| |
Lipoatrophy/Metabolics & Genetics: European mitochondrial DNA haplogroups and metabolic changes during antiretroviral therapy in AIDS Clinical Trials Group Study A5142 - pdf attached
|
| |
| |
Download the PDF here
AIDS:
2 January 2011
Hulgan, Todda; Haubrich, Richardb; Riddler, Sharon Ac; Tebas, Pablod; Ritchie, Marylyn Da; McComsey, Grace Ae; Haas, David Wa; Canter, Jeffrey Aa
aVanderbilt University, Nashville, Tennessee, USA. bUniversity of California-San Diego, San Diego, California, USA. cUniversity of Pittsburgh, Pittsburgh, USA. dUniversity of Pennsylvania, Philadelphia, Pennsylvania, USA. eCase Western Reserve University, Cleveland, Ohio, USA.
"These results demonstrate preliminary associations between a European mtDNA haplogroup and metabolic parameters in an HIV-infected population, both prior to and during ART exposure as part of a clinical trial. Many genetic and nongenetic factors likely influence risk for metabolic changes in this population; mtDNA variation may be one of these. Continued study is needed to replicate these associations and determine mechanisms by which mtDNA variation may be associated with different metabolic effects. More extensive mtDNA genotyping in large, well characterized populations is needed to identify additional functional variants (i.e. SNPs that change amino acid sequence and/or protein function) that may underlie this and other reported associations, and to expand these analyses to other racial/ethnic groups with different mtDNA lineages.......
.......When compared across mtDNA haplogroups, persons belonging to haplogroup I were more likely to have met the definition of lipoatrophy than those belonging to non-I haplogroups (50 vs. 20%; P = 0.04; Fig. 2a). No other haplogroup differed significantly with respect to lipoatrophy. In a subgroup analysis limited to persons receiving the thymidine analog NRTIs zidovudine or stavudine [N = 91; 30 (33%) developed lipoatrophy], haplogroup I remained significantly over-represented compared with non-I haplogroups (71 vs. 30%; P = 0.04; Fig. 2b)......{see Results below for further discussion}
......Logistic regression was used to determine ORs of baseline demographic, immunologic, and ART-related factors for lipoatrophy, and to adjust for possible confounding (Table 4). In unadjusted models.......Of mtDNA haplogroups, only haplogroup I was significantly associated with lipoatrophy [3.9 (1.1-14.4); P = 0.04]. In a model adjusted for randomization to the NRTI-sparing regimen, baseline extremity fat mass, and mtDNA haplogroup I, NRTI-sparing treatment remained protective of lipoatrophy [0.3 (0.1-0.9); P = 0.03], and the effect size of haplogroup I decreased and the P value no longer demonstrated statistical significance [3.0 (0.8-11.4); P = 0.11]. In separate models adjusting for individual thymidine analog NRTI use, stavudine remained significantly associated with lipoatrophy [5.2 (2.1-12.9); P < 0.001], zidovudine was not [1.4 (0.7-3.0); P = 0.3], and although haplogroup I retained similar OR [3.7 (0.9-14.6); P = 0.06 and 3.5 (0.9-12.9); P = 0.07, respectively; Table 4], these associations were no longer statistically significant. Baseline extremity fat was not associated with lipoatrophy in unadjusted or adjusted analyses.....
.....persons belonging to haplogroup I had significantly less median percentage increases in total (+3 vs. +23%; P = 0.005) and LDL (-5 vs. +21%; P = 0.01) cholesterol, and triglycerides (-4 vs. +55%; P = 0.04) over 96 weeks than those belonging to other haplogroups."
Abstract
Background: Mitochondrial DNA (mtDNA) influences metabolic diseases and perhaps antiretroviral therapy (ART) complications. We explored associations between European mtDNA haplogroups and metabolic changes among A5142 participants.
Methods: Seven hundred and fifty-seven ART-naive patients were randomized to one of three class-sparing ART regimens including efavirenz and/or lopinavir/ritonavir with or without nucleoside reverse transcriptase inhibitors (NRTIs). Nonrandomized NRTIs included stavudine, tenofovir, or zidovudine, each with lamivudine. Fasting lipid profiles and whole-body dual-energy X-ray absorptiometry (DEXA) were performed. Nine European mtDNA haplogroups were determined for 231 self-identified non-Hispanic white individuals. Metabolic changes from baseline to 96 weeks were analyzed by haplogroup.
Results: Median age was 39 years, 9% were women, and 37, 32, and 30 were randomized to NRTI-containing regimens with either efavirenz or lopinavir/ritonavir, and an NRTI-sparing regimen, respectively. Among NRTI-containing regimens, 51% included zidovudine, 28% tenofovir, and 21% stavudine. Compared with other haplogroups, mtDNA haplogroup I (N = 10) had higher baseline non-HDL cholesterol [160 mg/dl (interquartile range 137-171) vs. 120 mg/dl (104-136); P = 0.005], a decrease in non-HDL cholesterol over 96 weeks [-14% (-20 to 6) vs. +25% (8 to 51); P < 0.001], tended to have more baseline extremity fat, and had more extremity fat loss by DEXA [-13% (-13 to 12) vs. +9% (-13 to 26); P = 0.08] and lipoatrophy (50 vs. 20%; P = 0.04). Haplogroup W (N = 5; all randomized to NRTI-sparing regimens) had the greatest increase in extremity fat [+35.5% (26.8 to 54.9); P = 0.02].
Conclusions: Lipids and extremity fat were associated with European mtDNA haplogroups in this HIV-infected population. These preliminary results suggest that mitochondrial genomics may influence metabolic parameters before and during ART.
Introduction
Mitochondrial DNA (mtDNA) is distinct from nuclear DNA, encodes 13 subunits of the electron transport chain, and exhibits abundant genetic variation across more than 16 000 base pairs. Human mtDNA sequences have diverged over approximately the last 150 000 years due to natural selection and human migration, resulting in distinct patterns of single-nucleotide polymorphisms (SNPs), called haplogroups [1]. Evidence for functional differences among mtDNA haplogroups has been demonstrated in studies of human longevity [2], neurodegenerative disorders [3], and metabolic disease and cardiovascular risk [4-7].
Antiretroviral therapy (ART) improves HIV morbidity and mortality. Unfortunately ART also has treatment-limiting metabolic complications such as dyslipidemia, insulin resistance, abnormal fat accumulation, and peripheral fat loss (lipoatrophy). Inhibition of mtDNA replication by some nucleoside reverse transcriptase inhibitors (NRTIs) is believed to lead to a cascade of mitochondrial dysfunction with impaired oxidative phosphorylation, overproduction of free radicals, tissue injury, and ultimately symptomatic toxicity [8,9]. There is wide variation in the clinical manifestations of NRTI toxicities. This phenotypic variation together with their putative mechanism(s) suggests a role for host mtDNA variation in determining susceptibility. Our group has examined the role of mtDNA variation in susceptibility to NRTI-associated peripheral neuropathy, identifying preliminary associations with mtDNA haplogroups in US clinical trial participants of both European [10] and African [11] descent. We [12] and others [13,14] have also explored European mtDNA haplogroups and ART-associated metabolic effects, primarily lipoatrophy.
AIDS Clinical Trials Group (ACTG) study A5142 was a trial designed to determine virologic efficacy and adverse metabolic effects of three class-sparing ART regimens [15]. The important metabolic effects which have been attributed to various ART classes, including lipoatrophy, were key secondary study outcomes and have been presented elsewhere [16]. Briefly, A5142 confirmed that lipoatrophy was greatest in participants treated with thymidine analog NRTIs, and, more surprisingly, in those randomized to receive the non-NRTI (NNRTI), efavirenz. Lipid effects were greatest in the NRTI-sparing arm. We hypothesize that mtDNA variation as represented in haplogroups confers subtle differences in oxidative phosphorylation, mtDNA replication, and/or apoptotic regulation, and thus is a marker for relative susceptibility to ART complications. In the pharmacogenomic analyses reported herein we explore associations between European mtDNA haplogroups and metabolic changes during ART in non-Hispanic whites who participated in A5142.
Discussion
Metabolic complications of ART have become prominent treatment-limiting adverse effects that increase the complexity of management of HIV infection and likely contribute to excess morbidity and mortality due to cardiovascular disease. Accurately predicting persons at greatest risk of these complications is not possible at present, and though several studies have reported genetic predictors of ART-associated metabolic complications [19-24], none have been well validated or incorporated into prospective clinical trials. As the HIV-infected, ART-treated population ages, metabolic complications of ART are likely to become more prominent. Better strategies to prevent these complications are needed, including expanded studies of the role of pharmacogenomics in these drug toxicities.
The present study is the largest to explore associations between mtDNA haplogroups and changes in metabolic parameters among HIV-infected persons starting ART. We found differences in baseline serum lipids among persons belonging to mtDNA haplogroup I, with these persons having higher total, LDL, and combined non-HDL-cholesterol than persons from other haplogroups. Differences in baseline median body fat measurements by DEXA were also substantial (>2.5 kg more extremity fat and >3 kg more trunk fat), but marginally statistically significant. Persons belonging to haplogroup W demonstrated extremity fat gain and no cases of lipoatrophy. These observations should be interpreted cautiously, however, given that only five individuals had 96-week DEXA data available for analyses, and all five were randomized to NRTI-sparing treatment. To our knowledge, these metabolic parameters have not been reported based on mtDNA haplogroups in the general population, and reasons for these pre-ART differences are unknown. There are data demonstrating associations between a specific mtDNA SNP and lipid parameters in Asians [4,5]. On the basis of our hypothesis that mtDNA variation represented by haplogroups confers differences in oxidative phosphorylation, mtDNA replication, and/or apoptotic regulation, it is plausible that these differences may affect baseline lipid and fat metabolism prior to initiating ART.
Persons belonging to haplogroup I had more baseline fat by DEXA, but in contrast to the overall group and all other haplogroups, lost extremity fat over 96 weeks of ART. A significantly greater proportion of haplogroup I lost at least 20% of extremity fat during the study, meeting the prespecified definition of lipoatrophy. However, due at least in part to excess baseline extremity fat, the absolute extremity fat mass in I and non-I haplogroups did not differ at 96 weeks. In multivariate models, baseline extremity fat was not associated with lipoatrophy, and did not fully attenuate the association between haplogroup I and lipoatrophy, nor did adjustment for use of thymidine analog NRTIs. However (and perhaps not surprisingly), adjustment for randomization to NRTI-sparing treatment attenuated the association between haplogroup I and lipoatrophy, suggesting that the absence of NRTI-containing ART limited the influence of mtDNA variation on this outcome. Also of note was that despite baseline differences in serum lipids, persons belonging to haplogroup I still demonstrated significant differences in non-HDL cholesterol change from baseline. This led to a statistically and clinically significant difference in absolute week 96 values, with persons from haplogroup I having a median non-HDL cholesterol of 132 mg/dl, which was 20 mg/dl lower than non-I haplogroups.
These results did not confirm previous studies that had explored mtDNA haplogroups and metabolic changes in ART-treated persons of European descent. In ACTG study A5005 s, the metabolic substudy of ACTG 384, individuals belonging to haplogroup J had increases in extremity fat when compared to other haplogroups (+26 vs. -8%; P = 0.07) [12]. In the present study the increases in fat for haplogroup J did not reach statistical significance. There were only two persons with paired DEXA data available belonging to haplogroup I in the previous study. A cross-sectional study of more than 340 Italian HIV-infected patients referred to a metabolic clinic did not identify any associations between European mtDNA haplogroups and several metabolic outcomes [14]. Only four patients (1.2%) in this population belonged to haplogroup I, and they were excluded from analyses. More recently, data from more than 400 self-reported white participants in the Multicenter AIDS Cohort Study (MACS) underwent mtDNA haplogrouping and were analyzed for changes in peripheral lipoatrophy across three body regions (buttocks, legs, arms) determined by standardized physical exam assessment [13]. Persons belonging to haplogroup H had greater lipoatrophy severity in the arms and legs than persons from other haplogroups. The investigators noted that haplogroup J was associated with less severe lipoatrophy in all three regions, but these associations were not statistically significant. Haplogroup I was only analyzed as part of a combined IWX group, which was not statistically associated with lipoatrophy.
Our study has several limitations which should be noted. Although this is the largest published sample of HIV-infected patients with mtDNA haplogroups and longitudinal metabolic data from a pre-ART baseline, the sample size for many haplogroups is small, especially for subgroups of patients randomized to and/or exposed to various ART. This limited our power to detect associations with some haplogroups, and could explain the borderline statistical significance seen with extremity fat changes and lipoatrophy. The lack of replication of the possible association of haplogroup J with less extremity fat loss [12] in the study presented here (and in the MACS analysis [13]) could be due to different ART with higher rates of lipoatrophy (all participants included in the ACTG 384 study were randomized to include a thymidine analog NRTI, and half received stavudine and didanosine in combination), different phenotypes (longitudinal DEXA from pre-ART in a clinical trial vs. cross-sectional clinical lipoatrophy assessments during ART [13]), and/or a combination of these factors. Apart from the extremity fat gain seen with haplogroup W, persons belonging to haplogroup J (N = 17 with 96-week DEXA data) had the most extremity fat gain of any haplogroup [+14% (IQR -10 to +32%)]. Although this difference was not statistically significant, results from three independent populations have demonstrated either a gain in limb fat by DEXA or less severe lipoatrophy among persons belonging to haplogroup J. The lack of association between haplogroup H and lipoatrophy that was reported previously [13] may also be due to the issues noted above. Given the exploratory nature of these analyses, we did not correct for multiple comparisons, thus some associations based on the traditional P value threshold may be spurious. It should be noted, however, that 96-week percentage changes in both total and non-HDL cholesterol among haplogroup I all remain statistically significant with simple Bonferroni correction (P < 0.006). Lastly, our analyses did not include nuclear gene variants that have been associated with lipodystrophy [21-23] and dyslipidemia [19,20,24] in ART-treated populations, or assessments of facial lipoatrophy, and data on other nongenetic factors (e.g. diet and exercise) that may influence lipids and body fat were not available.
There is little biological evidence to explain associations observed between haplogroup I and metabolic phenotypes at baseline or during ART. Haplogroup I is relatively uncommon in persons of European descent (∼5%), and has not previously been associated with human diseases. This haplogroup is characterized by synonymous SNPs at mtDNA positions: 1719G-A located in the ribosomal RNA gene; 8251G-A in the cytochrome C oxidase subunit II gene; and 16391G-A in the D-loop/control region. Shared SNPs with haplogroups W and X at positions 8251 and 1719, respectively, have provided rationale for combining these relatively infrequent haplogroups into an IWX clade for analyses. This may in part explain why analyses using similar phenotypes have not reported associations with haplogroup I [13]. Haplogroup I is also related to haplogroups J and K in the sharing of a nonsynonymous SNP at position 10398A-G. This change leads to a Thr-Ala amino acid substitution in the NADH dehydrogenase subunit 3 gene and has been associated with neurodegenerative diseases and cancer [3,25]. Mechanisms by which mtDNA variation may influence baseline and/or ART-associated dyslipidemia are not known, but recent literature highlights the role of mitochondrial function in common metabolic and cardiovascular diseases [26-28], with studies examining the specific influence of mtDNA variation on risk of metabolic and cardiovascular disease in Asian [4-7,29,30] and non-Asian [31-37] populations. Obviously, findings from these studies in HIV-negative populations do not directly address the influence of mtDNA variation on acquired metabolic effects of ART.
These results demonstrate preliminary associations between a European mtDNA haplogroup and metabolic parameters in an HIV-infected population, both prior to and during ART exposure as part of a clinical trial. Many genetic and nongenetic factors likely influence risk for metabolic changes in this population; mtDNA variation may be one of these. Continued study is needed to replicate these associations and determine mechanisms by which mtDNA variation may be associated with different metabolic effects. More extensive mtDNA genotyping in large, well characterized populations is needed to identify additional functional variants (i.e. SNPs that change amino acid sequence and/or protein function) that may underlie this and other reported associations, and to expand these analyses to other racial/ethnic groups with different mtDNA lineages.
Methods
Study participants
The exploratory analysis included participants from ACTG study A5142 (NCT# 00050895), a phase III, multicenter, randomized, open-label trial that enrolled HIV-1-infected, ART-naïve volunteers in the US with plasma HIV-1 RNA of at least 2000 copies/ml [15]. Participants used self-identified categories for race/ethnicity. Due to the relative ease in genotyping and defining European mtDNA haplogroups compared with other racial/ethnic groups, this initial analysis focused on those self-identified as 'white, non-Hispanic.'
Study design
A5142 participants were randomized to one of three ART regimens: lopinavir/ritonavir and efavirenz (lopinavir-efavirenz arm) or two NRTIs and either lopinavir/ritonavir (lopinavir arm) or efavirenz (efavirenz arm). As described in detail elsewhere [15,16], investigators selected NRTIs (zidovudine, stavudine extended release, or tenofovir) to be given with lamivudine if the patient was randomized to the NRTI-containing arms. Randomization was stratified by NRTI choice. Patients were followed for 96 weeks after the last patient was enrolled. Changes in randomized regimen or NRTI were allowed toxicity or intolerance and were considered endpoints in the efficacy analysis. Body composition was determined by whole-body dual-energy X-ray absorptiometry (DEXA) at entry, 48 weeks, and 96 weeks. Scan results were determined at a central reading site (Tufts University, Boston, Massachusetts, USA) by readers blinded to treatment assignments. Fasting serum lipids were measured at each site by commercial laboratories at entry, weeks 12, 24, and every 24 weeks.
The present study was a retrospective analysis of non-Hispanic white participants with DNA available under ACTG protocol A5128 [17]. Both retrospective cohort (for changes in lipids and extremity fat) and case-control (for lipoatrophy) study designs were utilized. A5142 and A5128 protocols were approved by institutional review boards at each study site, and participants provided written informed consent. The Vanderbilt Committee for the Protection of Human Subjects and the ACTG approved the use of DNA.
DNA sequencing and mitochondrial haplogroup determination
DNA was isolated from whole blood using PUREGENE (Gentra Systems Inc., Minneapolis, Minnesota, USA). Genotyping was performed with the ABI PRISM 7900HT Sequence Detection System (Applied Biosystems Inc., Foster City, California, USA) using the 5' nuclease allelic discrimination Taqman assay. On the basis of the work of Torroni et al. [18], we characterized SNPs at positions 1719, 4580, 7028, 8251, 9055, 10398, 12308, 13368, 13708, and 16391. Probes and primers were based on previously reported genotyping [3] and have been presented in detail elsewhere [12]. Raw genotypic data were analyzed using ABI Sequence Detection System version 2.0 software and genotype calls were confirmed by manual inspection of the plots.
Statistical analysis
Simple proportions are used to describe demographic and genetic data. Medians and interquartile ranges (IQRs) are presented for continuous data. Fisher's exact or Pearson Chi-squared tests and Wilcoxon rank-sum (Mann-Whitney U) tests were used for comparisons of categorical and continuous covariates, respectively. Univariate and multivariate logistic regression were used to determine exact odds ratios (ORs) and 95% confidence intervals (CIs) for haplogroup associations with lipoatrophy. In the multivariate model, ART exposure was included as intent-to-treat based on randomization. In the primary metabolic study, results did not differ substantially in intent-to-treat vs. as-treated analyses [16]. Analyses were not corrected for multiple comparisons in this exploratory study. Stata SE version 10 (Stata Corp., College Station, Texas, USA) was used for statistical analyses.
Results
Study population demographics and baseline HIV disease parameters
A total of 245 self-identified white, non-Hispanic A5142 participants (32% of all study participants; 89% of white, non-Hispanic participants) had DNA available for analysis. Of these, 231 (94%) were classified into one of nine major European mtDNA haplogroups and included in subsequent analyses. Fourteen participants (6%) were excluded due to non-European mtDNA haplogroup classification (N = 11) or genotyping no-calls at haplogroup-defining base positions (N = 3). The 231 included 20 (9%) women, and had a median (IQR) baseline age of 39 (17-73) years, CD4+ T-cell count of 248 (83-357) cells/µl, and log10 plasma HIV-1 RNA of 4.9 (4.5-5.5) copies/ml.
Mitochondrial DNA haplogroups
European mtDNA haplogroup distributions were similar to those reported in other US populations. The majority (48%) belonged to haplogroup H, followed in frequency by haplogroups U (15%), T (11%), and J (7%). The remainder comprised haplogroups I, K (5% each), W, X (3% each), and V (<1%).
Baseline metabolic characteristics
Of the 231 participants evaluable for mtDNA haplogroups, 218 (94%) and 220 (95%) had baseline DEXA and fasting lipid data available, respectively. Baseline metabolic parameters and prescribed NRTIs are shown for the overall group and by randomized treatment arms in Table 1, and are similar to those in the primary study population [16]. Median baseline body mass index (BMI) was 25 kg/m2, extremity and trunk fat mass by DEXA were 7.1 and 9.2 kg, respectively, and non-HDL cholesterol was 121 mg/dl. None of the participant demographics, baseline HIV disease or metabolic parameters differed significantly by treatment arm. Among the study participants receiving NRTIs, 50% received zidovudine, 29% tenofovir, and 21% stavudine. There were no significant differences in NRTIs selected by randomized treatment arm.
Baseline metabolic parameters by mtDNA haplogroups are shown in Table 2 (haplogroup V is not shown due to only two individuals in this group). The only significant differences from the overall group were among persons belonging to haplogroup I who had significantly greater median total (195 vs. 152 mg/dl), non-HDL (160 vs. 120 mg/dl), and LDL (127 vs. 91 mg/dl) cholesterol (P ≤ 0.01 for each), and tended to have greater baseline extremity (9.9 vs. 7.1 kg; P = 0.09) and trunk fat (12.3 vs. 8.8 kg; P = 0.05) than those belonging to non-I haplogroups. Median baseline fasting triglycerides were elevated among persons belonging to haplogroup I (169 vs. 126 mg/dl among other haplogroups), but this difference was not statistically significant (P = 0.4). There were no statistically significant differences across haplogroups with respect to age, sex, baseline HIV-1 RNA level or CD4+ T cells (data not shown), baseline BMI, HDL cholesterol, or, among those randomized to NRTI-containing treatment arms, NRTIs selected (Table 2).
DEXA changes and lipoatrophy
Of 218 persons with baseline DEXA data, 178 (82%) had paired 96-week data. Overall, there was a median (IQR) percentage increase in extremity fat of +7.5% (-15.4 to 23.7), corresponding to a median absolute increase of +0.6 kg (-1.6 to 1.7) from baseline to 96 weeks (Table 3). This change was consistent with that seen in the primary study population [16]. Median week 48 and 96 changes in DEXA extremity fat stratified by mtDNA haplogroups are shown in Fig. 1a. The only statistically significant difference in percentage extremity fat change was in patients belonging to haplogroup W [N = 5; median (IQR) change +35.5% (26.8 to 54.9); P = 0.02]. Only haplogroup I demonstrated a trend toward an overall percentage decrease in extremity fat [N = 10; median change -13.2% (-31.2 to 11.6); P = 0.08].
As in the primary metabolic analysis, lipoatrophy was defined as at least 20% decrease in extremity fat from baseline at 96 weeks. Overall, 39 of 178 persons (22%) with 96-week DEXA data met this definition. This proportion was 32% of persons randomized to the efavirenz arm; 21% of those randomized to the lopinavir arm; and 11% of those randomized to the NRTI-sparing lopinavir-efavirenz arm (P = 0.02 for three-arm comparison). Not surprisingly, 96-week lipoatrophy rates were 50, 26, and 9% for persons receiving stavudine, zidovudine, and tenofovir, respectively (P = 0.002) and were similar to those in the primary study population [16]. When compared across mtDNA haplogroups, persons belonging to haplogroup I were more likely to have met the definition of lipoatrophy than those belonging to non-I haplogroups (50 vs. 20%; P = 0.04; Fig. 2a). No other haplogroup differed significantly with respect to lipoatrophy. In a subgroup analysis limited to persons receiving the thymidine analog NRTIs zidovudine or stavudine [N = 91; 30 (33%) developed lipoatrophy], haplogroup I remained significantly over-represented compared with non-I haplogroups (71 vs. 30%; P = 0.04; Fig. 2b).
Logistic regression was used to determine ORs of baseline demographic, immunologic, and ART-related factors for lipoatrophy, and to adjust for possible confounding (Table 4). In unadjusted models, randomization to the NRTI-containing efavirenz arm was associated with lipoatrophy [OR = 2.4 (95% CI 1.2-5.4); P = 0.02], and randomization to the NRTI-sparing arm was protective [0.3 (0.1-0.9); P = 0.03], consistent with the primary metabolic analyses (16). Among the selected NRTIs, stavudine use was associated with development of lipoatrophy [4.8 (2.0-11.6); P < 0.001], and tenofovir use tended to be protective [0.3 (0.1-1.1); P = 0.06). Of mtDNA haplogroups, only haplogroup I was significantly associated with lipoatrophy [3.9 (1.1-14.4); P = 0.04]. In a model adjusted for randomization to the NRTI-sparing regimen, baseline extremity fat mass, and mtDNA haplogroup I, NRTI-sparing treatment remained protective of lipoatrophy [0.3 (0.1-0.9); P = 0.03], and the effect size of haplogroup I decreased and the P value no longer demonstrated statistical significance [3.0 (0.8-11.4); P = 0.11]. In separate models adjusting for individual thymidine analog NRTI use, stavudine remained significantly associated with lipoatrophy [5.2 (2.1-12.9); P < 0.001], zidovudine was not [1.4 (0.7-3.0); P = 0.3], and although haplogroup I retained similar OR [3.7 (0.9-14.6); P = 0.06 and 3.5 (0.9-12.9); P = 0.07, respectively; Table 4], these associations were no longer statistically significant. Baseline extremity fat was not associated with lipoatrophy in unadjusted or adjusted analyses.
Fasting serum lipid changes
Of 220 persons with any baseline fasting lipid data, 174 (79%) had paired 96-week data. These sample sizes were 219 and 171 (78%), respectively, for non-HDL cholesterol. Overall, median percentage changes from baseline to 96 weeks for fasting total, LDL, non-HDL, HDL-cholesterol, and triglycerides were +22.5, +19.5, +21.8, +27.6, and +50.4%, respectively. Changes from baseline in the primary lipid outcome of non-HDL cholesterol stratified by mtDNA haplogroups are shown in Fig. 1b. Only haplogroup I demonstrated a percentage decrease in non-HDL cholesterol at 96 weeks [median -14.0% (IQR -19.9 to 6.2) vs. +25.5% (8.1 to 51.1) among combined non-I haplogroups; P < 0.001]. This change corresponded to a median decrease from baseline of 20 mg/dl to an absolute 96-week value of 132 mg/dl for haplogroup I compared to a median increase of 30 mg/dl from baseline to 152 mg/dl for non-I haplogroups. Both comparisons were statistically significant (P = 0.001 for absolute change and 0.05 for absolute 96-week values) despite the higher baseline non-HDL cholesterol among persons belonging to haplogroup I. In addition, persons belonging to haplogroup I had significantly less median percentage increases in total (+3 vs. +23%; P = 0.005) and LDL (-5 vs. +21%; P = 0.01) cholesterol, and triglycerides (-4 vs. +55%; P = 0.04) over 96 weeks than those belonging to other haplogroups.
Methods
Study participants
The exploratory analysis included participants from ACTG study A5142 (NCT# 00050895), a phase III, multicenter, randomized, open-label trial that enrolled HIV-1-infected, ART-naïve volunteers in the US with plasma HIV-1 RNA of at least 2000 copies/ml [15]. Participants used self-identified categories for race/ethnicity. Due to the relative ease in genotyping and defining European mtDNA haplogroups compared with other racial/ethnic groups, this initial analysis focused on those self-identified as 'white, non-Hispanic.'
Study design
A5142 participants were randomized to one of three ART regimens: lopinavir/ritonavir and efavirenz (lopinavir-efavirenz arm) or two NRTIs and either lopinavir/ritonavir (lopinavir arm) or efavirenz (efavirenz arm). As described in detail elsewhere [15,16], investigators selected NRTIs (zidovudine, stavudine extended release, or tenofovir) to be given with lamivudine if the patient was randomized to the NRTI-containing arms. Randomization was stratified by NRTI choice. Patients were followed for 96 weeks after the last patient was enrolled. Changes in randomized regimen or NRTI were allowed toxicity or intolerance and were considered endpoints in the efficacy analysis. Body composition was determined by whole-body dual-energy X-ray absorptiometry (DEXA) at entry, 48 weeks, and 96 weeks. Scan results were determined at a central reading site (Tufts University, Boston, Massachusetts, USA) by readers blinded to treatment assignments. Fasting serum lipids were measured at each site by commercial laboratories at entry, weeks 12, 24, and every 24 weeks.
The present study was a retrospective analysis of non-Hispanic white participants with DNA available under ACTG protocol A5128 [17]. Both retrospective cohort (for changes in lipids and extremity fat) and case-control (for lipoatrophy) study designs were utilized. A5142 and A5128 protocols were approved by institutional review boards at each study site, and participants provided written informed consent. The Vanderbilt Committee for the Protection of Human Subjects and the ACTG approved the use of DNA.
DNA sequencing and mitochondrial haplogroup determination
DNA was isolated from whole blood using PUREGENE (Gentra Systems Inc., Minneapolis, Minnesota, USA). Genotyping was performed with the ABI PRISM 7900HT Sequence Detection System (Applied Biosystems Inc., Foster City, California, USA) using the 5' nuclease allelic discrimination Taqman assay. On the basis of the work of Torroni et al. [18], we characterized SNPs at positions 1719, 4580, 7028, 8251, 9055, 10398, 12308, 13368, 13708, and 16391. Probes and primers were based on previously reported genotyping [3] and have been presented in detail elsewhere [12]. Raw genotypic data were analyzed using ABI Sequence Detection System version 2.0 software and genotype calls were confirmed by manual inspection of the plots.
Statistical analysis
Simple proportions are used to describe demographic and genetic data. Medians and interquartile ranges (IQRs) are presented for continuous data. Fisher's exact or Pearson Chi-squared tests and Wilcoxon rank-sum (Mann-Whitney U) tests were used for comparisons of categorical and continuous covariates, respectively. Univariate and multivariate logistic regression were used to determine exact odds ratios (ORs) and 95% confidence intervals (CIs) for haplogroup associations with lipoatrophy. In the multivariate model, ART exposure was included as intent-to-treat based on randomization. In the primary metabolic study, results did not differ substantially in intent-to-treat vs. as-treated analyses [16]. Analyses were not corrected for multiple comparisons in this exploratory study. Stata SE version 10 (Stata Corp., College Station, Texas, USA) was used for statistical analyses.
| |
| |
| |
|
|
|
