White matter integrity is reduced in bulimia nervosa.

The eating disorder (ED) bulimia nervosa (BN) is most distinctively characterized by episodic binge eating, followed by purging behaviors (1), although individuals with BN also frequently restrict food intake in between binges. BN is associated with anxious traits (2) and comorbid anxiety disorders (3), as well as depression (1), but also altered taste perception (4; 5) and brain reward response (6–8). The neurobiologic underpinnings of increased anxiety in BN are still unknown, although brain serotonin 1A, 2A and dopamine D2/3 receptors have been associated with anxious traits in the past (9–12). Recently, we showed that brain white matter (WM) in AN was related to anxiety (13). In that study we used diffusion tensor imaging (DTI), and found that patients with AN had reduced WM integrity in the bilateral fimbria fornix, an outflux tract of the medial temporal cortex and hippocampus, as well as in the inferior and superior fronto-occipital fasciculus and cingulum. Importantly, fornix WM integrity, expressed as the so-called fractional anisotropy (FA), predicted inversely harm avoidance and trait anxiety in the AN individuals. That finding raised the question whether BN would have similar WM alterations that would predict high trait anxiety in the disorder. For a long time it has been known that WM lesions in the brain may lead to disconnection syndromes, and studies over the past ten years now have increasingly implicated more subtle WM tract functionality in neuropsychiatric disorders (14). WM functionality can be studied using DTI, a relatively new tool in psychiatric research that uses magnetic resonance imaging (MRI) to assess WM functionality. DTI maps WM pathways by measuring water diffusivity along axons expressed as FA (15). FA is therefore a measure of the integrity of axons, their myelination, and density. A second measure is the apparent diffusion coefficient (ADC), which measures water diffusivity at the voxel level and is thought to be an indicator of the health of the axonal cells. A higher ADC indicates dispersed water diffusion reflecting disruption of these cells (15). Studies have investigated WM functionality in, for instance, mood and psychotic disorders (16), most commonly reporting reduced FA values across many brain regions, but increased WM FA has also been reported (17). The functionality of WM has not been studied previously in BN. Psychiatric disorders including BN are characterized by a complex interplay of cognitive and emotional behavior and rely heavily on the connections between brain regions (18). Thus, functional WM alterations in BN could be an important aspect of the pathophysiology of this disorder. Only few brain-imaging studies have investigated brain structure such as gray matter (GM) or WM in BN. Early studies analyzed total GM or WM and cerebrospinal fluid volumes. The first structural studies have shown potential brain atrophy (19; 20) or widening of the sulci and enlargement of the ventricles despite normal body weight (21–23). With advancement in technology using ‘voxel based morphology’ (VBM), that is a whole brain analysis of structural brain images that are normalized to a standard space and compared across groups, studies now can in much more detail investigate localized volume alterations that could be related to a specific brain disorder. One recent study found normal GM volumes in BN but “drive for thinness” was positively related to parietal cortex GM volume (24). Another study found increased orbitofrontal cortex and ventral striatal GM volume compared to controls (25). In that study purging frequency and body mass index (BMI, weight in kg/height in m2) were directly related to the volume of those regions. One study in recovered BN indicated normal WM volume (26). Most studies did not take comorbid depression or medication use into account, factors that have been shown to affect brain volumes. This and the differing analysis techniques most likely account for the conflicting results and more definite studies are still needed. Various neurotransmitter alterations were also found in BN, such as reduced serotonin (5HT) transporter binding in thalamus and hypothalamus (27). After recovery BN showed reduced midbrain but increased cingulate cortex 5HT transporter availability (28), reduced orbitofrontal 5HT2A (29), and increased 5HT1A receptor binding (10), most prominently in prefrontal, cingulate and parietal cortex regions. Altogether, those studies indicate brain alterations in BN in ill and recovered states, which may be related to altered food intake, mood states and anxiety, but their relationship to WM function is unclear. In this study we wanted to investigate whether we would find altered WM functionality in BN similar to results we previously reported in AN (13). Following on our AN results, we hypothesized that we would find reduced FA values in BN in the fornix area, fronto-occipital fasciculus and the cingulum, and those findings would not be due to WM volume alterations. We further hypothesized that lower FA would predict higher harm avoidance and trait anxiety. If BN were to have similar WM disturbances as the AN individuals, that might indicate a common vulnerability across eating disorder groups. Furthermore, we wanted to test whether BN specific behaviors such as binge/purge frequency would predict WM functionality, as it did predict brain reward response in BN in the past (8).