Impact of switching from zidovudine/lamivudine to tenofovir/emtricitabine on lipoatrophy: the RECOMB study

Authors


  • The data have previously been presented in part at the XVIIth International AIDS Conference, Mexico City, Mexico, 3−8 August 2008; the 5th International AIDS Society Conference on HIV Pathogenesis, Treatment, and Prevention, Cape Town, South Africa, 19−22 July 2009; the 9th International Congress on Drug Therapy in HIV Infection, Glasgow, UK, 9−13 November 2008; the 19th Conference on Retroviruses and Opportunistic Infections, Seattle, USA, 5−8 March 2012.

Correspondence: Dr Esteban Ribera, Servei de Malalties Infeccioses, Hospital Universitari Vall d'Hebron, Paseo Vall d'Hebron 119-129, Sexta planta, 08035 Barcelona, Spain. Tel: +34 934894497; fax: +34 932746204; e-mail: eriberap@gmail.com

Abstract

Objectives

Lipoatrophy is a long-term adverse effect of some antiretrovirals that affects quality of life, compromises adherence and may limit the clinical impact of HIV treatments. This paper explores the effect of tenofovir/emtricitabine (TDF/FTC) on the amount of limb fat in patients with virological suppression.

Methods

A randomized, prospective clinical trial was performed to compare continuation on a zidovudine/lamivudine (ZDV/3TC)-based regimen with switching to a TDF/FTC-based regimen in terms of the effect on limb fat mass as assessed by DEXA over a 72-week period.

Results

Eighty patients were included (39 in the TDF/FTC arm and 41 in the ZDV/3TC arm) and 73 completed the study (37 and 36, respectively). In the switch arm, limb fat increased by a median of 540 g from baseline (P = 0.022), while in the ZDV/3TC arm it decreased by a median of 379 g (P = 0.112; p between groups = 0.007). Subjects with baseline limb fat ≤ 7200 g, previous time on ZDV > 5 years or a body mass index > 25 kg/m2 experienced higher limb fat gains than other subjects, and these differences were statistically significant. Haemoglobin increased by a median of 1.0 g/dL in the TDF/FTC arm (P < 0.001) and remained unchanged in the ZDV/3TC arm (p between groups = 0.0002). There were no significant differences between groups in other secondary endpoints (body weight, total body and trunk fat content, total body bone mineral density, laboratory parameters, CD4 cell count and viral load).

Conclusions

Switching from a ZDV/3TC-based to a TDF/FTC-based regimen led to a statistically significant improvement in limb fat, in contrast to the progressive loss of limb fat in subjects continuing ZDV/3TC.

Introduction

Highly active antiretroviral therapy (HAART) has dramatically modified the course of HIV infection, thanks to the immunological and virological control it provides [1]. Nevertheless, long-term adherence to HAART (for the lifetime of the patient, according to current knowledge) without relevant adverse effects is required to achieve optimal results [2].

For HIV-infected patients, lipoatrophy, which is related to the use of HAART as well as to some host and viral factors, is a frequent problem with important consequences. It is characterized by a loss of subcutaneous fat in the face, buttocks and/or limbs, and is frequently associated with other metabolic complications (dyslipidaemia and insulin resistance), increased cardiovascular risk and lack of adherence to antiretroviral therapy (ART) [2-4]. It is also a source of great concern for patients as a result of stigmatization. The change in the amount and distribution of body fat must be considerable (loss of 30–50% of total body fat) before it becomes evident [5-7], and so cases of clinically obvious lipoatrophy are only the tip of the iceberg of a phenomenon that represents a generalized disorganization of the adipose tissue [8], starting when antiretrovirals (ARVs) are first used and continuing over time [9-11]. When the patient perceives the change in his/her body shape, the fat loss is difficult to regain [9, 12-14], and it is therefore important to prevent its appearance at a subclinical stage [15].

Agents associated with the appearance of lipoatrophy include nucleoside reverse transcriptase inhibitors (NRTIs), stavudine (d4T) being the main culprit [9, 10, 16-18] along with zidovudine (ZDV), treatment with which is an important cause of fat loss [17, 19]. Because of their toxicity, ZDV and d4T are no longer first-line options for HIV-infected patients, but they are still used by many patients who started their HAART regimens with the ZDV/3TC combination some time ago. The fixed dose combination of tenofovir/emtricitabine (TDF/FTC) is the preferred NRTI backbone option currently, and is unanimously recommended by all main international antiretroviral treatment guidelines [15, 20]. It is an easy to use and well-tolerated option, with a smaller fat loss effect [12, 21], and is also recommended for the treatment and prevention of lipoatrophy as a substitute for thymidine analogues [15].

The objective of this study was to compare switching from ZDV/3TC to TDF/FTC with continuing on ZDV/3TC in terms of limb fat changes, as well as changes in other clinically relevant parameters, in stable patients who were virologically suppressed.

Methods

Study design

This was a prospective, multicentre, randomized, open-label, phase 4 clinical trial, carried out in eight HIV clinics in Spain, between May 2006 and September 2008. The protocol was approved by the Ethics Committees of all the participating centres and by the Spanish Health Authorities. The study was conducted according to the principles of the Declaration of Helsinki 1975, 2000 revision. Each participant gave written informed consent.

Eligible subjects invited to participate were adult patients (> 18 years old) with confirmed HIV-1 infection who had been on stable therapy for at least 6 months with a HAART regimen containing ZDV+3TC, either as separate drugs or as the fixed dose combination Combivir® (ViiV Healthcare Ltd, Uxbridge, Middlesex, UK), plus a nonnucleoside reverse transcriptase inhibitor (NNRTI) or a protease inhibitor (PI), and with confirmed HIV-1 RNA < 50 copies/mL at the last two consecutive determinations.

Patients agreeing to participate were randomized to either continuing their HAART regimen (ZDV/3TC arm) or switching from ZDV+3TC to TDF+FTC (TDF/FTC arm), with the same third agent as previously used. In the ZDV/3TC arm, subjects received the study drugs either as a fixed dose combination [Combivir®; one tablet twice a day (bid)] or as separate components [3TC 150 mg (one tablet bid) plus ZDV 300 mg (one tablet bid)]; in the TDF/FTC arm, the study drugs were administrated as a fixed dose combination [Truvada® (Gilead Sciences Inc, Foster City, CA, USA); one tablet once a day (qd)]. ‘Treated subjects’ were those who were randomized and received at least one dose of the study drugs, and the ‘treated subjects’ data set was used for both efficacy and safety analyses. A centralized manual randomization procedure was used, whereby treatments were assigned to subjects at each study site, stratifying according to the current third agent of their HAART (NNRTI or PI) regimen.

Clinical examinations and laboratory analyses were conducted at each participating centre at baseline and every 12 weeks thereafter until week 72.Whole-body dual energy X-ray absorptiometry (DEXA) scans to measure body fat were performed at baseline and every 24 weeks, and were centralized in two clinics in Barcelona and Madrid. Scans were read centrally by a radiologist unaware of the patient's randomization arm. A DEXA Lunar Prodigy model was used and cross-calibration between centres was performed by a variable composition phantom. In order to provide a clinically appropriate range for instrument assessment, three different configurations of the phantom layers were applied, designed to evaluate a high, medium and low fat composition. The agreement among the different scanners was evaluated according to the Bland−Altman method and Lin's concordance correlation coefficient. Results showed a high concordance for body fat composition between the DEXA fan-beam systems with no significant differences in the fat component quantification.

Exclusion criteria were prior virological failure or intolerance to TDF- or FTC-containing therapy, treatment with investigational antiretroviral drugs, known history of drug abuse or chronic alcohol consumption, pregnancy or breast feeding, recent opportunistic infection (current or in the previous 4 weeks), documented active malignant disease (excluding cutaneous Kaposi sarcoma), renal disease with estimated glomerular filtration rate < 50 mL/min/1.73 m2, concomitant use of nephrotoxic or immunosuppressive drugs, systemic corticosteroids, interleukin-2 or chemotherapy, and patients who were not considered suitable for the study at the investigator's discretion. For women of childbearing potential, a negative urine pregnancy test at screening visit was required. Patients on lipid-lowering therapy were allowed to participate if their lipid-lowering therapy had been stable for at least the previous 8 weeks.

Endpoints

The primary endpoint of this study was the change from baseline in total limb fat measured by DEXA at weeks 24, 48 and 72. Total limb fat change was measured in grams, and was calculated as the sum of arm and leg fat. Additional subanalyses of total limb fat change were performed with stratification for baseline total limb fat ≤ 7200 g vs. > 7200 g (mean value for HIV-negative men [5, 6]), baseline body mass index (BMI) ≤ 25 kg/m2 vs. BMI > 25 kg/m2 (cut-off for overweight), and prior ZDV use ≤ 5 years vs. > 5 years.

Secondary endpoints were changes in trunk and total body fat, body weight, total bone mineral density (BMD), haemoglobin and haematocrit levels, fasting metabolic parameters [total cholesterol, high-density lipoprotein (HDL) cholesterol, LDL cholesterol, triglycerides and lactate], mitochondrial DNA/nuclear DNA (mDNA/nDNA) ratio in lymphocytes, and renal function measured by the glomerular filtration rate, calculated using the modification of diet in renal disease (MDRD) formula, the proportion of patients maintaining HIV-1 RNA < 400 and < 50 copies/mL, change in CD4 cell counts, safety (incidence of adverse events and laboratory abnormalities) and description of resistance mutations in the case of virological failure. Adherence to ART was assessed at each visit after baseline using the simplified medication adherence questionnaire (SMAQ), and the median number of days of compliance was calculated. The SMAQ is a validated instrument that classifies adherence as a dichotomous variable (adherent/nonadherent) [22].

Statistical analysis

Descriptive statistics were reported including number and percentage for categorical variables and number, mean, standard deviation, median and interquartile ranges (IQRs) for continuous variables. Two-sided 95% confidence intervals (CIs) for each treatment group were calculated for absolute values and changes from baseline values at each time-point. Changes from baseline within treatment groups were compared using a Wilcoxon signed rank test, and differences between treatment group changes were assessed using a Wilcoxon rank sum test. The proportions of patients with undetectable viral load were compared using a Fisher's exact test. The primary analysis was performed using a last post-baseline observation carried forward (LOCF) methodology for global and stratified data, and the global analysis was confirmed by a secondary missing equals excluded (M = E) analysis.

This trial was designed as a pilot exploratory study with a total of 80 subjects, randomized 1:1 to each treatment arm.

The study has been registered at clinicaltrials.gov (number NCT00324649).

Results

Baseline demographics and subject disposition

A total of 80 subjects were randomized, 39 subjects to the TDF/FTC group and 41 subjects to the ZDV/3TC group. Subjects were predominantly male (81%) with a mean age of 44 years, and groups were well matched for all variables except sex, creatinine and baseline CD4 cell count (Table 1).

Table 1. Baseline characteristics
 TDF/FTC armZDV/3TC armP value
(n = 39)(n = 41)
  1. ARV, antiretroviral; HAART, highly active antiretroviral therapy; IQR, interquartile range; NNRTI, nonnucleoside reverse transcriptase inhibitor; NS, not significant; SD, standard deviation; TDF/FTC, tenofovir/emtricitabine; ZDV/3TC, zidovudine/lamivudine.
  2. aSubjects could report more than one HIV risk factor.
Age (years) [mean (± SD)]44 (± 10.6)44 (± 7.4)0.927
Sex (male) [n (%)]28 (72)37 (90)0.046
HIV risk factor [n (%)]a   
Injecting drug use8 (21)12 (29)NS
Heterosexual16 (41)11 (27)
Men who have sex with men15 (38)18 (44)
Other2 (5)2 (5)
CD4 cell count (cells/μL) [median (IQR)]655 (505–789)504 (363–756)0.037
Undetectable HIV RNA [n (%)]   
<50 copies/mL38 (97)39 (95)0.588
50−400 copies/mL1 (3)2 (5)
ARV class of the third agent of HAART   
Protease inhibitor6 (15)7 (17)NS
NNRTI33 (85)34 (83)
Efavirenz21 (54)24 (59)
Nevirapine12 (31)10 (24)
Years on ZDV/3TC [median (IQR)]5.8 (4.2–7.2)6.2 (4.7–7.2)0.281
Body mass index (kg/m2) [mean (± SD)]24.6 (± 3.5)24.9 (± 3.9)0.969
Body weight (kg) [mean (± SD)]70.5 (± 10.4)74.0 (± 13.2)0.391
Limb fat content (kg) [median (IQR)]3.565 (2.695–6.525)4.304 (2.147–6.503)0.756

The majority of subjects completed the 72 weeks of the study. Only two subjects (5%) in the TDF/FTC group and five (12%) in the ZDV/3TC group discontinued study treatment prematurely. The two subjects in the TDF/FTC group discontinued treatment because of gastrointestinal intolerance and noncompliance, respectively. In the ZDV/3TC group, discontinuations were attributable to indinavir-related nephrolithiasis (n = 2), anaemia (n = 1), lactic acidosis (n = 1) and one subject withdrawing consent because of fear of lipoatrophy (n = 1).

The median study drug adherence was 100% for both groups (IQR for TDF/FTC: 99.8–100%; for ZDV/3TC: 99.6–100%).

Body composition

Changes in total limb fat are summarized in Figure 1. Subjects in the TDF/FTC arm had baseline median limb fat of 3565 g, which had increased significantly by 540 g at week 72 (P = 0.022), whereas subjects in the ZDV/3TC arm had baseline median limb fat of 4304 g, which had decreased by 379 g at the same time-point (P = 0.113). The difference between the two groups was statistically significant by LOCF analysis (P = 0.008).

Figure 1.

Median changes in limb fat and baseline values assessed by dual energy X-ray absorptiometry (DEXA) in the two treatment groups over 72 weeks of follow-up. ‘TDF/FTC − ZDV/3TC’ means the total difference between groups (grams gained/lost in one arm minus grams gained/lost in the other). All comparisons within groups are not significant unless otherwise indicated: aP < 0.05; bP < 0.01; cP < 0.001; dP < 0.0001. IQR, interquartile range; TDF/FTC, tenofovir/emtricitabine; ZDV/3TC, zidovudine/lamivudine.

Subgroup analyses showed significant differences in limb fat gain between groups at weeks 48 and 72 in subjects with baseline limb fat ≤ 7200 g (Fig. 2a), but not in those with > 7200 g of limb fat (Fig. 2b). Analyses using as cut-off the median value of limb fat in the study subjects (3642 g) showed similar results (data not shown). Differences in limb fat gain were also statistically significant between groups in individuals with previous time on ZDV > 5 years (Fig. 2c), but not in those with use of ZDV ≤ 5 years (Fig. 2d). The results of the subgroup analysis for subjects receiving an NNRTI as the third agent were similar to these primary results, and the analysis for subjects taking a PI was inconclusive because of the small sample size (n = 12).

Figure 2.

Median changes in limb fat according to baseline characteristics: (a) baseline total limb fat ≤7200 g; (b) baseline total limb fat >7.2 kg; (c) zidovudine duration > 5 years; (d) zidovudine duration ≤ 5 years; (e) baseline body mass index ≤25 kg/m2; (f) baseline body mass index >25 kg/m2. ‘TDF/FTC − ZDV/3TC’ means the total difference between groups (grams gained/lost in one arm minus grams gained/lost in the other). All comparisons within groups are not significant unless otherwise indicated: aP < 0.05; bP < 0.01; cP < 0.001; dP < 0.0001. IQR, interquartile range; TDF/FTC, tenofovir/emtricitabine; ZDV/3TC, zidovudine/lamivudine.

There were no statistically significant differences within or between treatment groups in the changes from baseline in either trunk or total body fat at week 72, and there were no clinically relevant changes from baseline in subjects' body weight in any arm during the study period (Table 2).

Table 2. Baseline values and changes at week 72 in body fat and laboratory parameters
ParameterTDF/FTC (n = 39)ZDV/3TC (n = 41)P value for comparison of change between groups
Baseline [median (IQR)]Week 72 change from baseline (median)Baseline [median (IQR)]Week 72 change from baseline (median)
  1. IQR, interquartile range; TC, total cholesterol; LDLc, low-density liprotein cholesterol; HDLc, high-density lipoprotein cholesterol; TDF/FTC, tenofovir/emtricitabine; ZDV/3TC, zidovudine/lamivudine.
  2. Within treatment group comparisons are not significant unless otherwise indicated: P < 0.05; bP < 0.01; cP < 0.0001.
  3. * Conversion factor for TC, HDLc and LDLc in mmol/L: mg/dL/39.
  4. † Conversion factor for triglycerides in mmol/L: mg/dL/89.
Trunk fat content (kg)10.1 (7.6–14.0)+0.3549.4 (6.0–13.1)+0.5260.897
Total body fat (kg)15.2 (10.7–21.5)+0.78613.5 (9.7–20.6)+0.3340.307
TC* (mg/dL)187.0 (157.0–206.0)−3.0189.0 (166.0–218.0)+1.00.249
TC/HDLc ratio*3.9 (3.2–4.7)+0.33.8 (2.8–4.9)+0.30.882
LDLc (mg/dL)*109.0 (86.0–136.0)−1.5107.5 (90.0–131.6)+12.00.043
HDLc (mg/dL)*49.0 (42.5–59.1)−4.0a49.5 (41.5–65.0)−1.00.372
Triglycerides (mg/dl)110.0 (76.0–154.0)+7.0120.0 (80.0–175.0)0.00.275
Haemoglobin (g/dL)14.6 (13.8–15.5)+1.0c14.5 (13.2–15.1)−0.10.0002
Haematocrit (%)42.6 (40.0–44.9)+2.3c42.0 (39.0–43.6)0.00.0014
Creatinine (mg/dL)0.80 (0.67–0.93)+0.050.90 (0.78–1.02)−0.020.001
Lactic acid (mmol/L)1.23 (0.89–1.60)−0.17b1.12 (0.83–1.99)0.070.024
Lymphocyte mDNA/nDNA ratio61.5 (35.0–83.0)+99.0c61.0 (43.0–82.0)+31.0c0.195

Regarding total BMD, baseline values were similar between treatment groups (1.07 ± 0.096 g/cm2 in the TDF/FTC group and 1.08 ± 0.127 g/cm2 in the ZDV/3TC group). After 72 weeks, there were no significant changes within the TDF/FTC (−0.95 ± 1.8%; P = 0.374) or the ZDV/3TC (−0.58 ± 1.1%; P = 0.837) group or between treatment groups (P = 0.307).

Lipid parameters

In the TDF/FTC arm, total cholesterol decreased from a baseline median of 187 mg/dL (4.79 mmol/L) to 182 mg (4.71 mmol/L) at week 12 (P < 0.001) and 178 mg/dL (4.60 mmol/L) at week 24 (P = 0.040), and then returned to 183 mg/dL (4.69 mmol/L) at week 72. A similar pattern was observed for LDL cholesterol, with an additional significant difference between groups in LDL cholesterol at week 72 [a median decrease of −1.5 mg/dL in the TDF/FTC arm (P = 0.553) and a median increase of +12.0 mg/dL in the ZDV/3TC arm (P = 0.081); P = 0.043 between groups]. There were no significant changes in the median total cholesterol value in the ZDV/3TC group at any time-point. Twelve subjects in the TDF/FTC arm and 16 in the ZDV/3TC arm had baseline total cholesterol values > 200 mg/dL (5.13 mmol/L), while at the end of follow-up 12 subjects in the TDF/FTC arm and 20 in the ZDV/3TC arm had values > 200 mg/dL. Three subjects (7%) in the ZDV/3TC arm and none in the TDF/FTC arm needed to start treatment with lipid-lowering drugs during the study period. At week 72 there were no significant differences within or between the groups for total cholesterol, total cholesterol/HDL ratio or triglycerides.

Other laboratory markers (Table 2)

Haemoglobin and haematocrit had increased in the TDF/FTC group by week 12 (median haemoglobin +0.9 g/dL; median haematocrit +3.0%; both P < 0.0001) and remained increased throughout the treatment period. There were no significant changes in these parameters within the ZDV/3TC group at any time-point. Differences between treatments were significant (P ≤ 0.015) at all post-baseline time-points for both haemoglobin and haematocrit. At the end of follow-up, six subjects in the ZDV/3TC arm and 11 in the TDF/FTC arm had increases in absolute haemoglobin of ≥ 1 g/dL, and six in the ZDV/3TC arm and three in the TDF/FTC arm had decreases of the same magnitude. Three subjects in the ZDV/3TC arm and one in the TDF/FTC arm needed specific treatment for anaemia during the study period.

Baseline serum creatinine concentration was significantly lower in the TDF/FTC arm than in the ZDV/3TC arm (median 0.80 vs. 0.90 mg/dL, respectively; P = 0.046), and slightly increased in the TDF/FTC group at week 72 (median 0.86 mg/dL; change from baseline +0.05 mg/dL; P < 0.001), with no significant changes in the ZDV/3TC group (median 0.84 mg/dL; change – 0.02 mg/dL; P = 0.12). At week 72, the median value for the glomerular filtration rate decreased from 99.2 to 92.2 mL/min/1.73 m2 in the TDF/FTC arm (median change −6.5 mL/min/1.73 m2; P < 0.001), and increased from 96.4 to 106.1 mL/min/1.73 m2 in the ZDV/3TC arm (median change +3.0 mL/min/1.73 m2; P = 0.110), the difference between groups being significant (P < 0.001). All creatinine and glomerular filtration rate values during the study period were within the normal range in both arms.

Lactic acid had decreased significantly by week 24 in the TDF/FTC group (−0.33 mmol/L; P = 0.0016), with no significant changes in the ZDV/3TC group.

HIV infection parameters

At week 72, 35 of 39 subjects (89.7%) in the TDF/FTC arm and 34 of 41 subjects (82.9%) in the ZDV/3TC arm (difference 6.8%; 95% CI −8.1 to 21.8%) maintained a sustained virological response [intent-to-treat (ITT), missing equals failure (M = F) analysis]. Only two patients in each group experienced viral rebound (two consecutive HIV RNA values > 50 copies/mL).

The median CD4 count over 72 weeks increased by 67 cells/μL in the TDF/FTC group and by 36 cells/μL in the ZDV/3TC group, with no significant differences between groups.

Safety

Treatment was well tolerated in both groups. Grade 3–4 adverse events occurred in one patient (2.6%) treated with TDF/FTC and five patients (12.2%) treated with ZDV/3TC. Of these, only two cases were considered to be possibly or probably related to the study drugs, both in the ZDV/3TC arm (lactic acidosis and anaemia). There were two cases of trauma-related bone fractures during the study (a lumbar vertebral fracture in the TDF/FTC arm and a radius fracture in the ZDV/3TC arm), neither of which was considered to be related to any of the study medications. The only renal adverse events were two cases of renal colic related to indinavir in the ZDV/3TC arm. No new or unexpected safety findings or deaths were reported during the study.

Discussion

In this randomized clinical trial conducted in virologically suppressed HIV-infected patients, switching from ZDV/3TC to TDF/FTC was associated with a statistically significant increase in limb fat compared with patients who continued on their original regimen. This difference was driven by a significant increase in limb fat relative to baseline in the TDF/FTC arm and a not statistically significant decrease in the ZDV/3TC arm. Changes were progressive over time and the difference between groups was significant from week 48 onwards. At the end of week 72, subjects in the TDF/FTC arm had almost 1 kg more limb fat that those who continued on ZDV/3TC. These results are consistent with previous findings which showed that switching from a regimen containing d4T or ZDV to a new regimen containing abacavir (ABC) [13, 14, 18, 23-25] or TDF [21, 25, 26] made it possible to halt the loss of subcutaneous fat and was associated with some subcutaneous fat regains.

In almost all the subgroups analysed, stratified by baseline limb fat, baseline BMI or previous time on ZDV, the trends in changes were the same: fat was preserved in the TDF/FTC arm and lost in the ZDV/3TC arm. In contrast to the findings of a similar study [27], in our study this benefit in the TDF/FTC arm was especially marked and statistically significant in subjects who had been exposed to ZDV for longer periods of time (> 5 years) and in those whose baseline limb fat was ≤ 7200 g (this value represents the mean value for limb fat in men without HIV infection) [5, 6]. This latter group would have represented both subjects who had already lost their limb fat as a consequence of their previous exposure to ZDV, and who therefore had the greatest deficit to be regained, and slim subjects, who are at known risk for lipoatrophy, as identified in previous series [9]. In contrast, differences within and between groups in strata with a shorter exposure to ZDV or with baseline limb fat > 7200 g showed similar trends but did not achieve statistical significance, probably because these subjects did not have a great fat loss to overcome. However, these results should be interpreted with caution as they were obtained in a small group of subjects. Finally, the group with the highest absolute increase in limb fat was that with a higher baseline BMI (> 25 kg/m2), this factor having been previously identified as an independent predictor of fat increase [13]. As has been reported before [27, 28], total body fat and trunk fat content showed a slight but not significant increase in both treatment groups during the study period. Regarding bone composition, there were no significant differences within or between study groups in total BMD after 72 weeks.

Virological and immunological control was maintained during the study period in both arms. TDF/FTC was well tolerated in this study population, with no new or unexpected safety findings reported during the study, as found in previous studies [12, 21, 25, 27, 29, 30]. Changes in the lipid profile were more favourable in the TDF/FTC arm, but they were small in magnitude in both groups. Haemoglobin values increased early in the TDF/FTC arm, suggesting that this switch strategy could be a good option in cases of symptomatic anaemia. Although there were significant changes from baseline in the serum creatinine concentration and in the glomerular filtration rate between groups, renal security was good in the TDF/FTC arm. The increase in serum creatinine levels and the decrease in the glomerular filtration rate were small and similar to those described in previous studies [27-29]. All patients had a renal function within the normal range and without significant renal adverse events.

One limitation of our study was the small sample size in certain subgroups of interest, which may have limited our chances of detecting important changes in specific subsets of subjects. Also, we do not have information about the subjective perception of lipoatrophy by subjects or by physicians. This information would have been of value in considering the implications of our findings, given that this phenomenon is often not clinically evident in its early stages, but has the capacity to seriously affect quality of life when it becomes visible. Previous studies have shown that, in patients with clinically evident lipoatrophy, an improvement in limb fat as measured by DEXA is not always perceived in the same way by patients and/or physicians [13, 14, 23]. Nevertheless, a recent study [31] has shown a statistically significant correlation between lipoatrophy scores, as determined by physicians and patients, and the measurements of limb fat changes by DEXA, and other authors [32, 33] have reported that changes in limb fat measured by DEXA have a positive and statistically significant correlation with the increase in malar volume as determined by a validated 3D laser imaging technique, suggesting that cosmetic attributes also improve when limb fat increases.

In summary, our study showed that switching from a HAART regimen based on ZDV/3TC to a regimen based on TDF/FTC led to a significant improvement in limb fat, in contrast to the progressive loss of limb fat seen in subjects who remained on their original regimen. This benefit was especially marked in subjects who had already lost limb fat. However, fat recovery was not complete and these results support a proactive change from ZDV/3TC to TDF/FTC.

Acknowledgements

We thank Ana María Caro-Murillo for editing the manuscript and for editorial assistance and Anthony Curran for English language editing. Funding was received from Gilead Sciences S.L. AC has received an unrestricted investigational grant from the Spanish Network for AIDS Research RETICS, RD12/0017/0003.

Author contributions: The study was designed by and data were analysed at Gilead Sciences. All authors had full access to the data and vouch for the accuracy and completeness of the data and the analyses. The manuscript was written by all the authors, each of whom contributed to the drafts and revisions and approved the final manuscript.

Ancillary