HIV and childhood maltreatment map onto different aging signals in South African women
In a cohort of 238 South African women, biological aging did not move as a single process. The pattern was fragmented, with some markers tracking HIV more clearly than others, and childhood maltreatment contributing far less than expected. Women living with HIV had shorter absolute telomere length, lower MRI-based brain-predicted age difference, and higher values on several epigenetic age acceleration measures, including HorvathAA, HannumAA, PhenoAA, and EEAA. The result is a reminder that biological aging is not one phenotype but several overlapping readouts, each shaped by different tissue compartments and molecular pathways.
The study brought together women with and without HIV and characterized childhood maltreatment severity using the Childhood Trauma Questionnaire–Short Form total score. Four aging platforms were assessed: absolute telomere length, mitochondrial DNA copy number, MRI-BPAD, and DNA methylation age clocks. Each captures a different biological layer. Telomeres reflect replicative history and stress biology in leukocytes. Mitochondrial DNA copy number is often interpreted as a marker of mitochondrial reserve or cellular stress adaptation. MRI-BPAD estimates whether the brain appears older or younger than expected on structural imaging. DNA methylation clocks compress decades of molecular change into age-adjusted epigenetic signatures.
The strongest signal in this dataset came from the epigenetic measures. DNAmAge clocks were moderately to strongly intercorrelated, with correlations ranging from r = 0.35 to 0.93. In practical terms, these clocks were often moving in the same direction, suggesting that several methylation-based aging tools may be reading a shared underlying biology. Their coherence also makes biological sense: methylation clocks are all derived from patterns of systemic cellular regulation and may be sensitive to similar exposures, including immune activation, oxidative stress, and chronic inflammatory burden.
By contrast, the other markers were less tightly linked. Absolute telomere length correlated weakly with mitochondrial DNA copy number, with ρ = 0.36, and showed weak inverse correlations with PhenoAA and EEAA, both at ρ = -0.18. MRI-BPAD correlated weakly with PhenoAA at r = 0.20. These small associations argue against a simple model in which all aging biomarkers rise and fall together. Telomere attrition, mitochondrial adaptation, brain structural aging, and methylation drift may be converging only partially, even in the setting of shared stress exposures.
One sentence captures the biology here: the markers are related, but they are not redundant.
Epigenetic clocks converged, while childhood maltreatment showed little independent signal
Childhood maltreatment did not emerge as a dominant driver of aging biomarkers in this analysis. Associations between CM and the measured aging outcomes were limited, and no robust two-way or three-way moderation effects survived false discovery rate correction. HIV status and CM severity therefore did not appear to interact in a way that consistently reshaped the biomarker profile after multiple-testing control.
That pattern matters because both HIV and maltreatment are chronic stressors, and both have been linked in other settings to accelerated aging biology. Yet association is not uniform across readouts. A person may show epigenetic acceleration without marked telomere shortening, or altered telomere biology without a parallel brain-aging signal. In women exposed to long-term adversity, the body may be recording stress in several molecular languages at once, but not every language carries the same information.
Women with HIV did show a clearer biological aging signature across several measures. Shorter telomeres and higher epigenetic age acceleration are biologically plausible findings in the setting of chronic immune activation, repeated inflammatory signaling, and cumulative exposure to infection-related stress. HIV is known to alter immune homeostasis in ways that can affect hematopoietic turnover, oxidative injury, and epigenetic programming. Those pathways can plausibly shorten telomeres and shift methylation patterns even when another marker, such as MRI-based brain age, behaves differently because it reflects a distinct tissue compartment and a different time scale of injury.
The brain finding deserves attention of its own. Lower MRI-BPAD in women with HIV suggests that the direction of the imaging-based age signal did not mirror the epigenetic pattern in a simple one-to-one fashion. Structural MRI measures can be influenced by tissue loss, developmental baseline, vascular injury, and analytic assumptions embedded in prediction models. A weak correlation between MRI-BPAD and PhenoAA reinforces the larger point: systemic molecular aging and brain-predicted aging overlap only partially.
The study’s main contribution is conceptual as much as statistical. It supports a multidimensional model of biological aging in which epigenetic clocks cluster together, but telomeres, mitochondria, and neuroimaging capture separate slices of the same underlying process. For clinicians and researchers working with women exposed to HIV and early-life adversity, that means one biomarker will rarely tell the full story. A short telomere count does not automatically imply a more aged brain, and a more accelerated methylation clock does not necessarily mean parallel change across every tissue.
Longitudinal follow-up will be needed to determine whether these markers diverge further over time, whether any of them better predicts later cardiometabolic, neurocognitive, or functional outcomes, and whether HIV treatment, psychosocial adversity, or both modify those trajectories. Cross-sectional biomarker snapshots can show concordance and mismatch, but they cannot establish which aging pathway moves first or which one most reliably foretells disease. In populations carrying overlapping burdens of infection and childhood trauma, that temporal question is central to understanding risk.