An initial ^99mTc-HMPAO-SPECT study in bulimia nervosa compared rCBF in 5 patients with bulimia (age 21.0 ± 2.9 years), 8 patients with anorexia (age 24.1 ± 7.8 years), and 9 healthy controls (age 20.3 ± 1.0 years). Blood flow was measured before and after eating a custard cake. Flow was expressed as ratio units, by comparing tracer uptake in a brain region to uptake in the cerebellum. Differences between the groups were observed only during the first scan. Whereas anorexics showed reduced flow in the left parietal region, bulimics demonstrated significantly increased flow in the bilateral inferior frontal and left temporal regions compared to the control group. Flow increases were noted in anorexics, and flow decreases in bulimics after eating; thus, any differences between the groups were abolished by the food stimulus (Nozoe et al. 1995). Since the frontal cortical area of the brain controls feeding together with the hypothalamus, flow differences in frontal regions may reflect disease-related differences in cortical function. Frontal lobe damage can result in hyperphagia; thus, dysfunction of this brain area could be related to binge eating in bulimics.

A case report examined rCBF in a male patient with bulimia nervosa (age 27 years), first during a period of purging and, 22 days later, during binge eating, using ¹²³I-IMP and SPECT. Global CBF was higher during the binge-eating phase than during the purge phase. In the purge phase, an asymmetric pattern was noted, with lower values for rCBF in the right temporal, parietal, and occipital lobe. This asymmetry disappeared during binge eating. Thus, rCBF differs between the two phases of bulimia nervosa, and flow asymmetry is dependent on the eating state (Hirano et al. 1999).

A subsequent study examined rCBF in 9 women with bulimia nervosa who had recovered from their disorder by showing stable food intake, normal weight, and regular menses for a period of more than 1 year. rCBF was measured with the tracer [¹⁵O]water and PET, and flow patterns were compared to those of an age-matched healthy control group (13 females). Significant differences between the groups were not observed, but rCBF in several cortical areas and left thalamus was significantly and inversely related to length of recovery in the patient group (Frank et al. 2000). Apparently, differences in rCBF between bulimics and controls are state-related and disappear during recovery.

A later [¹⁵O] water PET study compared rCBF in 10 women who had recovered from restrictive anorexia, 8 women who had recovered from binge/purge anorexia, 9 women who had recovered from bulimia, and 18 healthy control subjects. Partial volume-corrected rCBF values in the four groups were not significantly different in any brain region. Thus, rCBF appears to normalize after recovery not only in bulimics but also in subjects with anorexia nervosa (Frank et al. 2007).

An interesting PET study with ¹⁵O-water has suggested a vagal pathophysiology for bulimia nervosa and the accompanying depressive symptoms. Mechanical distention of the stomach with a balloon in female healthy volunteers and the associated vagal stimulation was shown to result in activation of several brain areas, including areas which are involved in the emotional aspects of eating (lateral inferior frontal and orbitofrontal cortex) and in the symptoms of depression (anterior cingulate cortex). The hypothesis that vagal afferent activity is involved in the cycles of binge eating and vomiting in bulimia nervosa with their associated symptoms of depression was subsequently tested in two ways: first, pain detection thresholds were examined in patients with BN and were found to fluctuate in association with bulimic episodes, suggesting fluctuation of vagal activity. A double-blind treatment protocol of bulimic individuals was then carried out with the serotonin 5-HT₃ antagonist ondansetron. This treatment significantly decreased binge eating and vomiting in BN patients, abolished the fluctuation in pain thresholds, and reduced the depressive symptoms. These findings were interpreted as evidence for the hypothesis that cyclic increases in vagal activity drive the urge to binge eat and vomit (Faris et al. 2006).

A large ^99mTc-ECD SPECT study examined rCBF in 67 female patients with eating disorders (31 restrictive anorexics, 16 binge/purge anorexics, and 20 bulimics). SPM analysis was applied to the SPECT data, and brain areas were identified in which perfusion covaried with symptoms measured by the Eating Disorder Inventory. The only significant correlation observed was a positive correlation between scores on body dissatisfaction and ineffectiveness and rCBF in the prefrontal and parietal cortex (Goethals et al. 2007b). Based on this finding, the authors argued that neurobiological findings in eating disorders, such as changes in the serotonergic system, may reflect not only emotional and behavioral factors (e.g., decreased impulse control) but also cognitive-evaluative features: attention, memory, and judgment being continuously affected by an overconcern with eating, body size, and shape.

This hypothesis was explored in a later study in which rCBF was measured with ^99mTc-HMPAO and SPECT in 34 subjects (9 restrictive anorexics, 13 bulimics, and 12 healthy controls) under 3 different conditions: at rest, after viewing a neutral stimulus (landscape video), and after viewing their own-filmed body image (positive stimulus). Anorexics showed a hyperactivation of the left parietal and right superior frontal cortex by the positive as compared to the neutral stimulus. Bulimics showed a hyperactivation of the right temporal and right occipital areas. Activation of the right temporal lobe may reflect an aversive response and abnormal activation of the left parietal lobe the storage of a distorted prototypical image of the body (Beato-Fernandez et al. 2009). In a follow-up study performed in the same subjects, the right temporal lobe activation in bulimics was shown to persist even after 1 year of participation in a treatment program for eating disorders (Rodriguez-Cano et al. 2009). Thus, although progress was made in the control of purging symptoms, mood (depression), and self-esteem, the aversive response of the patients towards their own body shape was still present after 1 year, and more specific long-term therapies are needed for the treatment of body dissatisfaction.

In summary, using either SPECT or PET, abnormal activation of certain brain areas has been detected both in BN and AN after presentation of various stimuli, related either to food intake or body shape (Table 24.2). These responses have been interpreted as symptoms of anxiety or phobia. Most abnormalities disappear after successful treatment, but abnormal activation of the right temporal lobe may persist in BN and reflect persistence of body dissatisfaction.

The first study of cerebral glucose metabolism in anorexia nervosa was published in 1987. Five female anorectic patients were scanned with PET and the tracer FDG, both during the anorectic state and after behavioral therapy. Scans were made in the resting state, with eyes closed and ears unplugged. Significant bilateral hypermetabolism in the caudate nucleus was observed in the anorectic state in comparison with results obtained after weight gain (Herholz et al. 1987). A subsequent study included nine patients with bulimia and seven patients with anorexia. Relative glucose metabolism in the caudate, compared to the rest of the brain, was significantly higher in anorexia than in bulimia (Krieg et al. 1991). These findings could be interpreted as high motor activity in the anorexic patients resulting in increased dopamine turnover in the caudate nucleus and metabolic hyperactivity.

A more extensive study appeared in 1995. FDG-PET scans were made during rest, with eyes closed and with low ambient noise, in 20 underweight anorectic girls and ten age-matched healthy female volunteers. Compared to controls, the patients showed a global hypometabolism, the most striking difference being present in the frontal and parietal cortex (Delvenne et al. 1995). The observed hypometabolism might reflect a primary cortical dysfunction underlying anorexia nervosa, but it could also be related to physiological or morphological changes as a consequence of starvation or to depression in the patient group. A subsequent study examined cerebral glucose metabolism in ten anorectic girls, both at the onset of therapy and after weight gain. Ten age-matched healthy females were used as controls. In the underweight state, patients showed the same hypometabolism as was observed previously, but after weight gain, cerebral glucose metabolism normalized, and patient data were no longer significantly different from those acquired in controls although a trend towards inferior metabolism in some brain areas was still apparent (Delvenne et al. 1996). For this reason, glucose hypometabolism appears to be state – rather than trait-related. A third FDG-PET study included ten underweight females with anorexia nervosa, ten underweight depressed patients, and ten depressed patients with normal weight (all age- and sex-matched). Absolute values for glucose metabolic rate were significantly correlated with body mass index in all subjects; the lowest values were observed in the anorexic group. Thus, glucose hypometabolism seems to be a consequence of low weight (Delvenne et al. 1997b). The hypothesis that cerebral hypometabolism of glucose is a consequence of starvation was confirmed in a further study which compared FDG-PET scans of ten young depressed patients with low weight without anorexia nervosa with those of ten age- and sex-matched healthy volunteers. Absolute global and regional metabolic rates of glucose were significantly lower in the patient group than in the control group (Delvenne et al. 1997a). One factor that could partially explain the described findings is the downregulation of glucose transporters under nutrient starvation (Merrall et al. 1993), since these proteins are involved in uptake of FDG from the blood.

A more recent PET study involved 14 women with anorexia nervosa, 20 aged-matched healthy control subjects, and the same group of anorexics after randomization to 3 weeks of low-dose replacement testosterone therapy or placebo. The study confirmed that cerebral glucose metabolism is significantly reduced in several cortical areas of anorexics as compared to controls. Testosterone therapy resulted in increases of metabolism in many areas including one region (posterior cingulate) which had previously shown hypometabolism (Miller et al. 2004). The clinical significance of this finding should be further examined.

In several PET studies (Delvenne et al. 1997b, 1999), relative glucose metabolism in the parietal cortex of anorexics was shown to be significantly decreased compared to controls and significantly increased in the caudate nucleus. Similar decreases of relative glucose metabolism were also noted in the parietal cortex of patients with bulimia; thus, it appears to be a common feature in both eating disorders (Delvenne et al. 1999).

Two PET studies have examined changes of cerebral glucose metabolism in an animal model of anorexia nervosa. In the first study, female Wistar rats received either free access to food or were severely restricted in their food intake until a 30 % weight loss occurred. Body weight was then maintained at 70 % of the control value by adjusting daily food intake and by providing free access to a running wheel. The tracer ¹⁸ F-FDG was administered intraperitoneally and was allowed to distribute in the body of the awake animals for 50 min before the rats were anesthetized and scanned. Absolute values for glucose metabolic rate could not be determined by this protocol (since an arterial input function was missing), but relative glucose metabolism was found to be significantly altered in the food-deprived animals, decreases being noted in hippocampus and striatum, and increases in the cerebellum (Barbarich-Marsteller et al. 2005). The second study used a somewhat different approach. Here, food restriction (1.5 h instead of 24 h/day) and running wheel access were combined from the beginning. Animals were scanned after 9 days, when body weight in the food restricted/exercised group had declined by 20 %. FDG was not allowed to distribute in awake but in pentobarbital-anesthetized rats, and the study used male animals rather than females. Decreases of glucose metabolic rate were observed in cortical areas and striatum, whereas increases occurred in mediodorsal thalamus, ventral pontine nuclei, and cerebellum. Brain metabolism in cingulate and the surrounding motor and somatosensory cortex were positively correlated to weight loss (van Kuyck et al. 2007). Both studies suggested that changes of cerebral metabolism can be detected with PET in animal models of anorexia nervosa and that these changes are related to loss of body weight.

A recent study evaluated how cerebral glucose metabolism correlates with clinical improvement after deep brain stimulation (DBS) in patients with anorexia nervosa. The authors showed that reversal of abnormalities seen in the anterior cingulate, insula, and parietal lobe at baseline (i.e., before DBS) is strongly correlated with the clinical benefits caused by this kind of therapy besides some adverse effects associated with DBS (Lipsman et al. 2013).

To summarize the findings in humans, most PET studies have reported cerebral hypometabolism in patients with AN as compared to controls, particularly in the frontal and parietal cortex (Table 24.3). Such hypometabolism appears to be a consequence of starvation rather than a trait leading to the development of anorexia. The ratio of metabolism in caudate nucleus to the rest of the brain is increased in anorexia. This may be a symptom of excessive motor activity in anorexics.

In an early FDG-PET study, cerebral metabolic rate of glucose was examined in eight women with bulimia and eight normal healthy females during the performance of a visual vigilance task. Healthy subjects showed asymmetry with higher glucose metabolism in the right than in the left hemisphere, but this asymmetry was absent in the patient group suggesting absence of the normal right activation and impaired processing of the visual task (Wu et al. 1990). In a subsequent publication, an additional group of eight women with major affective disorder was included. In contrast to the bulimics, depressed subjects showed normal activation in the right hemisphere during processing of the visual task, but they had decreased metabolism in the basal ganglia. Thus, although bulimics frequently suffer from symptoms of depression, their regional pattern of brain activation differs from that observed in major affective disorder (Hagman et al. 1990).

A later FDG-PET study examined the cerebral metabolic rate of glucose in 11 women with bulimia nervosa and 18 healthy age- and sex-matched control subjects. The bulimics were also tested for symptoms of major depression and obsessive-compulsive disorder. No group differences in orbitofrontal glucose metabolism were detected, but lower metabolism in the left anterolateral prefrontal cortex was correlated to greater depressive symptoms in the patient group (Andreason et al. 1992).

Another imaging study with PET and FDG examined cerebral glucose metabolism at rest (eyes closed, ears unplugged) in 11 normal-weight bulimic girls and 11 age- and sex-matched healthy volunteers. In contrast to the previous study, both global and regional levels of glucose metabolism were significantly lower in bulimics than in healthy controls. Relative levels of metabolism (compared to the rest of the brain) were reduced only in parietal cortex. No correlations were found between absolute or relative glucose metabolic rates, body mass index, anxiety scores, or scores of depression (Delvenne et al. 1997c). The observed reductions in glucose metabolism could either be a consequence of nutritional deficiencies or a brain dysfunction underlying eating disorders.

In summary, most PET studies have reported that cerebral glucose metabolism in bulimics is either decreased or not significantly different from that in healthy controls (Table 24.3). However, data from FDG studies using an activation paradigm suggest that the processing of visual tasks may be impaired in BN (Table 24.4).

Several observations suggest that eating disorders may be associated with altered serotonergic neurotransmission in the brain. Serotonergic signaling in the hypothalamus is known to be involved in the control of food intake and body weight, serotonin acting as an eating-inhibitory substance (Leibowitz 1986). Serotonin (5-HT) uptake in platelets of bulimia nervosa patients is increased compared to healthy controls (Goldbloom et al. 1990), and selective serotonin reuptake inhibitors (SSRIs) like fluoxetine can suppress bulimic symptoms (Freeman and Hampson 1987). Such observations (and many others, including the role of serotonin in regulation of mood and impulse control) have prompted imaging studies of 5-HT receptors and transporters in the brain of patients with eating disorders (Table 24.5; reviewed in Bailer and Kaye 2011; Barbarich et al. 2003; Kasper et al. 2002; Kaye et al. 2005a, b).