Body-Fat Declines Track Cardiometabolic Markers in Men With HIV

Key Takeaways

• Among men with virologically suppressed HIV receiving stable ART in Bratislava, Slovakia, within-person body-fat declines were significantly associated with selected lipid, glycemic, uric-acid and blood-pressure markers after FDR correction.
• Declines in the visceral fat index showed a narrower significant pattern that excluded uric acid.
• Greater estimated muscle mass, conditional on body weight, was significantly associated with lower total cholesterol, uric acid and diastolic pressure.
• Body-fat change added associations with lipid measures beyond BMI change, without established cross-validated predictive superiority.

Body mass index (BMI) summarizes weight relative to height but cannot distinguish changes in fat from changes in muscle during stable treatment for human immunodeficiency virus (HIV). Investigators enrolled 143 adult men in a single-center prospective observational body-composition study at University Hospital Bratislava, Slovakia.

All had unchanged combination ART for at least 12 months and a most recent HIV ribonucleic acid (RNA) level below 50 copies/mL. Eleven of 154 enrolled men were excluded after starting lipid-lowering medication between baseline and a second assessment; stable baseline statin use was adjusted for instead. At baseline and approximately 6, 12 and 18 months, fasting Omron BF511 bioimpedance estimated body fat, skeletal muscle and a unitless visceral fat index; investigators also measured fasting lipids, glucose and uric acid and office blood pressure.

Patient-random-intercept models separated within-person change from between-person differences, adjusting for visit, age, smoking, HIV duration, cluster of differentiation 4 (CD4) count, baseline statin use, diabetes or prediabetes, and ART regimen; muscle models additionally adjusted for concurrent weight. False discovery rate (FDR) correction covered ten outcomes separately per exposure, using available cases without imputation. Exploratory joint BMI/body-fat lipid models and patient-level cross-validation assessed the measures further.

Within-person body-fat declines tracked with several measured markers. Per one-percentage-point body-fat decrease, the total-cholesterol coefficient was β=−0.081 mmol/L (95% confidence interval [CI] −0.115 to −0.046). Eight outcomes survived FDR correction: total and low-density lipoprotein (LDL) cholesterol, triglycerides, calculated very-low-density lipoprotein (VLDL) cholesterol, fasting glucose, uric acid, and systolic and diastolic blood pressure. Calculated VLDL was derived from triglycerides, so those lipid findings are not independent. High-density lipoprotein (HDL) cholesterol and high-sensitivity C-reactive protein were not significantly associated.

A lower unitless visceral fat index had seven FDR-corrected associations with the same markers except uric acid; HDL cholesterol and high-sensitivity C-reactive protein were also not significantly associated. Greater estimated absolute skeletal-muscle mass, conditional on the same weight trajectory, was associated after correction with lower total cholesterol, uric acid and diastolic pressure; nominal LDL and systolic-pressure associations did not survive correction. In exploratory joint models, body-fat change retained associations with total, LDL and calculated VLDL cholesterol and triglycerides after accounting for BMI change, while BMI added no significant information. Patient-level cross-validation and paired bootstrap intervals did not establish superior prediction by body-fat models for those lipids; HDL favored BMI.

These observational associations cannot establish causation: diet, alcohol and activity were not systematically quantified, reverse causation remains possible, and bioimpedance estimates visceral adiposity indirectly. Thirty-nine scheduled 18-month assessments fell after administrative study closure, distinct from other missed or declined visits; missingness and lipid-therapy exclusions could select the observed cohort. Single-center enrollment of men in Slovakia limits extrapolation, including to North American populations.

The investigators concluded that within-person adiposity declines during unchanged ART accompanied more favorable measured cardiometabolic markers, while body-fat estimates added some explanatory information beyond BMI without proven better out-of-sample prediction. They called for validation of bioimpedance-derived measures and testing against clinical outcomes.

Clinician Questions

Were body-composition changes induced by a weight-loss intervention in men with HIV receiving stable ART?

Investigators administered no targeted pharmacologic treatment or structured lifestyle intervention to change body composition in men with HIV receiving stable antiretroviral therapy. Routine lifestyle counseling at baseline was usual care, not a research intervention, and ART regimens remained unchanged during follow-up.

Is a one-percentage-point body-fat change detectable with Omron BF511 in men living with HIV?

For men with HIV receiving stable antiretroviral therapy, one percentage point of Omron BF511 bioimpedance-derived body fat was a unit for a longitudinal model coefficient, not a validated minimum detectable individual change. Device-specific absolute measurement error and minimum detectable change were not established, so the modeled unit does not demonstrate that a small individual change can be distinguished reliably from measurement variability.