Randomized, Controlled, 48-Week Study of Switching Stavudine and/or Protease Inhibitors to Combivir/Abacavir to Prevent or Reverse Lipoatrophy in HIV-Infected Patients

Objective: HIV-1 protease inhibitors (versus no protease inhibitors) and stavudine (versus zidovudine) are independently associated with a higher risk of lipoatrophy in HIV-infected patients. We sought to determine whether the revision of stavudine and/or protease inhibitor-containing regimens to combivir/abacavir would result in prevention and/or reversibility of lipoatrophy in HIV-1-infected patients.

Design: The investigation was a prospective, randomized, controlled, open-label study.

Subjects: The subjects included 37 HIV-1-infected individuals with stable undetectable HIV-1 loads who were taking a regimen containing either stavudine or zidovudine with lamivudine and a protease inhibitor.

Intervention: Subjects were randomized to continue therapy or switch stavudine to zidovudine and protease inhibitor to abacavir, such that the universal switch regimen was combivir (zidovudine/lamivudine) and abacavir.

Main Outcome Measures: Total body, leg, and arm fat mass was measured at baseline, 24 weeks, and 48 weeks using whole-body dual-energy x-ray absorptiometry. Single-cut L4 computed tomography and assays of multiple metabolic parameters were also performed.

Results: There was an average gain in fat mass of 0.009 kg/(leg·mo) in switch patients versus a loss of 0.010 kg/(leg·mo) in controls (p = .04, on-treatment analysis) over 48 weeks. Significant arm fat restoration was observed in patients who switched regimens, with an average gain of 0.014 kg/(arm·mo) (p = .004), whereas controls did not have a significant change from baseline. Analyses of percentage changes in arm and leg fat masses showed similar findings. No significant effects on intraabdominal fat, blood lipid levels, glycemic indices, and lactate levels were detected, although most baseline mean values were normal in study subjects. Combivir/abacavir maintained virological control in all but one case, and three (13.6%) of 22 individuals had adverse reactions to abacavir therapy.

Conclusions: A switch to combivir/abacavir therapy was associated with objective evidence of limb fat-sparing and fat restoration compared with continued treatment with stavudine and/or protease inhibitor.

In the selection of an optimal antiretroviral regimen for treatment-naive and treatment-experienced HIV-infected patients, clinicians are increasingly taking account of the potential for particular antiretroviral drugs to induce, worsen, prevent, or improve lipoatrophy and other metabolic abnormalities. It is not yet known whether there are inherent biologic barriers to timely reversal of lipoatrophy even if the risk factors are removed. There is therefore a need to determine which combination(s) of currently available antiretroviral drugs will most likely slow progression of lipoatrophy (fat-sparing) and allow reversal of established lipoatrophy (fat restoration) from any point along the spectrum of baseline fat mass. The risk for lipoatrophy is increased in HIV-infected patients treated with stavudine compared with zidovudine and in those treated with protease inhibitors (PIs) compared with no PIs (^1,2). This study was therefore designed to examine the effect(s) of substitutions of both stavudine and PI on fat-sparing and fat restoration from any level of baseline body fat.

MATERIALS AND METHODS

This randomized, controlled study had objective endpoint measurements and a 48-week follow-up. The patients, enrolled between March 2000 and July 2001, were HIV-1-infected individuals who were taking a triple combination antiretroviral regimen containing either stavudine and a PI or zidovudine and a PI, had plasma HIV RNA levels of less than 400 copies/mL for at least the preceding 8 weeks, and had no prior treatment with abacavir. Clinically apparent changes associated with lipoatrophy were not required for enrollment. Subjects were randomized 1:1 to continue their therapy (control arm) or switch to open-label combivir (zidovudine [300 mg]/lamivudine [150 mg] twice daily) and abacavir (300 mg twice daily) (switch arm). Thus, patients taking stavudine, lamivudine, and PI at baseline would switch both stavudine (to zidovudine) and PI (to abacavir), whereas those taking zidovudine, lamivudine, and PI would only switch PI to abacavir. There were no patients taking didanosine, zalcitabine, or non-nucleoside reverse transcriptase inhibitors.

Total body, arm, and leg fat mass was measured by whole-body dual-energy x-ray absorptiometry at baseline, 24 weeks, and 48 weeks. The primary outcome measures of the study were changes from baseline in arm and leg fat masses per month expressed both as kilograms and as percentage of total arm and leg mass. Secondary end points included the change to visceral adipose tissue demonstrated by single-cut L4 computed tomography; fasting triglyceride, total cholesterol, and LDL and HDL cholesterol levels; HIV RNA concentration (Roche Amplicor Monitor assay version 1.0, Roche Diagnostics, Branchburg, NJ; lower limit of detection 400, copies/mL); and adverse events to week 48. Clinical assessments were conducted at weeks 2, 4, 8, 12, 24, 36, and 48 after enrollment.

Estimation and analysis of average trends over the 48 weeks subsequent to randomization were undertaken by the fitting of linear mixed effects models. Significance of the differences in these trends from zero and differences between switch and control groups were assessed by the application of Wald tests. Standard t tests were used for comparisons of mean baseline values. The analyses were carried out using the S-PLUS 2000 statistical package (MathSoft, Seattle, WA).

RESULTS

Of the 39 individuals initially enrolled in the study, two had less than 6 months of follow-up and were omitted from all analyses. Intention-to-treat analysis included data for those individuals with more than 6 months of follow-up (n = 37). There were 15 subjects randomized to the control arm and 22 to the switch arm. All 15 controls and 19 switch patients were male. Baseline regimens included zidovudine/lamivudine/indinavir, stavudine/lamivudine/nelfinavir, and stavudine/lamivudine/indinavir. There were no statistically significant differences in the following baseline mean values: percentage or kilograms of arm or leg fat mass; visceral adipose tissue area; fasting triglyceride, total cholesterol, LDL cholesterol, and venous lactate levels; and CD4 T cell count. However, the mean HDL cholesterol level ± standard deviation (SD) was higher in the switch arm than in the control arm at baseline (1.06 ± 0.22 vs. 0.91 ± 0.23 mmol/L, respectively;p = .05) (Table 1). With the exception of mean baseline triglyceride levels, which were elevated in both switch and control groups, mean levels of metabolic parameters were within the normal reference ranges at baseline.

Table 1

Over 48 weeks after randomization, the average leg fat mass in the control group decreased from baseline (loss of 0.010 kg/[leg·mo]), whereas the average leg fat mass increased in the switch group (gain of 0.007 kg/[leg·mo]). The mean difference between switch and control subjects was statistically significant (p = .05) (Figure 1). Similar trends were observed in percentage leg fat (increase of 0.014%/[leg·mo] in the switch group versus decrease of 0.046%/[leg·mo] in the control group, respectively;p = .06). The mean arm fat mass increased significantly from baseline in the switch group (0.012 kg/[arm·mo], p = .004; or 0.10%/[arm·mo], p = .007) but remained approximately constant in the control group (0.006 kg/[arm·mo], p = .25; or 0.002%/[arm·mo], p = .96).

Fig. 1

On-treatment analysis was limited to subjects who had taken at least 1 month of randomized therapy (n = 33). Results showed trends in leg and arm fat masses similar to those observed in the intention-to-treat analyses. Over 48 weeks, there was an average gain in leg fat mass of 0.009 kg/(leg·mo) in switch patients as opposed to a loss of 0.01 kg/(leg·mo) in controls (p = .04). Arm fat mass increased from baseline at a rate of 0.014 kg/(arm·mo) (p = .004) in those patients who switched to combivir/abacavir. As in the intention-to-treat analysis, there was no significant gain in arm fat mass in those patients who continued their baseline regimen (0.006 kg/[arm·mo], p > 0.2). Analyses of percentage rates of change of arm and leg fat masses were consistent with these results (data not shown). The average absolute gain in fat mass was 0.095 kg (0.23%) per leg and 0.154 kg (1.32%) per arm in patients switched to combivir/abacavir. Controls had an average fat loss of 0.110 kg (0.51%) per leg and an average fat gain of 0.062 kg (0.03%) per arm.

A statistically significant difference in the effects associated with switching from baseline zidovudine versus baseline stavudine was not detected. However, an average leg fat gain of 0.014 kg/(leg·mo) was observed in the eight patients who switched their PI alone (baseline zidovudine-based regimen), compared with a loss of 0.020 kg/(leg·mo) in the corresponding seven controls (p = .04). The gain in mean percentage arm fat was greater in the 10 individuals who switched both stavudine and PI than in the eight who continued their baseline therapy (0.192% versus 0.062%, respectively;p = .034). Notably, the switch and control groups were well matched for baseline fat mass values and duration of HAART prior to randomization, but subgroups taking zidovudine/lamivudine/PI versus stavudine/lamivudine/PI were not. Those subjects taking zidovudine at baseline had a significantly higher mean fat mass in the legs than did those taking stavudine (3.56 kg versus 2.51 kg, respectively;p = .04), although they had a longer duration of HAART (43.5 months vs. 36.2 months, respectively;p = .02).

Single-cut L4 computed tomography scans were obtained at baseline, 24 weeks, 48 weeks, and after randomization. The mean visceral adipose tissue values ± SD were 120.7 ± 75.4 cm² in the switch group and 137.7 ± 56.7 cm² in the control group at baseline, and subsequent changes were not significantly different between control and switch groups. Similarly, there were no statistically significant differences between control and switch groups in linear trends over the 48 weeks after randomization for mean blood triglyceride, total cholesterol, HDL and LDL cholesterol, or lactate levels between control and switch groups.

There was one patient previously treated with zidovudine, lamivudine, and indinavir who had an increase in viral load of more than 400 copies/mL at weeks 4 and 8 after switching to combivir/abacavir. There were three subjects in the switch group who reported adverse reactions to abacavir therapy. Two of these patients fulfilled diagnostic criteria for abacavir hypersensitivity, with development of generalized rash, fever, and gastrointestinal symptoms within 2 weeks of starting combivir/abacavir treatment, which resolved after revision of therapy. The third patient reported symptoms of lethargy, loss of appetite, and abdominal discomfort without rash or fever, which did not fulfill diagnostic criteria for abacavir hypersensitivity, but these symptoms resolved after abacavir treatment was ceased 2 weeks after its introduction. There was one subject who developed severe headaches and nausea 8 days after switching therapy from stavudine/lamivudine/nelfinavir to combivir/abacavir. These symptoms resolved after replacing zidovudine with stavudine therapy. There were no serious adverse events reported by controls.

DISCUSSION

In this study, revision of antiretroviral therapy to combivir/abacavir resulted in an average gain in leg fat compared with fat loss in controls who continued therapy containing stavudine and/or PI over 48 weeks. Significant arm fat restoration was also observed in patients who switched regimens, whereas controls did not have a significant change from baseline.

To date, the switching strategies tested in randomized, controlled settings have usually been either a PI class switch with continuation or intensification of the backbone nucleoside reverse transcriptase inhibitors or a nucleoside reverse transcriptase inhibitor switch with PI or nonnucleoside reverse transcriptase inhibitor continuation (^2-9). The former strategy has not resulted in objective improvements in lipoatrophy in up to 1 year of observation in moderate- to large-sized studies (^3,4), and indeed, intensification of non-PI therapies has been associated with worsening of lipoatrophy and poor short-term tolerability (⁵). In contrast, replacement of stavudine with abacavir (⁶) or (nonrandomized) with zidovudine (^7-9) while continuing PI or nonnucleoside reverse transcriptase inhibitor has led to small but significant increases in dual-energy x-ray absorptiometry-measured or computed tomography-measured limb fat mass in the 24-week or 48-week follow-up. Such data support the findings of clinical studies that identified stavudine as a strong risk factor for developing lipoatrophy but appear at odds with a putative additional role for PIs. A previous analysis of the Western Australian HIV Cohort Study, from which subjects of this study were recruited, estimated an excess risk of lipoatrophy associated with stavudine of 1.9 for the first 18 months of therapy. Use of PIs increased the baseline nucleoside reverse transcriptase inhibitor-associated risk, after adjustment for stavudine use (¹). Greater falls in limb fat on dual-energy x-ray absorptiometry scans have been associated with both stavudine/didanosine (versus zidovudine/lamivudine) and nelfinavir (versus efavirenz) in a more recent randomized study (¹⁰). It is therefore possible that ongoing nucleoside reverse transcriptase inhibitor exposure was the critical impediment to fat restoration in past studies of PI switching.

This study was designed to incorporate switching of stavudine to zidovudine as well as PI to abacavir on the basis that combivir/abacavir would optimally reduce total lipoatrophy risk. We found that patients who switched to combivir/abacavir did have a significant increase in their limb fat mass from baseline levels. Although this absolute increase 48 weeks after substitution was small, switch patients also avoided the further decline in (leg) fat mass associated with continued stavudine and/or PI use. Thus, the substitution was both fat restoring and fat sparing.

The average amount of fat restoration observed in this study did not exceed that found in studies of stavudine revision alone (^6-9). However, we did detect modest (leg) fat restoration in the subgroup who only switched PI in a zidovudine-containing regimen. There was insufficient statistical power in this study to compare the effects associated with switching from baseline zidovudine (PI switch alone) versus baseline stavudine (stavudine and PI switch). The magnitude of the independent or interactive effects of PIs on lipoatrophy has been difficult to quantify in nonrandomized clinical studies because of the potential confounding influence of concurrent stavudine in PI-based regimens (²). Larger trials with sufficient statistical power to compare single switches of stavudine or PI with switching both drugs are needed.

A critical issue in comparing the magnitude of reversibility across different switching strategies is the natural limits to increases or decreases in limb fat mass imposed by the level of fat at baseline. Individuals with milder lipoatrophy will need less fat restoration to reach normal and level off, as will naturally lean individuals. More substantial fat restoration may only be possible in individuals with at least moderate to severe lipoatrophy. Similarly, the amount of ongoing fat loss in controls, which is a measure of the fat sparing afforded by switching, would be very limited in those who have already lost most of their limb fat. Notably in this study, the fat sparing (magnitude of fat loss in controls) was most evident in those subjects taking baseline zidovudine/lamivudine/PI, whereas fat restoration was dominant in those taking baseline stavudine/lamivudine/PI. This finding may be related to the fact that subjects taking zidovudine at baseline had a significantly higher mean fat mass in the legs than did those taking stavudine, despite a longer duration of HAART. The distribution of baseline arm and leg fat levels in patients recruited to recent switching trials has varied, and this alone could account for variations observed between studies (^6,7). In addition, the rate of fat restoration may not be constant. There may be increased biologic barriers to fat restoration in those patients with very severe lipoatrophy and longer established pathology, particularly if adipocyte proliferative capacity is impaired. In this study, there was a larger incremental percentage leg fat gain between 24 and 48 weeks than between 0 and 24 weeks. If fat restoration follows an exponential or other nonlinear trajectory, longer-term follow-up may reveal still larger increases. It also appears that arm fat shows earlier restoration than leg fat, as demonstrated in this and other studies (^6,7). This could reflect differential regional rates of initial fat loss, the reason for which is unknown at this time, with consequent effects on the rate of fat restoration.

The results of this study are consistent with those of another study of similar design (¹¹) and add further support to the hypothesis that stavudine and PIs have an etiologic role in progressive lipoatrophy. In terms of treatment strategy, combivir/abacavir maintained virological control in all but one case, and three of 22 individuals had adverse effects due to abacavir. Our findings provide objective evidence that the benefits of using combivir/abacavir in virologically suppressed individuals include a greater level of prevention and reversal of lipoatrophy than regimens containing stavudine and/or PI. However, as fat restoration is modest in absolute terms compared with the amount of early fat loss, it is preferable to prevent the progression of lipoatrophy than to try to reverse it.

Acknowledgments:

The authors thank Alison Cain and the Royal Perth Hospital HIV Clinical Trials Group, the division of Imaging Services at Royal Perth Hospital, and all the study subjects. This study received funding from GlaxoSmithKline.