While previous studies have consistently shown HIV-related white matter (WM) microstructural alterations in adults, adolescents, children and animal models, studies in youth have typically included wide age ranges during which both white matter volume and FA increase significantly [1-5]. There are no studies examining whether early ART (before 12 weeks of age) could mitigate these effects. Here we examine WM alterations in HIV infected children at age 7 years and compare those who initiated ART before and after 12 weeks of age.

POPULATION: One hundred and twenty-one (Xhosa and Cape Coloured) 7-year-old children from a longitudinal neurodevelopmental sub study of the Children with HIV Early Antiretroviral (CHER) trial [6] in Cape Town, South Africa were studied. The group included HIV-infected children, stable on antiretroviral therapy (ART) and age-matched controls from a parallel vaccine study [7]. After exclusion of subjects due to motion artifacts we divided our subjects into the following groups for comparison: controls (CTRL, 48 children, age 7.24±0.13 yrs (mean age ± standard deviation), 26 males, 39 Xhosa) and HIV infected (HIV, 68, 7.21±0.13 yrs, 33 males, 59 Xhosa). Within controls there were HIV-exposed, uninfected (HEU) (19 children, age 7.22±0.11 yrs, 11 males, 10 Xhosa) and HIV-unexposed, uninfected (HUU) (29 children, age 7.25±0.15 yrs, 15 males, all Xhosa) children. The HIV group was also divided into two treatment subgroups: children who initiated ART at or before 12 weeks of age (EARLY, 53 children, age 7.21±0.10 yrs, 25 males, 46 Xhosa) and those who began ART treatment after 12 weeks (LATE, 15 children, age 7.17±0.20 yrs, 8 males, 13 Xhosa).

SCANNING PROCEDURE: Children in the study were scanned on a 3T Siemens Allegra (Erlangen, Germany) with a single channel head coil according to protocols that had been approved by the Human Research Ethics Committees of participating institutions. Children were scanned with structural T₁ imaging followed by 2 DTI acquisitions with opposite phase encoding (AP-PA) directions using a prospectively motion-corrected navigated twice-refocused spin echo sequence with 5 reacquisitions [8]. Acquisition parameters for diffusion were: TR/TE 10100/86ms, 72 slices, 2×2×2mm³, FOV 224mm, 30 non-collinear diffusion directions, b=1000s/mm², and four non-diffusion-weighted (b₀) acquisitions.

DIFFUSION DATA PROCESSING: DTI data were preprocessed in TORTOISE [9]: for each subject the set of AP-PA phase-encoded data sets was corrected for motion and eddy current distortions. FSL and AFNI were used to co-register all of the DTI results to the Haskins pediatric template [10]. Voxel-wise group comparisons were performed in FSL [11] based on a general-linear model (gender and race included as confounders). Clusters showing statistical group differences were identified from uncorrected p-value maps for 3 sets of group comparisons (CTRL vs HIV; HEU vs HUU; EARLY vs LATE). Cluster size thresholding at p_th=0.005 and alpha=0.05 (where alpha is the probability of random noise at p_th) yielded a minimum cluster size of 102mm³ for CTRL vs HIV, and 110mm³ for HEU vs HUU and EARLY vs LATE comparisons.

One region (in left (L) inferior longitudinal fasciculus) showed lower FA and 5 regions (in right (R) corticospinal tract, bilateral inferior fronto-occipital fasciculus, R superior longitudinal fasciculus, and R superior fronto-occipital fasciculus) higher MD in infected children compared to controls (Figures 1 and 2). Three regions were identified in 2 tracts (bilateral corticopontine tract and middle cerebellar peduncle) with higher FA in HEU children compared to HUU children, and bilateral regions in the superior longitudinal fasciculus showed lower MD in HEU children (Figures 3 and 4). No regions showed FA or MD differences between children in EARLY and LATE groups.

CTRL vs HIV: Lower FA and higher MD in HIV-infected children were largely attributable to higher radial diffusivity (L23), indicative of poorer myelination in the affected regions. Regions with increased MD also showed increased axial diffusivity (L1). Both axial and radial diffusivity have been shown to decrease from neonates to 1-year olds [12] and throughout childhood [13]. The higher levels in HIV-infected children seen here could be indicative of altered developmental trajectories.

HEU vs HUU: Unexpectedly, we only found regions showing higher FA in HEU children compared to HUU children. These effects were attributable both to higher L1 and lower L23, suggesting improved myelination in these children. Moreover, lower MD in the superior longitudinal fasciculus in HEU children was due to reduced L1 and L23.

EARLY and LATE: The absence of differences between these groups may be due to the small sample size of the latter.