a previous report that this is the case in only a small number of subjects (5). Instead, central lipohypertrophy was associated with peripheral lipohypertrophy, and peripheral lipoatrophy was associated with reduced trunk fat mass. Because the loss of trunk fat was smaller than the loss of extremity fat mass in the lipoatrophy group, the trunk to extremity fat ratio increased in both subtypes of lipodystrophy. Thus, this ratio does not provide an adequate description of body fat redistribution in HIV-infected individuals.
To our surprise, we did not see metabolic complications in subjects who did not develop lipoatrophy. Even the large increases in total, trunk, and intraabdominal fat in those subjects developing lipohypertrophy without lipoatrophy were not associated with an increase in fasting triglycerides or a reduction in insulin sensitivity. Plasma concentrations of HDL-cholesterol actually increased in this group, probably due to improved control of subjects’ HIV infections over the 12 months of the study. It is likely that preservation of peripheral sc adipocytes in subjects with lipohypertrophy provided a storage location for fatty acids, which would have otherwise been deposited in liver and muscle. In support of this hypothesis, it had been reported that the severity of insulin resistance in patients with HAART-associated lipodystrophy is related to the extent of fat accumulation in liver and muscle cells (16,17). Whereas it is plausible that the loss of sc adipose tissue in itself could cause a passive shunting of fat into liver and muscle cells, an increased storage of triglycerides in hepatocytes and myocytes could also be the result of the hypoleptinemia and hypoadiponectinemia that is associated with lipoatrophy (18,19,20).
In our study, the overlap in total and trunk fat mass between the lipoatrophy and lipohypertrophy groups was extensive at the 12-month time point. Had we based our categorization only on the 12-month data, we would have incorrectly attributed metabolic complications to lipohypertrophy. Whereas it had been established before that protease inhibitors to a varying extent reduce insulin sensitivity (21,22) and raise plasma triglycerides (23,24), our study now demonstrates that the HAART-induced body fat distribution change associated with these metabolic complications is lipoatrophy but not lipohypertrophy.
Bastard et al. (11) reported previously that the expression levels of SREBP-1, PPARγ, C/EBPα, and C/EBPβ in sc adipose tissue were lower in lipoatrophic subjects after an average of 37 months of protease inhibitor-based HAART than in controls. Their study was unable to determine whether this reduction in expression of these key genes for adipocyte differentiation was the result or a cause of lipoatrophy. Also, these authors were unable to assess whether the expression levels actually changed with HAART over time because they compared lipoatrophic patients with HIV-uninfected healthy controls. Our study demonstrates that expression of genes involved in adipocyte differentiation, lipid uptake, and local cortisol production is actually decreased relative to baseline at the 12-month time point when lipoatrophy is present. Furthermore, this reduction in gene expression was seen already at 2 months, before a loss of extremity fat mass occurs. Because all of our target genes encode proteins that could reasonably affect the efficient storage of fat and because changes in gene expression can be expected to result in changes in the concentration of the encoded proteins, our finding suggests that a direct or indirect effect of HAART on adipogenic gene expression potentially contributes to the later loss in fat mass. It remains unclear why changes in gene expression were evident in thigh but not abdominal sc fat.
Although subjects in the lipoatrophy group demonstrated marked reductions in adipogenic gene expression in thigh sc adipose tissue between baseline and 2 months, their loss of extremity fat mass was not evident until the 12-month time point. This contrasts with the lipohypertrophy group in which increases in trunk and extremity fat mass were clearly present at 6 months. The loss in extremity fat mass in the lipoatrophy group in the second half of the study period did not appear to be due to changes in antiretroviral drug intake because type and dose of the drugs in the subjects’ HAART regimen were stable throughout the study. A possible explanation for this observation is that the process leading to extremity fat loss began early during the study but was obscured by a reduction in inflammation and improved cellular nutrition after initiation or modification of the subjects’ HAART regimens. Favorable changes in HIV viral load, CD4+ cell counts, and proinflammatory cytokine levels between baseline and 2 months support this possibility (supplemental Table 3). Presumably, at some time between 6 and 12 months, the lipoatrophic process prevailed over the early improvement in the subjects’ infections, and a loss in extremity fat became detectable.
It has been suggested by Sutinen et al. (25) that increased expression of 11β-HSD1 in abdominal adipose tissue might explain the pseudo-Cushing’s features in patients with HAART-associated lipodystrophy. This is consistent with our study in that we also observed an increase in the expression of 11β-HSD1 in abdominal adipose tissue in HIV-infected subjects on HAART. However, the fact that in our study the 11β-HSD1 expression differed distinctly between abdominal and thigh adipose tissue as well as between subjects developing and not developing lipoatrophy suggests that the mechanism linking antiretroviral drugs and the different types of lipodystrophy is more complex and not solely due to increased adipose tissue 11β-HSD1 expression. In particular, our finding of a reduction in the expression of genes critical for adipocyte differentiation in adipose tissues that later undergo lipoatrophy suggests that antiretroviral drugs, directly or indirectly through changes in immune status or response, interfere with an early step in the differentiation cascade, for example the hormonal stimulation of C/EBPβ and -δ expression.
Limitations of this study include the small sample size and the fact that HIV-infected subjects were treated by a large number of different HAART regimens. These limitations made it impossible to determine whether lipoatrophy and/or metabolic complications were associated with specific drugs or drug classes. Another potential concern is that subjects who developed lipoatrophy had more total and trunk fat mass at baseline than subjects developing lipohypertrophy. It would therefore be possible that both groups simply moved closer to the means in subsequent measurements. We performed multiple regression analysis using all body composition variables to test whether this might have been the case in our groups of subjects and found that the pattern of body composition changes in the three groups of HIV-infected subjects was significantly different from that observed in the HIV-uninfected control subjects. For example, the lipoatrophy group gained intraabdominal fat and lean body mass in the face of distinct losses of extremity and sc abdominal fat mass. This analysis supports our contention that the body composition changes seen in the HIV-infected subjects on HAART were highly specific.
The gene expression data were normalized to the amount of total RNA added to the RT-PCR instead of a housekeeping gene. Our rationale for doing so was that normalizing for total RNA content has generally been recommended for samples obtained in vivo (15), whereas under conditions in which adipose tissue mass is undergoing large changes, there is no housekeeping gene that is known to be expressed at a stable level. We initially measured GAPDH expression for this purpose but found GAPDH expression changes to be highly correlated to those of all of our target genes. Furthermore, a review of the literature had revealed that one of our target genes, C/EBPα, plays an important role in regulating GAPDH expression (26). The fact that we analyzed thigh and abdominal adipose tissue simultaneously and ran a standard curve with each reaction set to correct for possible variation in reaction efficiency makes it extremely unlikely that our results were an artifact of normalization strategy.
In conclusion, our study demonstrates that the metabolic complications of HIV-associated lipodystrophy are restricted to individuals developing lipoatrophy and that these individuals might be identified long before they lose extremity fat mass by finding reduced expression of adipogenic genes in thigh sc adipose tissue. These observations could form the basis for a clinical test to determine whether a patient is at risk for developing lipoatrophy in time to change the HAART regimen and possibly prevent this complication of therapy. More research must be done to determine the feasibility of this approach.
Supplementary Material
Acknowledgments
We thank Pamela Y. Yang, Steven I. Hashimoto, and Carol A. Isaac for excellent technical assistance and Heidi Crane for critically reading the manuscript.
Footnotes
This work was supported by individual National Institutes of Health Grants 1 R01 DK55460 and 1 K24 DK02860 (to D.S.W.). A portion of this work was conducted through the Clinical Nutrition Research Unit at the University of Washington (National Institutes of Health Grant P30-DK035816) and the General Clinical Research Center at the University of Washington (National Institutes of Health Grant M01-RR-00037). Gene expression analyses were performed in the Molecular and Genetics Core of the Diabetes Endocrinology Research Center, National Institutes of Health Grant 5-P30-DK17047. M.K. was supported by a Dick and Julia McAbee Endowed Fellowship in Diabetes Research from the Diabetes Endocrinology Research Center of the University of Washington. S.E.K. is the recipient of a Distinguished Clinical Scientist Award from the American Diabetes Association.
Disclosure Statement: The authors have nothing to disclose.
First Published Online December 18, 2007
Abbreviations: C/EBP, CCAAT/enhancer binding protein; CT, computed tomography; DEXA, dual-energy x-ray absorptiometry; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; HAART, highly active antiretroviral therapy; HDL, high-density lipoprotein; 11β-HSD1, hydroxysteroid 11-β dehydrogenase 1; LPL, lipoprotein lipase; PPAR, peroxisome proliferative activated receptor; SREBP, sterol regulatory element binding protein.
References
- Yeni P 2006 Update on HAART in HIV. J Hepatol 44:S100–S103 [DOI] [PubMed] [Google Scholar]
- Koutkia P, Grinspoon S 2004 HIV-associated lipodystrophy: pathogenesis, prognosis, treatment, and controversies. Annu Rev Med 55:303–317 [DOI] [PubMed] [Google Scholar]
- Kamin DS, Grinspoon SK 2005 Cardiovascular disease in HIV-positive patients. AIDS 19:641–652 [DOI] [PubMed] [Google Scholar]
- Lichtenstein KA 2005 Redefining lipodystrophy syndrome: risks and impact on clinical decision making. J Acquir Immune Defic Syndr 39:395–400 [DOI] [PubMed] [Google Scholar]
- Bacchetti P, Gripshover B, Grunfeld C, Heymsfield S, McCreath H, Osmond D, Saag M, Scherzer R, Shlipak M, Tien P 2005 Fat distribution in men with HIV infection. J Acquir Immune Defic Syndr 40:121–131 [DOI] [PMC free article] [PubMed] [Google Scholar]
- Domingo P, Matias-Guiu X, Pujol RM, Francia E, Lagarda E, Sambeat MA, Vazquez G 1999 Subcutaneous adipocyte apoptosis in HIV-1 protease inhibitor-associated lipodystrophy. AIDS 13:2261–2267 [DOI] [PubMed] [Google Scholar]
- Grigem S, Fischer-Posovszky P, Debatin KM, Loizon E, Vidal H, Wabitsch M 2005 The effect of the HIV protease inhibitor ritonavir on proliferation, differentiation, lipogenesis, gene expression and apoptosis of human preadipocytes and adipocytes. Horm Metab Res 37:602–609 [DOI] [PubMed] [Google Scholar]
- Dowell P, Flexner C, Kwiterovich PO, Lane MD 2000 Suppression of preadipocyte differentiation and promotion of adipocyte death by HIV protease inhibitors. J Biol Chem 275:41325–41332 [DOI] [PubMed] [Google Scholar]
- Jones SP, Janneh O, Back DJ, Pirmohamed M 2005 Altered adipokine response in murine 3T3–F442A adipocytes treated with protease inhibitors and nucleoside reverse transcriptase inhibitors. Antivir Ther 10:207–213 [PubMed] [Google Scholar]
- Lenhard JM, Furfine ES, Jain RG, Ittoop O, Orband-Miller LA, Blanchard SG, Paulik MA, Weiel JE 2000 HIV protease inhibitors block adipogenesis and increase lipolysis in vitro. Antiviral Res 47:121–129 [DOI] [PubMed] [Google Scholar]
- Bastard JP, Caron M, Vidal H, Jan V, Auclair M, Vigouroux C, Luboinski J, Laville M, Maachi M, Girard PM, Rozenbaum W, Levan P, Capeau J 2002 Association between altered expression of adipogenic factor SREBP1 in lipoatrophic adipose tissue from HIV-1-infected patients and abnormal adipocyte differentiation and insulin resistance. Lancet 359:1026–1031 [DOI] [PubMed] [Google Scholar]
- Shuman WP, Morris LL, Leonetti DL, Wahl PW, Moceri VM, Moss AA, Fujimoto WY 1986 Abnormal body fat distribution detected by computed tomography in diabetic men. Invest Radiol 21:483–487 [DOI] [PubMed] [Google Scholar]
- Beard JC, Bergman RN, Ward WK, Porte Jr D 1986 The insulin sensitivity index in nondiabetic man. Correlation between clamp-derived and IVGTT-derived values. Diabetes 35:362–369 [DOI] [PubMed] [Google Scholar]
- Bergman RN, Ider YZ, Bowden CR, Cobelli C 1979 Quantitative estimation of insulin sensitivity. Am J Physiol 236:E667–E677 [DOI] [PubMed] [Google Scholar]
- Bustin SA 2002 Quantification of mRNA using real-time reverse transcription PCR (RT-PCR): trends and problems. J Mol Endocrinol 29:23–39 [DOI] [PubMed] [Google Scholar]
- Sutinen J, Hakkinen AM, Westerbacka J, Seppala-Lindroos A, Vehkavaara S, Halavaara J, Jarvinen A, Ristola M, Yki-Jarvinen H 2002 Increased fat accumulation in the liver in HIV-infected patients with antiretroviral therapy-associated lipodystrophy. AIDS 16:2183–2193 [DOI] [PubMed] [Google Scholar]
- Gan SK, Samaras K, Thompson CH, Kraegen EW, Carr A, Cooper DA, Chisholm DJ 2002 Altered myocellular and abdominal fat partitioning predict disturbance in insulin action in HIV protease inhibitor-related lipodystrophy. Diabetes 51:3163–3169 [DOI] [PubMed] [Google Scholar]
- Addy CL, Gavrila A, Tsiodras S, Brodovicz K, Karchmer AW, Mantzoros CS 2003 Hypoadiponectinemia is associated with insulin resistance, hypertriglyceridemia, and fat redistribution in human immunodeficiency virus-infected patients treated with highly active antiretroviral therapy. J Clin Endocrinol Metab 88:627–636 [DOI] [PubMed] [Google Scholar]
- Nagy GS, Tsiodras S, Martin LD, Avihingsanon A, Gavrila A, Hsu WC, Karchmer AW, Mantzoros CS 2003 Human immunodeficiency virus type 1-related lipoatrophy and lipohypertrophy are associated with serum concentrations of leptin. Clin Infect Dis 36:795–802 [DOI] [PubMed] [Google Scholar]
- Lee JH, Chan JL, Sourlas E, Raptopoulos V, Mantzoros CS 2006 Recombinant methionyl human leptin therapy in replacement doses improves insulin resistance and metabolic profile in patients with lipoatrophy and metabolic syndrome induced by the highly active antiretroviral therapy. J Clin Endocrinol Metab 91:2605–2611 [DOI] [PubMed] [Google Scholar]
- Murata H, Hruz PW, Mueckler M 2000 The mechanism of insulin resistance caused by HIV protease inhibitor therapy. J Biol Chem 275:20251–20254 [DOI] [PubMed] [Google Scholar]
- Hruz PW, Murata H, Qiu H, Mueckler M 2002 Indinavir induces acute and reversible peripheral insulin resistance in rats. Diabetes 51:937–942 [DOI] [PubMed] [Google Scholar]
- Periard D, Telenti A, Sudre P, Cheseaux JJ, Halfon P, Reymond MJ, Marcovina SM, Glauser MP, Nicod P, Darioli R, Mooser V 1999 Atherogenic dyslipidemia in HIV-infected individuals treated with protease inhibitors. The Swiss HIV Cohort Study. Circulation 100:700–705 [DOI] [PubMed] [Google Scholar]
- Purnell JQ, Zambon A, Knopp RH, Pizzuti DJ, Achari R, Leonard JM, Locke C, Brunzell JD 2000 Effect of ritonavir on lipids and post-heparin lipase activities in normal subjects. AIDS 14:51–57 [DOI] [PubMed] [Google Scholar]
- Sutinen J, Kannisto K, Korsheninnikova E, Nyman T, Ehrenborg E, Andrew R, Wake DJ, Hamsten A, Walker BR, Yki-Jarvinen H 2004 In the lipodystrophy associated with highly active antiretroviral therapy, pseudo-Cushing’s syndrome is associated with increased regeneration of cortisol by 11β-hydroxysteroid dehydrogenase type 1 in adipose tissue. Diabetologia 47:1668–1671 [DOI] [PubMed] [Google Scholar]
- Rolland V, Le L, X, Houbiguian ML, Lavau M, Dugail I 1995 C/EBPα expression in adipose tissue of genetically obese Zucker rats. Biochem Biophys Res Commun 207:761–767 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.