Dr Asao Ogawa, Psychiatry and Behavioral Science, Osaka University Graduate School of Medicine, D3, 2-2 Yamadaoka, Suita City, Osaka 565-0871, Japan. Email: email@example.com
Abstract Eating disorders are common psychiatric disorders in young women. The aim of the present study was to evaluate the gastric electrical acitivity of patients with eating disorders and its relation to their symptoms. The electrogastrography (EGG) was performed before and after a water load test for outpatients with eating disorders (n = 36; 14 anorexia nervosa, 14 bulimia nervosa, eight eating disorder not otherwise specified) and healthy women (n = 19). A structured interview (Eating Disorder Examination) was used to assess clinical symptoms. The percentage of normal gastric myoelectrical power was significantly smaller in the eating disorder patients (44.5% vs 74.2%; P < 0.05), while the percentage of bradygastric power was significantly greater, both before and after the water load test compared with the control subjects (30.4% vs 10.4%; P < 0.05). In addition, moderate correlation was found between the duration of illness and the percentage of bradygastria (P < 0.05). In conclusion, it is suggested that longstanding abnormal eating in patients with eating disorders may induce disturbances to gastric motor function, resulting in their abnormal, eating-related behavior, and form a symptomatic vicious circle. The EGG may be a promising method for determining the pathophysiology of eating disorders and for developing effective therapeutic approaches.
Eating disorders are characterized by abnormal eating (excessive food restraint, bulimia) and its compensating behavior (i.e. behavior aimed at preventing even the slightest weight gain, such as purging behavior and excessive exercise). According to Diagnostic and Statistal Manual of Mental Disorders (4th edn; DSM-IV) criteria, eating disorders are classified into two major subtypes, anorexia nervosa and bulimia nervosa. Their prevalence rates among young women are quite high: somewhere between 0.1 and 0.7% for anorexia and 1.3–2.3% for bulimia.1,2 Other problems are their chronicity and recurrence. Twenty to 50% of patients with bulimia nervosa continue to show symptoms at 5 year follow up.3
Previous studies have suggested that different subtypes of eating disorders share some biological mechanisms eliciting and maintaining abnormal, eating-related behavior. For instance, decreased cerebrospinal fluid levels of the serotonin metabolite 5-hydroxyindole acetic acid (5-HIAA) and phagostimulant β-endorphin have been reported for both anorexia and bulimia nervosa.4,5 Among many other factors presumed to be involved are vagal nerve hyperactivity and abnormal gastrointestinal function.6–8
Electrogastrography (EGG) is a method for recording gastric electrical activity captured through electrodes attached to the abdominal wall. The EGG non-invasively measures regulating processes of gastrointestinal motor activity and is utilized to assess function of the autonomic nervous system, which is known to be affected in various diseases. Previous studies have found EGG abnormalities in patients with eating disorders.6,8 The subjects in those studies were limited to persons exhibiting specific subtypes of eating disorders (i.e. anorexia nervosa or bulimia nervosa). No previous studies exist that have examined both subtypes simultaneously. In the present case–control study, EGG of both subtypes was studied in relation to other clinical features.
Of 48 consecutive patients (46 women and two men) diagnosed with eating disorders according to DSM-IV criteria at the outpatient clinic of Osaka University Graduate School of Medicine between March and November 2001, we studied 38 female patients between the ages of 16 and 30 years with no prior gastrointestinal symptoms. The group consisted of 15 anorexia nervosa patients (AN; mean ± SD: (22 ± 5 years), 15 bulimia nervosa patients (BN; 23 ± 3 years), and eight patients with eating disorder not otherwise specified (EDNOS; (22 ± 4 years). All AN and BN patients belonged to purging subtypes, with the exception of only two restricting AN patients. Nineteen sex- and age-matched healthy volunteers, with no history of psychiatric illness or gastrointestinal/neurological disturbances, were recruited through acquaintances of the staff of Osaka University Graduate School of Medicine to serve as the control group. Prior to participation in the study, written informed consent was acquired from the patients and the controls
Interview and questionnaires
For the patients and the controls, the following interviews were administered at the time of their first visit to the clinic at Osaka University Graduate School of Medicine. The Eating Disorder Examination (EDE) is a structured interview for diagnosing and assessing eating disorders and their related behavior and attitudes.9 In the present study we used EDE to determine diagnosis and assess clinical symptoms (duration of illness, days and numbers of episodes of objective bulimia and purging behavior during the 3 month period before the examination). In order to assess the presence of gastrointestinal symptoms, we interviewed the subjects following the checklist we developed, which asked about various gastrointestinal symptoms such as nausea, heartburn, upper abdominal pain and dysphasia.
Electrogastrography was examined along with the results of the aforementioned interviews. Subjects were instructed to abstain from food and beverage for 3 h before the examination. Baseline EGG activities were recorded for 20 min, after the subjects had rested lying on their backs for 10 min. Subjects were then given 250 mL of 23°C non-caloric water and instructed to drink it down within a minute. After this water load, EGG activity was recorded for 25 min.
A portable-type EGG recorder (EG; Nipro, Tokyo, Japan) was used for measurement of EGG. The EGG was recorded at a sampling rate of 1 Hz and analyzed by two ad hoc computer programs egs2 (Gram, Saitama, Japan) and DaDisp (DSP Development, Neuton, MA, USA).
After gentle skin abrasion to enhance electrical conditions, five electrodes were fixed as follows: reference electrode (Ch5) at the midpoint between sternum and navel on the midline of the body; electrodes 1 (Ch1) and 2 (Ch2) at the midpoints between sternum and reference electrode on right and left midclavicular line; and electrodes 3 (Ch3) and 4 (Ch4) at the midpoints between the reference electrode and navel on the right and the left midclavicular lines. The EGG was recorded by unipolar lead with electode 5 as reference. In the present study we analyzed EGG derived from electrode 2, which is considered to capture gastric action potential most accurately.
There are several electrically active organs close to the stomach, including heart, diaphragm, abdominal wall muscles, duodenum, and small and large intestines. We took the following measures to prevent electrical activities of these organs from affecting EGG recording. First, the influence of the abdominal wall muscles was minimized by having the subject lie quietly. Second, the influence of the electrical activity of cardiac, respiratory, and intestinal muscles was removed by setting a filter, based on the differences of electrical activity cycle between these and gastric muscles. Low- and high-cut filters were set at 0.03 and 0.09 Hz, respectively. Before analyzing the data we checked the raw data visually and excluded the periods of time contaminated by artifacts. The data thus obtained were analyzed, using fast-Fourier transformation, running power spectral analysis with 256 s ‘windows’, and the results were examined in terms of wave forms and changes along the time axis.
The EGG data were divided into three periods: 1–10 min before water load; 1–10 min after water load; and 10–20 min after water load. Fast-Fourier transformed spectral analysis was performed for the signal data of each period. The baseline EGG represents electrical control activity (ECA), which regulates constricting movement of gastric smooth muscles, while the EGG after water load represents the electrical response activity (ERA), which arises as a response to water load and causes actual constriction of the gastric smooth muscles. The EGG signals were divided into three major frequency bands: 1–2.4 cycles per min (c.p.m.; bradygastria); 2.4–3.6 c.p.m. (normal range; normogastria); and 3.6–5.4 c.p.m. (tachygastria). Their ratio to total power present was calculated and analyzed. Normal gastro-electrical signals (normogastria) were distributed around 3 c.p.m. (2.4–3.6 c.p.m.), which represents the ECA and potential frequency of pacesetter cells.10,11 Bradygastria and tachygastria represent abnormal gastro-electrical activities.11–13
Based on power spectrum analysis of EGG data, we calculated percentages of bradygastria, normogastria, and tachygastria; anova was used for the between-group comparison. Pearson correlation was used for calculating correlation between duration of illness, frequency of bingeing/purging and the frequency of EGG signals. Statistical software spss version 10.0 (SPSS, Chicago, IL, USA) was used for data analysis.
Figures 1 and 2 show the sample raw EGG and the results of spectral analysis for controls and patients before the water load. For the controls, normogastria was predominant. Normal EGG pattern was confirmed for 18 out of 19 controls. In contrast, no 3 c.p.m. peak was observed for the eating disorder patients. Spectral analysis of basic electrical activity revealed one to two peaks around the 1–2 c.p.m. wave ranges, with the peak of normal range waves (3 c.p.m.) being diminished or absent.
Figure 3 shows the mean percentage distribution of bradygastria, normogastria, and tachygastria to total EGG power at baseline, for both the control group and the eating disorder patients. For healthy controls, regular normogastria was dominant, and spectral analysis produced a peak at 3 c.p.m. In contrast, bradygastria was predominant for the eating disorder patients, while normogastria was decreased or absent. Normogastria was significantly lower for the eating disorder patients (44.5 ± 17.7%), while bradygastria was siginificantly higher (30.4 ± 14.2%) compared with that of the controls (normogastria 74.2 ± 10.6%, F3,54 = 13.9, P < 0.001; bradygastria 10.4 ± 5.8%, F3,54 = 14.5, P < 0.001; tachygastria 9.1 ± 6.5%, F3,54 = 2.42, P = 0.08, NS). Among the subtypes of patients (AN, BN, EDNOS), no significant differences were found in terms of spectral distribution of baseline EGG.
Water load test
For the controls, normogastria was predominant both before and after water load. The 3 c.p.m. peak increased 3–5 min after water load. Normal EGG pattern was confirmed for 18 out of 19 controls. In contrast, no. 3 c.p.m. peak was observed for the patients, either before or after water load. Spectral analysis of basic electrical activity revealed one to two peaks around the 1–2 c.p.m. wave ranges, with the peak of normal range waves (3 c.p.m.) waves being diminished or absent. For all the subtypes (AN, BN, EDNOS), a 1–2 c.p.m. peak was observed in the first 10 min after water load, both for the raw data and spectral analysis. Alhough 3 c.p.m. waves eventually appeared, their timing was delayed to 15 min after water load for AN and EDNOS, and to 20 min for BN.
Figure 4 shows the mean percentage distributions of bradygastria, normogastria, and tachygastria to total EGG power during the 10 min period after water load for both the controls and the eating disorder patients. As with baseline electrical activity, normogastria was significantly decreased (41.6 ± 19.9%), while bradygastria was significantly increased (29.2 ± 13.2%) for the patients, compared with the controls (normogastria 65.6 ± 13.3%, F3,54 = 7.27, P < 0.001; bradygastria 17.4 ± 7.3%, F3,54 = 4.30, P < 0.01). These differences held even 10–20 min after water load (normogastria 61.2 ± 15.1%, F3,54 = 4.92, P < 0.01, bradygastria 15.2 ± 6.5%, F3,54 = 2.90, P < 0.05; Fig. 5). For tachygastria, no significant differences were found between the controls and the eating disorder patients. Also, no significant differences in EGG spectrum distribution were found among the subtypes of patients (AN, BN, EDNOS).
Correlation between the duration of illness and EGG
Figure 6 shows the correlation between the duration of illness and percentage of bradygastria to total EGG power for the patients. Moderate correlation was found between the duration of illness and percentage of bradygastria (r = 0.446, P = 0.005). No correlation was found between baseline EGG activity and frequency of bulimia and vomiting for the 3 month period before the examination.
The present study has demonstrated (i) that the patients with eating disorders had a different pattern of gastric myoelectrical activity from the healthy controls, both during rest and after water load; (ii) that no significant differences in pattern of gastric myoelectrical activity were found among the subtypes of the patients; and (iii) that the degree of EGG abnormality during the rest period positively and significantly correlated with the duration of illness.
Gastric peristalsis is regulated by an intrinsic, electrical activity of gastric pacemaker cells, the interstitial cells of Cajal.14,15 The gastric pacemaker area is located at the junction of the fundus and body on the greater curvature16,17 and the pacemaker cells depolarize and repolarize at a frequency of approximately 3 c.p.m. Electrical activity is transmitted annularly toward the pylorus and can be recorded trans-serosally or transcutaneously.16,18
For actual gastric constriction, this baseline electrical activity needs to be superposed with the electrical activity of a spike potential at the time of depolarization. These two types of electrical activity are, respectively, the ECA and ERA.16
The water load test is a highly reproducible method for assessing gastric capacity, adaptability, and neuromuscular rhythmicity.17 The rhythmic pattern of EGG evocated by the physical volume of the water reflects pure, gastric neuromuscular response, uncontaminated by the effects of carbohydrates, protein, fat, and hormones activated by these nutrients.
In gastric dysrhythmia, normal 3 c.p.m. gastric peristalsis is disturbed, leading to ineffective gastric constriction because of the inefficiency of the multifocal electromotive forces it generates, with resultant deterioration of its excretory capacity.17 Bradygastria and tachygastria are unable to cause normal constriction at the antrum.19 Clinically, changes in EGG rhythm from normogastria to bradygastria and/or tachygastria often cause epigastric distress, especially nausea.10,17,20 Studies examining the capacity of gastric emptying by assessing the movement of gastric contents found that 3 c.p.m. waves were predominant among subjects with normal gastric emptying, whereas 3 c.p.m. waves were diminished or absent, and various types of abnormal waves emerged, among patients with delayed gastric emptying.
Along with the development of EGG studies, the distinction between gastrointestinal motility disorders (GIMD) and gastrointestinal functional disorders (GIFD) has been established. The latter is considered to be strongly related to indefinite complaints of gastrointestinal tract. This is supported by the fact that promotile agent such as neostigmin and prostaglandin F2α facilitate gastrointestinal motility without improving gastrointestinal symptoms, while prokinetic agents such as cisapride, itopride, mosapride, and domperidone improve gastrointestinal symptoms without necessarily facilitating gastrointestinal motility.
In the present study bradygastria was predominant for both AN and BN patients. This finding may indicate that AN and BN share gastric rhythm abnormality. Also, we found that the EGG abnormality existed for AN and BN patients not only after water load but also during the resting period. The instability of gastric motor function suggested by EGG abnormality is known to disturb gastric compliant relaxation (i.e. relaxation responses of the fundus occurring immediately after food intake). Disturbances of this compliant relaxation (e.g. excessive relaxation in reaction to food intake) may elicit symptoms of premature feelings of fullness and distention, as well as nausea.21 Patients with eating disorders often complain of indefinite symptoms of upper gastrointestinal tract, such as nausea, pain, epigastric fullness. Moreover, they have high rates of comorbid functional abnormality of the upper gastrointestinal tract.22–24
We found that the degree of EGG abnormality increased with the duration of the illness. This finding suggests that longstanding abnormal eating in patients with eating disorders may induce disturbances to gastric motor function, leading to digestive complaints such as feelings of fullness, and resulting in their characteristic eating-related behavior such as self-induced vomiting, thus forming a symptomatic vicious circle. The longer the duration of illness, the more severe the gastric functional disturbances, and the more refractory the illness may become.
Previous studies have indicated that autonomic nervous system and hormones contributed to EGG change.25,26 Normal, post-prandial increase of EGG amplitude was suppressed in patients who had undergone surgical vagotomy.27 Post-prandial EGG changes correlated with high-amplitude components of ECG change, which is an indicator of parasympathetic nervous activity of the heart.28 The EGG changes were demonstrated in patients with diabetes, one of the representative diseases causing disturbances to the autonomic nervous system.29,30
Functional abnormality of autonomic nervous system was reported to exist in patients with eating disorders. Garner noted that autonomic nervous system is a factor contributing to cardiovascular system-related death/sudden death in patients with anorexia nervosa.31 Other researchers stress the importance of vagal abnormalities in AN.32–34
With respect to the neuroendocrine system, leptin participation is suggested to be responsible for the pathophysiology of eating disorders. Decrease in leptin blood concentration was reported for both AN and BN.35 Leptin was discovered in the gastric epithelium of rats as well as in their adipocytes.36 Leptin-sensitive, afferent vagal endings were identified in the stomach of a rat.37 Satiety effect of cholecystokinin (CCK) was enhanced by leptin.38,39 Vagal stimulation induced leptin secretion.40 Together these findings suggest that stomach-residing leptin may contribute to the control of food intake through CCK, which is released into the blood from the gastrointestinal tract by the stimulus of food intake, and that insufficient secretion of stomach-residing leptin is responsible for the pathophysiology of eating disorders both in its elicitation and its maintenance.
Although both ‘involuntary (driven)’ and ‘voluntary’ factors have been considered to be implicated in abnormal eating behavior in eating disorders, previous studies have tended to focus only on ‘voluntary’ factors, with relative neglect of the pathophysiology of ‘involuntary’ aspects of abnormal eating.
Recently, ondansetron, a peripheral serotonin-selective antagonist, which inhibits activities of the vagal nerves distributed along the gastrointestinal tract, was reported to improve severe bulimic symptoms.41 In contrast, antidepressants, including recently spotlighted selective serotonin re-uptake inhibitors (SSRI), are not necessarily effective for chronic and/or severe eating disorders.42,43 Although further study is required, these findings indicate that disturbances of the peripheral gastrointestinal tract may be responsible for the elicitation and maintenance of abonormal eating in eating disorders.
The present study had certain limitations. First, the number of subjects was limited. In order for these findings to be generalized, more cases need to be examined from various perspectives, including detailed comparison among subtypes. For example, we were unable to statistically compare the binge-eating/purging and restricting subtypes because only two of the AN and BN patients belonged to the latter. This comparison would be necessary to clarify specific effects of binge-eating/purging on EGG. Second, assessment of primary symptoms such as duration of illness, bulimia, and self-induced vomiting was solely based on subjective self-report at the time of the examination. Longitudinal observation using more objective assessment methods are needed in order to ascertain our findings. Third, there are technical issues in terms of EGG measurement. Although we analyzed EGG data derived from electrode 2, which is generally considered to capture gastric action potential most accurately, it is not entirely clear if this selection was a truly adequate one. Also, different results might have been obtained had more time been given to recording, considering that as many as three distinct phases were observed during the resting period. Nonetheless, we believe that the finding of our study, that disturbances in gastrointestinal functioning increase as a function of the duration of illness, lends strong support to the possibility that ‘involuntary’ factors (i.e. disturbances in gastrointestinal function) contribute to the pathophysiology of eating disorders, in addition to ‘voluntary’ factors. Electrophysiological study of gastrointestinal dysfunction using EGG may be a promising route to elucidating the pathophysiology of eating disorders, and to developing effective therapeutic approaches.