Reliability of White Matter Microstructural Changes in HIV Infection: Meta-Analysis and Confirmation.

BACKGROUND

Diffusion tensor imaging has been widely used to measure HIV effects on white matter microarchitecture. While many authors have reported reduced fractional anisotropy and increased mean diffusivity in HIV, quantitative inconsistencies across studies are numerous.

PURPOSE

Our aim was to evaluate the consistency across studies of HIV effects on DTI measures and then examine the DTI reliability in a longitudinal seropositive cohort.

DATA SOURCES

Published studies and investigators.

STUDY SELECTION

The meta-analysis included 16 cross-sectional studies reporting fractional anisotropy and 12 studies reporting mean diffusivity in the corpus callosum.

DATA ANALYSIS

Random-effects meta-analysis was used to estimate study standardized mean differences and heterogeneity. DTI longitudinal reliability was estimated in seropositive participants studied before and 3 and 6 months after beginning treatment.

DATA SYNTHESIS

Meta-analysis revealed lower fractional anisotropy (standardized mean difference, -0.43; < .001) and higher mean diffusivity (standardized mean difference, 0.44; < .003) in seropositive participants. Nevertheless, between-study heterogeneity accounted for 58% and 66% of the observed variance ( < .01). In contrast, the longitudinal cohort fractional anisotropy was higher and mean diffusivity was lower in seropositive participants (both, < .001), and fractional anisotropy and mean diffusivity measures were very stable during 6 months, with intraclass correlation coefficients all >0.96.

LIMITATIONS

Many studies pooled participants with varying treatments, ages, and disease durations.

CONCLUSIONS

HIV effects on WM microstructure had substantial variations that could result from acquisition, processing, or cohort-selection differences. When acquisition parameters and processing were carefully controlled, the resulting DTI measures did not show high temporal variation. HIV effects on WM microstructure may be age-dependent. The high longitudinal reliability of DTI WM microstructure measures makes them promising disease-activity markers.