Maternal ART throughout gestation prevents caudate volume reductions in neonates who are HIV exposed but uninfected

Global programs have reduced perinatal HIV transmission in South Africa to below 1%, creating a growing population of infants who are HIV-exposed but uninfected (HEU). The long-term effects of in utero exposure to HIV and antiretroviral therapy (ART) on brain development remain unclear. Prior work shows that children who are HEU may have greater neurodevelopmental delays than HIV-unexposed and uninfected (HUU) peers. This study investigates whether prenatal HIV/ART exposure is associated with volumetric differences in neonatal subcortical structures and cerebellum, focusing on basal ganglia and cerebellum due to their roles in motor and cognitive functions. The authors hypothesized smaller basal ganglia and cerebellar volumes in HEU vs. HUU neonates and, based on prior behavioral data, anticipated no effect of ART initiation timing on volumes.

The paper reviews imaging and non-imaging studies of HEU children. Prior MRI/MRS/DTI studies reported mixed findings: some found altered regional volumes, metabolite levels, and white matter integrity in HEU vs. HUU, while others found no differences. Only one prior infant structural study (Wedderburn et al., 2022a) reported smaller bilateral caudate volumes in HEU neonates but did not examine ART duration effects. Non-imaging studies show deficits in language, motor development, and adaptive behavior among HEU children, particularly in Southern Africa (e.g., Wedderburn 2019b; Chaudhury 2017; Ntozini 2020; Le Roux 2018; Springer 2012; Madlala 2020). Mechanistic literature suggests maternal HIV and inflammation, and potential ART neurotoxicity, may affect fetal brain development, with oxidative stress and maternal immune activation implicated.

Design and participants: Prospective cohort in Cape Town, South Africa (2017–2021). Recruited 226 low-risk pregnant Xhosa women (144 living with HIV; 82 HIV-negative controls) ≤29 weeks’ gestation. After exclusions, 187 mother-infant pairs were enrolled; imaging was acquired in 166 infants, and 120 high-quality scans were included in manual tracing analyses (79 HEU, 41 HUU). Among HEU, 40 had maternal ART initiated preconception (HEU-pre) and 39 post-conception (HEU-post); mean gestational age at ART start in post-conception group: 15.4±5.7 weeks. All mothers with HIV received fixed-dose Tenofovir/Efavirenz/Emtricitabine. Infant prophylaxis per risk included Nevirapine (± Zidovudine). Ethics approval obtained; informed consent from mothers for themselves and infants. Inclusion/exclusion: Excluded maternal chronic disorders, poor obstetric history, TB, pregnancy complications or medications beyond supplements/ART, non-adherence or non-standard ART, high alcohol (≥7 drinks/week or ≥4/occasion) or illicit drug use, language barrier. Infant exclusions: preterm <36 weeks GA, neonatal admission, birth weight <2500 g, positive HIV-1 PCR, significant congenital/chromosomal anomalies, neonatal asphyxia, persistent hypoglycemia, severe jaundice. Maternal assessments: Demographics, monthly health visits, trauma (Harvard Trauma Scale) and depression (CES-D) screening, timeline follow-back for alcohol/drugs/smoking, urine drug screens (~20 and 33 weeks GA). GA estimated via LMP, fundal height, early ultrasound, adjusted post-delivery. Clinical HIV data: Maternal viral load (VL) and CD4 within 6 months of pregnancy and at delivery were abstracted from clinic records; VL measured post-ART start for newly diagnosed, with no pre-ART VL in HEU-post mothers. ART adherence assessed at each visit. Infant MRI preparation: Pediatrician (blinded to exposure) examined infants, performed Dubowitz Neurological Examination ~1 hour pre-scan. Infants fed, swaddled, protected with earplugs/MiniMuffs, monitored by pulse oximeter, scanned unsedated. MRI acquisition: 3T Siemens Skyra with 16-channel pediatric head/neck coil at mean GA-equivalent 41.5±1.0 weeks. High-resolution T1-weighted 3D EPI-navigated multi-echo MPRAGE (MEMPRAGE): FOV 192×192 mm2, TR 2540 ms, TI 1450 ms, TEs 1.69/3.55/5.41/7.27 ms, BW 650 Hz/px, 144 sagittal slices, 1.0 mm isotropic; GRAPPA 2. Image quality checked; 120 high-quality scans retained. Manual segmentation: Target ROIs included caudate, putamen, globus pallidus, thalamus, cerebellar hemispheres, cerebellar vermis, and corpus callosum (CC). Segmentation performed in Freeview (FreeSurfer v7.1.0). A single expert neuroanatomist (AI) manually traced subcortical and cerebellar ROIs slice-by-slice in coronal plane with axial/sagittal corrections; CC traced (by FW) on two contiguous midline sagittal slices after AC-PC alignment, averaging areas. Tracers blinded to exposure status. Senior neuroanatomist (FW) verified all tracings. Intra-rater reliability: 10 randomly selected brains re-traced; ICCs for absolute agreement and consistency >0.86 (p<0.002) for all regions. Total brain volume estimated via infant FreeSurfer pipeline; head circumference at scan recorded. Statistical analysis: Conducted in R. Group comparisons for demographics via t-test/ANOVA and chi-square. Linear regression tested volumetric differences: HEU vs HUU; HEU-pre vs HUU; HEU-post vs HUU. Potential confounders considered: maternal age at delivery, maternal weight change per week, average alcohol (oz absolute alcohol/day across pregnancy), education (highest grade), infant sex, GA-equivalent at scan, infant weight at scan, and head circumference (preferred over total brain volume due to stronger associations). Variables correlated with any regional volume at p<0.10 were included as covariates. Associations among HEU infants between regional volumes and maternal clinical/treatment variables (CD4, peak VL [log-transformed], ART exposure duration, maternal VL detectability status during pregnancy) were examined with Pearson correlation and adjusted linear regression. For HEU-post mothers lacking pre-ART VL data, VL was assumed detectable until first undetectable measurement (<20 copies/mL). Multiple comparisons noted (Bonferroni for 12 regions).

- Sample: 120 neonates (59 female); 79 HEU (40 HEU-pre; 39 HEU-post), 41 HUU. Mean scan GA-equivalent 41.5±1.0 weeks. - Reliability: Intra-rater ICCs >0.86 (p<0.002) for all ROIs. - Covariates: Infant sex, maternal age at delivery, and maternal education were not associated with regional volumes; head circumference showed stronger associations than total brain volume and was used as a covariate. - Group volumetric differences (adjusted linear regression with selected covariates): • Left putamen: HEU < HUU, β (SE) = −90.3 (45.3), p = 0.05 (approx. 4.4% smaller; HEU mean 2022±229 mm3 vs HUU 2115±270 mm3). Within subgroups, HEU-pre vs HUU β = −107.2 (55.9), p = 0.06; HEU-post vs HUU β = −86.8 (56.8), p = 0.13. • Caudate nuclei: Significant bilateral reductions only in HEU-post vs HUU: - Left caudate: β (SE) = −145.5 (45.1), p = 0.002; HEU-post 1508±174 mm3 vs HUU 1648±227 mm3 (−8.5%). Significant after Bonferroni correction across 12 regions. - Right caudate: β (SE) = −135.7 (49.7), p = 0.008; HEU-post 1503±183 mm3 vs HUU 1626±238 mm3 (−7.6%). - HEU-pre showed no caudate reductions vs HUU (left 1666±253 mm3; right 1650±244 mm3; both p>0.5). • No significant group differences for pallidum, thalamus, cerebellar hemispheres, vermis, or corpus callosum. - Maternal clinical/treatment associations among HEU infants (n≈74–79): • ART exposure duration: Positively associated with caudate volume bilaterally (r = 0.38, p<0.001 left; r = 0.35, p=0.002 right); associations remained significant after adjustment [left β = 8.60 (p = 0.004); right β = 8.45 (p = 0.007)]. • Peak maternal VL during pregnancy (log-transformed): Higher VL associated with smaller left putamen, cerebellar hemispheres, and vermis (e.g., right cerebellum remained significant after adjustment: β = −0.20, p = 0.05). Trends for caudate (negative correlation) did not remain significant after adjustment. • Maternal CD4 within 6 months of pregnancy: Trend toward positive association with left caudate (r = 0.21, p = 0.07). - Interpretation: Maternal ART throughout gestation appears protective for caudate development; putaminal reductions persist in HEU regardless of ART timing.

Findings address the study’s hypothesis by demonstrating subcortical vulnerability in HEU neonates: significant reductions in left putamen across HEU and bilateral caudate reductions specifically in HEU-post infants. Contrary to the initial hypothesis regarding ART timing, duration of in utero ART exposure was strongly related to caudate volumes—greater duration associated with larger caudate—indicating that maternal ART from conception confers neuroprotection to the fetal caudate. The persistent left putamen reduction across HEU, independent of ART timing, suggests either later-gestation vulnerability, differential susceptibility unrelated to ventricular proximity, or altered lateralization processes. Associations between higher peak maternal VL and reduced volumes in putamen and cerebellar regions further implicate maternal viral burden and inflammation in fetal brain development. Mechanistic considerations include maternal immune activation, inflammatory cytokine exposure at the materno-fetal interface, and oxidative stress affecting basal ganglia and cerebellum. The selective caudate vulnerability may relate to proximity to ventricles and timing of basal ganglia ontogenesis. Clinically, early and continuous ART may mitigate specific subcortical risks in HEU infants, potentially influencing later neurodevelopmental outcomes, especially in motor and cognitive domains linked to basal ganglia function.

HEU neonates exhibit subcortical volume reductions despite absence of HIV infection. Maternal ART initiated before conception and sustained throughout pregnancy appears to protect bilateral caudate development, whereas left putamen volume is reduced across HEU irrespective of ART timing. These results underscore the importance of early ART initiation in pregnancy to reduce fetal neuroanatomical risk. Future research should longitudinally assess whether neonatal volumetric differences persist or resolve, link volumes to neurobehavioral outcomes, evaluate white matter microstructure, and include immunologic biomarkers (e.g., migratory monocytes) to elucidate mechanisms by ART exposure timing.

- Cross-sectional design limits causal inference and trajectory assessment. - Migratory monocytes/neuroinflammation biomarkers were not quantified. - No concurrent assessment of functional neurodevelopmental domains linked to regional volumes. - White matter microstructure was not evaluated in this analysis. - Limited pre-ART viral load data for mothers initiating ART post-conception required assumptions about detectability until first undetectable measurement. - Manual tracing, while reliable, and selection of high-quality scans (n=120) may limit generalizability to the full cohort.