Abstract
Background: Anorexia nervosa (AN) is a debilitating illness whose neural basis remains unclear. Studies using tract-based spatial statistics (TBSS) with diffusion tensor imaging (DTI) have demonstrated differences in white matter (WM) microarchitecture in AN, but the findings are inconclusive and controversial. Objectives: To identify the most consistent WM abnormalities among previous TBSS studies of differences in WM microarchitecture in AN. Methods: By systematically searching online databases, a total of 11 datasets were identified, including 245 patients with AN and 246 healthy controls (HC). We used Seed-based d Mapping to analyze fractional anisotropy (FA) differences between AN patients and HC, and performed meta-regression analysis to explore the effects of clinical characteristics on WM abnormalities in AN. Results: The pooled results of all AN patients showed robustly lower FA in the corpus callosum (CC) and the cingulum compared to HC. These two regions preserved significance in the sensitivity analysis as well as in all subgroup analyses. Fiber tracking showed that the WM tracts primarily involved were the body of the CC and the cingulum bundle. Meta-regression analysis revealed that the body mass index and mean age were not linearly correlated with the lower FA. Conclusions: The most consistent WM microstructural differences in AN were in the interhemispheric connections and limbic association fibers. These common "targets" advance our understanding of the complex neural mechanisms underlying the puzzling symptoms of AN, and may help in developing early treatment approaches.