Dynamic Structural Brain Changes in Anorexia Nervosa: A Replication Study, Mega-analysis, and Virtual Histology Approach

Research output: Contribution to journalResearch articleContributedpeer-review


OBJECTIVE: Several, but not all, previous studies of brain structure in anorexia nervosa (AN) have reported reductions in gray matter volume and cortical thickness (CT) in acutely underweight patients, which seem to reverse upon weight gain. The biological mechanisms underlying these dynamic alterations remain unclear.

METHOD: In this structural magnetic resonance imaging study, we first replicated and extended previous results in (1) a larger independent sample of 75 acutely underweight adolescent and young adult female patients with AN (acAN; n = 54 rescanned longitudinally after partial weight restoration), 34 weight-recovered individuals with a history of AN (recAN), and 139 healthy controls (HC); and 2) a greater combined sample compiled of both our previous samples and the present replication sample (120 acAN [90 rescanned longitudinally], 68 recAN, and 207 HC). Next, we applied a "virtual histology" approach to the combined data, investigating relations between interregional profiles of differences in CT and profiles of cell-specific gene expression. Finally, we used the ENIGMA toolbox to relate aforementioned CT profiles to normative structural and functional connectomics.

RESULTS: We confirmed sizeable and widespread reductions of CT as well as volumes (and, to a lesser extent, surface area) in acAN and rapid increases related to partial weight restoration. No differences were detected between either short- or long-term weight-recovered patients and HC. The virtual histology analysis identified associations between gene expression profiles of S1 pyramidal cells and oligodendrocytes and brain regions with more marked differences in CT, whereas the remaining regions were those with a greater expression of genes specific to CA1 pyramidal, astrocytes, microglia, and ependymal cells. Furthermore, the most affected regions were also more functionally and structurally connected.

CONCLUSION: The overall data pattern deviates from findings in other psychiatric disorders. Both virtual histology and connectomics analyses indicated that brain regions most affected in AN are also the most energetically demanding.


Original languageEnglish
Pages (from-to)1168-1181
Number of pages14
Journal Journal of the American Academy of Child and Adolescent Psychiatry : JAACAP
Issue number9
Publication statusPublished - Sept 2022

External IDs

Scopus 85129688756
unpaywall 10.1016/j.jaac.2022.03.026
Mendeley 20f4540d-d5c3-33ff-8935-a5ee40405ef6
ORCID /0000-0002-2864-5578/work/142233476
ORCID /0000-0003-2132-4445/work/142236351
ORCID /0000-0002-5112-405X/work/142242684
ORCID /0000-0002-5413-0359/work/142248933


Research priority areas of TU Dresden

Sustainable Development Goals


  • Anorexia Nervosa/diagnostic imaging, Brain/diagnostic imaging, Female, Humans, Magnetic Resonance Imaging, Thinness/psychology, Weight Gain, cerebral cortex, structural MRI, longitudinal study, virtual histology, anorexia nervosa