| |
Changes in Body Composition and Mitochondrial Nucleic Acid Content in Patients Switched from Failed Nucleoside Analogue Therapy to Ritonavir-Boosted Indinavir and Efavirenz
|
| |
| |
The Journal of Infectious Diseases 2006;194:642-650
Mark A. Boyd,1,3,5 Andrew Carr,4 Kiat Ruxrungtham,1,2 Preeyaporn Srasuebkul,1,3 Darl Bien,6 Matthew Law,3 Somjai Wangsuphachart,2 Anchali Krisanachinda,2 Sakalaya Lerdlum,2 Joep M. A. Lange,1,7 Praphan Phanuphak,1,2 David A. Cooper,1,3 and Peter Reiss1,7
1HIV Netherlands Australia Thailand Research Collaboration, Thai Red Cross AIDS Research Centre, and 2Faculty of Medicine, Chulalongkorn University, Bangkok, Thailand; 3National Centre in HIV Epidemiology and Clinical Research, University of New South Wales, and 4St. Vincent's Hospital, Sydney, and 5Department of Microbiology and Infectious Diseases, Flinders Medical Centre and Flinders University, Bedford Park, Australia; 6University of Denver, Denver, Colorado; 7Department of Infectious Diseases, Tropical Medicine and AIDS, Academic Medical Centre and International Antiviral Therapy Evaluation Centre, Amsterdam, The Netherlands
ABSTRACT
Background. Body composition changes complicate antiretroviral therapy. Improvements in lipoatrophy after a switch in nucleoside reverse-transcriptase inhibitors (NRTIs) have been demonstrated. We investigated 60 patients switching from failed NRTIs to ritonavir-boosted indinavir and efavirenz.
Methods. Body composition (assessed by dual-energy x-ray absorptiometry scan and by single-slice computed tomography of the abdomen through the level of the fourth lumbar vertebra [L4] and the mid-right thigh) and fasted metabolics were measured at the baseline time-point at switch and at weeks 48 and 96 thereafter. Mitochondrial DNA and RNA were extracted from right-thigh subcutaneous fat and peripheral-blood mononuclear cells (PBMCs) at weeks 0 and 48. The primary end point was the change in mean limb fat over 48 weeks.
Results.
At week 96, we observed increases in mean (standard deviation [SD]) limb fat (+620 [974] g; P = .003), L4 subcutaneous adipose tissue (+20 [35] cm2; P < .001), mid-thigh subcutaneous adipose tissue (+5 [10] cm2; P < .001), and L4 visceral adipose tissue (+11 [34] cm2; P = .01), but we also observed reduced lean limb mass (-831 [1100] g; P = .3).
Mean (SD) mtDNA content in subcutaneous fat and in PBMCs increased (+109 [274] and +45 [100] copies/cell, respectively). Improved virological control or immune recovery did not explain the results.
Triglyceride, total cholesterol, estimated low-density lipoprotein cholesterol, ratio of total cholesterol to high-density lipoprotein cholesterol, and blood glucose levels deteriorated (i.e., had increased by 206%, 67%, 58%, 19%, and 6%, respectively, at week 96).
Conclusions. This regimen was associated with statistically significant but clinically modest increases in peripheral fat, visceral fat, and mitochondrial nucleic acid content. A predominantly adverse metabolic profile developed.
Background
Although HIV-related morbidity and mortality have dramatically declined since the advent of potent combination antiretroviral therapy, there has been an increased recognition of its toxicities. Prominent among these are peripheral lipoatrophy, central visceral fat accumulation, and metabolic disturbances [1, 2]. Mitochondrial toxicity as an explanation for the pathogenesis of these changes in body fat was first proposed in 1999 [3]. Attempts to demonstrate that patients who are considered to display lipoatrophy have depletion of mtDNA have generally had confirmatory results [4-7], although this has not been consistently the case [8]. Several clinical studies have demonstrated that peripheral lipoatrophy is at least partially reversible after a switch in nucleoside reverse-transcriptase inhibitor (NRTI)-based therapies [9-15]. These studies have demonstrated improvements in subcutaneous fat observed on dual-energy x-ray absorptiometry (DEXA) scan imaging after a switch from stavudine (d4T) or zidovudine (AZT) to abacavir or tenofovir.
The HIV-NAT 009 trial was designed for patients who had experienced failure of therapy consisting of either dual or triple NRTI combinations and who were switched to an NRTI-sparing combination of ritonavir-boosted indinavir (800 mg of ritonavir and 100 mg of indinavir twice per day) and efavirenz (600 mg once daily) [16]. The trial provided an opportunity to examine the effect of this switch on body fat distribution (assessed by DEXA and computed tomography [CT]), as well as on mtDNA and mitochondrial RNA (mtRNA) content in subcutaneous fat and peripheral-blood mononuclear cells (PBMCs).
RESULTS
Main study. Sixty-one patients were enrolled in the main study, of whom 60 underwent baseline assessment for lipodystrophy. At baseline, the median (interquartile range) NRTI exposure was 4.4 (3.9-4.7) years, HIV RNA level was 4.09 (3.75-4.61) log10 copies/mL, and CD4 cell count was 169 (60-277) cells/L. The mean (SD) duration of virological failure (defined as the time from a confirmed HIV RNA level of >1000 copies/mL during combination NRTI therapy to enrollment in the HIV-NAT 009 study) in those for whom virological data were available (80% [48/60] of the cohort) was 2.72 (0.76) years. The mean (SD) change in the time-weighted average HIV RNA level from baseline to weeks 48 and 96 was -2.1 (0.7) and -2.1 (0.8) log10 copies/mL, respectively, which resulted in 87% and 69% of patients having an HIV RNA level <50 copies/mL at weeks 48 and 96, respectively. The time-weighted average change in CD4 cell count over 96 weeks was 103 cells/L [16].
Lipodystrophy substudy. Forty-four (73%) of the 60 patients gave separate consent to participate in the MITOX substudy over the first 48 weeks. Of these 44 patients, 35 contributed paired mtDNA samples successfully extracted from fat biopsy samples at weeks 0 and 48, and 36 contributed mtDNA samples extracted from PBMCs at weeks 0 and 48. For mtRNA analysis, 32 and 33 paired samples were successfully extracted from fat and PBMCs, respectively.
Baseline characteristics of the 60 patients who had lipodystrophy assessments are summarized in table 1. The disposition of patients over the course of the study and the reasons for permanent study withdrawal are shown in figure 1. All patients continued follow-up in accordance with the trial protocol, whether they had withdrawn from the study or not.
Body composition assessment: DEXA scan. We observed a significant increase over 48 weeks in the mean (SD) total limb fat (+380 [820] g; P < .001). Despite this, the mean (SD) body weight of the cohort failed to increase (at baseline, 56 [9.7] kg; at week 48, 56 [10.5] kg; P = .2). A post hoc analysis of lean tissue changes was undertaken to determine whether lean tissue loss might account for this discrepancy. A significant loss was observed in lean limb mass over the 48-week period (mean [SD] change, -955 [1120] g; P < .001). In the second 48-week period of observation, there was a continued increase in total limb fat (+240 [610] g; P = .003), an increase in total lean limb mass (+120 [890] g; P = .3), and, therefore, an overall net increase in total limb mass (+360 [1260] g; P = .03). This was reflected in an increase in body weight (+1.31 [0.34] kg). An additional post hoc analysis of the database was undertaken to determine whether GI adverse events experienced during the first 48 weeks of the study might account for the lean tissue loss. We found no association between changes in fat or lean compartments observed on DEXA scans and GI toxicity associated with the antiretroviral regimen (all P values for association were >.25; data not shown).
Body composition assessment: CT scan. At week 48, we observed increases in mean (SD) mid-thigh SAT (+4 [8] cm2; P < .001) and L4 SAT (+9 [32] cm2; P = .04) as well as L4 VAT (+8 [30] cm2; P = .05). The similar increases in SAT and VAT resulted in a stable VAT : SAT ratio (mean [SD] change, -0.2 [1.4]; P = .2). These increases at week 48 continued to week 96 for mid-thigh SAT (+2 [6] cm2; P = .02) and L4 SAT (+12 [23] cm2; P = .002), but L4 VAT remained stable (+3 [22] cm2; P = .2). As a result, the VAT : SAT ratio decreased over the second 48-week period (-0.1 [0.2]; P = .002). The absolute changes in DEXA and CT assessments between baseline and weeks 48 and 96 are expressed in table 2.
Mitochondrial assessments. mtDNA and mtRNA content increased in both mid-thigh fat and PBMCs (table 3). For PBMCs, the only statistically significant changes in mtDNA and mtRNA content occurred between week 0 and week 12 (P = .005 and P = .02, for mtDNA and mtRNA changes, respectively; data not shown), with a leveling off thereafter. When the data were stratified according to whether the patient had ceased d4T or AZT therapy at baseline, we observed a greater rise from baseline to week 48 in mtDNA content in fat for those ceasing d4T therapy (mean [SD] change, +239 [194] mtDNA copies/cell vs. +56 [291] mtDNA copies/cell for AZT therapy; P = .06). We also observed an increase from baseline to week 48 in mtDNA in PBMCs for those ceasing AZT therapy (+55 [550] mtDNA copies/cell), whereas we found a decrease for those ceasing d4T therapy (-17 [129] mtDNA copies/cell; P = .2) (table 3). These patterns of recovery were not reflected in changes in mtRNA content. Results for mtDNA and mtRNA content in fat and PBMCs, including stratification for the last thymidine-analogue NRTI received, are expressed in table 3.
Correlations between body composition and mitochondrial nucleic acid content. We found no correlation between changes in limb fat observed on DEXA and changes in mtDNA content in fat (r = -0.11; P = .5). We found no correlation between changes in mtDNA content in fat and changes in SAT or VAT observed on CT. We observed a significant association between changes in mtDNA and mtRNA content in fat (r = 0.45; P = .01) but saw no association between changes in mtDNA and mtRNA content in PBMCs (r = 0.13; P = .5).
Fasted metabolic parameters. Substantial, deleterious changes occurred in triglyceride, TC, estimated LDL cholesterol, and blood glucose levels. The HDL cholesterol fraction demonstrated an improvement, but not to an extent that compensated for the increase in TC level; therefore, the TC : HDL cholesterol ratio increased. Table 4 depicts these metabolic parameters over 48 and 96 weeks of the study.
DISCUSSION
In patients who experienced failure of combination NRTI therapy and who switched to an NRTI-sparing regimen of ritonavir-boosted indinavir and efavirenz, limb fat mass showed a clinically moderate-albeit statistically significant-improvement over 96 weeks, a finding consistent with the results of previous NRTI-switch studies [9-15]. These changes occurred in the absence of changes in diet or exercise and were not related to either virological suppression or immune recovery. In addition, we observed a modest increase in visceral abdominal tissue. Despite these clinically relatively trivial changes, we did observe a substantial deterioration in the metabolic profile of the cohort, with significant increases in triglyceride, TC, and estimated LDL cholesterol levels. Although there was a concomitant increase in HDL cholesterol level, it was not of the same magnitude as that seen in TC level; therefore, the overall TC : HDL cholesterol ratio increased. This ratio is considered to be a powerful predictor of future vascular events [22]. The increase in HDL cholesterol level may be related to an effect of efavirenz, although part of this increase might also be accounted for by improved virological control and, therefore, the reversal of the low level of HDL cholesterol associated with uncontrolled HIV infection [23, 24]. We found that 45 (74%) of the 61 originally enrolled patients would qualify for lipid-lowering therapy on the basis of dyslipidemia alone [25]. Clearly, this adverse effect on metabolic parameters should be carefully weighed in any consideration of using this particular NRTI-sparing regimen for primary or secondary prevention of changes in body composition.
A separate analysis of lean-tissue changes was not a component of the original plan for analysis but was done after the discovery that, despite the significant increases in subcutaneous and visceral fat mass over the first 48 weeks of the study, body weight appeared not to have increased from baseline. We found that the improvements in limb fat were accompanied by a decrease in lean limb tissue mass, accounting for the failure to gain weight. At least 2 other studies have demonstrated a similar finding in regard to loss of lean tissue following changes to the NRTI aspect of the regimen [26, 27]. We had initially hypothesized that the explanation for this phenomenon might lie in the GI tolerability of the regimen, but a correlation analysis based on a score for GI toxicity observed in the cohort over the first 48 weeks of the study did not support such an association. We are therefore unable to explain this phenomenon. An important caveat in the interpretation of the findings related to changes in body composition in this study is that all patients were experiencing failure of combination nucleoside therapy at the time of the switch (minimum HIV RNA level of 1000 copies/mL); therefore, the improvements in fat mass might simply reflect improved health status in this cohort, who generally experienced an excellent response to therapy (87% and 69% of the intention-to-treat cohort had HIV RNA levels <50 copies/mL at weeks 48 and 96, respectively). However, if this were the case, we should also have observed concomitant improvements in lean tissue, which we did not.
Over the first 48 weeks of the study, we found that mtDNA and mtRNA content in both subcutaneous fat and PBMCs improved significantly after NRTI withdrawal, consistent with the hypothesis that NRTIs may cause mitochondrial toxicity associated with mtDNA depletion and impairment in DNA transcription and that a recovery of mtDNA and mtRNA content occurs after withdrawal of NRTIs. It is important to remember, however, that the observed improvements might be explained by restored health from newly established virological control. Nonetheless, we were unable to demonstrate any correlation between HIV suppression or immunological improvement (which may act as a surrogate of restored health) and changes in mitochondrial nucleic acid content. The improvements in mtDNA and mtRNA content in PBMCs were most marked immediately after the switch of therapy, suggesting that recovery, at least in this compartment, occurs relatively soon after withdrawal of the NRTIs. This brisk recovery is inconsistent with the recent findings of Mussini et al., who observed improvements in mtDNA content in PBMCs only after 6 months had elapsed since interruption of all antiretroviral therapy [28]. The likely explanation for this discordance may be found in the very different strategies employed in the 2 studies, in that our study examined the effect of the introduction of a new and potent NRTI-sparing therapy in patients experiencing failure of combination NRTIs, whereas the Mussini et al. study employed the opposite strategy (i.e., their patients interrupted successful antiretroviral therapy, with subsequent recrudescence of HIV viremia). A relationship between uncontrolled HIV replication and decreased mtDNA content in PBMCs has been described elsewhere [6, 29].
When we stratified the analysis according to whether patients had ceased taking d4T or AZT at baseline, we found a lower mtDNA copy number in fat at baseline and a greater increase in mtDNA content in fat after 48 weeks in patients who had switched from d4T than in those who had switched from AZT. Conversely, we found a lower baseline mtDNA copy number in PBMCs and a greater rise in mtDNA content in PBMCs in the patients who had switched from AZT. These observations need to be treated with caution in this unrandomized study but are nevertheless consistent with clinical and molecular observations that different NRTIs exhibit different toxicities in different cell lines [30]. The results suggest that increases in mtDNA content in patients may be more profound in one tissue compartment than in another, depending on the last thymidine-analogue NRTI used. A finding from another study, in which demonstrable improvements in objective measurements of subcutaneous fat mass after a switch from either AZT or d4T to abacavir were not reflected by changes in mtDNA content in PBMCs over the same study period, supports this hypothesis [31]. An alternative explanation might be that patients switching from a d4T-containing regimen had a relatively greater reduction of mtDNA from baseline than did those switching from an AZT-containing regimen and, therefore, had a greater capacity for gain after withdrawal of the NRTI [32].
We found that increases in mtDNA content were associated with increases in mtRNA content in both the fat and the PBMC compartments. However, although we found a correlation between increases in mtDNA and mtRNA content in fat, we did not find the same clear correlation between mtDNA and mtRNA content in PBMCs. Whether these discordant results reflect a lack of statistical power, differential toxicities of NRTIs in different compartments, discord between mtDNA recovery and mtDNA transcription, or another explanation is unknown [33]. We were also unable to demonstrate any association between the changes in mtDNA and mtRNA content in fat and PBMCs over the first 48 weeks of the study and the changes seen in measurements of peripheral lipoatrophy in the same time period. This finding might demonstrate either that there is no direct association between improvements in mitochondrial toxicity, as measured by mtDNA copy number, and improvements in peripheral fat or that the improvements in fat are more closely associated with other aspects of mitochondrial function that are not well reflected in measurement of mtDNA copy number per se [34]. It must also be acknowledged that these correlation analyses could be performed only on the subset of the cohort for whom paired mtDNA and mtRNA samples were available. Therefore, the analysis may simply lack the power to detect associations between these markers.
There are a number of limitations to our study. First, as already acknowledged, it differs from studies of the effects of an antiretroviral regimen switch in which patients were experiencing stable virological control at the time of switch and, in general, continued to experience good virological control after the switch. Therefore, restoration of health associated with response to the new antiretroviral regimen may be a plausible explanation for some of our findings. Second, our study was not randomized, and we are thus unable to judge the extent to which the described changes differ from those that might have occurred if patients had remained on the same failing regimen or had switched to a new NRTI-containing or a different NRTI-sparing regimen. Finally, we were unable to generate paired samples of mtDNA and mtRNA for analysis in a proportion of the cohort. However, a comparative analysis of those who had samples of mtDNA and mtRNA successfully amplified for analysis and those who did not failed to demonstrate any meaningful differences between the populations in terms of baseline characteristics or assessments of body composition (data not shown).
In conclusion, in this study, we switched a cohort of patients experiencing failure of various combinations of NRTI analogues to a nucleoside-sparing regimen of ritonavir-boosted indinavir (800 mg of ritonavir and 100 mg of indinavir twice per day) and efavirenz (600 mg every day). After the switch, we observed recovery of virological control, which, in the majority, was associated with clinically modest but statistically significant increases in peripheral and visceral fat as well as in mitochondrial nucleic acid content. However, we also noted a substantial deterioration in overall metabolic profile, which might dispose patients to accelerated cardiovascular disease. These observations suggest that the various aspects of toxicity associated with combination antiretroviral therapy may be relatively independent phenomena that have explanatory mechanisms dependent on either the different classes of antiretrovirals or, as now seems more likely, specific agents within these classes. Antiretroviral combinations employing protease inhibitors that display minimal lipid or glycemic effects (particularly atazanavir or saquinavir) or NRTIs that demonstrate less intrinsic toxicity to mitochondria are attractive for the design of future studies that may avoid or ameliorate the complication of lipoatrophy and adverse metabolic changes.
METHODS
Study population. Patients enrolled in the main study were asked to give separate informed consent to participate in the substudies of changes in body fat and mitochondrial nucleic acid. All studies were approved by the Ethics Committee of the Faculty of Medicine, The King Chulalongkorn Memorial Hospital, Bangkok, Thailand.
Assessment of body composition. Consenting patients underwent a single-slice abdominal CT scan through the level of the fourth lumbar vertebra (L4), from which the areas of visceral adipose tissue (VAT) and subcutaneous adipose tissue (SAT) were calculated. Patients also underwent a single-slice CT scan through the mid-right thigh, from which SAT area was calculated. DEXA was performed to measure total and regional body fat and lean tissue. All imaging studies were performed at baseline and at weeks 48 and 96 after the switch. We asked patients on a 6-monthly basis for a self-assessment of changes in exercise level (sedentary or low, moderate, or high activity level) or in the purpose of their diet (to maintain weight, lose weight, or increase weight) that might affect changes in body composition, by means of a standardized questionnaire. Patients were weighed at each visit with the same scales, which were calibrated at the beginning of each visit day. Patients were weighed while clothed but with shoes removed.
Blood collection. Blood was drawn from patients in a fasted state (minimum fasting time, 8 h) at all time points, and measurements included total cholesterol (TC), high-density lipoprotein (HDL) cholesterol, triglyceride, and glucose levels. Low-density lipoprotein (LDL) cholesterol values were estimated using the Friedewald equation (except when triglyceride values were >4.52 mmol/L). Whole blood for PBMC collection was drawn, at 12-week intervals, into Vacutainer cell-preparation tubes containing sodium heparin (Becton Dickinson). PBMCs were separated on lymphocyte separation medium (Isoprep) and were washed twice in PBS. PBMCs were then counted, were viably cryopreserved in 2-mL cryovials, were transferred into liquid nitrogen for storage, and were later transferred to Primagen in The Netherlands for analysis.
Fat biopsy methods. Patients underwent 4-mm subcutaneous fat punch biopsies from the mid-medial right thigh. Specimens were immediately snap-frozen in liquid nitrogen, were stored on site, and were later transferred to Primagen, where they underwent preparation and analysis.
mtDNA and mtRNA quantification. PBMC samples were thawed by immersion of the cryovials in a 37 C water bath and were microscopically checked for platelet contamination. If the ratio of platelets to PBMCs exceeded a factor of 5, the cells were washed once or twice more with PBS plus 2% fetal calf serum at a speed of 100 relative centrifugal force, to remove excess platelets. The platelet : PBMC ratio was maintained at a factor <5, a level of contamination that has been shown not to alter the results of mtDNA quantification in PBMCs [17, 18]. Subcutaneous fat biopsy samples were treated with collagenase, to separate adipose cells from other cells [19]. The cells were spun for 5 min at 350 g, which resulted in the adipose cells floating in aqueous solution and all other cells being deposited as a pellet in the base of the tube. Adipose cells were collected, and nucleic acid was isolated from them in accordance with the method of Boom et al. [20]. The amplification of mtDNA, mtRNA, and nuclear DNA from both PBMCs and adipose cells was performed using a real-time, duplex, nucleic acid sequence-based amplification in a single tube. Detection of the amplification products occurred in real time by mitochondrial- and nuclear-specific molecular beacons in a thermostat-regulated fluorimeter at 41 C. The assay determined mtDNA and mtRNA copies relative to the nuclear DNA, and the results were expressed as mtDNA or mtRNA copies per cell [21].
Statistical analysis. The primary end point of the study was the change in mean limb fat over 48 weeks, by use of intention-to-treat (ITT) analysis. This ITT population was defined as those patients who commenced therapy and had at least a baseline assessment of lipodystrophy. In the case of DEXA and CT imaging, the ITT population and the population receiving treatment were equivalent (60 patients) for the period between baseline and week 48. At week 96, fifty-eight patients underwent CT and DEXA. Missing scan data at week 96 were managed with a last-observation-carried-forward imputation. Changes in DEXA, CT, body weight, mtDNA and mtRNA in fat biopsy samples, and fasting lipid values from baseline to weeks 48 and 96 were analyzed using paired t tests. The sample t tests were used to compare differences in CD4 cell count, fasting lipid results, CT results, and DEXA results according to the last thymidine-analogue NRTI received. The gastrointestinal (GI) tolerability of the regimen was assessed by calculating a GI toxicity score, which was defined as the sum of the number of GI adverse events multiplied by the severity of each event (severity scores: 1, mild; 2, moderate; and 3, severe). Associations between variables were investigated using Spearman's rank correlation coefficient. All data were analyzed using Stata (version 8.0; StataCorp). A 2-sided statistical significance level of = 0.05 was used throughout.
|
|
| |
| |
|
|
|
