Lipoatrophic men 44 months after the diagnosis of lipoatrophy are less lipoatrophic but more hypertensive
...Clinical lipoatrophy seemed partly reversible, as 37% of patients no longer fulfilled the clinical criteria for lipoatrophy after 44 months. In the follow-up study, one significant risk factor for atrophy reversal was found: less than 36 months of stavudine treatment. In addition, we found a disturbingly high prevalence of hypertension in the lipoatrophic patients. However, the number of patients in the follow-up study was limited, and prospective studies in larger cohorts are required to confirm the findings... Elevated blood pressure in patients on ART has been demonstrated in earlier studies. As hypertension in HIV-infected patients is associated with older age, male gender, white race, elevated waist/hip ratio, cholesterol and triglyceride levels, a high prevalence of hypertension was expected in our cohort.....
BM Bergersen1, L Sandvik2, I Ellingsen3 and JN Bruun1
1Department of Infectious Diseases, 2Centre for Clinical Research, and 3Department of Preventive Cardiology, Ullevål University Hospital, Oslo, Norway
INTRODUCTION. Peripheral lipoatrophy in HIV-infected patients is part of the lipodystrophy syndrome, first described in 1998 . Lipodystrophy was initially attributed to protease inhibitors (PIs), the new group of drugs introduced 2 years before, but replacement of PIs with non-nucleoside reverse transcriptase inhibitors (NNRTIs) [2,3] or nucleoside reverse transcriptase inhibitors (NRTIs)  has had no effect on fat atrophy. Lipoatrophy has been associated with certain host, disease and drug factors such as older age, white ethnicity, female gender, use of antiretroviral therapy (ART) for longer than 2 years, low body mass index (BMI), advanced HIV disease and use of thymidine analogues [5-9].
Recent studies [10-12] have demonstrated that replacement of stavudine or zidovudine with abacavir improves limb fat mass as measured by dual-energy X-ray absorptiometry (DEXA) over time. None the less, in most of these studies increased limb fat mass was not subjectively apparent when evaluated by patients and physicians, and thus lipoatrophy seemed rather irreversible. To date, no observational studies have been published in which a lipoatrophic cohort is followed for more than 2 years. The main aim of this study was to identify clinical factors related to HIV-associated lipoatrophy. In addition, we wished to address the question of the reversibility of lipoatrophy and to evaluate body composition changes, blood pressure and lipid levels in lipoatrophic subjects 3-4 years after the atrophy diagnosis.
Objectives: To identify clinical factors associated with HIV-associated lipoatrophy and to evaluate body composition changes, blood pressure and lipid levels in lipoatrophic subjects 3-4 years after the atrophy diagnosis.
Methods: Clinical signs of lipoatrophy were assessed in 308 ambulant HIV-positive patients in 2000-2001. Possible clinical risk factors, such as age, gender, race, wasting, duration of HIV infection, presence or absence of AIDS diagnosis, viral load and CD4 count, and detailed information about drug treatment were analysed and explored in a multivariate model. Lipoatrophic white males with triceps skin fold <10 mm were re-examined after 44 months. Signs of lipoatrophy and associated factors, blood pressure, lipid levels, diet and level of exercise at first and second visits were compared.
Results: In the multivariate analysis, significant clinical risk factors for lipoatrophy were weight loss >7 kg compared to normal weight [odds ratio (OR) 3.76; 95% confidence interval (CI) 1.80-7.82; P<0.001], current and/or previous use of stavudine (OR 3.72; 95% CI 1.57-8.83; P=0.003) and duration of HIV infection >80 months (OR 2.28; 95% CI 1.13-4.59; P=0.021).
Forty of 47 lipoatrophic white males with skin fold<10 mm were available for re-examination. Of these, 15 (38%) no longer fulfilled the atrophy diagnosis (P<0.001). The prevalence of arm atrophy fell from 63 to 28% (P=0.001) and facial atrophy from 55 to 43% (P=0.23). Use of stavudine for<36 months was significantly associated with lipoatrophy reversal (OR 5.00; 95% CI 1.15-21.80; P=0.032), but weight gain and increased CD4 count were not.
Prevalence of hypertension increased from 28 to 50% (P=0.035), mean systolic blood pressure from 130±14 to 136±19 mmHg (P=0.021) and diastolic blood pressure from 82±10 to 87±12 mmHg (P<0.001). In spite of increased use of lipid-lowering drugs (from two to nine patients), levels of total cholesterol, high-density lipoprotein (HDL) cholesterol and triglycerides were unchanged.
Conclusions: In this study, we found that weight loss >7 kg, use of stavudine and long duration of HIV infection were significant risk factors for clinical lipoatrophy. Clinical lipoatrophy was partly reversible, and <36 months on stavudine was significantly associated with atrophy reversal. The prevalence of hypertension and the yearly increase of mean blood pressure were disturbingly high in these patients. However, the number of patients in this study was limited, and prospective studies in larger cohorts are required to confirm these findings.
In this study, we found that significant risk factors for clinical lipoatrophy were weight loss, current and/or previous use of stavudine and long duration of HIV infection. Clinical lipoatrophy seemed partly reversible, as 37% of patients no longer fulfilled the clinical criteria for lipoatrophy after 44 months. In the follow-up study, one significant risk factor for atrophy reversal was found: less than 36 months of stavudine treatment. In addition, we found a disturbingly high prevalence of hypertension in the lipoatrophic patients. However, the number of patients in the follow-up study was limited, and prospective studies in larger cohorts are required to confirm the findings.
In the initial OHIV cohort study, lipoatrophy defined by this clinical method was significantly associated with weight loss. However, we did not find any association between atrophy reversal and weight gain in the follow-up study. This discrepancy may be explained by the fact that 100% of the six patients who died had a weight loss>7 kg in the initial study, compared to 20% of the 40 examined patients. Consequently, the high mortality rate among the lipoatrophic subjects with weight loss>7 kg (six of 15) introduces a selection bias to our study. This also illustrates an important issue when evaluating lipoatrophic patients: the need to exclude weight loss and possible underlying diseases before considering ART adjustments. The lack of an objective definition of lipoatrophy is clearly a limitation of this study. Unfortunately, the only objective definition of lipodystrophy  does not split lipodystrophy into subgroups of fat atrophy and fat accumulation. We applied the same clinical method for body composition evaluation as used in the lipodystrophy case definition study . Although undoubtedly biased, the individual subjective scores were remarkably similar in the two studies [14,20]. However, this method does not discriminate clearly between patients with lipoatrophy who usually do not lose weight and patients with wasting.
Some of the atrophy reversal in the follow-up study can probably be explained by decreased use of stavudine. In the initial cohort study, we found an association between lipoatrophy and current and/or previous stavudine experience. This finding was confirmed in the follow-up study, where atrophy reversal was associated with less time on stavudine. In the mitochondrial toxicity (MITOX) Extension study , limb fat measured with a DEXA scan increased 2 years after a switch from zidovudine or stavudine to abacavir, but no difference was found in self-assessed lipodystrophy. The shorter observation time and the heterogeneity of the lipodystrophy definition may explain the discrepancy with our results. In the TARHEEL (trial to assess the regression of hyperlactatemia and to evaluate the regression of established lipodystrophy in HIV-1-positive subjects) study , about 20% of the lipoatrophic patients reported 'some' or 'a lot' of fat gain in the arms, legs, buttock and/or face 48 weeks after replacing stavudine with abacavir or zidovudine. Whether stavudine discontinuation may cause BMI changes is still unclear.
There is growing evidence that the core problem in HIV-associated lipodystrophy is disturbances in the metabolism of peripheral adipocytes. One possible mechanism is mitochondrial toxicity, which was early proposed as an explanatory factor for multiple HIV-induced symptoms . NRTIs may induce depletion of mtDNA and mitochondrial dysfunction by inhibition of DNA polymerase gamma . Another explanation for apoptosis of peripheral adipocytes is inhibition of adipogenic differentiation factors . Pro-inflammatory cytokines such as tumour necrosis factor-alpha may inhibit some of these factors [27,28], and thus levels of various pro-inflammatory cytokines should be measured in future cohort studies of lipoatrophy.
The atrophy reversal in our study could also partly be attributable to regression-to-the-mean, because lipoatrophy was an inclusion criterion in the study. However, the reduction of face and arm atrophy was not parallel and it seems unlikely that regression-to-the-mean explains the entire effect.
Elevated blood pressure in patients on ART has been demonstrated in earlier studies [29-32]. As hypertension in HIV-infected patients is associated with older age, male gender, white race, elevated waist/hip ratio, cholesterol and triglyceride levels , a high prevalence of hypertension was expected in our cohort. Nevertheless, the systolic blood pressure (SBP) elevation of 1.7 mmHg/year was markedly higher than that in healthy, middle-aged Norwegian men, where SBP increased by 0.43 mmHg/year . To validate our findings, we compared the OHIV measurements in 2000-2001 (median date 7 July 2000) with recent random sphygmomanometer measures made by the physicians at the ambulatory unit (median date 10 June 2004). In the 47 of 66 male LA patients who had blood pressure measurements from 2003 to 2004 (including nonwhite subjects and those with skin fold <=10 mm), SBP had increased by 1 mmHg/year (P=0.15). In 121 of 177 male non-LA patients, SBP had increased by 1.5 mmHg/year (P=0.016). We may speculate about an increased risk of blood pressure elevation in HIV-infected subjects, but this analysis had many limitations and the hypothesis should be tested prospectively in larger studies. In a retrospective study, Chow et al.  found a SBP elevation of 2.8 mmHg/year in patients on stable ART regimens. In a cohort including 5504 HIV-positive men, Seaberg et al.  found that ART treatment for more than 2 years increased the risk of hypertension significantly. Although no correlation between blood pressure in HIV-positive patients and BMI was found in our initial study , some of the blood pressure elevation in the follow-up study could be attributable to the increased weight, but BMI was still 3 kg/m2 below that of HIV-negative controls . Another potential mechanism is an unfavourable long-term effect of HIV and/or ART on the kidneys or the vascular bed. Endothelial dysfunction , increased arterial stiffness  and increased intima thickness  have been demonstrated in HIV-positive patients, and thus a vascular long-term effect of HIV infection itself may aggravate the effects of ART. Our study was not designed to explore risk factors for increased blood pressure in HIV-positive patients. Larger studies are needed incorporating all known risk factors for hypertension, detailed ART histories and markers of kidney and vascular dysfunction.
No change in diet, alcohol consumption or physical activity that could explain the changes in lipoatrophy or blood pressure was found. A limiting factor in the diet questionnaire was the lack of questions about salt consumption, but it seems unlikely that ingestion of salt should have increased significantly between the two visits.
Lipoatrophy and associated factors in the Oslo HIV Cohort 2000-01
Of the 308 OHIV patients, 67 (22%) had lipoatrophy. The lipoatrophy (LA) patients were older, included more men and more white individuals, and had been HIV-infected for a longer period than the non-lipoatrophic (non-LA) patients. They also had greater previous weight loss, but the prevalences of BMI<23 kg/m2 were similar (LA patients 60% vs. non-LA patients 52%; P=0.28). This discrepancy was explained by a higher initial weight in the LA group compared to the non-LA group (76±12 vs. 72±13 kg, respectively; P=0.026).
The prevalences of an AIDS diagnosis (P=0.065) and HIV wasting syndrome (LA group, five patients; non-LA group, three patients) were similar. Rather more subjects in the non-LA group had a high viral load, but the difference was not statistically significant (32% vs. 25%, respectively; P=0.088). The percentages of patients with current CD4 count<100 cells/muL were similar in the two groups (P=0.41), but the prevalence of nadir CD4 count<160 cells/muL was higher in the LA than in the non-LA group. Ninety-four per cent of the LA group and 75% of the non-LA group were ART-experienced (P<0.001).
The prevalence of the use of all ART drugs that differed significantly between the groups is presented in Table 1. Efavirenz was included in the multivariate analysis because the difference was nearly significant (P=0.051). In the OHIV cohort, 52% of patients were stavudine-experienced, 78% in the LA group and 44% in the non-LA group (P<0.001), and the LA subjects had used stavudine for a longer time (mean 24±12 vs. 17±11 months, respectively). In the LA group, 60% still used stavudine at the first visit, compared to 30% in the non-LA group (P<0.001). Older age [OR 1.79; 95% confidence interval (CI) 1.04-3.11; P=0.037], male gender (OR 3.30; 95% CI 1.36-8.13; P=0.009), white race (OR 3.08; 95% CI 1.26-7.50; P=0.013), duration of HIV infection>80 months (OR 3.30; 95% CI 1.83-5.96; P<0.001), low nadir CD4 count (OR 3.08; 95% CI 1.72-5.52; P<0.001), weight loss>7 kg (OR 2.93; 95% CI 1.59-5.41; P=0.001), current and/or previous use of stavudine (OR 4.34; 95% CI 2.32-8.14; P<0.001) and the other ART drugs shown in Table 1 (data not presented) were all significantly associated with lipoatrophy and hence included in the multivariate model.
In the final analysis, three factors were significantly associated with lipoatrophy; weight loss>7 kg, current and/or previous use of stavudine and duration of HIV infection>80 months.
Lipoatrophy in the follow-up study
To include the most lipoatrophic subjects in the follow-up study, we chose white males with triceps skin fold<10 mm. Forty-seven of the 308 patients examined in 2000-2001 fulfilled the inclusion criteria. Forty-six of the patients in the ART-treated group were included. None of the ART±patients had both lipoatrophy and triceps skin fold<10 mm. In the ART-naïve group, three male patients had lipoatrophy, but only one had triceps skin fold<10 mm. This patient was more than 70 years old and died before the follow-up study was initiated. By December 2003, five of the 47 patients had died. The causes of death were HIV-related heart failure, ART-induced pancreatitis, AIDS with cytomegalovirus (CMV) infection, coronary heart disease and unknown in one case. One patient did not respond to the invitation letter and one was hospitalized and died later of AIDS/CMV infection. In the OHIV study 2000-01, all the above-mentioned six patients who died had a weight loss>7 kg (range 8-35 kg). This group had a mean nadir CD4 count of 17 cells/mul, three of them had an AIDS diagnosis and one was diagnosed with HIV wasting syndrome.
The 40 remaining patients were examined a median of 44 months after the first visit [interquartile range (IQR) 40-46 months]. In 90% of the patients, the HIV infection was homosexually acquired. Median age at baseline was 47 years (IQR 38-50 years). They had been HIV-infected for a long time: median 13.1 years (IQR 9.7-17.5 years) at the second visit. Fifteen patients (38%) no longer fulfilled the criteria of lipoatrophy. The prevalence of arm and buttock atrophy was particularly reduced. Eighteen patients (45%) reported an increase and nine (23%) a reduction in peripheral or central fat. Thirty patients (75%) had visible veins at the first visit, as opposed to twenty (50%) at the second visit (P=0.013). Two patients with a Buffalo hump and one with atrophic cheeks had undergone cosmetic surgery. Twelve of the 17 patients with facial atrophy would consider cosmetic surgery if accessible.
Mean CD4 count had increased between the first and second visits; from 394±180 to 528±285 cells/muL (P=0.004), but the prevalences of HIV RNA<50 copies/mL were similar; 52% vs. 55%, respectively. Mean BMI and waist and hip circumferences increased between the first and the second visits. Twenty-five patients had put on weight, 20 more than 2 kg. Nine patients (23%) had lost more than 7 kg. In the 12 patients who had prominent veins at the first but not at the second visit, mean triceps skin fold thickness had increased from 6.1±1.7 to 7.4±2.0 mm (P=0.021). No statistically significant association was found between atrophy reversal and weight gain>0.63 kg (OR 2.16; 95% CI 0.54-8.68; P=0.27) or CD4 count increase>170 cells/muL (OR 2.25; 95% CI 0.61-8.31; P=0.22).
The patients were heavily treatment-experienced, with a median of 9.1 (IQR 7.3-11.3) years on ART. Sixty per cent had started with monotherapy (one NRTI), 20% with dual therapy (two NRTIs) and 20% with highly active antiretroviral therapy (HAART) (two NRTIs and at least one PI or NNRTI). The median time on HAART at the second visit was 7.3 (IQR 6.4-7.6) years. Thirty-seven of the 40 patients were stavudine-experienced, with a median time of 36 months (range 5-84 months). Only three patients used stavudine at the second visit. All three had still lipoatrophy, and their mean total atrophy grade had increased from 14±4 to 15±1 (mean±SD). Most of the others had stopped stavudine because of lipoatrophy, with a median time since stavudine discontinuation of 3 years. Use of stavudine for <=36 months at the second visit was significantly associated with lipoatrophy reversal (OR 5.00; 95% CI 1.15-21.80; P=0.032). The use of didanosine had increased from 23% to 39% (P=0.15). All the 40 patients were PI-experienced, but current use of PI fell from 75% to 45% from the first to the second visit. The use of NNRTIs also fell; from 50% to 43%. A probable explanation is that seven patients had stopped ART, five of them as participants in a vaccination project  and two for individual reasons. Median time since ART discontinuation was 11.6 (IQR 8.6-13.3) months.
Blood pressure and blood lipids in the follow-up study
The prevalence of hypertension increased from 28% to 50% in 44 months. Mean systolic blood pressure increased by a mean of 6 mmHg and diastolic blood pressure by 5 mmHg, and six patients had been put on antihypertensive medication. Forty per cent of the patients had a parent or sibling with hypertension, but the prevalences of hypertension were similar in patients with and without hypertensive relatives (eight of 16 patients with hypertensive relatives vs. 12 of 24 patients without hypertensive relatives). In spite of an increased use of lipid-lowering drugs (from two to nine patients), blood lipid levels were stable, with mean (±SD) values as follows on the first and second visits, respectively: total cholesterol, 6.3±1.2 vs. 6.5±1.6 mmol/L (P=0.79); LDL cholesterol, 3.8±1.3 vs. 3.8±1.3 mmol/L (P=0.83); HDL cholesterol, 1.15±0.35 vs. 1.22±0.35 mmol/L (P=0.50); and the median concentrations of triglycerides were 2.9 mmol/L (IQR 1.7-3.9) vs. 2.4 mmol/L (IQR 1.4-4.5), respectively (P=0.84).
One patient used prednisolone at the first visit and two at the second visit, and two patients had previous use of anabolic steroids for less than 3 months. None of the patients used nonsteroidal anti-inflammatory drugs at the first or second visit.
Diet and physical activity
No dietary advice was given at the first OHIV visit. None the less, somewhat less sugar and more polyunsaturated fatty acids were found to be present in the diet on the second visit compared with the first visit, although the changes were not statistically significant. The consumption of chocolate or sweets 4-6 times per week fell from 23% to 10% (P=0.23), and consumption of one or more glasses of cola or lemonade with sugar daily fell from 18% to 8% (P=0.13). Juice consumption was unchanged (one or more glasses of juice daily, 40% vs. 43%, respectively). Rather more patients (75% vs. 80%; P=0.72) used soft or light margarine and oil instead of hard margarine or butter. Consumption of oily fish once or more per week and/or daily cod liver oil had increased from 55% to 65% (P=0.39). Questions about table salt or salty food were not included, apart from intake of potato crisps, which was somewhat lower at the second visit; consumption of crisps 'rarely or never' increased from 28% to 43% (P=0.11). Alcohol consumption fell from a median of 36.0 g (IQR 0-64.5 g) per week to 24.0 g (IQR 0-60.0 g) per week (P=0.42). The level of physical activity did not change.
METHODS. From April 2000 to April 2001, body composition changes and cardiovascular risk factors were recorded for 308 ambulatory HIV-positive patients. The prevalence of various body composition changes, hypertension and lipid disorders in the Oslo HIV Cohort (OHIV) is reported in previous papers [13-16]. The OHIV study included 244 ART-experienced patients: 219 patients who had been on ART for more than 6 months (ART-treated) and 25 patients who either had used ART before or had used ART for less than 6 months (ART±). Sixty-four patients had never used ART (ART-naïve). From December 2003 to February 2004, a follow-up study including all lipoatrophic white men in the OHIV cohort with a triceps skin fold <10 mm was carried out. The study protocol was evaluated by the Regional Committee for Medical Research Ethics and approved by the Norwegian Data Inspectorate. The study has been conducted in full accordance with the World Medical Association Declaration of Helsinki.
The study physician (BMB) performed all the clinical examinations (waist, hip, weight, triceps skin fold, blood pressure and lipodystrophy assessments) with the same equipment and at the same method at both visits. Waist circumference was measured at the umbilical level, and hip circumference at the upper level of the trochanter major. Triceps skin fold thickness was measured according to the Anthropometric Standardization Reference Manual . The details of the blood pressure measurement and body change evaluation are given elsewhere [13,14]. Hypertension was defined as systolic blood pressure >=140 and/or diastolic blood pressure>=90 mmHg or self-reported medication for high blood pressure . Information about coronary heart disease risk factors was sampled using standardized questions from The Oslo Health Study 2000-01  regarding diet (question 7), physical activity (question 10), family history (question 12; hypertension added) and medication (question 13).
The Lipodystrophy Questionnaire developed by Andrew Carr and his group  was used for body change assessment. The degree of fat change at eight sites (face, arms, legs, buttocks, abdomen, breasts, neck and dorso-cervical spine) was rated independently both by the patient and by the physician as absent (score 0), mild (score 1), moderate (score 2) or severe (score 3). Lipoatrophy was defined as the presence of at least one of four signs of peripheral atrophy (in the face, arms, legs and buttocks) rated by the patient as moderate or severe and independently confirmed by the physician as moderate or severe [14,20]. 'No lipoatrophy' (non-LA) subjects had no or mild signs of lipoatrophy. 'Wasting' was defined as weight loss >7 kg compared to normal weight before HIV infection. We chose 10% of mean weight in the cohort (72±13 kg; mean±standard deviation) as the threshold for significant weight loss. We used the Centers for Disease Control and Prevention (CDC) definition of HIV wasting syndrome: involuntary loss of at least 10% of body weight with chronic fever, weakness, or diarrhoea in the absence of other related illnesses contributing to the weight loss .
Blood sample analyses were performed by the same technicians, at the same place (Ulleval University Hospital) and with the same methods at both visits. Serum for lipid analyses was drawn between 0800 and 1100 h after at least 8 h of fasting. Concentrations of total cholesterol (normal range for those aged>40 years: 4.1-8.7 mmol/L, ideally<5 mmol/L), high-density lipoprotein (HDL) cholesterol (normal range for men aged 35-55 years: 0.80-2.0 mmol/L) and triglycerides (normal range for men aged 35-55 years: 0.3-1.8 mmol/L) were measured enzymatically. The concentration of low-density lipoprotein (LDL) cholesterol was calculated using the Friedwald formula. CD4 cell counts were determined by flow cytometry (TrïTest CD4; Becton Dickinson, Santa Jose; CA) and plasma HIV RNA levels by the Cobas Amplicor HIV-1 Monitor assay, version 1.5 (Roche Diagnostics, Rotkreutz, Switzerland).
When comparing a continuous variable in two groups, a t-test was used. If the distribution of a continuous variable was markedly skewed, the comparison was performed with the Mann-Whitney test. When comparing a dichotomous variable, Fisher's exact test was used. Odds ratios (ORs) were used to describe the relationship between lipoatrophy and possible risk factors: age>42 years, male gender, white race, AIDS diagnosis, HIV wasting syndrome, HIV RNA level>10 000 HIV-1 RNA copies/mL, current CD4 count<100 cells/muL, nadir CD4 count<160 cells/muL, weight loss>7 kg compared to normal weight, BMI<23 kg/m2 and duration of HIV infection>80 months (estimated by the interval between the first serological test positive for HIV and the examination day of the study). Continuous variables, apart from viral load, CD4 count and weight loss, were dichotomized by the median values of the cohort. All the individual ART medicaments [current and/or previous use (ever)] were also included in the analysis. Logistic regression analysis was used to determine the association between the possible risk factors and lipoatrophy.
In the follow-up study, related samples tests were used when comparing variables from first and second visits: the Wilcoxon signed-rank test for continuous variables and the McNemar test for binary variables. When anthropometrical measurements were analysed, one single outlier with 13 kg weight loss because of depression was excluded. Odds ratios were used to describe the relationship between atrophy reversal (lipoatrophy at the first but not the second visit) and possible risk factors: use of stavudine <=36 months at the second visit, weight gain>0.63 kg and CD4 count increase>170 cells/muL. The level of significance used was 5%. Most results are given as mean±standard deviation (SD). If markedly skewed values are given as median (interquartile range).