| |
Prevalence of, Evolution of, and Risk Factors for Fat Atrophy and Fat Deposition in a Cohort of HIV-Infected Men and Women
|
| |
| |
Clinical Infectious Diseases June 15 2005;40:1837-1845
Denise L. Jacobson,1 Tamsin Knox,1 Donna Spiegelman,2 Sally Skinner,1 Sherwood Gorbach,1,3 and Christine Wanke1,3
1Department of Public Health and Family Medicine, Tufts University School of Medicine, 2Departments of Epidemiology and Biostatistics, Harvard School of Public Health, and 3Department of Medicine, Tufts-New England Medical Center, Boston, Massachusetts
ABSTRACT
Background. At present, no uniform definition of human immunodeficiency virus (HIV)-associated lipoatrophy exists. The risk factors for fat atrophy (FA) and central fat deposition (FD) are multifactorial. We assessed the evolution and predictors of FA and FD in HIV-infected men and women.
Methods. Participants (n = 452) were evaluated at baseline (starting in November 1998) and 1 year later. FA was defined as triceps skin-fold measurement less than the 10th percentile on the National Health and Nutrition Examination Survey for sex and age. FD was defined as a waist-to-hip ratio of >0.95 for men and of >0.85 for women. Predictors of the baseline prevalence of FA and FD and new cases of each syndrome after 1 year were determined.
Results. The baseline prevalences of FA, FD, and combined FA and FD were 35%, 44%, and 14%, respectively. Twenty-two percent of subjects had newly developed FA at 1 year, and 16% of subjects with FA at baseline did not have it at 1 year. Also, 23% of subjects had newly developed FD at 1 year, and 15% of those with FD at baseline did not have it at 1 year. The risk of developing new FA was increased among participants with low triceps skin-fold values (P < .001), smaller hips (P < .001), higher nadir HIV load (P = .006), abacavir use (P < .001), stavudine use (P < .001), and use of highly active antiretroviral therapy (P = .002). The risk of developing new FD was higher among women (P < .001) and among participants with greater body fat levels (P = .005) and higher triglyceride levels (P < .001), and it was lower among those with a high school education (P = .003) and higher triceps skin-fold values (P = .026).
Conclusions. FA and FD are common in HIV-infected patients, but may change over time in the individual. FA and FD appear to be different syndromes, because risk factors for the development differ, and the prevalence of the combined syndrome differs from the prevalences of the 2 independent syndromes.
AUTHOR DISCUSSION
In contrast with previous reports, our study used objective measurements based on population norms to determine prevalence, cumulative incidence, loss of syndrome, and risk factors for fat atrophy and fat deposition over a 1-year period in a diverse cohort of HIV-infected men and women receiving different treatments in a real-world setting. We believe that our definitions are clinically valid, because participants who developed fat atrophy had subcutaneous fat loss at all skin-fold sites measured. They also lost more abdominal and hip girth, compared with subjects who did not develop fat atrophy. Those who developed fat deposition did not have marked changes in subcutaneous fat (as determined by skin-folds), but they did have changes in body circumferences.
Fat atrophy and fat deposition were common in our cohort, and we believe that they are distinct syndromes. First, the proportion of participants with mixed FAD syndrome actually equaled the probability of having 2 independent syndromes. Moreover, in our cohort, risk factors for fat atrophy differed from those for fat deposition, as in the literature. Development of fat atrophy was more likely to occur in participants with less fat at baseline (i.e., those with low triceps skin-fold measurements and smaller hips), increased disease severity (those with a higher nadir viral load), and stavudine and abacavir use. Continuous HAART use may indicate a longer duration or greater severity of HIV disease. The risk of incident fat deposition was higher in subjects with more body fat at baseline, including for those with more subcutaneous body fat, for women, for those with a high school education, and for those with a higher triglyceride level.
In this study, the prevalences of fat atrophy (35%) and fat deposition (44%) were in the upper range of those reported in the literature. Other studies reported that 15%-36% of patients had fat atrophy [1, 3, 5, 6, 8, 9, 12-16, 19, 21, 24] and 4%-33% of patients had fat deposition [1, 5, 6, 8, 9, 12-16, 19, 20, 24]. Only 1 study of PI recipients reported higher estimated prevalences [6]. Other studies did not separate fat atrophy from fat deposition.
Few studies of the incidence of lipodystrophy have examined fat atrophy and fat deposition separately. The 1-year incidences of fat atrophy (22%) and fat deposition (23%) in our study were in the middle of the range of estimates from 4 incidence studies [2, 4, 7, 20]. Our estimates may differ because we studied an ethnically diverse group of men and women who were or were not receiving different antiretroviral therapies and who may have been healthier and more overweight than were the subjects in other reports and because we used objective, anthropometric definitions of each syndrome.
Few studies have used quantitative measurements to define body-shape changes in HIV-infected patients [3, 5, 16, 22, 36] or as covariates [5]. Most investigators used patients' self-reported findings and/or findings from physical examinations [1-21, 37]. Our inexpensive, quantitative definitions [26] could be used clinically. Triceps skin-fold values in less than the 10th percentile of the NHANES values were used to define fat atrophy. Subcutaneous fat atrophy is characterized by a loss of extremity fat on dual-energy X-ray absorptiometry (DXA) [38]; however, DXA cannot distinguish visceral trunk fat from subcutaneous trunk fat. Subcutaneous fat atrophy can also be quantified by CT [15]. Triceps skin-fold values correlated highly with extremity fat noted on CT scans [15]. In our cohort, triceps skin-fold values had a correlation of 0.77 (P < .001) with extremity fat noted by DXA, and skin-fold values were correlated at all sites (range, 0.42-0.84), suggesting that our definition does indeed detect generalized subcutaneous fat atrophy. Waist-to-hip ratio has been used in HIV-associated lipodystrophy studies [15], and both waist-to-hip ratio and waist circumference are significant predictors of cardiovascular disease and other outcomes [27]. HIV-associated visceral fat deposition is characterized by increased intra-abdominal fat noted on CT scans. Abdominal circumference correlates well with intra-abdominal fat deposition [15] and may be more specific than waist-to-hip ratio, which may also reflect loss of hip fat, gain in waist circumference, or both. However, in our cohort, subjects who developed high waist-to-hip ratios did not have a decrease in buttock fat, as estimated by hip size. High waist-to-hip ratios may indicate less-severe central fat deposition than does waist circumference alone. In our data, only 4 of 51 people developed high waist circumferences alone, whereas 17 of 51 developed both high waist circumference and high waist-to-hip ratio, and 30 of 51 developed a high waist-to-hip ratio alone. Patients who developed high waist circumferences had increases in skin-fold values at all sites, whereas those who developed high waist-to-hip ratios did not. High waist circumferences may indicate obesity better than does a high waist-to-hip ratio.
We determined host-, virus-, and treatment-related risk factors for body-shape changes: large triceps, an indication of fat stores, was protective against development of fat atrophy. Both disease severity (i.e., high nadir viral load) and treatment were predictive of increased risk of fat atrophy. We also observed a strong association between NRTI use-particularly stavudine use-(i.e., its impact on mtDNA) and fat atrophy. Abacavir use was associated with development of fat atrophy, which has not been previously reported, and may reflect an impact on mtDNA. Lifetime duration of HAART use may indicate a longer period of HIV infection and/or increased severity of disease. In other studies, heterosexual transmission of virus, increased age, male sex, white race, non-injection drug use-related transmission, lower CD4+ cell count, lower BMI, AIDS, longer duration of therapy, use of stavudine, use of any NRTI, and hepatitis C predicted development of fat atrophy [2, 4, 7, 20].
Among subjects without either fat atrophy or fat deposition at baseline, the risk factors for new development of fat deposition differed from those that predicted new fat atrophy, except for triceps skin-fold value, which predicted both. Host factors that predicted development of fat deposition (sex, education level, higher body fat level, and higher triglyceride level) may reflect poor dietary quality. In a case-control study involving our cohort, lower dietary fiber intake was associated with development of fat deposition [22]. Martinez et al. [7] showed that greater increases in the CD4+ cell count predicted development of any type of lipodystrophy. Others have found that female sex increased risk and injection drug use decreased risk [2].
We carefully chose measurements of fat atrophy and fat deposition based on population standards and studies. One study found that skin-fold measurements had lower reliability than did waist measurements, especially among overweight/obese people [39]. To obtain good reliability, an experienced nutritionist recertified all interviewers semiannually. We averaged 3 measurements within 3 mm for each skin-fold measurement, and the intraclass correlations were 0.99. Thus, with standardized training, anthropometrics could be used clinically. Because overweight or obese people were at lower risk of developing fat atrophy and had higher triceps skin-fold values, it is unlikely that they were misclassified as having fat atrophy. A limitation of the study was the lack of collection of thigh and neck measurements.
Our data suggest that fat atrophy and fat deposition are 2 unique syndromes that occur more frequently as independent syndromes than together. This agrees with the Fat Redistribution and Metabolic Change in HIV Infection study, which suggests that fat atrophy is unique to HIV infection but that fat deposition occurs with equal or greater frequency in HIV-uninfected persons [40]. Important avenues of future study include adherence to antiretroviral treatment and body-shape changes, as well as the consequences of body-shape changes.
Fat atrophy and fat deposition are not static processes, and an individual does not develop either syndrome at the time of antiretroviral therapy initiation and simply maintain it. In our study, both syndromes evolved over a 1-year period, as in the HIV Outpatient Study [4]. Body-shape changes-and, potentially, their metabolic and clinical outcomes-are dynamic processes deserving additional longitudinal study.
RESULTS
Baseline Findings
Baseline characteristics of subjects are shown in table 1, and baseline prevalences of fat atrophy, fat deposition, and mixed FAD are shown in table 2. Among the 117 men and 37 women with fat atrophy, 10% and 32% had a BMI <20, respectively. Twenty-three percent of men with fat deposition and 35% women with fat deposition had a BMI of >30.
Incidence and Evolution of Body-Shape Changes over a 1-Year Period
The evolutions of fat atrophy and fat deposition over a 1-year period are shown in figures 1 and 2, respectively. More men than women had newly developed fat atrophy (P = .020), and more women than men had newly developed fat deposition (P = .013). Among those who developed fat atrophy, 11% had a BMI of <20, and among those who developed fat deposition, 21% had a BMI of >30. At baseline, among 33 men and 12 women who developed mixed FAD syndrome, 55% and 42% had fat atrophy, respectively, and 30% and 33% had fat deposition, respectively. Among 44 men and 19 women with mixed FAD syndrome at baseline, 12 men (27%) and 7 women (37%) no longer had it after 1 year.
Determinants of Prevalence
In multivariate analysis, the baseline prevalence of fat atrophy (data not shown) was 55% lower for subjects in the highest tertile of subscapular skin-fold value, compared with those in the middle and lowest tertiles (PR, 0.45; 95% CI, 0.27-0.75; P = .002). The prevalence was greater among those with higher albumin levels (88% of all participants had normal levels) (PR, 1.5 per g/dL; 95% CI, 1.1-2.1; P = .018) and longer continuous use of stavudine (Zerit; Bristol-Meyers Squibb) until baseline (PR, 1.009 per month of use; 95% CI, 1.002-1.017; P = .018). There was interaction between BMI and race (P = .005). Higher BMI was more protective of fat atrophy among white subjects (PR, 0.08 per 1-unit increase in BMI; 95% CI, 0.01-0.55; P = .011) than among nonwhite subjects (PR, 0.86 per 1-unit increase in BMI; 95% CI, 0.79-0.93; P < .001). (This model was adjusted for confounding due to changes in BMI since diagnosis of HIV infection.)
In multivariate analysis, prevalence of fat deposition was higher among subjects with a high percentage of body fat (PR, 2.0; 95% CI, 1.2-3.3; P = .005), and it was 2 times higher among those in the middle tertile of BMI than among those in the lowest tertile of BMI (PR, 2.0 per tertile; 95% CI, 1.6-2.5; P < .001) and 3.9 times higher among those in the highest tertile of BMI than among those in the lowest tertile of BMI. The prevalence of fat deposition increased with increasing age (PR, 1.03 per year; 95% CI, 1.02-1.05; P < .001) and was higher among white subjects (PR, 1.4; 95% CI, 1.1-1.8; P = .004). The prevalence decreased as hip circumference increased: it was 35% lower among subjects in the middle tertile of hip circumference, compared with those in the lowest tertile (PR, 0.65 per tertile; 95% CI, 0.52-0.83; P < .001), and it was 57% lower among those in the highest versus those in the lowest tertile. Higher intake of insoluble fiber was associated with a lower prevalence of fat deposition (PR, 0.96 per g; 95% CI, 0.94-0.99; P = .011). (This model was adjusted for confounding due to baseline triceps skin-fold value.)
Determinants of Incidence
Fat atrophy. In multivariate analysis, among subjects without fat atrophy or fat deposition at baseline, the risk of new development of fat atrophy was lower among those with larger triceps and larger hips. The risk was greater in subjects who had a higher nadir viral load, who received abacavir, who received stavudine, and who used HAART for a longer time. CD4+ cell count, peak viral load, triglyceride level, albumin level, other types and durations of therapy, testosterone use, marijuana use, race, sex, injection drug use, transmission category, education level, and poverty were not independent predictors.
In a separate model (data not shown) in which subjects with fat deposition at baseline were included, there was increased risk of new development of fat atrophy for white subjects, subjects with higher baseline alanine aminotransferase levels (64% of all participants had normal levels), abacavir use throughout the follow-up period, and any use of nucleoside reverse-transcriptase inhibitors (NRTI) at baseline. There was decreased risk among subjects who had a larger waist, had a higher baseline BMI, engaged in greater physical activity, and used efavirenz throughout the follow-up period.
Fat deposition. In multivariate analysis (table 3), the risk of newly developing fat deposition among those without fat atrophy or fat deposition at baseline was higher among female subjects, those with a greater percentage of body fat, and those with higher triglyceride levels. It was lower among those with a high school education. Viral load, CD4+ cell count, testosterone level, marijuana use, age, race, injection drug use, lack of stable housing, and dietary fiber intake were not significant independent predictors.
In a separate model (data not shown) in which subjects with fat atrophy at baseline were included, the risk of fat deposition was increased with age, greater increases in the CD4+ cell count during follow-up, receipt of nonnucleoside reverse-transcriptase inhibitor-based HAART at baseline, and weight gain during follow-up. The risk decreased with increasing duration of saquinavir use during follow-up. Among men only, subjects in the lowest tertile of subscapular skin-fold measurements had a lower risk of fat deposition. Men who were receiving PI-based HAART at baseline had a higher risk of developing fat deposition. For women, there was no association between receipt of PI-based HAART and fat deposition. For white subjects, abacavir use increased the risk of developing fat deposition. Among nonwhite subjects, there was no association between abacavir use and fat deposition.
Change in Anthropometrics for Each Syndrome
Compared with participants who did not develop fat atrophy at 1 year, those who developed fat atrophy had greater decreases in weight (-2.05 vs. +0.8 kg; P < .001), triceps skin-fold value (-3.0 vs. -1.0 mm; P < .001), suprailiac skin-fold value (-5.1 vs. -2.3 mm; P < .001), subscapular skin-fold value (-5.1 vs. -1.3 mm; P < .001), waist circumference (-1.4 vs. +0.6 cm; P < .001), and hip circumference (-2.0 vs. 0.0; P = .004). There was no difference in waist-to-hip ratios. Compared with subjects who continued to have fat atrophy at 1 year, those who no longer had fat atrophy had significant increases in weight (+0.3 vs. -0.6 kg; P = .005), triceps skin-fold value (+3.2 vs. -0.5 mm; P < .001), suprailiac skin-fold value (+2.3 vs. -0.8 mm; P = .004), and subscapular skin-fold value (+3.2 vs. -1.2 mm; P = .005) but no differences in waist circumference, hip circumference, or waist-to-hip ratio.
Subjects who developed fat deposition had significantly greater increases in waist-to-hip-ratio (+0.05 vs. +0.004; P < .001), waist circumference (+4.5 vs. -0.05 cm; P < .001), and weight (+0.95 vs. -0.55 kg; P = .014) than did participants who did not develop fat deposition. There were no differences in changes to hip circumference or skin-fold value. Subjects who no longer had fat deposition after 1 year had significantly greater decreases in waist-to-hip ratio (-0.05 vs. +0.01; P < .001), waist circumference (-2.3 vs.-0.15 cm; P = .023), and hip circumference (+2.1 vs. -0.75 cm; P = .003) than did those who maintained their fat deposition. These participants did not differ with regard to changes in skin-fold values or weight.
Comparison of Subjects with a High Waist-to-hip Ratio and/or a Large Waist
There were 217 participants at baseline without a high waist-to-hip ratio or high waist circumference. Among those who developed either of these conditions over the course of 1 year, 17 of 51 developed both, 30 of 51 developed only a high waist-to-hip ratio, and 4 of 51 developed only a high waist circumference (P < .001).
Compared with subjects who did not develop a high waist circumference, participants who developed a high waist circumference had significantly greater increases in waist-to-hip ratio (+0.04 vs. +0.01; P < .001), waist circumference (+8.0 vs. +0.5 cm; P < .001), hip circumference (+3.8 vs. -0.3 cm; P < .001), weight (+4.8 vs. -0.5 kg; P < .001), subscapular skin-fold value (+5.3 vs. -1.3 mm; P < .001), triceps skin-fold value (+1.3 vs. -0.8; P < .001), and suprailiac skin-fold value (+3.7 vs. -1.0; P < .001).
|
|
| |
| |
|
|
|
