Background/Purpose: Neuropsychiatric SLE in children and adolescents presents diagnostic challenge due to limitations of conventional magnetic resonance imaging (MRI) to detect clinically relevant brain changes. We examined structural brain abnormalities in pediatric-onset SLE (pSLE) utilizing advanced MRI techniques.

Methods: We conducted a cross-sectional analysis of clinically-obtained brain MRI images from subjects with pSLE, and compared them to existing brain MRI images from age and sex-matched controls from another study.  All images were obtained between 2007 and 2015 on the same scanner at 3T using a T1-weighted MPRAGE (magnetization-prepared rapid, acquisition gradient echo) protocol. We used an advanced multi-atlas segmentation algorithm to divide the brain into 154 anatomical regions of interest (ROIs), organized hierarchically within larger brain structures. We calculated volumes of individual ROIs and larger brain structures, and compared volumetric measurements from pSLE and control subjects using univariate paired t-tests. ROIs with significant group differences after Bonferroni correction for multiple comparisons (q<0.01, where q=p/#ROIs tested) were reported. A neuroradiologist, blinded to the volumetric results, performed conventional re-reads of MRIs for pSLE subjects.

Results: We matched 29 SLE adolescents to 29 controls, comprised of 90% females with a mean age of 15.9 (SD=3.6). Median disease duration for SLE subjects was 1.1 years (interquartile range, IQR= 0.2, 2.8), and there was a history of nephritis in 11 (39%), seizures and/or stroke in 5 (18%), anti-phospholipid syndrome in 5 (18%), depression and/or anxiety in 15 (54%). Glucocorticoids were used by 86% of pSLE subjects at the time of MRI (median prednisone dose=10 mg, IQR=6,25). Conventional pSLE MRI reads indicated T2 hyperintensities in 14 (48%) and mild diffuse volume loss in 8 (28%). Using advanced segmentation, total brain volume did not differ between groups, but volumes in specific gray matter ROIs were significantly decreased in pSLE subjects compared to controls (q<0.01) (Figure). These ROIs are involved in: decision-making, memory, and social cognition (right medial frontal cortex); empathy, emotional memory and processing (right temporal pole, left anterior cingulate gyrus); topographical and facial recognition (right lingual gyrus); language processing (left parietal lateral cortex, bilateral planum polare, frontal operculum); visual and spatial attention (bilateral cuneus). SLE subjects also had significantly larger lateral ventricles.

Conclusion: Compared to healthy peers, children and adolescents with SLE have decreased gray matter volumes in regions involved in executive function, social cognition, language and emotional processing. Future study should examine the functional correlates of these structural brain abnormalities.