Association between childhood maltreatment and cortical folding in women with eating disorders

Childhood maltreatment (CM) is associated with distinct clinical and biological characteristics in people with eating disorders (EDs). The measurement of local gyrification index (lGI) may help to better characterize the impact of CM on cortical structure. Thus, the present study investigated the association of CM with lGI in women with EDs. Twenty‐six women with anorexia nervosa (AN) and 24 with bulimia nervosa (BN) underwent a 3T MRI scan. All participants filled in the Childhood Trauma Questionnaire. All neuroimaging data were processed by FreeSurfer. LGI maps underwent a general linear model to evaluate differences between groups with or without CM. People with AN and BN were merged together. Based on the Childhood Trauma Questionnaire cutoff scores, 24 participants were identified as maltreated and 26 as non‐maltreated. Maltreated people with EDs showed a significantly lower lGI in the left middle temporal gyrus compared with non‐maltreated people, whereas no differences emerged in the right hemisphere between groups. The present study showed that in people with EDs, CM is associated with reduced cortical folding in the left middle temporal gyrus, an area that could be involved in ED psychopathology. This finding corroborates the hypothesis of a ‘maltreated ecophenotype’, which argues that CM may allow to biologically, other than clinically, distinguish individuals with the same psychiatric disorder.


| INTRODUCTION
People with eating disorders (EDs) exhibit a prevalence of childhood maltreatment (CM) higher than general population (Caslini et al., 2016). In people with either Abbreviations: AN, anorexia nervosa; BMI, body mass index; BN, bulimia nervosa; CM, childhood maltreatment; CTQ, Childhood Trauma Questionnaire; EDs, eating disorders; lGI, local gyrification index. anorexia nervosa (AN) or bulimia nervosa (BN), CM has been found to be associated with an earlier age at onset of the disorder, a greater clinical severity, a more frequent psychiatric comorbidity (Molendijk et al., 2017) and a poorer treatment response (Castellini et al., 2018). Moreover, compared with ED people without CM, maltreated people with EDs have been shown to exhibit reduced grey matter volume in several brain areas (Monteleone et al., 2019), cortical thinning (Cascino et al., 2022) and alterations in both the basal and stressinduced activity of the hypothalamus-pituitary-adrenal axis . Indeed, it has been suggested that CM may act as a stressor to produce a cascade of physiological and neuro-endocrine reactions that alter brain-development trajectories (Teicher et al., 2016), because the human brain is a highly plastic organ whose development is regulated by genes but also shaped by environmental factors (Lim et al., 2014). Therefore, several findings reporting an association between CM and alterations in brain structure in clinical and non-clinical populations support the existence of a 'maltreated ecophenotype' biologically and clinically different from the non-maltreated one (Teicher & Samson, 2013).
Local gyrification index (lGI), which is the ratio between the total surface area including the cortex hidden in sulci and exposed cortical surface, is a surfacebased measure of the brain cortex, which reflects cortical folding (Schaer et al., 2008). Although the lGI was considered to remain stable throughout brain maturation after birth (Armstrong et al., 1995), it has been shown that there is a continuous molding of the brain surface morphology throughout childhood and adolescence with a decrease of lGI likely associated with the decrease of underlying grey matter volume (Hogstrom et al., 2013;Su et al., 2013). Previous studies have shown a reduced grey matter volume in the right paracentral lobule and in the left inferior temporal gyrus (Monteleone et al., 2019) and a cortical thinning in the left superior frontal and in right caudal middle frontal and superior parietal gyri in maltreated people with EDs compared with non-maltreated ones (Cascino et al., 2022). Therefore, the measurement of lGI may help to better characterize the impact of CM on cortical structure in people with EDs. Compared with healthy controls, people with EDs have been shown to exhibit a lower lGI ) that has been found associated with greater clinical severity or with a poor clinical outcome in AN (Collantoni et al., 2019). To our knowledge, so far, no study explored the association between the degree of cortical folding and the history of CM in people with EDs, whereas one single study has reported a reduction of lGI in the lingual gyrus and the insula in a community sample of maltreated children compared with non-maltreated peers (Kelly et al., 2013).
Therefore, in the present study, we investigated the association of CM with lGI in women with EDs. Our study hypothesis was that maltreated women with EDs would show a reduced cortical folding compared with non-maltreated people with EDs.

| METHODS
Study participation was proposed to patients consecutively admitted to the adult EDs outpatient centre of the University of Salerno. Participants with EDs had to meet the following inclusion/exclusion criteria: (1) diagnosis of current AN or BN, according to DSM-5 criteria, confirmed by the Structured Clinical Interview for DSM-5 Disorders-Research Version; (2) age ≥ 18 years; (3) absence of severe physical disorders, comorbid schizophrenia or bipolar disorder; (4) no history of head trauma; (5) no use of psychotropic medications, psychoactive substance or oral contraceptives; and (6) right handedness. Diagnostic assessments were performed by trained psychiatrists. Participants gave their written consent after being fully informed of the nature and procedures of the study. The study was approved by the local Ethics Committee and performed in accordance with the Declaration of Helsinki and its later amendments. Participants completed the short form of the Childhood Trauma Questionnaire (CTQ; Bernstein et al., 2003). The CTQ is a 28-item questionnaire investigating childhood experience across five types of CM: emotional neglect, emotional abuse, sexual abuse, physical neglect and physical abuse. Validated cutoff scores indicating the occurrence of maltreatment have been provided for each subscale (Walker et al., 1999). We classified as 'maltreated' those participants who scored higher than the threshold in at least one subscale of the CTQ and the others as 'non-maltreated'.
All subjects underwent a 3T scanner (MAGNETOM Skyra, Siemens, Erlangen, Germany). The image protocol consisted of the acquisition of T1-weighted 3D Magnetization Prepared RApid Gradient Echo (MPRAGE), sagittal orientation, matrix size of 256 Â 240, FOV of 240 Â 256 mm, 136 slices, slice thickness of 1.2 mm, inplane voxel size of 1 Â 1 mm, flip angle of 9 and TR/TE of 2300/2.98 ms, one average. All participants were outpatients and underwent the MRI scanning within 2 weeks from the first assessment, before starting any type of treatment. Patients with comorbid depression or anxiety disorder were scanned before starting specific antidepressant or anxiolytic drug treatments.
All images were processed using FreeSurfer, version 6.0 (https://surfer.nmr.mgh.harvard.edu/). The lGI was measured according to previously validated algorithms (Schaer et al., 2008). Maps of lGI were computed in order to perform a vertex-by-vertex analysis.
LGI maps underwent a general linear model (GLM) analysis to evaluate differences between maltreated and non-maltreated groups. The analyses were carried out in accordance with the instructions available at https://surfer.nmr.mgh. harvard.edu/fswiki/FsTutorial/GroupAnalysisV6.0.
Age, body mass index (BMI) and illness duration were included as nuisance covariates. Statistical maps were then corrected for multiple comparisons using Monte Carlo simulation with 10,000 iterations and a statistical threshold of p = 0.05 to retain only clustered vertices (Hagler et al., 2006). We extracted from each cluster showing significant differences in the mean lGI values to plot the trend of these measures across groups.

| RESULTS
Twenty-six women with AN (19 restrictive subtypes and seven binge-purging subtypes) and 24 with BN participated in the study. Six patients (12%) reported a diagnosis of comorbid anxiety disorder, whereas four patients (8%) reported a diagnosis of comorbid major depression.
Based on CTQ cutoff scores, 12 participants with AN and 12 with BN were identified as maltreated and 14 participants with AN and 12 with BN as non-maltreated. The groups did not differ in age, age at onset, BMI and illness duration (Table 1). The comparison between participants with AN and BN is reported in the Supporting Information.
Maltreated people with EDs showed a significantly lower lGI in the left middle temporal gyrus (cluster dimension: 302.47 mm 2 ; Talaraich coordinates, x: À49.5, y: À14.3, z: À27.6) compared with non-maltreated people, whereas no significant differences emerged in the right hemisphere between the two groups ( Figure 1).

| DISCUSSION
According to our study hypothesis, women with EDs and history of CM showed a lower lGI in the left middle temporal gyrus compared with non-maltreated women with EDs.
Although the association between deranged neural structure in several brain regions and CM in people with EDs was already reported (Cascino et al., 2022;Monteleone et al., 2019), this is the first study showing an association between reduced cortical folding and CM in people with EDs. Indeed, previous studies have explored lGI in people with EDs without taking into account their history of CM. Bernardoni et al. (2018) reported a widespread reduction of GI in patients with AN that normalized after weight restoration, suggesting that the reduced lGI in people with AN could be considered a state-dependent alteration related to malnutrition. Instead, Miles et al. (2018) reported reductions in bilateral frontal lGI in women with both acute and remitted T A B L E 1 Demographic characteristics of maltreated and no maltreated participants with eating disorders (EDs).

Maltreated EDs (n = 26)
No maltreated EDs (n = 24) t p Age ( F I G U R E 1 Differences in local gyrification index between maltreated and non-maltreated patients with eating disorders. AN. Therefore, the lack of CM assessment might at least partially justify the inconsistency between those studies. When lGI has been assessed in community children exposed to maltreatment, a reduced cortical folding was found in two clusters of the left hemisphere, the first located in the lingual gyrus and the second extending across the insula into the pars opercularis (Kelly et al., 2013). In the same study, a reduction of cortical surface area was found in the left middle temporal gyrus. These findings are consistent with reports of reduced grey matter volume in this same region in maltreated children (Hanson et al., 2010) and support the idea that CM has an impact on cortical development.
As for the possible pathophysiological significance of our results, it is worth mentioning that the middle temporal gyri are related to higher order visual perception. Increased activity in the inferior and middle temporal gyri have been specifically found in response to own face stimuli in healthy individuals (Sugiura et al., 2008). People with AN displayed increased activity in the lingual, and inferior and middle temporal gyri to images of their own face, compared with neutral control images, as well as to healthy individuals (Phillipou et al., 2015). Moreover, people with AN and BN have been shown to display a weaker activation of the left middle temporal gyrus to looking at one's own body in comparison with healthy controls, whereas people with AN displayed a higher bilateral activation in this brain area to looking at another woman's body compared with controls (Vocks et al., 2010). Therefore, the lower lGI in the temporal region of maltreated ED patients may contribute to some aspects of their psychopathology, although it remains to be demonstrated that maltreated people with EDs exhibit different activation of this region to the perception of their own face and body than those without CM.
There are some limitations of our study that deserve to be acknowledged. First, the relatively small size of our sample, although consistent with most of the neuroimaging studies in people with EDs, did not allow to identify differences between ED diagnostic groups. To overcome this issue, BMI and age were included as nuisance covariates in the GLM analysis. Second, the lack of a control group did not allow us to disentangle the effect of psychiatric disorder from that of CM. Finally, the retrospective self-report of childhood experiences, as assessed by the CTQ, does not exclude recall bias.
In conclusion, the present study showed that CM is associated with reduced cortical folding in the left middle temporal gyrus in people with EDs. This finding further corroborates the hypothesis of the existence of a 'maltreated ecophenotype', which argues that CM may allow to biologically, other than clinically, distinguish individuals with the same psychiatric disorder (Teicher & Samson, 2013). Further studies with larger samples are needed to confirm whether changes in lGI represent a biological marker of CM exposure in women with EDs and whether these alterations could mediate the risk to develop an ED in the adulthood among maltreated individuals.