4.1 Global brain volume alterations
We found no significant differences between BN patients or controls in total GMV, WMV or intracranial volume, which is consistent with some previous studies in female patients[18, 19]. Our patients did show increased CSF volume, consistent with results obtained with one group of German patients [20] but not with a second German group [18]. These differences may reflect differences in methodology. Our observation of increased CSF volume but unchanged TIV suggests that the reduced GMV in BN is replaced by increased CSF.
4.2 Regional GMV alterations
Patients showed reduced GMV in the SFGmed.L, STG.R, DCG.L, DCG.R and SFGdor.L, and no regions with increased GMV. These results are consistent with some studies but not with others, which have reported reduced GMV in the superior frontal gyrus and cingulate cortices [6], increased GMV in the medial orbital frontal cortex and the ventral striatum [4], or increased GMV in the paracentral lobule, precuneus, left putamen and insula, but reduced GMV in the caudate nucleus and thalamus [19]. Thus, different studies of BN patients vary considerably in the regions showing GMV alterations and the direction of the alteration (increase or decrease). These discrepancies may be due to differences in disease course, disease stage, medication history, ethnicity and other factors. Future studies should strive to examine larger, multi-center populations that may help reduce the influence of clinico-demographic factors on VBM.
We found reduced GMV in the SFG and STG of BN patients, which was also reported in a study of Japanese women with anorexia nervosa (AN) [21]. The SFG is located in the superior part of the prefrontal cortex, which plays essential roles in executive control [22], including self-regulation of eating behavior[23]. Functional MRI studies have shown the SFG to be deactivated in BN in the presence of the expected reward during a reward-based learning task [24], during a Simon spatial incompatibility task[25]. The GMV reduction in the STG in the Japanese study [21] was associated with body dissatisfaction. Reduced STG volume may cause abnormal body image perception and excessive concern about body shape and weight, leading to restrictive or binge eating. A functional MRI study has shown deactivation in STG when the individual was thinking about eating food [26]. Indeed, the structural and functional abnormalities in the SFG and STG in patients with BN or other eating disorders have been associated with deficits in self-regulation and reward processing[25]. These alterations are likely lead to binge eating behaviors, thereby contributing to the development and maintenance of BN. The prefontal cortex is an important part of the fronto-striatal circuits, which is involved in self-regulatory control. Previous study has found deficient activity in the fronto-striatal circuits associated with impaired self-regulatory processes [25, 28]. The prefrontal cortex is also an important component of the forebrain system, which may contribute to eating dysregulation by driving maladaptive over- and undereating [29].
Patients with eating disorders such as BN typically show emotional symptoms such as depressive and anxiety disorders [14, 30]. How these symptoms relate to inappropriate eating behaviors is controversial. Many studies have suggested that inappropriate eating behaviors are maladaptive strategies to manage negative feelings [31]. However, other studies have suggested that the eating behaviors alter mental state, such as by activating serotonin projections of the dorsal raphe to the prefrontal cortex; as a result, normalizing eating behavior can normalize psychiatric symptoms [32]. We found reduced GMV volume in the STG of our BN patients, and such a reduction has been linked to more severe symptoms of depression and anxiety disorders[33, 34]. We also found reduced GMV in the right superior temporal gyrus of our BN patients, and such a reduction has been linked to risk of suicide attempts in community samples of adolescents [35]. Further studies should examine whether such GMV reductions precede or follow the onset of inappropriate eating behaviors.
We observed reduced GMV in the cingulate cortex, and such a reduction has also been reported in AN [36]. Reduced cortical thickness in the cingulate cortex has been reported in BN [36]. In addition, BN patients in one study showed reduced activation of the anterior cingulate cortex when confronted with disease-specific food stimuli [37]. The cingulate cortex is involved in pain perception [38], so the observed GMV reduction may be associated with decreased pain sensitivity in BN, which even persists after recovery [39]. This decrease may be a mechanism to compensate for the physical discomfort of binge/purge behavior, creating feelings of calm and euphoria during such behavior [39].
4.3 Correlations between GMV alterations and clinical variables
BN-associated GMV reductions did not vary significantly with BMI, duration of illness, or total scores on the RSES or CSES. Age did, however, correlate negatively with GMV in the SFGmed.L, DCG.R, DCG.L and SFGdor.L.
As expected, RSES and CSES scores were lower in BN patients than in controls, reflecting that a frequent symptom of BN is body image dissatisfaction [40], which is linked to low self-esteem [41, 42]. At the beginning of this study, we hypothesized that the RSES and CSES scores of BN patients would be associated with GMV alterations. However, we did not observe such an association in our patients. In contrast, a study of BN patients in the US did detect an association between brain functional alterations and self-esteem: activation of the right temporoparietal junction, precuneus and dorsal anterior cingulate cortex was weaker in BN patients than in controls during execution of social and self-knowledge tasks [43]. These brain regions are associated with self-knowledge and social processing. The discrepancy between that study and ours may be due to methodological differences. Whether the GMV reductions in our BN patients correlate with functional changes affecting self-esteem and self-evaluation should be explored further.
4.4 Study limitations
This study analyzed only women, even though some men are also affected by eating disorders. In addition, some of our patients had mild anxiety and depressive symptoms, which may have confounded our results but which may also make our sample more representative of typical BN patient populations [14]. The cross-sectional nature of our study means that further work is needed to clarify whether the observed GMV alterations are the result of BN pathology or the consequence of binge-eating and compensatory behaviors. Our results should be verified and extended with much larger samples. In our study, we explored only brain structural alterations in BN; future work should explore changes in brain functional connectivity and networks in order to understand how brain areas interact with one another.