Symptoms of gastroesophageal reflux disease (GERD) including

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2007;5:439 444 The Effects of Dietary Fat and Calorie Density on Esophageal Acid Exposure and Reflux Symptoms MARK FOX,*, CAROLE BARR, SUZANNE NOLAN, MIRANDA LOMER, ANGELA ANGGIANSAH, and TERRY WONG* *Department of Gastroenterology, Esophageal Laboratory, St Thomas Hospital, London, United Kingdom; and Nutrition and Dietetics, King s College London, London, United Kingdom Background & Aims: The effects of diet on gastroesophageal reflux disease are not well understood. This study assessed the effects of dietary fat and calorie density on esophageal acid exposure and reflux symptoms in patients with reflux symptoms. Methods: Patients referred for the investigation of reflux symptoms were recruited (most with nonerosive disease). A catheter-free system provided esophageal ph monitoring over 4 days in 4 dietary conditions. A high-fat (50%) vs low-fat (25%) diet (calorie-controlled), and a high-calorie (1000 kcal) vs low-calorie (500 kcal) diet (fat-controlled) were provided in randomized order, and meal volume was controlled. The effects of meal consistency also were studied. Results: Complete data were available for 15 patients (6 men, 9 women; age, 48 y; range, 26 70 y; body mass index, 26 kg/m 2 ; body mass index range, 21 35 kg/m 2 ). Demographic variables and meal sequence had no effect on reflux parameters. Dietary composition had effects on esophageal acid exposure (F statistic [analysis of variance] 7.4, P <.005) and symptoms (Friedman test 24.2, P <.001). No effect of meal consistency was present. Esophageal acid exposure was greater during the high-calorie than the low-calorie diet (mean, 8.6% 2.0% vs 5.2% 1.4% time ph < 4/24 h; P <.01). No difference was observed between the high-fat and low-fat diets (mean, 8.6% 2.0% vs 8.2% 1.6% time ph < 4/24 h; P NS). In contrast, the frequency of reflux symptoms was not affected by calorie density (median, 6; range, 2 12 vs median, 8; range, 2 13; P NS) but was increased by the high-fat compared with the low-fat diet (median, 11; range, 5 18 vs median, 6; range, 2 12; P <.05). Conclusions: Calorie density determines the severity of esophageal acid exposure in gastroesophageal reflux disease after a meal; however, the percentage fat content of the diet has important effects on the frequency of reflux symptoms. Symptoms of gastroesophageal reflux disease (GERD) including heartburn, acid regurgitation, and noncardiac chest pain are common complaints in the general population with important effects on patient perception of health and quality of life. 1 The prevalence of GERD has increased in recent years, with a major impact on direct and indirect health costs. Moreover, there is a strong and probably causal relation between acid reflux and the development of esophageal adenocarcinoma. 2 These concerns highlight the importance of identifying potentially modifiable risk factors for GERD. One such lifestyle factor may be diet. Patients with heartburn and acid regurgitation are often told to avoid fatty foods. This advice is based on patient complaints of increased symptoms after eating high-fat foods, 3 and epidemiologic surveys that report a link between high fat intake and the presence of GERD. 4 Classic studies have shown that fat decreases lower esophageal sphincter pressure, increases the rate of transient lower esophageal sphincter relaxation, and delays gastric emptying, all factors that would be expected to increase esophageal acid exposure in GERD. 5 In clinical studies, esophageal ph measurement provides direct evidence of an association between dietary fat and acid reflux; however, results have been inconsistent. 6 10 Moreover, interpretation of these studies is difficult because many were small, of short duration, performed in healthy volunteers rather than patients with GERD, and failed to provide appropriate controls for meal volume and calorie density. In addition to the nutritional composition of the diet, the physical characteristics of the meal may also affect reflux. Large meals cause more reflux than small meals, 11 13 however, the physical characteristics of food may be relevant because solid food is stored in the upper stomach (fundus) whereas liquid food tends to settle in the lower stomach (antrum). These factors may alter the distension of the fundus or the rate of gastric emptying and thus the severity of GERD. Previous research has been limited by the difficulty of recruiting subjects for prolonged ph studies because of uncomfortable, socially intrusive, catheter-based ph monitoring. Recently a wireless ph monitoring system was introduced (Bravo; Medtronic Inc., Shoreview, MN). 14,15 A small ph monitor (capsule) is attached to the esophageal mucosa by endoscopy and transmits ph data to a receiver on the patient s belt. In general, the patient experiences no discomfort except for a mild foreign body sensation on swallowing. 14,15 In clinical practice, 48-hour Bravo ph measurements are standard; however, the capsule remains in place for a median of 5 6 days, 14 and longer recordings can be acquired with no additional discomfort to the patient. This study investigated the effects of dietary fat and calorie density on gastroesophageal reflux in patients referred for 24-hour ph studies. In addition, the effects of meal consistency were assessed. The improved tolerability and capability for prolonged (4-day) ph recording of the Bravo system provided an ideal method with which to study these issues. Methods Subjects Patients with typical reflux symptoms referred for Bravo ph monitoring were invited to participate in the study. Exclusion criteria included previous upper-gastrointestinal surgery Abbreviations used in this paper: ANOVA, analysis of variance; F, F statistic (analysis of variance); Fr, Friedman test; GERD, gastroesophageal reflux disease; SAP, symptom-associated probability. 2007 by the AGA Institute 1542-3565/07/$32.00 doi:10.1016/j.cgh.2006.12.013

440 FOX ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 4 (or interventions), gastrointestinal symptoms other than reflux, ingestion of medication that affects either gastrointestinal function (acid suppression was stopped 4 days before the study), special dietary requirements (eg, diabetes) or behavior (eg, vegetarian), and pregnancy. Subjects provided written informed consent after explanation of the study, which was in accordance with the Declaration of Helsinki and was approved by the local ethics committee. Prospective power calculations determined that a sample size of 17 patients was necessary to provide an 80% chance of detecting a 20% difference in esophageal acid exposure in response to dietary intervention for an value of.05 (or 12 patients to detect a 30% difference). These calculations were based on published evidence of lifestyle management in GERD 5,16 and the day-to-day variability of ph measurements of approximately 20% reported for the Bravo system 14 (reproduced in-house 17 ). Extra patients were recruited in case of technical failures and early withdrawals. Bravo Capsule Placement and ph Recording Upper endoscopy under moderate sedation with 2 8 mg of midazolam and 0 50 mcg of fentanyl was performed with a standard gastroscope in the left lateral decubitus position to localize the Z-line. After calibration the Bravo delivery system was passed orally and the capsule was deployed 6 cm proximal to the Z-line. 14 The subject consumed 100 ml of orange juice (ph 4.3; Calipso, Wrexham, UK) as an in vivo calibration and to confirm capsule placement. The subject was provided with test meals (see later) and instructed to continue normal activities. Participants returned after 48 hours to upload ph data and replace receiver batteries. A second glass of orange juice was consumed to check that the system was still functioning. The subject then was provided with further test meals and underwent an additional 48 hours of Bravo ph monitoring. Subjects were instructed to document meal times, position change, and symptoms (heartburn, chest pain, and regurgitation) on the receiver. Follow-up evaluation was arranged with an unblinded physician to review the patient and the test results. Test Meals Patients were provided with test meals prepared in-house by nutrition research scientists (C.B., S.N.) under the guidance of a senior dietician (M.L.) as follows: high-fat (58% energy), highcalorie (1000 kcal) mixed solid/liquid; low-fat (25% energy), highcalorie (1000 kcal) mixed solid/liquid; low-fat (25% energy), lowcalorie (500 kcal) mixed solid/liquid; and high-fat (58% energy), high-calorie (1000 kcal) liquid. Meals were controlled for volume ( 800 ml). For the mixed meals, the volume, calorie content, and macronutrient composition was divided equally between the solid and liquid parts (Appendix). Test meals were provided in a randomized order as directed by prospective balanced randomization. Patients were not told the dietary composition of the meal. During each 24-hour period the same meal was provided for lunch and dinner. A standard breakfast was recommended. Subjects consumed each meal within 30 minutes and left at least 3 hours after the evening meal before bed. Acid-suppressant medications were not allowed. The use of antacid rescue medication was allowed but discouraged. Data Analysis Detection of acid reflux was performed as for a clinical study. The manufacturer s software for the Bravo system (PolygramNet; Medtronic Inc.) registered the number of acid reflux events and other reflux parameters during each 24-hour period. Reflux event detection was set at a sampling frequency of every 6 seconds and defined as 2 consecutive ph measurements of less than 4 (ie, 12 seconds of acid exposure). Data were collected by investigators blinded to the test conditions. The primary outcome measure was the esophageal acid exposure (percentage of time ph 4/24 h). Secondary outcome measures included the number of acid reflux events, number of long reflux episodes ( 5 min), longest reflux episode, duration of acid reflux exposure, and the composite DeMeester score. 18 The frequency of reflux-related symptoms during each 24-hour period were recorded (heartburn, acid regurgitation, and chest pain), together with an aggregate symptom score to compensate for the fact that patients report different symptoms in response to reflux events. The aggregate score did not count symptoms more than once when multiple symptoms were reported within a short time (ie, triggered by a single reflux event). The symptom-associated probability (SAP) was calculated to quantify the association between symptoms and reflux episodes. 19 Statistical Analysis The experimental design was partially balanced (not fully balanced because it is not possible to produce a high-fat, low-calorie meal that is isovolumic with the other test meals), using the 2 factors of meal percentage fat content (high fat vs low fat) and calorie density (high calorie vs low calorie), with a further factor of meal consistency (mixed solid/liquid vs liquid). The interpretation of the findings takes into account that for the different meals and food preparations, 6 possible pair-wise comparisons were possible. To limit an accumulation of errors as a result of multiple comparisons, analysis of variance (ANOVA) tested for group effects (ie, dietary composition and meal consistency). Pair-wise comparisons of individual meals were not performed if a group effect was not present. In addition, a Bonferroni correction was applied. To avoid an accumulation of errors, a subanalysis of the data was not performed (eg, effects of diet on nocturnal/diurnal reflux, duration of effects on acid exposure, and symptoms). Demographic data are presented as medians and ranges. The data were analyzed using a standard statistical package (Prism; GraphPad, San Diego, CA). Before analysis, a formal Kolmogorov Smirnov test was used to determine whether that data conformed to a normal parametric distribution. After this assessment, parametric data such as esophageal acid exposure (% time ph 4/24 h), were summarized as mean SD or standard error of the mean as appropriate. Nonparametric data, such as symptom events, were summarized by median and interquartile range. Repeated-measurement ANOVA was used for paired parametric data with subject as the random variable and meal composition and meal consistency as fixed group variables for repeated between- and within-subject comparisons vs reflux parameters. The Friedman test for repeated comparisons of paired nonparametric data was used when appropriate. The strength of association was expressed using the F statistic (ANOVA) or Fr statistic (Friedman test) for parametric and nonparametric data, respectively. An value of less than.05 was considered significant.

April 2007 DIET AND REFLUX DISEASE 441 Results Study Progress Forty-six patients with reflux symptoms waiting for Bravo ph monitoring were contacted between May and September 2005, 22 patients accepted an invitation for screening. One patient was excluded because of dietary restrictions and 2 patients withdrew consent before the study began. Endoscopy was performed in 19 patients, 4 had mild erosive esophagitis and 2 had a small hiatus hernia (acid suppression was stopped only 4 days before the study began). The Bravo capsule failed to deploy in 1 patient (despite 2 attempts). One patient refused to continue the diet after 2 days. Two capsules detached before study completion after 3 and 48 hours (as evidenced by the ph recording). Thus, complete data from 15 subjects were available for analysis (9 women, 6 men; median age, 48 y; range, 21 70 y; body mass index, 26 kg/m 2 ; body mass index range, 21 35 kg/m 2 ). None were active smokers, drank more than 3 cups of coffee per day, or 14 units of alcohol per week. The median use of antacids during the study was less than 1 dose/day/patient. Patients were instructed to continue normal activities of daily living; all reported a sedentary lifestyle and none engaged in sports (or strenuous activity) during the study. There were no differences in baseline characteristics between the subjects who did and did not complete the study. Ambulatory ph Measurements In vivo calibration confirmed that all Bravo capsules functioned normally throughout the study with no evidence of ph drift. The majority of patients had mild-to-moderate reflux disease on ambulatory ph measurements, with most reflux events occurring in the upright position and, in some cases, shortly after assuming a supine position. ANOVA showed a group effect of dietary composition on esophageal acid exposure (F, 7.4; P.005). There was no interaction with demographic variables and no sequence effect was present (ie, no effect of meal order). No interaction with meal consistency was found. Comparisons of esophageal acid exposure for different dietary compositions by ANOVA revealed that the percentage of time that ph was less than 4/24 h was similar after intake of high-fat and low-fat meals (calorie content and volume controlled); 8.5% 1.9% vs 7.1% 1.3% time that ph was less than 4/24 h (P NS). Esophageal acid exposure was increased significantly in the high-calorie compared with the low-calorie meal (fat content and volume controlled); 8.9% 1.6% vs 5.5% 1.5% (P.001). Pair-wise comparisons of the individual meals (Figure 1) confirmed that esophageal acid exposure after the high-fat, high-calorie meal was very similar to the low-fat, high-calorie meal (8.2% 1.6% vs 8.6% 2.0%; P NS), whereas acid exposure after the low-fat, high-calorie meal was about 40% higher compared with the low-fat, low-calorie meal (8.6% 2.0% vs 5.2% 1.4%; P.02). Pair-wise comparisons revealed similar effects on individual reflux parameters during different dietary conditions (Table 1). With respect to diagnostic yield for GERD, 11 of 15 patients had pathologic reflux on both high-calorie diets ( 4.2% time ph 4/24 h and/or DeMeester score 14.7), whereas 6 of 15 patients had pathologic reflux on the low-calorie diet. Reflux Symptoms The Friedman test (nonparametric repeated-measures ANOVA) showed group effects of dietary composition on individual reflux symptoms of heartburn (Fr, 9.3; P.03), regurgitation Figure 1. Effect of different types of meals on esophageal acid exposure. Percentage time of esophageal acid exposure was similar on the high-fat (HF), high-calorie (HC) and the low-fat (LF), high-calorie (HC) diets, but there was an approximately 40% decrease in esophageal acid exposure on the low-fat, low-calorie (LC) diet. No effect of meal consistency was present. (Fr, 18.4; P.01), chest pain (Fr, 17.2; P.01), and the aggregate symptom score (Fr, 24.2; P.001). The frequency of reflux symptoms (aggregate score) was greater during the high-calorie than the low-calorie diet (24 [range, 9 39] vs 7.5 [range, 1.5 12]); P.01); however, this effect was dominated by an important interaction with the percentage fat content of the meal (P.01). Thus, patients reported more symptoms during the high-fat diet than the low-fat diet independent of the calorie content (18 [range, 4.5 25] vs 7.5 [range, 2 16.5]; P.01). The effect of percentage fat content on reflux symptoms was confirmed by pair-wise comparisons for the individual meals (Figure 2). Reflux symptoms (aggregate score) were more frequent during the high-fat, high-calorie meal than the low-fat, high-calorie meal (11 [range, 5 18] vs 6 [range, 2 12]; P.05), but there was no difference in the frequency of reflux symptoms between the low-fat, high-calorie meal and the low-fat, low-calorie meal (6 [range, 2 12] vs 8 [range, 2 13], P NS). Pair-wise comparisons revealed similar effects on the frequency of individual symptoms reported during different dietary conditions (Table 2). The strength of the association (as assessed by SAP) between symptoms and reflux events was higher for the high-fat, high-calorie meal than for the low-fat, high-calorie meal or the low-fat, low-calorie meal (median SAP, 100 [range, 66 100] vs 94 [range, 62 99] vs 83 [range, 68 100]). Thus, 11 of 15 patients had a positive SAP ( 95%) on a high-fat, high-calorie diet, whereas only 7 of 15 and 5 of 15 patients had a positive SAP on the low-fat, high-calorie and low-fat, low-calorie diets, respectively. Gastroesophageal Reflux and Meal Consistency ANOVA and pair-wise comparisons did not reveal significant effects of meal consistency on ph measurements (P.17) or reflux symptoms (P.35); there was no interaction with calorie density or fat content. The high-fat, high-calorie meal produced similar acid reflux in the liquid and mixed solid/liquid consistency (8.6% 2.0% vs 8.2% 1.6% time ph 4/24 h; P NS). Discussion Prolonged ph recordings by wireless ambulatory ph measurements provided novel information about the effects of diet in

442 FOX ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 4 Table 1. The Effects of Meal Composition on Gastro-Oesophageal Reflux Disease HF/HC (mixed) LF/HC (mixed) LF/LC (mixed) ANOVA (group effect) No. reflux events 50.0 8.4 55.3 9.6 45.1 9.6 P.33 No. long reflux ( 5 min) 4.6 1.2 4.6 1.1 3.2 1.1 P.04 Longest reflux (min) 29.6 7.6 30.3 9.3 8.6 1.7*** P.0001 Time ph 4 (min) 114.5 22.3 120.3 22.0 72.6 19.4** P.004 Fraction time ph 4 (%) 8.2 1.6 8.6 1.6 5.2 1.3*** P.0005 DeMeester score 29.8 5.4 30.9 6.3 17.1 4.4*** P.004 NOTE: Results are presented as mean and standard error of the mean derived from ANOVA. ** P.01 and *** P.001 for pair-wise comparison of LF/LC vs other meals. patients with GERD, both in terms of objective esophageal acid exposure and subjective reflux symptoms. This study also provided information about the effects of food consistency on gastroesophageal reflux. The protocol allowed the effects of calorie density, percentage fat content, and food consistency on GERD to be assessed independently. Other factors that may have effects on gastroesophageal reflux such as meal volume were controlled as far as possible. The order in which meals were provided was randomized and the sequence had no measurable effect on results. The exposure of the esophagus to acid reflux (percentage of time esophageal ph 4/24 h) was related to the calorie density of the meal. In contrast, it was not affected by the percentage fat content of the meal independent of the calorie density. The effects were statistically significant and clinically relevant with approximately 40% less acid reflux measured during the low-fat, lowcalorie diet than under any of the high-calorie conditions (Figure 1), a similar effect was observed on the DeMeester score. 18 This finding is consistent with a recent study by Colombo et al 10 that monitored the effects of calorie density and fat content on postprandial gastroesophageal reflux (6 hours) in 13 healthy volunteers. The percentage of time at ph less than 4 after the high-fat, high-calorie meal was almost double that after the low-fat, lowcalorie meal (3.0% vs 1.6%; P.05), but was similar to the Figure 2. Effect of different types of meals on reflux symptoms. Patients reported reflux symptoms 40% more frequently while on the high-fat (HF) diets than the low-fat (LF) diets, independent of calorie content. The frequency of reflux symptoms was similar during the lowfat, high-calorie (HC) and low-fat, low-calorie (LC) diets. No effect of meal consistency was present. isocaloric low-fat, high-calorie meal (3.0% vs 2.5%; P NS). In contrast, Pehl et al 9,20 did not show effects of calorie density (n 12) or fat content (n 12) on gastroesophageal reflux in healthy volunteers; however, the latter investigations were of short duration (3 hours), and the small number of recorded events impacted on the power to produce significant results. Becker et al 6 used ambulatory 24-hour ph measurement to assess gastroesophageal reflux in healthy subjects (n 10) and patients with GERD (n 10) after isocaloric low- and high-fat meals. In this study, high-fat diet increased acid reflux in healthy subjects (only in the upright position), but not in patients with GERD. 6 The effect of calorie density on gastroesophageal reflux was not studied, and thus these results are difficult to interpret. In contrast, the current study provides clear evidence that esophageal acid exposure in patients with GERD is determined by the calorie density of the meal independent of its percentage fat content. To study the physiologic basis of this effect would have required invasive monitoring; however, calorie density had effects on both the number and duration of reflux events (Table 1), parameters that provide an indirect assessment of GERD pathophysiology. Any factor that increases gastric distension is likely to increase the number of transient lower esophageal sphincter relaxations and reflux events. 11 Any factor that impairs esophageal clearance will increase the duration of acid exposure. Finally, any factor that delays gastric emptying prolongs the period of time during which reflux can occur. In this study the number of reflux events was higher during the high-calorie than the low-calorie diets; however, this did not reach statistical significance (the number of reflux events is highly variable and more subjects would have been required to show an effect). The increase in overall reflux duration was significant. As in previous studies, prolonged reflux episodes were most common soon after the patients went to bed, when esophageal clearance is less efficient. 21 This suggests that the delay in gastric emptying after high-calorie meals is an important mechanism by which diet affects GERD, a hypothesis that is supported by classic studies from Hunt et al 22,23 that showed that the rate of gastric emptying is associated with the calorie density of the meal and not macronutrient composition. As would be predicted from the ph data, reflux symptoms were reported less frequently in the low-calorie than the high-calorie diets. However, there was an important interaction with the fat content of the meal such that patients reported more symptoms during the high-fat diet than the low-fat diet independent of calorie content. Pair-wise comparisons of the individual meals confirmed that the effect of percentage fat content on reflux symptoms was more pronounced than that of calorie density (Figure 2). Moreover, the probability with which reflux events were

April 2007 DIET AND REFLUX DISEASE 443 Table 2. The Effects of Meal Composition and Consistency on Reflux Symptoms HF/HC LF/HC LF/LC Friedman (group Median IQR Median IQR Median IQR effect) Heartburn 3 (0 10) 2 (0 8) 1 (0 6) P.03 Chest pain 2* (0 4) 1 (0 2) 0 (0 4) P.01 Regurgitation 1* (0 11) 1 (0 7) 0 (0 9) P.01 Aggregate score 11** (5 18) 6 (2 12) 8 (2 13) P.001 NOTE: The median number of reflux symptoms appears disproportionately higher in the aggregate score compared to the individual symptoms because of the highly non-parametric nature of the data. In addition certain patients always recorded symptoms as heartburn rather than chest pain or vice versa. Thus for any given symptom a number of patients recorded zero episodes producing a negative skew (low median). This behaviour does not affect the aggregate score. Results are presented as median inter-quartile range. * P.05 and ** P.01 for pair-wise comparison of HF/HC vs other meals. associated with reflux symptoms (SAP) was increased in the highfat compared with the low-fat diet. Thus, dietary fat appeared to increase visceral sensitivity to reflux events; an effect above and beyond that expected from any increase in esophageal acid exposure. This finding is consistent with a study by Meyer et al 24 in which controlled acidification of the esophagus was applied during duodenal infusion of lipid in patients with GERD (n 11). In this experiment duodenal fat shortened the latency to onset of typical reflux symptoms and increased the intensity and severity of heartburn (these results were not reproduced when lipid was delivered into the stomach or esophagus during esophageal acidification). 24,25 In a review by Fried and Feinle, 26 the role of smallbowel fat receptors in modulating visceral sensitivity was described. A series of elegant, carefully controlled studies has established lipids as a major trigger of dyspeptic symptoms such as nausea, bloating, and pain. 27 29 The limitation of this experimental work is that unphysiologic, highly invasive techniques were used. In contrast, spontaneous patient reports of reflux symptoms during ambulatory ph monitoring are likely to provide reliable and clinically relevant measurements. The current study shows that dietary fat has effects on visceral sensitivity to acid reflux in the clinical setting. Moreover, it explains why patients with GERD complain of increased symptoms after fatty food even though ph studies do not consistently detect an increase in esophageal acid exposure. No effect of meal consistency (mixed solid/liquid vs liquid) on esophageal acid exposure or reflux symptoms was found. The large volume of refluxogenic nutrients may have obscured an effect of meal consistency; however, if such an effect is present it is likely to be small. The findings also may have been different if the nutritional composition of the solid and liquid components had not been the same (as is more typical in real life). Recent studies using magnetic resonance imaging have shown that altering the intragastric distribution of fat (and thus the passage of nutrients into the small bowel) by position change or meal order has important effects on gastric emptying. 30,31 Prolonged ambulatory ph studies would provide an opportunity to investigate the effect of these factors in patients with GERD. Compared with previous research into the effects of diet on gastroesophageal reflux using catheter-based ph measurement, the Bravo system is less invasive and has less impact on patient behavior. 14,17 The food supplied represented variations on a typical, highly palatable Western diet. Patients with reflux symptoms were assessed over 4 days, with 24 hours in each of the 4 dietary conditions, producing a large data set. Patient reports of reflux symptoms were spontaneous and were recorded immediately on the Bravo receiver. All these factors contributed to the acquisition of reliable, clinically relevant information. Possible limitations of the study include the recruitment of patients on the basis of reflux symptoms rather than endoscopic findings or esophageal acid exposure. Most patients did not have mucosal disease on endoscopy (although acid suppression was withheld for only 4 days before the study). Most had positive ph studies on the high-calorie diets; however, patients with negative ph studies and functional heartburn were not excluded. All patient groups appeared to respond in the same way to dietary interventions; however, it cannot be excluded that the effects on symptoms are most prominent in those with endoscopy-negative reflux disease (in whom visceral sensitivity often is heightened). The use of antacids as rescue treatment could have masked the effect of diet on esophageal acid exposure; however, the median use was less than 1 dose/day/ patient without evidence of greater use in high-fat and low-fat diets. Every effort was made to control factors unrelated to calorie density or fat content that could affect the refluxogenic potential of the diet; however, perfect control is not possible using real food. For example, variation in the amount of indigestible carbohydrate (eg, fiber) between the different meals is potentially confounding because in the short-term fiber reduces reflux, 1 but in the midterm colonic fermentation of indigestible carbohydrates increases the rate of transient lower esophageal sphincter relaxation and acid exposure. 32 Another issue is that it is not possible to control both the percentage fat content and the absolute fat content of a meal while varying calorie density and maintaining meal volume (considered essential). Finally, it remains possible that the long-term effects of diet on GERD may be different from the short-term effects reported in this study. If present, these effects are likely to be complex involving long-term adaptation to a diet and the effects of weight change on GERD (ie, interactions between obesity, fat intake, acid reflux, and symptoms may be present). 4,33 In conclusion, this study showed that the calorie density of the diet determines the severity of esophageal acid reflux in patients with GERD; whereas the percentage fat content of the diet has significant effects on visceral sensitivity and the likelihood that reflux events will cause symptoms. This was not a study of dietary management for GERD; however, changes in calorie density and percentage fat content had large, potentially important effects on esophageal acid exposure and symptoms that suggest clinical trials are warranted, especially in patients with endoscopy-negative reflux disease in whom therapy is focused on symptom control. 34,35

444 FOX ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 4 References 1. Fox M, Forgacs I. Gastro-oesophageal reflux disease. BMJ 2006; 332:88 93. 2. Lagergren J, Bergstrom R, Lindgren A, et al. Symptomatic gastroesophageal reflux as a risk factor for esophageal adenocarcinoma. N Engl J Med 1999;340:825 831. 3. Nebel OT, Fornes MF, Castell DO. Symptomatic gastroesophageal reflux: incidence and precipitating factors. Am J Dig Dis 1976;21:953 956. 4. El-Serag HB, Satia JA, Rabeneck L. Dietary intake and the risk of gastro-esophageal reflux disease: a cross sectional study in volunteers. Gut 2005;54:11 17. 5. Castell DO, Murray JA, Tutuian R, et al. Review article: the pathophysiology of gastro-esophageal reflux disease esophageal manifestations. Aliment Pharmacol Ther 2004;20(Suppl 9):14 25. 6. Becker DJ, Sinclair J, Castell DO, et al. A comparison of high and low fat meals on postprandial esophageal acid exposure. Am J Gastroenterol 1989;84:782 786. 7. Iwakiri K, Kobayashi M, Kotoyori M, et al. Relationship between postprandial esophageal acid exposure and meal volume and fat content. Dig Dis Sci 1996;41:926 930. 8. Penagini R, Mangano M, Bianchi PA. Effect of increasing the fat content but not the energy load of a meal on gastro-esophageal reflux and lower esophageal sphincter motor function. Gut 1998; 42:330 333. 9. Pehl C, Waizenhoefer A, Wendl B, et al. Effect of low and high fat meals on lower esophageal sphincter motility and gastroesophageal reflux in healthy subjects. Am J Gastroenterol 1999;94: 1192 1196. 10. Colombo P, Mangano M, Bianchi PA, et al. Effect of calories and fat on postprandial gastro-esophageal reflux. Scand J Gastroenterol 2002;37:3 5. 11. Holloway RH, Hongo M, Berger K, et al. Gastric distention: a mechanism for postprandial gastroesophageal reflux. Gastroenterology 1985;89:779 784. 12. Salvia G, De Vizia B, Manguso F, et al. Effect of intragastric volume and osmolality on mechanisms of gastroesophageal reflux in children with gastroesophageal reflux disease. Am J Gastroenterol 2001;96:1725 1732. 13. Emerenziani S, Zhang X, Blondeau K, et al. Gastric fullness, physical activity, and proximal extent of gastroesophageal reflux. Am J Gastroenterol 2005;100:1251 1256. 14. Pandolfino JE, Richter JE, Ours T, et al. Ambulatory esophageal ph monitoring using a wireless system. Am J Gastroenterol 2003;98:740 749. 15. Hirano I, Zhang Q, Pandolfino JE, Kahrilas PJ. Four-day Bravo ph capsule monitoring with and without proton pump inhibitor therapy. Clin Gastroenterol Hepatol 2005;3:1083 1088. 16. Nilsson M, Johnsen R, Ye W, et al. Lifestyle related risk factors in the aetiology of gastro-esophageal reflux. Gut 2004;53:1730 1735. 17. Anggiansah A, Fox M, Anggiansah R, et al. Day to day variability of 48 hour esophageal ph monitoring by bravo capsule (abstr). Gut 2005;54(Suppl 2):A22. 18. Johnson LF, Demeester TR. Twenty-four-hour ph monitoring of the distal esophagus. A quantitative measure of gastroesophageal reflux. Am J Gastroenterol 1974;62:325 332. 19. Weusten BL, Roelofs JM, Akkermans LM, et al. The symptomassociation probability: an improved method for symptom analysis of 24-hour esophageal ph data. Gastroenterology 1994;107: 1741 1745. 20. Pehl C, Pfeiffer A, Waizenhoefer A, et al. Effect of caloric density of a meal on lower esophageal sphincter motility and gastroesophageal reflux in healthy subjects. Aliment Pharmacol Ther 2001;15:233 239. 21. Hila A, Castell DO. Nighttime reflux is primarily an early event. J Clin Gastroenterol 2005;39:579 583. 22. Hunt J, Stubbs RJ. The volume and energy content of meals as determinants of gastric emptying. J Physiol 1975;245:209 225. 23. Hunt J, Smith J, Jiang C. Effect of meal volume and energy density on the gastric emptying of carbohydrates. Gastroenterology 1985;89:1326 1330. 24. Meyer JH, Lembo A, Elashoff JD, et al. Duodenal fat intensifies the perception of heartburn. Gut 2001;49:624 628. 25. Mangano M, Colombo P, Bianchi PA, et al. Fat and esophageal sensitivity to acid. Dig Dis Sci 2002;47:657 660. 26. Fried M, Feinle C. The role of fat and cholecystokinin in functional dyspepsia. Gut 2002;51(Suppl 1):i54 i57. 27. Feinle C, D Amato M, Read NW. Cholecystokinin-A receptors modulate gastric sensory and motor responses to gastric distension and duodenal lipid. Gastroenterology 1996;110:1379 1385. 28. Feinle C, Rades T, Otto B, et al. Fat digestion modulates gastrointestinal sensations induced by gastric distention and duodenal lipid in humans. Gastroenterology 2001;120:1100 1107. 29. Feinle C, Christen M, Grundy D, et al. Effects of duodenal fat, protein or mixed-nutrient infusions on epigastric sensations during sustained gastric distension in healthy humans. Neurogastroenterol Motil 2002;14:205 213. 30. Boulby P, Gowland P, Adams V, et al. Use of echo planar imaging to demonstrate the effect of posture on the intragastric distribution and emptying of an oil/water meal. Neurogastroenterol Motil 1997;9:41 47. 31. Kunz P, Feinle-Bisset C, Faas H, et al. Effect of ingestion order of the fat component of a solid meal on intragastric fat distribution and gastric emptying assessed by MRI. J Magn Reson Imaging 2005;21:383 390. 32. Piche T, des Varannes SB, Sacher-Huvelin S, et al. Colonic fermentation influences lower esophageal sphincter function in gastro-esophageal reflux disease. Gastroenterology 2003;124: 894 902. 33. Ruhl CE, Everhart JE. Overweight, but not high dietary fat intake, increases risk of gastro-esophageal reflux disease hospitalization: the NHANES I Epidemiologic Follow-up Study. First National Health and Nutrition Examination Survey. Ann Epidemiol 1999; 9:424 435. 34. North of England Dyspepsia Guideline Development Group. Dyspepsia: management of dyspepsia in adults in primary care: National Institute of Clinical Excellence. Newcastle upon Tyne, UK: Center for Health Care Services; 2004. 35. Talley N, Vakil N, and the Practice Parameters Committee of the American College of Gastroenterology. Guidelines for the management of dyspepsia. Am J Gastroenterol 2005;100:2324 2337. Address requests for reprints to: Mark Fox, MD, Esophageal Laboratory, St Thomas Hospital, London, SE1 7EH, United Kingdom. e-mail: markfox_2@hotmail.co.uk; fax: (0044) 208-2972315. Supported by the Guy s and St. Thomas Charitable Trust. The authors are grateful to Mark Traffas and Medtronic Inc. (Shoreview, MN) for the loan of an extra Bravo ph monitoring system during this study.

444.e1 FOX ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 5, No. 4 Appendix Commercially prepared foods formed the basis of the test meals to maintain the consistency of the diets, ease of transport and storage, and ease of preparation and consumption. The meals were designed to avoid unusual foods or strong flavors to suit the tastes of as many patients as possible, and a choice of sweet or savory liquid was also given for this reason. High fat, high calorie meal Pork sausages in a beef and gravy with a creamy mashed potato Solid 450 g 644 32.4 21.6 66.6 3.6 Liquid 1 - Soup 600 g 410 28.8 9.6 28.2 4.8 Liquid 2 - Milkshake 500 ml 553.6 36 15.7 42.1 N/A Total (solid liquid 1) 1050 g 1054 61.2 31.2 94.8 8.4 Total (solid liquid 2) 950 g 1198 68.4 37.3 108.7 N/A Solid liquid 1 (61.2 9)/1054 52% Solid liquid 2 (68.4 9)/1198 51% High-calorie, high-fat liquid meal Liquid 1 - soup 500 ml 410 28.8 9.6 28.2 4.8 Liquid 2 - milkshake 500 ml 553.6 36 15.7 42.1 N/A Total (liquid 1 liquid 2) 1000 ml 963.6 64.8 25.3 70.3 4.8 Liquid 1 liquid 2 (64.8 9)/963.6 60% Low-fat, high-calorie meal Penne pasta mixed with a tomato and herb sauce and cheddar cheese Solid 450 g 915.5 24.8 36 137.6 7.3 Liquid 1 - soup 600 g 212 5.4 4.8 36 2.4 Liquid 2 - milkshake 500 ml 323.6 8.5 17 44.6 N/A Total (solid liquid 1) 1050 g 1127.5 30.2 40.8 173.6 9.7 Total (solid liquid 2) 950 g 1239.1 33.3 53 182.8 N/A Solid liquid 1 (30.2 9)/1127.5 24% Solid liquid 2 (33.3 9)/1239.1 24% Low-fat, low-calorie meal Chicken with a honey and mustard sauce, vegetables, and cheddar cheese Solid 425 g 391.5 11.4 21.6 66.6 3.6 Liquid 1 - soup 600 g 212 5.4 4.8 36 2.4 Liquid 2 - milkshake 400 ml 274.6 6.8 13.6 39.6 N/A Total (solid liquid 1) 1025 g 603.5 16.8 26.4 102.6 6.0 Total (solid liquid 2) 1000 g 666.1 18.2 34.2 105.9 3.6 Solid liquid 1 (16.8 9)/603.5 25% Solid liquid 2 (18.2 9)/666.1 25%