Between June 2000 and September 2002, 1183 HIV-infected persons and 297 controls were enrolled in the study of Fat Redistribution and Metabolic Change in HIV Infection (FRAM) study. The FRAM study was cross-sectional. One of the primary aims of the FRAM study was to understand the association of regional adipose tissue distribution in HIV infection with metabolic outcomes, including serum amino-transferases; ALT has been suggested as a marker of steatosis in the general population. HIV-infected participants were selected from randomly ordered coded lists of patients seen in 16 HIVor infectious disease clinics or cohorts during 1999. Of the 1183 HIV-infected participants, 30% (n = 350) were women. Control subjects were recruited from 2 centers of the Coronary Artery Risk Development in Young Adults (CARDIA) study.20,21
CARDIA subjects were originally recruited as a sample of healthy 18- to 30-year-old white and African American men and women from 4 cities in 1985 to 1986 for a longitudinal study of cardiovascular risk factors, with population-based recruitment in 3 cities and recruitment from the membership of a prepaid health care program in the fourth city. Participants in the CARDIA study were stratified for the 2 races and genders in each center. The recruitment and data collection procedures for the entire cohort have been described elsewhere.22
HIV-infected participants in FRAM study were nationally representative of HIV-infected patients in care,22
and control participants were representative of the general population.20
Institutional review boards at all participating sites approved the study protocol and consent process.
FRAM study participants were asked about their physical activity, alcohol intake, smoking, illicit drug use, and adequacy of food intake using standardized instruments.23–26
Medical history was also assessed. Research associates interviewed HIV-infected participants and reviewed medical charts to determine the dates of use of individual ARV medications.
Height and weight were measured. Using standardized protocols, body composition was measured using regional anthropometry, including waist and hip circumferences, and by MRI. MRI scans were segmented using image analysis software (Tomovision Inc., Montreal, Quebec, Canada).16,27,28
The volume of each tissue for the space between 2 consecutive slices was calculated by means of a mathematic algorithm.29
Using these methods, we quantified adipose tissue volume in the leg, lower trunk (abdomen and back), upper trunk (chest and back), arm, and abdominal viscera.
Blood was drawn and sent to a central laboratory (Covance) for determination of CD4 cell counts and HIV RNA levels by polymerase chain reaction (PCR) in HIV-infected participants and for determination of AST and ALT levels in HIV-infected and control participants For women, the upper limit of normal (ULN) for ALT and AST was 34 U/L, and for men, the ULN was 43 U/L for ALT and 34 U/L for AST, based on age-specific normal ranges provided by Covance laboratory. Stored serum samples were tested centrally for HCV RNA level by branched DNA (bDNA) using the Bayer Versant HCV RNA 3.0 assay (Bayer HealthCare-Diagnostics, Tarrytown, NY) and for hepatitis B surface antigenemia using the Auszyme monoclonal enzyme immunoassay (Abbott Laboratories, Abbott Park, IL) on all FRAM study participants.
Among the FRAM study HIV-infected participants, 1149 had available ALT or AST measurements and known HCV status (by HCV RNA). An additional 30 participants with an OI or malignancy within the same or previous calendar month as the examination were excluded because they may have had acute changes in fat. Among the control participants, 292 had available ALT or AST measurements; those with known HCV infection (n = 5) or HIV infection (n = 3) were excluded. Therefore, 1119 HIV-infected and 284 control participants were included in the analysis.
ALT and AST levels were compared between HIV/HCV-coinfected, HIV-monoinfected, and control participants using the Mann-Whitney U test.
Multivariable linear regression analysis was first conducted using stepwise regression to examine the association of HIV status with ALT and AST after adjustment for potential confounding factors. This analysis was restricted to those between the ages of 33 and 45 years (n = 594 HIV-infected participants), because the control population did not include participants outside this age range. Gender, age, ethnicity, HIV status, and HCV status (by HCV RNA level) were forced to be included in the model, and other candidates were included with a criterion of P
≤ 0.05 for entry and retention. Interactions between gender, HIV/HCV status, and other factors in the model were assessed and included in models if they had a P
value <0.05. Because we saw substantially stronger effects of HIV/HCV coinfection on ALT and AST levels compared with HIV monoinfection, HIV/HCV-coinfected and HIV-monoinfected participants were compared separately with controls. Because of their skewed distribution, ALT and AST were log-transformed in all linear regression analyses; results were back-transformed to produce estimated percentage effects of each factor. Confidence intervals (CIs) were determined using the bias-corrected accelerated bootstrap method,29
with the P
value defined as that minus the highest confidence level that still excluded 0; this was necessary because the error residuals seemed to be non-Gaussian.
Non–HIV-related candidate variables tested in the multivariable models included MRI measurements of adipose tissue volume from 5 anatomic sites (plus total SAT and total fat), diabetes (defined as fasting glucose ≥126 mg/dL or hypoglycemic medication use), current lipid-lowering medication use, tobacco use, alcohol intake, adequate food intake, level of physical activity, and current illicit drug use (marijuana, speed, heroin, crack, cocaine [which is distinguished from crack in our self-administered questionnaire when participants are asked whether they have used other forms of cocaine that are not crack, including powder, freebase, and coca paste], and combination use of crack and cocaine). The linearity assumption for continuous predictors was also tested. The 5 adipose tissue sites considered were visceral, lower trunk, upper trunk, arm, and leg. Measurements were normalized by dividing by height-squared, analogous to body mass index (BMI), and summaries were back-transformed to 1.75 m of height. Because of their skewed distribution, the adipose tissue measures were also log-transformed.
A second set of multivariable regression analyses were stratified by HIV and HCV status to examine the associations of adipose tissue depots with ALT and AST. These models controlled for gender, age, and ethnicity, with other candidates tested as described previously.
Finally, multivariable regression analyses were performed only among HIV-infected subjects to determine the factors independently associated with ALT and AST levels in HIV infection. In addition to the predictors listed previously, these models included HIV RNA level (log10
) and CD4 cell count (log2
) at the time of study visit as well as hepatitis B surface antigen (HBsAg) status. Other candidates related to HIV-infection included AIDS diagnosis by a CD4 count <200 cells/μL or OI, HIV duration (from self-reported date of HIV diagnosis), days since last OI, and HIV risk factors. In multivariable models controlling for the previous factors that were independent predictors, we evaluated current use and duration of each individual ARV drug and ARV class: nucleoside reverse transcriptase inhibitor (NRTI; which included stavudine, zidovudine, lamivudine, didanosine, zalcitabine, and abacavir), nonnucleoside reverse transcriptase inhibitor (NNRTI; which included efavirenz, nevirapine, and delavirdine), protease inhibitor (PI; which included indinavir, lopinavir/ritonavir, nelfinavir, amprenavir, saquinavir, and ritonavir), and highly active antiretroviral therapy (HAART), as previously defined.15
We also performed exploratory analyses to examine the association of BMI and anthropometry measures (ie, waist and hip circumference) in place of MRI-measured regional adipose tissue.
All analyses were conducted using the SAS system, version 9.1 (SAS Institute Inc., Cary, NC).