Brain Structural Changes following HIV Infection: Meta-Analysis

Abstract

Background: Numerous studies have used structural neuroimaging to measure HIV effects on brain macroarchitecture. While many have reported changes in total brain volume, gray matter volume, white matter volume, CSF volume, and basal ganglia volume following HIV infection, quantitative inconsistencies observed across studies are large.

Purpose: Our aim was to evaluate the consistency and temporal stability of serostatus effects on a range of structural neuroimaging measures.

Data sources: PubMed, reference lists, and corresponding authors.

Study selection: The meta-analysis included 19 cross-sectional studies reporting HIV effects on cortical and subcortical volume from 1993 to 2016.

Data analysis: Random-effects meta-analysis was used to estimate individual study standardized mean differences and study heterogeneity. Meta-regression was used to examine the effects of the study publication year.

Data synthesis: Meta-analysis revealed standardized mean differences related to the serostatus of -0.65 (P = .002) for total brain volume, -0.28 for gray matter volume (P = .008), -0.24 (P = .076) for white matter volume, and 0.56 (P = .001) for CSF volume. Basal ganglia volume differences related to serostatus were not significant. Nevertheless, estimates of between-study heterogeneity suggested that much of the observed variance was between studies. Publication year was associated with recent reductions in many neurostructural effects.

Limitations: Many studies pooled participants with varying durations of treatment, disease, and comorbidities. Image-acquisition methods changed with time.

Conclusions: While published studies of HIV effects on brain structure had substantial variations that are likely to result from changes in HIV treatment practice during the study period, quantitative neurostructural measures can reliably detect the effects of HIV infection during treatment, serving as reliable biomarkers.

Fig 1.

Meta-analysis flow diagram for study selection.

Fig 2.

Total brain volume standardized mean differences across studies comparing seropositive with seronegative participants. A, Forest plot. B, Funnel plot. C, Radial plot.

Fig 3.

Gray matter standardized mean differences across studies comparing seropositive with seronegative participants. A, Forest plot. B, Funnel plot. C, Radial plot.

Fig 4.

White matter standardized mean differences across studies comparing seropositive with seronegative participants. A, Forest plot. B, Funnel plot. C, Radial plot.

Fig 5.

CSF standardized mean differences across studies comparing seropositive with seronegative participants. A, Forest plot. B, Funnel plot. C, Radial plot.

Fig 6.

Forest plots of basal ganglia gray matter showing standardized mean differences across studies comparing seropositive with seronegative participants A, Caudate volume. B, Putamen volume. C, Pallidal volume.

Fig 7.

Publication year influence on serostatus effects. These are plots of structural brain measure standardized mean differences and their corresponding regression lines, for total brain volume (A) P = .001, gray matter volume (B) P = .086, white matter volume (C) P < .001, and CSF volume (D) P < .001. The plot circle area is inversely proportional to the estimated SMD variance.