Fat Distribution in Relation to Drug Use, Human Immunodeficiency Virus (HIV) Status, and the Use of Antiretroviral Therapies in Hispanic Patients with HIV Infection
- Department of Family Medicine and Community Health, Tufts University School of Medicine, Boston, Massachusetts
- J. E. Forrester, Dept. of Family Medicine and Community Health, Tufts University School of Medicine, 136 Harrison Ave. (Stearns 203A), Boston, MA 02111 (janet.forrester{at}tufts.edu).
Abstract
Human immunodeficiency virus (HIV)—associated fat-redistribution syndrome is still a subject of controversy. There is, as yet, little agreement on the definition, etiology, and prevalence of the syndrome. Many studies have examined medication or disease-related factors. Fewer studies have examined patient-related factors. Illicit drug use is an important risk factor for HIV infection, yet the role of drug use in fat distribution has not been well described. We examined fat distribution, measured by dual energy x-ray absorptiometry, in relation to drug use, smoking, and alcohol use in Hispanic patients with HIV infection and control group of HIV-negative drug users. Our results suggest that neither drug use nor alcohol consumption are predictors of fat distribution. However, among men, smoking was independently associated with less total fat, less trunk fat, and more appendicular fat. The role of patient-specific factors in the etiology of HIV-associated fatredistribution syndrome warrants further investigation.
Although there is, as yet, no agreed-on definition of HIV-associated fat redistribution (or lipodystrophy) syndrome, there is general agreement that this syndrome may include ⩾1 features, including the accumulation of fat in the visceral area of the abdomen, a loss of subcutaneous fat from the limbs and face (lipoatrophy), and, less commonly, increases in dorsocervical fat and the enlargement of breasts in women [1–11]. Also observed are changes in the serum lipid profile, with increases in triglyceride levels, very low density lipoprotein and low-density lipoprotein cholesterol, decreases in high-density lipoprotein cholesterol, and impaired glucose tolerance [1, 2, 4, 5, 7–13].
The causes of fat redistribution are not clear, although accumulating evidence points to multiple risk factors, including specific antiretroviral medications [1, 5, 9, 10] as well as HIV infection per se [7, 9, 11, 14]. There are few data on the patient-specific factors that may influence the development of lipodystrophy and dyslipidemia. Most studies have combined data from persons of different races and HIV transmission groups, to focus on treatment- and disease-related risk factors. However, differences among studies in the sorts of HIV-positive patients involved may explain, in part, the conflicting data on the relative prevalence of lipodystrophy syndrome and the factors that predict its occurrence.
Illicit drug use is an important risk factor for HIV infection. Previous studies have suggested that drug use may contribute to malnutrition and weight loss in HIV-positive persons [15–17]. Drug use is associated with the use of other harmful substances, such as smoking and alcohol consumption. At present, there is little understanding of the interactions among these behavioral factors, antiretroviral treatment, and the presentation of fat-redistribution syndrome. Here we present cross-sectional data on body composition, including regional body fat distribution as measured by dual energy x-ray absorptiometry (DXA), in relation to current drug use, smoking, the consumption of alcohol, and the use of antiretroviral therapies.
Subjects and Methods
Study design. The BIENESTAR study is a prospective cohort study of the relationship between drug use and nutritional status in Hispanic patients infected with HIV. The cohort is composed of 3 groups: HIV-positive current drug users, HIV-positive persons who do not use drugs ("non–drug users"), and HIV-negative current drug users. Study recruitment began in August 1999. Potential participants were recruited through HIV organizations and clinics, needle-exchange programs, homeless shelters, and street outreach. Adults aged ⩾18 years who identified themselves as Hispanic, with Spanish as their first language, were invited to participate in the study. Participants were excluded if they fulfilled any of the following criteria: persons with plans to leave Boston within 3 years of enrollment, homeless persons who were unable to provide the telephone numbers of at least 3 contacts, women who were pregnant at the time of recruitment, the presence of a non–HIV-related malignancy, and persons who refused to sign permission to have medical records released. Two individuals who had a history of hormone treatments for sex reassignment were also excluded from the study.
After screening to determine eligibility, participants were seen in the General Clinical Research Center of the New England Medical Center for baseline and semiannual follow-up visits. All visits were conducted in Spanish by Hispanic personnel. At each visit, data were collected, by standardized interview, on sociodemographic characteristics, medical history, current medication use, and a complete history of current and past illicit drug use. A participant was classified as a current drug user if he or she reported using heroin, cocaine, methadone, or amphetamines by any route within the preceeding 6 months. Persons who reported past drug use (last use prior to the 6 months preceding the baseline visit) were classified as non–drug users for the purposes of the present analysis. The frequency of drug use was defined as the highest frequency reported among all the types of drugs used by any individual participant. Marijuana was not considered to be an illicit drug in the present study because it is often used as an appetite stimulant among persons with HIV infection. We used ELISA for HIV antibodies to confirm HIV infection.
Body-composition measures. Body composition was measured by DXA at baseline and once per year thereafter. Transverse whole-body scans were obtained using QDR2000 scanner (Hologic) in the array mode. DXA phantoms were scanned daily, to minimize instrument drift, as recommended by the manufacturer. DXA was conducted with the subject in a supine position, wearing a hospital gown, and after having emptied his or her bladder. Data were collected in 0.120-pixel elements per transverse scan, with a pixel size of 5 × 10 mm. Percentage fat and lean mass were derived for the whole body and for regions of interest using whole-body software (version 7.10A), which was provided by the manufacturer (Hologic). A series of DXA cutoff lines positioned at anatomical markers defined the regions of interest. The trunk was isolated to include the neck, chest, abdomen, and the central portion of the pelvis. Longitudinal cutoff lines positioned between the head of the humerus and the scapula at the glenoid fossa isolated the arms (to include deltoid muscles). A horizontal line positioned at the iliac crest defined the top of the pelvis, with diagonal cutoff lines below the pelvis to bisect both femoral necks and to isolate the legs (to include the upper thighs and the lateral portion of the gluteus maximus). Appendicular composition was analyzed by summing the regional data from 4 extremities. DXA measurements were not collected until May 2000. Therefore, not all study participants had DXA measurements done at their baseline visit. For the purposes of the present analysis, data were included from the first DXA measurement for every participant who had at least 1 DXA measurement done within the first 2 years of study followup.
We did not have enough data to look at individual medications. Therefore, HIV therapy regimens were divided into categories, and participants were classified as taking or not taking medications within each of the categories. The categories of HIV medications used in the analyses were nucleoside reverse-transciptase inhibitors (NRTIs), nonnucleoside reverse-transciptase inhibitors (NNRTIs), protease inhibitors (PIs), or HAART. The definition of HAART used in the present analysis was any of the following combination of therapies: (1) ⩾2 PIs, NNRTIs, or ⩾3 NRTIs; (2) 2 NRTIs plus 1 PI; (3) at least 1 PI plus at least 1 NNRTI plus another PI, NRTI, or NNRTI; or (4) at least 1 NNRTI plus at least 2 NRTIs.
Statistical analyses. The data included in these analyses were those available from the initiation of the study in August 1999 until 21 May 2002. Data from men and women were analyzed separately and stratified by study group. Because of the small numbers of some groups, the data were slightly skewed in distribution; thus, the median and 25th and 75th percentiles were used for some data in table 1 and for all data in tables 2 and 3. Statistical analyses were done on triglyceride values transformed on a log scale. For this reason, average triglyceride values are not reported in table 4.
Table 1
Baseline characteristics of the BIENESTAR cohort.
Table 2
Dual energy x-ray absorptiometry measurements, in men, of regional body composition in relation to HIV infection, drug use, and HAART.
Table 3
Dual energy x-ray absorptiometry measurements, in women, of regional body composition in relation to HIV infection, drug use, and HAART.
Table 4
Adjusted mean trunk and appendicular fat in patients.
The analyses were done in 2 ways. In the first approach, the 3 study groups were further stratified into 5 groups on the basis of HAART: (1) HIV-positive drug users receiving HAART, (2) HIV-positive drug users not receiving HAART, (3) HIV-positive non-drug users receiving HAART, (4) HIV-positive non–drug users not receiving HAART, and (5) HIV-negative drug users. Total and regional body composition differences among the groups were examined, while controlling for potentially confounding variables using analysis of covariance. Because “treatment with HAART” does not distinguish between regimens that include PIs and those that do not, in the second approach to the analyses, each person was categorized as receiving NRTI, NNRTI, or PI therapies. These medication categories were then included in regression models as 3 separate binary variables (yes/no), to examine drug use, alcohol consumption, and smoking as independent predictors of differences in trunk fat or appendicular fat, while controlling for the effects of these specific categories of HIV therapies. The HIV-negative group was set aside in this second analysis, because that group did not have a category for medication use. All differences were tested for significance using procGLM in SAS software (version 8.0; SAS Institute), in which the contrast groups were treated as class variables or coded as indicator groups. All analyses were adjusted for age. Analyses in which the outcome variable was either total lean mass or total fat mass were adjusted for height. Analyses in which the outcome was a measurement of regional fat were adjusted for total fat. Analyses in which the outcome was regional lean were adjusted for total lean. Length of time with HIV infection, lowest reported CD4 cell counts, and current CD4 cell counts were also assessed as potential confounders. We did not have information on virus load for these analyses. Results were considered statistically significant if the 2-tailed P value was <.05. The protocol was approved by the Human Investigations Review Committee of Tufts University, and written informed consent was obtained from each participant.
Results
There were 235 participants enrolled in the study as of 21 May 2002: 183 men and 52 women. The baseline characteristics of first 235 participants, stratified by study group, are shown in table 1.
Among the men, the HIV-negative drug users were significantly younger than the HIV-positive drug users (P < .05). The large difference in rates of homelessness between those who were HIV-positive and HIV-negative was a result of how participants were recruited into the study: many HIV-positive men and women were recruited at community-based organizations that work with HIV-positive persons, whereas HIV-negative participants were recruited mainly from street outreach or shelters for the homeless. The majority of the participants in all groups, both HIV-positive and -negative, had some sort of publicly funded medical insurance (Medicare or Medicaid).
The drug users reported the frequencies of the various types of drugs used as follows: heroin or methadone alone, 44%; cocaine alone, 13%; and some combination of opiates and cocaine, 42%. Fewer than 2% of the drug users reported using amphetamines, and this was always in combination with opiates or cocaine. Seventy-nine percent of the drug users reported using drugs at least twice per week.
Among the men, 72% were current smokers, and 59% reported drinking alcohol. Rates of smoking and drinking were lower among women. Fifty-four percent of the women smoked, and 39% reported current consumption of alcohol.
DXA data were available for 161 (88%) of 183 men and for 46 (88%) of 52 women. Data on total and regional body composition, stratified by sex, drug use, HIV status, and use of HAART, are shown in tables 2 (men) and 3 (women). There were no significant between-group differences in body mass index, percentage body fat, total fat, total lean, trunk lean, or appendicular lean in the men (table 2).
When the regional distribution of fat was examined, it was found that, after adjustment for differences in age and total fat, both the HIV-positive drug users and HIV-positive non–drug users who were receiving HAART had higher adjusted levels mean trunk fat than those not receiving HAART (table 4). HIV-positive drug users who were receiving HAART had 1.3 kg more trunk fat than HIV-positive drug users not receiving HAART (P = .0003) or HIV-negative drug users (P < .0001); HIV-positive non–drug users receiving HAART had 1.0 kg (P = .003) more trunk fat than HIV-positive drug users not receiving HAART and 1.0 kg more trunk fat than HIV-negative drug users (P = .0002). The HIV-positive drug users not receiving HAART had adjusted mean levels of trunk fat that were more similar to the HIV-negative drug users than to any of the other HIV-positive groups (7.9 kg trunk fat in both groups). However, the 0.5-kg difference in trunk fat between the HIV-positive drug users receiving HAART and the non–drug users not receiving HAART did not reach statistical significance (P = .26).
There were also significant between-group differences in appendicular fat when the DXA results were adjusted for age and total fat (table 4). The HIV-positive groups receiving HAART had lower levels of appendicular fat compared with the other 3 groups. HIV-positive drug users who were receiving HAART had 1.3 kg less appendicular fat than HIV-positive drug users not receiving HAART (P = .0003) and 1.3 kg less appendicular fat than HIV-negative drug users (P < .0001). HIV-positive non–drug users receiving HAART had 1.0 kg (P = .003) less appendicular fat than HIV-positive drug users not receiving HAART and 1.0 kg less appendicular fat than HIV-negative drug users (P = .0003). Again, the HIV-positive drug users not receiving HAART had adjusted mean levels of appendicular fat that were more similar to the HIV-negative drug users than to any of the other HIV-positive groups (8.5 kg appendicular fat in both groups).
The levels of serum triglycerides in the groups, analyzed on the log scale, mirrored the patterns of fat distribution (data not shown). Compared with the HIV-negative drug users, both the HIV-positive drug users receiving HAART and HIV-positive non–drug users receiving HAART had higher log triglyceride values (P = .0004 and P = .0006, respectively). The point estimates for the log-transformed triglyceride levels of the HIV-positive drug users not receiving HAART and the HIV-negative drug users were lower than the estimates for the other 3 HIV-positive groups. Controlling for length of time with HIV infection, lowest reported CD4 cell count, or current CD4 cell count did not alter the conclusions. The number of women was too small to conduct meaningful analyses on differences in regional fat; however, the pattern of trunk and appendicular fat in relation to drug use and HAART was different in the women than in the men (table 4).
In the second analysis of the data from men, medication use was redefined as the use of 3 specific types of treatment: use of NRTIs, NNRTIs, or PIs. This was done to remove the possible confounding effects of the specific components of HAART. Trunk fat, appendicular fat, and triglycerides were examined as functions of drug use, smoking, and alcohol consumption, while controlling age, total fat, and the use of NRTIs, NNRTIs, or PIs. The results of these regression analyses indicated that drug use was not a significant independent predictor of trunk fat, appendicular fat, or triglyceride levels (table 5), nor was intravenous drug use a predictor of these outcomes (data not shown). Current smoking was, however, a predictor of fat distribution in the men. Current smokers had 0.62 kg less trunk fat (P = .02) than nonsmokers when potentially confounding variables, including total fat and specific HIV therapies, were included. Smoking was also associated with more appendicular fat (0.64 kg, P = .02) in the men. Alcohol was not a significant predictor of any of the outcomes. Length of time with HIV infection, lowest reported CD4 cell counts, and current CD4 cell counts were neither predictors of the outcomes nor confounders of the relations examined. Among the women, smoking was of borderline significance (P = .05) in predicting higher triglyceride levels.
Table 5
Adjusted mean trunk and appendicular fat in patients.
Discussion
Our data suggest that, once differences in the use of HIV medications and total fat are accounted for, drug use is not a predictor of regional fat distribution. The data do support a role for smoking as a predictor of fat distribution. Smoking was associated with less trunk fat and more appendicular fat. Alcohol consumption was not a significant explanatory factor for differences in trunk fat, appendicular fat, or triglyceride levels among the men.
Our finding of a negative association between smoking and trunk fat contradicts previous data on the role of smoking as a determinant of central obesity; several previous studies have shown a positive association between smoking and central obesity [18–23]. There are several plausible explanations for our contrary findings. Previous studies have generally used skin-fold thickness, waist circumference, or the waist : hip ratio (WHR) as a measurement of central obesity. DXA, the method used in the present study, is a better measurement of regional body composition than skin-fold thickness, waist measurements, or WHR. Also, previous studies may have inadequately controlled for differences in total fat between smokers and nonsmokers by using measurements such as skin-fold thickness at an alternate site or body-mass index as surrogates for total body fat. Our analyses differed from these studies in that total fat was measured using DXA. Thus, the amount of trunk fat was examined while adjusting all subjects' data to the same average total fat. A final explanation for our contrary findings is that the effect of smoking on trunk fat differs between persons who are HIV positive and those who are HIV negative. Smoking was not a significant predictor of trunk fat in the 78 HIV-negative men, but because 67 (86%) of 78 were smokers, we had limited statistical power to detect a difference. However, male and female smokers had less fat overall than nonsmokers in our study (-3.3 ± 1.6 kg, P = .05).
Our study has several strengths over previous studies of the predictors of HIV-associated fat distribution. First, DXA provides an objective and unbiased measurement of body composition and regional fat distribution that is not subject to the reporting bias possible when fat redistribution is patient or physician reported. Another strength of our study was our detailed measurement of current drug use. Many other studies have relied on HIV transmission categories to identify drug users. However, transmission category may not describe current drug use. Many of the non–drug users in our study, especially the men, had a history of past drug use; thus, our results relate to current drug use only.
Our study also had a number of weaknesses, including the cross-sectional design of the analysis. We did not have a complete HIV medication history, and what information we had was based on self-report. We also lacked statistical power in some groups among the men and overall in the women. Finally, DXA describes trunk fat but not visceral fat. Visceral fat is the compartment of primary concern in fat-redistribution syndrome. However, if there were important between-group differences in the amount of fat located in the trunk region in relation to drug use, these likely would have been detectable in our analyses.
In summary, our data suggest that patient-specific factors may play a role in the development of lipodystrophy. Specifically, the data suggest that, although drug use appears not to be an important factor in fat distribution, smoking may inhibit the development of truncal obesity and the loss of appendicular fat. These and other patient-specific factors warrant further investigation, because they may explain differences among studies in the prevalence and risk factors for fat redistribution in HIV infection.
Acknowledgements
We thank the participants for their time and dedication. We also thank the research team at BIENESTAR for their diligence.
Footnotes
-
Financial support: National Institute on Drug Abuse (grants DA 11598 and DA14501); Center for AIDS Research (grant 1-P308142853); National Institute of Diabetes and Digestive and Kidney Diseases (grant DK4 5734–07). The General Research Center of the New England Medical Center, Boston, is supported by the Division of Research Resources of the National Institutes of Health (grant M01-RR00054).
- © 2003 by the Infectious Diseases Society of America
References
This Article
- AbstractFree
- » Full Text (HTML)Free
- Full Text (PDF)Free
Classifications
Services
Citing Articles
Google Scholar
PubMed
Related Content
Share
- Email this article
-
CiteULike
Connotea
Delicious
Facebook
Google+
Mendeley
Twitter
Published on behalf of
IDSA Members: For your free access to this journal, log in via the IDSA members area.
The Journal
Impact Factor: 8.186
5-Yr impact factor: 7.898
For The Media
For Authors
- Services for authors
- Self-archiving Policy
- IDSA Conflict of Interest policy
- Instructions for Authors
- ICMJE Form
- Submit Now!
Open access options for authors visit Oxford Open
This journal enables compliance with the NIH Public Access Policy
Corporate Services
Alerting Services
Related IDSA Publications
