Digestive Diseases and Sciences

, Volume 51, Issue 8, pp 1307–1312

A Very Low-Carbohydrate Diet Improves Gastroesophageal Reflux and Its Symptoms

Authors

    • Division of Gastroenterology and Hepatology and the Center for Esophageal Diseases and SwallowingUniversity of North Carolina
    • Bioinformatics Building CB#7080UNC-CH
  • Michelle T. Thiny
    • Division of Gastroenterology and Hepatology and the Center for Esophageal Diseases and SwallowingUniversity of North Carolina
  • Eric C. Westman
    • Division of General Internal MedicineDuke University
  • William S. YancyJr.
    • Division of General Internal MedicineDuke University
  • Nicholas J. Shaheen
    • Division of Gastroenterology and Hepatology and the Center for Esophageal Diseases and SwallowingUniversity of North Carolina
Original Paper

DOI: 10.1007/s10620-005-9027-7

Cite this article as:
Austin, G.L., Thiny, M.T., Westman, E.C. et al. Dig Dis Sci (2006) 51: 1307. doi:10.1007/s10620-005-9027-7

Abstract

Obese patients with gastroesophageal reflux disease (GERD) may experience resolution of symptoms utilizing a very low-carbohydrate diet. The mechanism of this improvement is unknown. This studied aimed to prospectively assess changes in distal esophageal acid exposure and GERD symptoms among obese adults initiating a very low-carbohydrate diet. We studied obese individuals with GERD initiating a diet containing less than 20 g/day of carbohydrates. Symptom severity was assessed using the GERD Symptom Assessment Scale—Distress Subscale (GSAS-ds). Participants underwent 24-hr esophageal pH probe testing and initiated the diet upon its completion. Within 6 days, a second pH probe test was performed. Outcomes included changes in the Johnson-DeMeester score, percentage total time with a pH<4 in the distal esophagus, and GSAS-ds scores. Eight participants were enrolled. Mean Johnson-DeMeester score decreased from 34.7 to 14.0 (P=0.023). Percentage time with pH<4 decreased from 5.1% to 2.5% (P=0.022). Mean GSAS-ds score decreased from 1.28 to 0.72 (P=0.0004). These data suggest that a very low-carbohydrate diet in obese individuals with GERD significantly reduces distal esophageal acid exposure and improves symptoms.

Keywords

Gastroesophageal reflux diseaseVery low-carbohydrate dietObesity24-hr ambulatory esophageal pH monitoring

Gastroesophageal reflux disease (GERD) is a common disorder in the United States, with 5%–20% of individuals experiencing symptoms at least once a week [1]. GERD is the most costly of all chronic gastrointestinal disorders; direct annual medical costs of GERD exceed $9 billion [2]. The prevalence of obesity continues to increase in this country, and previous research has found an association between GERD and obesity [3, 4]. Dietary modifications, such as avoidance of fatty foods, and lifestyle modifications, such as weight loss in obese individuals, are routinely recommended for patients with GERD [5]. However, reviews of the evidence of these lifestyle modifications have shown conflicting results for their efficacy [69].

Over the last several years, there has been considerable interest in using very low-carbohydrate diets as a means of weight loss. While many individuals have achieved significant levels of weight loss using these diets, the effect of these diets on other outcomes often associated with obesity, such as hyperlipidemia, cardiovascular risk, diabetes, and GERD, are still being investigated. A previous small case series showed a significant, almost immediate resolution of GERD symptoms in obese individuals initiating one type of a very low-carbohydrate diet [10]. However, it is unclear whether these symptom improvements were due to decreased distal esophageal acid exposure, some other physiological change, or a placebo effect. The purpose of this study was to evaluate the influence of a very low-carbohydrate diet on GERD symptoms and to objectively measure the changes in acid exposure in the esophagus within 1 week of diet initiation. We hypothesized that the initiation of a very low-carbohydrate diet would reduce distal esophageal acid exposure and improve GERD symptoms.

Methods

Study Design. This study was a prospective cohort study of individuals with obesity and GERD symptoms initiating a very low-carbohydrate diet. This study was approved by the University of North Carolina Committee for the Protection of the Rights of Human Subjects.

Subjects. Participants were recruited via advertisements placed in the internal medicine and family practice clinics at the University of North Carolina. Eight volunteers with GERD symptoms and obesity were enrolled. Inclusion criteria were the following: [1] age 18–70 years, [2] body mass index (BMI) >30 kg/m2, [3] desire to use a very low-carbohydrate diet for weight loss, and [4] score of >0.5 on the Distress Subscale portion of the Gastroesophageal Reflux Disease Symptom Assessment Scale (GSAS-ds). Exclusion criteria were the following: [1] known history of Barrett's esophagus or esophageal adenocarcinoma, [2] previous surgical antireflux procedure, [3] pregnancy or breastfeeding, [4] diabetes, [5] any history of heart disease, [6] history of any weight loss surgery, including gastric bypass, and [7] unwillingness to discontinue use of proton pump inhibitors (PPIs) or H2-receptor antagonists per study protocol. A brief, initial phone interview was conducted to assess eligibility criteria and participant interest. All individuals meeting eligibility criteria were then seen at a clinic where the study protocol was explained and informed consent was obtained.

Diet. All participants received formal one-on-one instruction on the required diet from one of the investigators (G.A.). The very low-carbohydrate diet was defined as a daily carbohydrate intake of less than 20 g a day. Participants were allowed to eat unlimited portion sizes of meat and eggs as well as limited portions of hard cheeses. Certain low-carbohydrate vegetables were allowed and their total daily quantity was specifically delineated in a three-page handout that each participant was given as a reference guide upon enrollment into the study. They were instructed that any food not listed as permissible was not allowed. All participants were required to keep a food diary starting 2 days prior to the initial pH probe through the duration of the study. Alcohol and caffeine were also prohibited throughout the duration of the study. Participants were encouraged to drink at least six 8-oz glasses of water a day during the study. Dietary compliance was assessed through review of participant food diaries and by interview. Additionally, we measured urinary ketones with a dipstick method on the day of insertion of the second pH probe.

Study Protocol. After obtaining informed consent, participants were scheduled for an initial 24-hr pH probe and a follow-up pH probe 3 to 6 days later. This window was chosen because it would allow flexibility for participants whose second visit would fall on a weekend and, also, because significant weight loss was unlikely to occur in such a short interval. Participants were instructed to continue their usual diet through the completion of their first pH probe. They were instructed to discontinue any use of proton pump inhibitors and H2-receptor antagonists for at least 12 days prior to their initial pH probe and until study completion. Only one of the eight participants was taking a PPI at the time of study enrolment, and none were taking an H2-receptor antagonist. All study participants were also required to avoid use of over-the-counter antacids during the study period. After completion of their first 24-hr pH probe, all participants initiated the very low-carbohydrate diet and remained on the diet through the completion of their second pH probe. Participation in the study was complete following the second 24-hr pH assessment.

Symptom Questionnaire. The GSAS-ds is a self-administered questionnaire used to evaluate the frequency and severity of 15 GERD-related symptoms within the past week [11, 12]. The symptoms evaluated are the following: (1) heartburn or burning pain inside the chest or breast bone, (2) a feeling of pressure or discomfort inside the chest, (3) food coming back into the mouth, (4) an acid or sour taste in the mouth, (5) frequent gurgling in the stomach or belly, (6) a feeling of pressure or lump in the throat, (7) nausea, (8) burning pain in the throat, (9) bloating or a feeling like the subjects had to loosen their belt or unbutton their pants/skirt, (10) belching, (11) flatulence or passing gas from below, (12) feeling full after eating a little, (13) bad breath, (14) coughing, and (15) hoarseness. Participants who responded yes to an individual symptom were then asked to rate how distressful or bothersome that particular symptom was over the past week on a 4-point scale (0 = not at all, 1 = somewhat, 2 = quite a bit, and 3 = very much). Participants completed the GSAS-ds questionnaire on the days of their two pH probes. The GSAS-ds score was calculated by averaging the severity scores (from 0 to 3) for the 15 GERD-related symptoms.

Ambulatory 24-hr pH Monitoring. Ambulatory 24-hr pH monitoring was performed using the Sandhills pH monitoring system (Sandhills Scientific, Highlands Ranch, CO), as previously described [13]. All probes were placed by the motility unit at our institution, and all studies were read by a single gastroenterologist (N.J.S.). The primary outcome of the 24-hr pH probe was the Johnson-DeMeester score [14, 15]. The Johnson-DeMeester score is calculated by an algorithm from the following data obtained from the distal esophageal sensor of the 24-hr pH probe: (1) percentage total time with pH<4.0, (2) percentage time with a pH<4.0 in an upright position, (3) percentage time with pH<4.0 in a supine position, (4) number of reflux episodes, (5) number of reflux episodes lasting longer than 5 min, and (6) length of the longest reflux episode.

Data Analysis. All data were entered into and analyzed using Stata 8.2 statistical software (STATA Corp., College Station, TX). Paired t-tests were used to analyze the primary outcomes of interest, which were changes in Johnson-DeMeester score, percentage total time with pH<4.0 in the distal esophagus, and GSAS-ds after initiating the very low-carbohydrate diet. A two-sided P value of <0.05 was considered statistically significant.

Results

A total of eight individuals were enrolled. Demographic characteristics are presented in Table 1. The mean number of complete days on the very low-carbohydrate diet prior to placement of the second pH probe and completion of the second GSAS-ds questionnaire was 4. Two individuals completed 3 days, two completed 4 days, three completed 5 days, and one completed 6 days on the diet. After beginning the diet, participants reported good compliance with the prescribed diet, with no reports of eating prohibited foods after reviewing participant food diaries.
Table 1

Baseline Characteristics of the Study Participants

Characteristic

Mean

SD

Minimum

Maximum

Age

40

10

23

54

Height (m)

1.63

0.08

1.49

1.74

Initial weight (kg)

117

32

66

158

Initial BMI (kg/m2)

43.5

9.2

30.2

58.2

Average weight change (kg)

–1.7

2.1

–4.5

1.1

Gender (% female)

100

   

Race (African American/Caucasian)

4/4

   
After initiating a very low-carbohydrate diet, participants demonstrated a significant decrease in Johnson-DeMeester score (mean±SE of 34.7 ± 10.1 before the diet vs. 14.0 ± 3.7 after initiating the diet; P=0.023). Of the five individuals who had abnormal baseline studies, all showed a substantial decrease in Johnson-DeMeester score following initiation of the diet as shown in Fig. 1. The three individuals who had normal baseline Johnson-DeMeester scores showed very little variation on their second pH probe. Participants also exhibited a significant decrease in the percentage time with a pH<4.0 in the distal esophagus (5.1 ± 1.3% before the diet vs. 2.5 ± 0.6% afterward; P=0.022) as shown in Fig. 2.
https://static-content.springer.com/image/art%3A10.1007%2Fs10620-005-9027-7/MediaObjects/10620_2005_9027_Fig1_HTML.gif
Fig. 1

Individual Johnson-DeMeester scores before and after initiating the very low-carbohydrate diet

https://static-content.springer.com/image/art%3A10.1007%2Fs10620-005-9027-7/MediaObjects/10620_2005_9027_Fig2_HTML.gif
Fig. 2

Percentage time with pH<4.0 in the distal esophagus before and after initiating the very low-carbohydrate diet

All eight individuals had evident improvement in their GSAS-ds score as shown in Fig. 3. Mean (±SE) GSAS-ds score decreased from 1.28 ± 0.15 prior to the diet to 0.72 ± 0.12 after initiation of the diet (P=0.0004). Assessing changes in the components of the GSAS-ds, the severity score of the symptom of “an acid or sour taste in the mouth” improved significantly, from 1.88 ± 0.35 prior to the diet to 0.75  ± 0.31 following initiation of the diet (P=0.007). The severity score of the symptom of “heartburn or burning pain inside the chest or breast bone” improved significantly, from 1.88 ± 0.23 prior to the diet to 0.88 ± 0.23 following initiation of the diet (P=0.019). The severity score of the symptom of “nausea or feeling like you were going to vomit” was also significantly improved, from 1.75 ± 0.31 before the diet to 0.13 ± 0.13 following initiation of the diet (P=0.002). The severity score of the symptom of “bloating or feeling like you had to loosen your belt or unbutton your pants/skirt” improved significantly, from 1.63 ± 0.50 before the diet to 0.88 ± 0.35 after the diet was initiated (P=0.048). The mean severity score of the symptom of “burning pain in the throat” also improved significantly, from 1.25 ± 0.31 before the diet to 0.38 ± 0.26 after initiating the diet (P=0.021). None of the other symptoms were statistically significantly changed. Two of the participants had positive urinary ketones at the time of placement of their second 24-hr pH probe.
https://static-content.springer.com/image/art%3A10.1007%2Fs10620-005-9027-7/MediaObjects/10620_2005_9027_Fig3_HTML.gif
Fig. 3

Individual GSAS-ds symptom scores before and after initiating a the very low-carbohydrate diet

Discussion

The purpose of this study was to assess the effect of a very low-carbohydrate diet on distal esophageal acid exposure and GERD symptoms in obese individuals with symptomatic GERD. Based on a previous case series suggesting an improvement in GERD symptoms immediately following the initiation of the diet, we hypothesized that a very low-carbohydrate diet would decrease distal esophageal acid exposure and improve the symptoms of GERD. We found preliminary evidence to support both hypotheses. Five of the eight symptomatic participants had abnormal Johnson-DeMeester scores at baseline. All five of these participants showed a substantial decrease in their Johnson-DeMeester score and percentage total time with a pH<4.0 in the distal esophagus. The magnitude of the decrease in Johnson-DeMeester scores following initiation of the very low-carbohydrate diet is similar to what has been reported with PPI treatment [16, 17].

We retained individuals who had normal baseline pH probes to evaluate their symptoms as well as to assess whether there might be significant random variation in distal esophageal acid exposure. Additionally, if only individuals with abnormal Johnson-DeMeester scores were retained and had follow-up pH probes, improvement might occur in mean scores regardless of the diet the participants followed because of regression to the mean. The three individuals who had normal baseline pH probes also had normal pH probes after having been on the very low-carbohydrate diet for 3 to 6 days. There was very little variation in their Johnson-DeMeester scores or in the percentage total time with a pH<4.0 in the distal esophagus. The other primary outcome of this study was the change in GERD symptoms following initiation of the very low-carbohydrate diet. All individuals, regardless of baseline pH probe results, had an improvement in their overall GSAS-ds symptom severity scores, and the mean decrease was highly statistically significant.

There is an epidemic of obesity in the United States, and the prevalence of GERD is higher among obese individuals compared to those with a normal BMI [3, 4]. Very low-carbohydrate diets have become popular among those pursuing weight loss. Therefore, it is concerning that we know so little regarding the effect of a very low-carbohydrate diet on gastrointestinal symptoms [18, 19]. Given that individuals who follow a very low-carbohydrate diet often consume diets enriched in fat, and that fat slows gastric emptying, one might postulate that a very low-carbohydrate diet would actually promote reflux. Indeed, a recent cross-sectional study found an increased prevalence of GERD among those consuming a high-fat, normal carbohydrate diet [20]. To the contrary, the findings of our prospective study confirm a previous case series that found a very low-carbohydrate diet improved symptoms related to GERD. Yancy et al. reported a case series of five patients who experienced almost-complete resolution of their GERD symptoms within a few days of initiating a very low-carbohydrate diet [10]. These patients noted return or worsening of their GERD-related symptoms when they attempted to return to a normal level of carbohydrate consumption. However, the study by Yancy and colleagues did not have an objective measure of distal esophageal acid exposure.

To our knowledge, this is the first study to document decreased distal esophageal acid exposure using 24-hr pH probes in participants with GERD initiating a very low-carbohydrate diet. The demonstration of improved Johnson-DeMeester scores and decreased percentage total time with a pH<4.0 lends support to a physiological rather than a placebo effect for the improved symptoms seen in individuals initiating a very low-carbohydrate diet. It is less clear why those participants with GERD symptoms, but normal acid exposures on pH monitoring, also had a significant improvement in their GERD symptoms. This observation might be consistent with a placebo effect. However, previous studies have demonstrated that 24-hr pH testing does not identify a substantial proportion of individuals with clinically significant reflux [21, 22].

The mechanisms underlying the improvement in our participants' symptoms and distal esophageal acid exposure are not understood. Previous work has suggested a potential mechanism for the improvement of GERD symptoms with alteration of dietary carbohydrate intake. In healthy volunteers, ingestion of lactose results in an increased number of transient lower esophageal sphincter relaxations (TLESRs), increased reflux episodes, higher esophageal acid exposures, and more severe GERD symptoms [23]. Oral administration of fructo-oligosaccharides reproduces manometric and pH probe findings similar to those seen after the administration of lactose [24]. Fructo-oligosaccharides are generally poorly absorbed in the small intestine but completely fermented in the colon and are frequently found in many processed food products. Therefore, it has been suggested that colonic fermentation of malabsorbed carbohydrates leads to the production of short-chain fatty acids that might be partially responsible for the increased GERD seen in individuals ingesting these carbohydrates. This hypothesis is enhanced by the observation that intracolonic infusion of short-chain fatty acids produces similar physiological changes with respect to GERD as those seen with oral administration of lactose and fructo-oligosaccharides.

Other studies have focused on hormonal effects as a potential explanation for changes in GERD symptoms induced by diet. Both cholecystokinin (CCK) and intestinal peptide YY (PYY) have been investigated without a clear consensus as to their role in GERD. A study by Clave found that endogenous CCK promoted postprandial GERD by reducing LES pressure and increasing the rate of TLESRs [25]. Investigators have also demonstrated that infusion of exogenous CCK increased the rate of TLESRs and that this effect was significantly reduced after administration of loxiglumide, a selective CCK-A receptor antagonist [25, 26]. Other authors have also found a potential role for CCK in the control of LES tone [27]. Several other hormones have been investigated, including somatostatin and γ-aminobutyric acid-B (GABAB) [28, 29]. However, there are no studies evaluating the effect of a very low-carbohydrate diet on any of these potential mediators.

There are other possible mechanisms for the finding of improved GERD with a very low-carbohydrate diet. It is possible that the improvement is related to reduced gastric distension. The effect of gastric distension may be mediated through CCK, but there are direct physical alterations of the LES with increased gastric distension as well. It might be expected that an isocaloric diet with a higher proportion of calories from carbohydrates would occupy a greater gastric volume than one with a higher fat content. Interestingly, increased fat intake, as may be seen with a very low-carbohydrate diet, should delay gastric emptying. It is reasonable to expect that delayed gastric emptying would increase GERD, a hypothesis that has been borne out in previous studies and is the basis for the current recommendations for patients with GERD to avoid high-fat foods [30, 31]. Paradoxically, our participants improved remarkably following a diet that most authorities would consider “refluxogenic.” Additionally, it does not appear as though achievement of ketosis is required for improvement of GERD. Despite good compliance with the diet as evidenced by their food diaries, only two of the eight participants, and only one of the five with an abnormal baseline pH probe, had positive urinary ketones. Because it can take several days of the low-carbohydrate diet before the development of ketosis, this finding does not imply noncompliance with the diet. Weight loss is also unlikely to explain the changes we observed. The participants were specifically studied after only a short time on the diet to avoid the confounding effect that weight loss might have on reflux. Indeed, the average weight loss during the study period was <2 kg, or <2% of these participants' initial body weight.

This study has several limitations. Foremost is study size. Our study represents the experience of only eight individuals and must be confirmed in larger numbers. Also, it is unclear whether the effects of the very low-carbohydrate diet are specific to that diet plan. Because we did not assess participants pursuing other weight loss diets, it is possible that a more traditional calorie-restricted diet with a standard proportion of calories from carbohydrates may have produced similar results. Additionally, while all participants did complete a food diary, the true extent of compliance with the diet is unknown. However, one might expect that noncompliance with the diet would bias the results toward the null. Next, while we did recommend control of food content, we made no attempt to control food volume. If the altered diet of the participants featured decreased total volume, some portion of the effect observed may be secondary to this. Finally, all of the individuals in the study were female. It is unknown whether these effects would be observed in a male population.

Despite these limitations, this study found objective evidence that obese individuals initiating a very low-carbohydrate diet had a profound reduction in their distal esophageal acid exposure. This finding requires further investigation to help understand the mechanisms of how diet influences reflux. This research may help elucidate additional pharmacologic methods for managing patients with GERD. Perhaps more importantly, this may also help develop more effective nonpharmacologic methods for managing GERD in some patients.

In conclusion, obese individuals with GERD symptoms initiating a very low-carbohydrate diet demonstrated improved Johnson-DeMeester scores, a lower percentage total time with a pH<4.0 in the distal esophagus, and improved GSAS-ds scores. We are currently planning a larger study with a control arm to assess whether these findings are specific to a low-carbohydrate diet, as well as the potential hormonal, manometric, and dietary mediators underlying this observation.

Copyright information

© Springer Science+Business Media, Inc. 2006