Bulimia nervosa (BN) is a mental illness characterized by repetitive overeating and purging behaviors to prevent weight gain, and is commonly seen in young women. Patients with BN experience significant psychological, physical, and social suffering related to their ongoing eating disorder (ED), including BN-specific psychopathology. Behavioral abnormalities driven by craving for foods are a prominent feature of BN, and neuroimaging studies using food-related tasks are thus an important approach for elucidating the correlations between these behaviors and their underlying neural substrates (1, 2).

In clinical settings, BN patients describe eating behaviors including anhedonia surrounding mild flavors, pathological consumption of highly palatable foods (e.g., high sugar/fat foods), and diminished feelings of emotional and/or physical satiety (3). Several observational and investigative studies have clarified some skewed preferences in BN-related eating habits. For example, overeating and purging in BN patients are associated with consuming more sucrose and fat, and less protein (4, 5). Also, breads and cereals are the primary carbohydrate sources for healthy eaters, whereas fruits and vegetables are the main carbohydrate sources for BN patients (6). Several studies have indicated that food selection in BN or in postpubertal youth with bulimic symptoms is heavily biased toward increased sugar/fat caloric intake at the expense of protein calories (7, 8). Another study also found high levels of consumption of artificial sweeteners, which provide few calories but excessive sweetness, in BN patients (9). Furthermore, BN patients were unable to enjoy the taste and smell of foods, or to feel distress when they overate (10). These results suggest that BN patients might have some dysfunction in the perception and evaluation of tastes that affects a wide range of foods.

In addition to sweetness, the five basic tastes include saltiness, sourness, bitterness, and umami (11). Umami was discovered in 1908 by isolating monosodium glutamate (MSG) from kelp (12) and is a preferred taste (13). Facial expression studies on neonatal human infant responses to stimulation with different tastes noted that umami-tasting stimuli induced eager sucking and licking movements, as well as triggering facial expressions very similar to those induced by sweet taste (14). MSG, thought to be indicative of the protein content of food (15), is widely used as a flavor-enhancing seasoning that also increases food intake (16). In fact, MSG has several physiological actions. In rats, MSG has obesity-controlling effects (17), and unlike glucose, it does not induce the expression of Fos-related protein in the nucleus accumbens (18). It also induced vagally mediated brain activity when administered intragastrically in rats (19), and promoted gastric motility when added to protein-rich meals in human (20). Behaviorally, the addition of MSG to low-calorie meals enhanced appetite and postprandial satisfaction (21), and when added to essential foods MSG improved balanced food selection in older adults and patients with diabetes (22). Thus, umami is a taste with a certain level of preferentiality, and MSG can potently influence human physiology and nutrition.

Of the many functional magnetic resonance imaging (fMRI) studies of BN patients, those using taste stimuli are considered particularly useful because they directly examine dysfunction in the perception and evaluation of food as a reward (23). To date, fMRI studies of BN patients using taste or flavor stimuli have found abnormal responses to sweet and high-fat cream in specific regions of the brain including the insula, anterior cingulate cortex, orbitofrontal cortex, and ventral striatum (24–28). Although differences across the presented tasks limit comparisons of these studies, Kaye et al. have proposed that patients with EDs have a disorder in the ventral neural pathway formed by the insula, orbitofrontal cortex, amygdala, anterior cingulate cortex, and related regions of the brain, leading to abnormalities in food-related reward evaluation (29). However, those studies were limited by using only highly pleasant tastes, sweetness, or fats.

On another front, it is assumed that the neurophysiology of disgusting stimuli processing may also be important for elucidating the pathogenesis of aberrant eating behaviors in EDs (30, 31). In fact, feelings of disgust influence the modulation of food intake in human (32). Furthermore, BN patients or women with abnormal eating attitudes showed altered disgust sensitivity to food-related stimuli (33–35). More recently, Monteleone et al. investigated the brain response to bitter taste of quinine concomitantly with sweet taste in BN patients (36). Their study focused on the processing of universally aversive taste stimulus, which showed decreased brain response in the amygdala and insula to the bitter taste in BN patients.

Monosodium glutamate has higher hedonic impact than water, but lower impact than sucrose (37). So, it is reasonable to consider that umami does not present high pleasantness on equal terms to sweetness, but it does offer a certain level of pleasantness, contrary to the bitterness that is widely considered as disgusting. In the present study, for the first time, we aimed to investigate the representations of umami in brain of BN patients compared with healthy women (HW). We hypothesized that BN patients show alterations of brain responses to taste based not only on sweetness and fat (obviously pleasant tastes) or bitterness (obviously aversive taste), but also umami (moderately pleasant taste). We thus investigated the brain responses to MSG stimuli in BN patients to determine whether these patients have lost the ability to properly perceive and evaluate umami tastant or its rewarding/satisfying feelings.

To the best of our knowledge, this is the first study to investigate brain activation in response to tasting umami tastant in BN patients. We expected that the brain regions related to food-reward processing, such as the insula, orbitofrontal cortex, anterior cingulate cortex, and ventral striatum (50, 51), would show aberrant activation in BN patients, based on previous studies showing altered activation in these regions in BN with regard to sweet taste (24–27) or high-fat cream (28). Our data showed higher activation in the insula and anterior cingulate cortex, but no difference in the orbitofrontal cortex and ventral striatum, in BN patients compared with HW participants.

Monosodium glutamate tasting-related activations were strongly seen in the right insula in all participants. This is consistent with previous fMRI studies showing umami-related brain activation in healthy individuals in the insula (40, 52, 53). Thus, the MSG tasting task we employed appeared appropriate as a taste-presenting task.

The most important result in our study is that BN patients showed higher activation in the right insula compared with HW participants. The insula consolidates the sensations of taste, temperature, viscosity, and texture (54), and it is interconnected with the orbitofrontal cortex, anterior cingulate cortex, amygdala, and other regions of the brain responsible for evaluating the comfort and emotional relevance of taste stimuli in relation to previous experiences (50, 51). In addition, it has multimodal roles for the integration of affective, cognitive, and physiological information (55). Thus, the insula is a pivotal structure in the corticolimbic circuits involved in appetite regulation, integrating both internal and external information linked to reward processing, whereas the homeostatic aspects of the regulation of food intake are mediated by the hypothalamus (56). The altered activation in the middle insula in BN in the present study might be interpreted as the altered representation of MSG taste in the primary taste cortex of the human brain. It has been regarded that the primary taste cortex in the human brain is composed of the anterior insula and adjoining frontal operculum (48, 49). In fact, these areas were activated in response to each prototypical taste stimulus, sweet, and salty (57), bitter and sour (53), and umami (40). However, recent findings from neuroimaging studies suggest that gustatory information may be represented in the more posterior parts of the human insular region (58–60). In fact, multiple studies demonstrated sufficient activation to taste stimuli in the middle or posterior insula, as well as in the anterior insula, in humans (57, 61–63). Considering these data, the location of the primary taste cortex in the human insular cortex is less precisely defined. Thus, our data might indicate altered representation of MSG taste in the primary taste cortex in BN patients. The primary taste cortex acts as the entry point into the corticolimbic circuits involved in appetite regulation that represent the quality and intensity of the taste of food (64). In our study, the subjective ratings of intensity to the MSG solution were not significantly different between the two groups. Thus, our finding in the insula might be interpreted as the representation of altered perception and evaluation to quality of the MSG solution in BN.

Although the brain activity in the right insula in the HW group increased with decreasing subjective pleasantness for the MSG stimuli, that in the BN group was not affected. Interestingly, several studies reported that brain activation to tastes in the right insula decreased as unpleasantness of taste decreased in healthy subjects (65, 66). Our result is consistent with those findings. On the other hand, BN patients showed deregulated insula activation regardless of the scores of pleasantness of the MSG solution. These results also reinforce our hypothesis of an inability in BN patients to evaluate umami taste properly.

In our results, BN patients showed higher activation in the dorsal anterior cingulate cortex, whereas the orbitofrontal cortex and pregenual anterior cingulate cortex, regions most often reported in co-activation with the orbitofrontal cortex in taste-related studies (67), were not significantly activated in the two groups. The dorsal anterior cingulate cortex is known to reflect the unpleasantness of taste stimuli (68, 69). Higher activation in the dorsal anterior cingulate cortex to MSG stimuli in BN patients compared with HW participants might reflect the perceived relatively low pleasantness or liking of the taste in BN patients.

As previously mentioned, some fMRI studies have investigated the brain responses to taste stimuli in BN with regard to food-reward processing errors in BN. Patients with current BN showed reduced anticipatory responses in the insula to a (sweet) chocolate milkshake (24), and in the insula and striatum to the sweet taste of sucrose (26). Individuals who had recovered from BN showed increased responses in the insula to sucrose (27), and in the striatum to high-fat flavor (whipping cream) (28). These studies showed reduced anticipatory and exaggerated consummatory responses to taste stimuli with hedonic impact, suggesting that this pattern of responding to food-reward might account for deficits in reward evaluating before eating, and excessive reward evaluating in food consumption, which might support the disturbed eating behavior in BN. The enhanced insula activity in the consummatory phase found in our study is consistent with the results of these previous studies, although there are conflicting results among them. Frank et al. reported that participants with recovered BN showed decreased activity in the right anterior cingulate cortex and left cuneus when tasting glucose (25). Bohon and Stice also reported reduced activity in the right insula of patients with current BN related to consummatory, and not just anticipatory, responses to chocolate milkshake (24). To establish a consistent theory on the taste response pattern of BN patients with regard to food-reward processing errors, more data must be accumulated for the future studies.

From a different perspective, it is notable that the insula might work as the key neural structure responsible for disgust processing (70, 71). Vicario et al. suggests that the insula might be the hub structure linking each domain of disgust, not just food-related disgust, in consort with other interconnected brain regions such as the orbitofrontal cortex, striatum, and anterior cingulate cortex (41). Although little is known concerning BN, the involvement of insula dysfunction to altered disgust processing is well-known in AN, and in people with hyperphagia such as obesity (41). Furthermore, as previously mentioned, a recent fMRI study revealed altered insula response to the highly aversive stimulation that is bitterness with quinine in BN patients (36). Since the possibility exists that BN patients might perceive the MSG solution as an aversive stimulus, our results which showed higher activation in the right insula and more negative hedonic scores in response to MSG stimuli might indicate altered processing of disgust not only to highly aversive taste stimuli but also to a wide range of taste stimuli in the current BN patients.

Sweetness, a widely preferred taste, has intense hedonic impact (72), and it may lead to binging or overconsumption (73). On the other hand, MSG also has a greater hedonic impact than water, but a lower impact than sucrose (37). The present study indicates that BN patients might have impaired function in evaluating not only highly pleasant tastes but also tastes with mild pleasantness such as umami, with the possibility that they tend to perceive a wide range of tastes as disgust. Thus, our study suggests the importance of not only controlling excessive palatable foods, but also supporting restoration of appreciation for moderate taste in the therapeutic practice for patients with BN.

Several limitations of the present study should be noted. First, because of a lack of accurate and objective examination methods, we were unable to investigate the actual eating habits and preferences of the participants in this study. Also standardized meals prior to the scanning and assessment of subjective ratings on the distilled water and satiety level during scanning would allow for a more accurate analysis of the brain responses to umami tastant. Second, this fMRI study focused only on umami tastant; sweetness and other taste stimuli were not tested. If sweetness and the other tastes were compared simultaneously in the same participants, the significance and specificity of brain responses to umami tastant would be clearer. Furthermore, we did not measure disgust sensitivity and disgust propensity in our participants. Considering the involvement of their factors in insula activation to aversive stimuli (74), assessment of disgust sensitivity and disgust propensity might help us to attain more reasoned interpretation of our results. In addition, the participants in the study were all Japanese women, who are obviously familiar with Japanese diet and dietary culture, which is deeply involved in applying umami flavoring to various dishes. There are clear inter-individual differences in sensitivity to MSG, and in familiarity to the umami taste, in European cultures (75). Our design limited the generalizability of our conclusions, especially in people of other cultures. Further, we did not exclude participants with depression and those taking medication, although they comprised a small number. Although significant neural response difference in the right insula between the BN group and HW groups survived after control for depression and anxiety (Tables S1 and S2 in Supplementary Material, respectively), our results could have been influenced by these factors. Another limitation relates to our task not excluding the effects of anticipation for taste stimuli on neural responses. Future studies that investigate independently both consummatory and anticipatory phases will help understand neural substrates in patients with BN more clearly.

In conclusion, this is the first study that investigated brain responses to umami tastant in BN patients. The results demonstrated altered perception and hedonic evaluation to umami tastant in BN patients, which might contribute to imbalance in food choices and disordered feelings of satisfaction and pleasure when eating a well-balanced diet.

The Research Ethics Committee of Chiba University Hospital approved all procedures, and written informed consent was obtained from all participants prior to scanning. All experiments were performed in accordance with the Helsinki Declaration.

RS, YH, and MN designed the project and RS, MT, and TT prepared and provided the MSG solution. RS, YH, NN, KM, and YM. performed the experiment and collected the data. YH and RS analyzed the data. RS and YH drafted the manuscript. MN, DM, MI, and ES supervised this study.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.