Table 2.

Structural imaging studies—methods, main findings and clinical interpretations.

Study	Method and Procedure	Data Analysis	Hydration before Imaging for DTI Studies	Presentation of the Main Findings	Clinical Interpretations
MRI
Katzman (1996)	MRI 1.5T Tested at one time point.	Not mentioned		Significantly larger total CSF volume and reduced total GM and WM volumes. Alterations correlated with BMI and cortisol levels. No correlation with disease duration.	No clinical interpretations. Deficits in GM volume were associated with severity but not disease duration and were related to hypercortisolemia.
Olivo (2018)	MRI 3T Tested at one time point.	Voxel based morphometry (VBM)		Total GM, WM, and CSF volumes were not significantly different between groups.	Τhe preservation of GM volume might indeed differentiate atypical AN from AN. Alternatively, there may be a weight cut-off under which GM alterations become obvious.
Myrvang (2018)	MRI 3T Tested at one time point.	Magnetization Prepared—RApid Gradient Echo (MPRAGE)-sequence		Statistically significant volume reduction in GM, total hippocampal volume and in all hippocampal subfields apart from fissure.	Hippocampal atrophy may be attributed to hypercortisolemia due to high levels of stress.
King (2015)	MRI 3T Tested at one time point.	Source based morphometry (SBM)		Significant GM thickness reduction in a total of 86% of the cortical surface, apart from bilateral temporal pole and entorhinal cortex. Reduced volume of nucleus accumbens, amygdala, cerebellum, hippocampus, putamen and thalamus.	A correlation was found between cortical thickness and “drive for thinness” in a broad region of the right lateral occipitotemporal cortex. The normal neurodevelopmental trajectory of cortical thickness (CT) across adolescence and young adulthood may be interrupted in AN.
Yue (2018)	MRI 3T Tested at one time point.	Not mentioned		Significantly reduced total GM volume and ventricular enlargement. Reduced thalamus volume CT in the left precuneus and a larger ratio of caudate volume.	The relative preservation of caudate volume and reduced CT of the left precuneus may be involved in body image distortion.
Fujisawa (2015)	MRI 3T Tested at one time point.	VBM		Significant volume decreases in total GM as well as in bilateral inferior frontal gyrus (IFG) (19,1% left and 17,6% right). Significant correlations were found between regional reduction of GM in the bilateral IFG and age, BMI and age at disease onset.	Volumetric decreases in the IFG might explain the impulsive behaviors observed in patients with AN.
Neumärker (2000)	MRI 1.5T Testedat three time points: at admission (T1), with 50% weight restoration (T2), with normal weight (T3).	Not mentioned		T1: Significant larger lateral ventricles and wider fissures of Sylvius bilaterally. Mesencephalon was also markedly reduced. T2&T3: Reduced mesencephalon size persisted.	Volumetric alterations were related to the degree of impairment in arithmetic performance. Intact number processing abilities may be a good predictor for weight restoration.
Castro-Fornieles (2009)	MRI 1.5T Tested at two time points: before treatment (T1) and after weight recovery (T2).	VBM		T1: Lower global GM and higher CSF volumes and not statistically significant differences in WM. In regional VBM study, significantly decreased GM was observed in bilateral parietal, right temporal cortex and cingulum. T2: Decreased GM volume remained in cingulum, not to the same extent as in the first assessment.	Overall, GM reduction at first assessment correlated with Rey Complex Figure Test copy time, indicating a relationship to slowness in complex mental processing.
Monzon (2017)	MRI 3T 26 AN(T1) patients evaluated at the beginning, 10 AN(T2) patients re-examined after reaching at least 85% of expected body weight.	VBM		T1: Significantly reduced GM volume in OFC, dlPFC, mPFC, insular cortex and hippocampus, anterior cingulate cortex (ACC), medial cingulate cortex (MCC), posterior cingulate cortex (PCC) and the precuneus bilaterally. Additionally, in bilateral amygdala and thalamus. No significant difference in total brain volume between groups. T2: Significantly reduced GM volume remained in ACC, caudate nucleus and right hippocampus. GM volume increase after weight gain in thalamus was negatively correlated to the presence of eating concern symptoms, while in left OFC was negatively correlated to shape-concern symptoms evaluated by the EDE-Q.	Alterations found in PFC, insular and cingulate cortices, hippocampal region, amygdala and parietal cortex could explain distorted body image, emotional disturbances and cognitive deficits.
Golden (1996)	MRI 1T Tested at two time points: before treatment (T1) and after weight gain (T2).	Not mentioned		T1: Ventricular enlargement, especially of the third ventricle. T2: Significantly decreased total ventricular volume. An inverse relationship was found between ventricular volume and BMI.	Atrophy of the cerebral cortex may occur as a result of decreased protein synthesis caused by malnutrition. Structural changes and cognitive functioning seem to improve with weight gain.
Akgül (2016)	MRI 1.5T-MTI Tested at two time points: T1 at admission, T2 after weight recovery.	Regions of interest (ROIs)		T1: Magnetization Transfer Ratio (MTR) did not differ. MRI identified widening of the cerebral sulci in 7 patients with no other gross abnormalities. (ROIs: Left dlPFC, left cerebellar hemisphere, thalamus, amygdala, pons, corona radiata). T2: MTR did not differ.	No clinical interpretations. Adiposity-related variations in phospholipid composition of brain lipids during adolescence could be related to the reversibility of functional impairment.
Bernardoni (2016)	MRI 3T Tested at two time points: T1 at admission, T2 after weight recovery.	SBM-ROIs		T1: Global cortical thinning. AN(T1) vs. AN(T2): 84% of CT restored. AN(T2) vs. HC: CT normalised apart from left temporal pole and enthorhinal cortex. Subcortical GM volume was increased in all ROIs apart from pallidum where a decrease was observed.	Normalization of CT following partial weight restoration is independent of improvements in psychopathology.
Katzman (1997)	MRI 1.5T Tested at two time points: at low weight (T1) and at normal weight (T2) 2–3 years later.	Not mentioned		T1: GM and WM volume decrease and ventricular enlargement. T2: Findings persisted apart from WM volume decrease. Increase of GM volume correlated with BMI increase.	Hypercortisolemia may lead to neuronal damage and persistent brain abnormalities.
MRS
Blasel (2012)	MRI 3T and MRS Tested at one time point.	Separate analysis of region1: anterior region rostral of the anterior commissure & region 2: posterior region dorsal of the anterior commissure.		No difference between GM fraction. WM fraction was significantly lower to region 2. Significant differences in metabolite concentrations were determined in GM with higher concentrations of tCho, tCr, tNAA, Glx. No difference was found in WM metabolites. MI concentrations did not differ between patients and controls.	The Glx increase may indicate a psychiatric or neurodegenerative origin of AN rather than the result of nutrition depletion.
Castro-Fornieles (2007)	MRI 1.5T and MRS Tested at two time points: T1 before treatment and T2 after weight recovery.	Not mentioned		T1: Significantly lower NAA, Glx and mI. No difference was found in the concentration of Cr and Chol. A positive correlation was reported between NAA & T3 and NAA & Wechsler Intelligence Scale for children (WISC). No difference in metabolites concentration between males and females. T2: A statistically significant increase in NAA and a non-significant increase in Glx in frontal GM.	No clinical interpretations.
Schlemmer (1997)	MRI 1.5T and MRS Tested at one time point.	Two ROIs: the parieto-occipital WM and the thalamus.		A 25% elevation of Cho/Cr and a 25% depression of NAA/Cho were observed in the parieto-occipital WM. No statistically significant differences were found in thalamus. No correlations were found between the metabolic ratios and age, weight or BMI.	No clinical interpretations. The abnormal phospholipid metabolism of membranes might be responsible for brain atrophy.
MRI-DTI
Pfuhl (2016)	DTI, MRI Tested at one time point.	Global tractography	Urine specific gravity	No significant volumetric differences or microstructural abnormalities in 18 WM tracts. All four diffusivity indices were evaluated (FA, MD, AD, RD).	The preserved WM microstructure may explain why adolescents often do not show marked impairment in executive functioning.
Hu (2017)	DTI Tested at one time point.	VBM	At least 1 week of supervised meals and hydration.	Decreased FA in the left superior frontal gyrus, medial frontal gyrus, ACC, middle frontal gyrus, IFG, thalamus and bilateral insula. Positive correlations between the FA of the left IFG, insula, thalamus and BMI.	WM alterations in prefrontal cortex, parietal lobe and subcortical regions may be associated with impaired cognitive functions.
Gaudio (2017)	DTI Tested at one time point.	VBM	Not assessed	Decreased FA in the left anterior and superior corona radiata and in the SLF. Decreased AD in the left superior and anterior corona radiata and in the SLF bilaterally, external capsule, posterior limb of the internal capsule and posterior thalamic radiation. No differences in MD, RD. No significant correlations.	WM alterations may be involved in impaired cognitive flexibility and body image distortion.
Travis (2015)	DTI Tested at one time point.	Tractography, relaxometry	Not assessed	Twenty-six WM tracts were identified, 9 bilateral cerebral and 8 subdivisions of the corpus callosum. FA was found decreased in 4 of 26 tracts (including bilateral fimbria—fornix and right SLF and motor subdivisions of corpus callosum) and increased in 2 (including right anterior thalamic radiation and left SLF). R1 was decreased in 11 of 26 tracts mainly in corticospinal tracts and subdivisions of the corpus callosum—body and splenium. No significant associations between BMI and clinical measures.	WM alterations seem to be related to myelin quality, affecting cognitive, emotional and social functions.
Vogel (2016)	DTI Tested at one time point.	TBSS	Urine specific gravity	T1: Increased FA in bilateral frontal, parietal and temporal areas, including bilateral superior corona radiata, corpus callosum, anterior and posterior thalamic radiation, anterior and posterior limb of internal capsule and left inferior longitudinal fasciculus. FA increase due to reduced RD, not altered AD. Most areas with FA increase exhibited reduced MD. T2: No differences in FA after weight rehabilitation. Higher FA was associated with faster weight loss.	The different pattern of WM microstructural changes in adolescents compared to adults may reflect a different susceptibility and reaction to semi starvation in the still developing brain or a time-dependent pathomechanism differing with extent of chronicity.
Olivo (2018)	DTI Tested at one time point.	TBSS	Patients were instructed to eat before the scanning.	No differences detected in diffusivity indices (FA, MD, RD, AD).	Preserved WM microstructure in patients with atypical AN suggests that alterations observed in full syndrome may constitute state-related consequences of severe weight loss.
Von Schwanenflug (2019)	DTI Tested at two time points: At baseline (T1) and after partial weight restoration (T2).	TBSS	Urine specific gravity	T1: In acAN significantly decreased FA and increased MD, AD, RD in corpus callosum, mainly in the body and increased FA in the right corticospinal tract. Additionally, increased FA in the right SLF. T2: After partial weight restoration significantly increased FA and dicreased MD, AD, RD in the fornix extending into bilateral optic radiation. No clinical correlations.	The decreased FA in corpus callosum may contribute to the distorted body image.