Heart Positive (Clinicaltrials.gov ID NCT00246376) was a large multiethnic study of healthy patients with HIV infection on stable HAART[13]. The primary outcomes have been reported[13]. The protocol was approved by the Institutional Review Boards of Baylor College of Medicine and the two recruitment centers. Written informed consent was obtained (in Spanish or English) from all subjects prior to their participation.
Subjects
HIV patients with no history of diabetes were recruited from two HIV centers in Houston: Legacy Community Health Services and Thomas Street Clinic, Harris County Hospital District. The sample comprised 202 patients on stable HAART with hypertriglyceridemia who underwent complete screening for the Heart Positive study. Racial/ethnic distribution of the subjects included NHWs, African Americans, Hispanics, and Native Americans. Three Native American subjects were excluded from this analysis due to their small numbers. All subjects were hypertriglyceridemic (fasting plasma triglycerides >150 mg/dL), on no lipid-lowering or anti-diabetic medications, on stable HAART for at least 6 months, with stable CD4 count and viral load, normal kidney and thyroid functions, and transaminase (AST / ALT) levels ≤ 2 times upper limit of normal.
Measures
Demographic and clinical variables
Age, gender, ethnicity, number of years since HIV diagnosis, protease inhibitor (PI) use, smoking, alcohol use, drug abuse, and duration of HAART.
Body composition
Was measured by bioelectrical impedance analysis (Quantum II Analyzer, RJL Systems, Clinton Township, MI). Body Mass Index and waist and hip circumference were measured.
Glucose and insulin
Subjects ingested 75 g glucose (SunDex, Fisher Health Care, Houston, TX) in the morning after a 10 h overnight fast, with blood sampling at 0 (fasting), 30, 60, 90 and 120 minutes. Plasma glucose was measured by the glucose oxidase method and insulin by radioimmunoassay (Linco, St. Louis, MO). HbA1c was measured by HPLC at The Methodist Hospital clinical laboratory, Houston, TX.
CVD risk markers
Total cholesterol, HDL-C and triglyceride profiles were determined by an automated enzymatic method (Olympus AU400e Analyzer, Center Valley, PA) at the Baylor Atherosclerosis Core Laboratory (CLIA-CAP certified). LDL-C was calculated using the Friedewald formula. Fasting plasma total cholesterol, HDL-C and triglycerides were measured in the Baylor Atherosclerosis Clinical Laboratory using an Olympus AU400e automated analyzer. The median interassay c.v.’s for these assays were < 2–5%, and the median intra-assay c.v.’s were < 2–8%.
Energy expenditure
Resting energy expenditure (REE) was measured at the Baylor General Clinical Research Center (GCRC) by indirect calorimetry (Deltratrac, Sensormedics, Fullerton, CA) and calculated as REE = VO2 x [4.686 + (RQ-0.707) * 0.361/0.293[14]. For subjects who could not visit the GCRC (n = 68), REE was measured using a MedGem device (Microlife USA, Dunedin, FL) that measures VO2 and calculates REE using a modified Weir equation.
CD4+ T cell count and HIV-1 viral load
CD4 count was measured by flow cytometry. To assess the relationship of CD4 levels to glucose abnormalities, subjects were divided into two groups: “low” (<300/cc; n = 53), and “moderate-high” (≥300/cc; n = 147). The cutoff of 300/cc was based on the World Health Organization’s current recommendation to the increase the threshold for the use of antiretroviral therapy (ART) for HIV-infected individuals to a CD4 count of ≤350 cells/cc (from the previous level of ≤200 cells/μL)[15]. In view of these changes that occurred during the period of the Heart Positive study, we used a cutoff slightly lower than that of the WHO to define the CD4 strata. HIV-1 viral load (VL) was measured by LabCorp or Quest Diagnostics (Madison, NJ) using either of two quantitative real-time PCR assays (routine, with a lower limit of 400 copies/cc, or ultrasensitive, with a lower limit of 50 copies/cc).
VL values measured as <400/cc by the first assay were assigned a value of 200/cc and all the VL values were log-transformed prior to analysis.
Statistical analysis
Descriptive measures, chi-square (χ2) analysis and ANOVA were used to examine associations and differences in metabolic, hormonal, anthropometric and clinical parameters by ethnicity and CD4 strata. When differences in groups were observed, Bonferroni Post-Hoc analysis was performed to determine which groups were different from each other. Repeated measures ANOVA was used to assess the relationship of ethnicity with glucose measures following the oral challenge. Covariates were age, sex, BMI, lean mass, REE/lean mass, HAART duration, PI usage, HIV duration, fasting glucose level and fasting insulin level. Two-way interactions between ethnicity and glucose, ethnicity and gender, ethnicity and age, and ethnicity and CD4 count were included in the model and removed if non-significant (p > 0.05). Linear regression was performed with fasting serum glucose, 2-hour glucose, and HbA1c as outcome variables, and age, sex, BMI, waist, lean mass, REE/lean mass, HAART duration, PI usage, HIV duration, fasting glucose, HOMA-IR, systolic and diastolic blood pressure, family history of diabetes and interactions as independent variables. To avoid colinearity, BMI and lean mass (and not waist-hip-ratio and fat mass) were included in the final model. All statistical analyses were performed using the Statistical Package for Social Sciences (SPSS) (version 16 .0, SPSS Inc, Chicago).