Acidic gastro-oesophageal reflux is common following meals, and this may be attributed partly to the increased frequency of transient lower oesophageal relaxations during the postprandial period.1^–4 However, the source of the postprandial acidic refluxate has been unclear. During periods of fasting, the gastric juice is highly acidic, with a pH of less than 2. However, on eating, the intragastric pH rises due to the buffering effect of the food, and this persists during the postprandial period. Acidic gastro-oesophageal reflux occurring during periods of postprandial gastric buffering presents a paradox.

Fletcher et al recently reported the presence of an unbuffered acidic region in the gastric cardia during the postprandial period which often traversed the squamocolumnar junction.5 This small acid pocket has been postulated to be the source of the acidic refluxate occurring after a meal. Further work by Pandolfino et al has confirmed less meal-related buffering in the gastric cardia of both normal controls and symptomatic gastro-oesophageal reflux disease patients.6 Both groups propose that this acidified zone could be the source of postprandial acid reflux events as the nadir cardia pH is similar to the nadir pH of acidic refluxate. The existence of a postprandial acid pocket has also been confirmed in studies by other groups.7 8

Several questions regarding this unbuffered acidic zone and its possible role in gastro-oesophageal reflux disease remain to be answered. It is unclear whether the postprandial acid pocket extends across the squamocolumnar junction or is confined to the gastric cardia region. The relationship of the acid pocket to the high pressure zone (HPZ) of the lower oesophagus is also unknown. The most important question, however, is whether the acid pocket contributes to the aetiology of reflux disease and/or to the high prevalence of carditis and intestinal metaplasia of the gastro-oesophageal junction. If the acid pocket is of pathological significance then it should be different in subjects with compared with those without gastro-oesophageal reflux disease.

Aims

• To define the location of the acid pocket in healthy subjects relative to the squamocolumnar junction and HPZ.

• To compare the acid pocket in patients with severe reflux disease versus healthy subjects.

SUBJECTS AND METHODS

The study population consisted of 12 healthy subjects (3 females) with no history of gastro-oesophageal reflux symptoms and 16 patients (7 females) with ⩾grade 3 reflux oesophagitis (n = 4) or Barrett’s oesophagus (n = 12). Patients and subjects were similar with respect to intragastric acidity (table 1), and thus the mean age of the healthy subjects was lower than that of the reflux group (30.9 years (range 21–58) vs 62.6 years (range 28–74)). Helicobacter pylori status was documented in all patients prior to the study (one healthy volunteer and one reflux patient were found to be Helicobacter positive). Fourteen of the reflux patients were on proton pump inhibitors at enrolment and the medication was discontinued a mean of 8 days prior to the study.

Study design

Each subject had two combined pH/manometric station pull-through studies, one fasted and one after a meal. The pH and manometric recordings were related to the anatomy by the use of endoscopically placed radio-opaque clips visualised with lateral chest x-ray. Fasting and postprandial barium studies were also performed to delineate the anatomy of the gastro-oesophageal junction.

Study protocol

The study took place over two consecutive days. On day 1 the subjects reported fasted and underwent upper gastrointestinal endoscopy. This was performed either with xylocaine throat spray or under conscious sedation with intravenous midazolam. Two standard haemostatic metal clips (HX-600-090, Olympus, UK) were deployed, using an endoscopic clip-fixing device (HX-5LR-1,Olympus, UK), 180° apart at the proximal margin of the gastric folds (which in volunteers equated to the squamocolumnar junction). In patients with endoscopic appearances consistent with a hiatus hernia, two further clips were placed at the diaphragmatic hiatus. One hour after the endoscopy, a barium swallow was performed to delineate the fasting morphology of the gastro-oesophageal junction and the position of the clips relative to the gastro-oesophageal junction and hiatal hernia sac, if present.

On the following day, the subjects again reported fasted from midnight, and had a combined apparatus, consisting of an antimony dual-channel pH catheter (Synectics Medical, Enfield, UK) and solid state manometer (Gaeltec CTO-3, Isle of Skye, UK), passed nasally into the stomach. The pH catheter had two sensors 7 cm apart with an external reference electrode. The electrodes were calibrated in buffer (Synectics, UK) before the procedure, and subsequent data were collected using Polygram Net software (Medtronic, Denmark). Calibration at room temperature was corrected by the computer software for pH measurements at body temperature. The solid state manometer had three sensors arranged 120° radially and was calibrated before the procedure, and subsequent data were collected using Polygram Net software (Synectics). The pH catheter was attached to the manometer so that the distal pH sensor was 1 cm proximal to the three radial manometry sensors (and the proximal pH sensor was 8 cm proximal to the manometry sensors).

Position in the stomach was confirmed by acidic gastric pH recorded by both pH sensors and an increased pressure measured by the manometer on deep inspiration. The patient was placed in a semi-recumbent position and, after 15 min with the apparatus in the stomach, the apparatus was withdrawn at 1 cm intervals every 1 min. This pull-through continued until the manometry sensors recorded oesophageal pressure—that is, when the end-expiratory pressure in all three sensors was lower than the intragastric pressure. At this point, the apparatus was immediately taped to the nose and lateral x-rays were performed immediately in the semi-recumbent study position in an adjoining room. To minimise patient movement, the subjects were wheeled over to the x-ray area on the trolley used for the pull-through study, thus maintaning the position held during the pull-through study. These semi-recumbent lateral chest x-rays were performed to establish the position of the clips relative to the pH/manometry apparatus. The x-rays were taken on inspiration and expiration, and the mid-respiratory position of the clips was calculated. The measured distance between the pH sensors (known to be 7 cm apart) on x-ray was used as an internal scale for other radiographic measurements.

The apparatus was re-introduced into the stomach and the position confirmed as before. The subjects were then given a meal of fried fish and French fried potatoes. A second pull-through study was started 15 min after finishing the meal, again withdrawing the apparatus at 1 cm intervals every 1 min. As before, the apparatus was fixed to the nose and the patient underwent a lateral chest x-ray to establish the position of the clips relative to the apparatus. The pH/manometry apparatus was then removed and the patient underwent a second barium swallow to delineate the morphology of the postprandial gastro-oesophageal junction.

Analyses

For each pull-through study, the pH was recorded at each position of both the proximal and distal pH electrode. The mean pH for the time interval at each catheter position was then calculated. The fasting pH step-up point was defined as the most distal catheter position where mean pH was >4. In the postprandial state, the pH step-up point could only be identified if there was an unbuffered acidified zone in the proximal stomach and the pH step-up point was the catheter position where the pH rose above 4 proximal to this acidic region.

The lower border of the HPZ was defined as the first point where there was an increase of >2 mm Hg in end-expiratory pressure above gastric baseline measured in one of the manometry sensors. The proximal border of the HPZ was identified by a step-down in the recorded end-expiratory pressure to the intrathoracic pressure.

The position of the proximal margin of the gastric folds was calculated by measuring the distance between the end of the manometry sensor and the mid-point of the two endoclips placed at the proximal gastric folds on lateral chest x-ray. This distance was then added to the distance the manometry sensor was from the nares, allowing calculation of the position of the proximal gastric folds from the nares.

To correct for error in measuring the position of the pH step-up and lower border of the HPZ using 1 cm increments, 0.5 cm was added to the distance from the nares for the position of each mean pH, and the mean end-expiratory pressure measured.

The postprandial maximal intragastric pH was defined as the highest mean pH recorded distal to the position of the proximal gastric fold clips after the meal, and reflected postprandial buffering.

A region of highly acidic pH at the gastro-oesophageal junction was observed in both normal and reflux patients after the meal. To compare these two groups directly, the number of consecutive 1 cm increments (from the pull-through study), where the mean pH was <2 immediately distal to the location of the proximal gastric folds, was recorded. For each patient there were two pH pull-through recordings available for analysis (one from the proximal pH sensor and one from the distal pH sensor).

Statistical analysis

Statistical analysis was performed using one-sample Wilcoxon test or Mann–Whitney U test unless specified otherwise. Results are given as medians and ranges unless otherwise specified

Ethics

The study was approved by the North Glasgow University NHS Trust Ethics Committee. All subjects participating gave written informed consent.

RESULTS

Movement of radio-opaque clips with respiration

Distal movement of the endoscopically placed clips was seen radiologically with inspiration in both groups. In the healthy subjects, the respiratory movement of the clips at the proximal gastric folds relative to the pH/manometry apparatus was 2.5 cm (range 0.2–4.6). The proximal gastric folds moved 0.2 cm (range −0.2 to 2.2) distally with inspiration in the reflux group. There was a strong trend for increased movement with respiration in the healthy subjects compared with the reflux patients (p = 0.08). There was no significant respiratory movement of the diaphragmatic clips between the inspiratory and expiratory phases of respiration measured in the reflux group.

Healthy subjects—fasting

In the healthy subjects, the lower oesophageal HPZ had its proximal border at 44.5 cm (range 39.5–45.5) from the nares and it extended 4 cm (range 3–5) distal to this. The proximal margin of the gastric folds was 2.8 cm (range −0.1 to 4.6; p = 0.005) distal to the proximal border of the HPZ.

The maximal fasting intragastric pH for the 12 normal subjects was 1.8 (range 1.0–6.0). The fasting pH step-up occurred 3.5 cm (range 2–5) distal to the proximal border of the HPZ. The fasting pH step-up point was thus slightly distal to the proximal gastric folds (median 0.8 cm; range 3.0 to −1.5; p = 0.08), and 1 cm (range −1 to 2; p = 0.08) proximal to the lower border of the HPZ (table 1).

Healthy subjects—after the meal

Following the meal, the proximal border of the HPZ was unchanged from its fasting position but the length of the HPZ was 1.5 cm shorter than its fasting length (p = 0.006) (table 1). Following the meal, there was therefore loss of the distal 1.5 cm portion of the HPZ. The proximal margin of the gastric folds was 2.6 cm (range −1.1 to 4.3) distal to the proximal border of the HPZ, which was not significantly different from its fasting position.

Following the meal, the maximal intragastric pH was increased at 6.1 (range 2.4–7.0) compared with the fasting maximal pH of 1.8 (table 1). An unbuffered acid pocket (a region of intragastric pH of ⩽2, distal to the clips at the proximal margin of the gastric folds) was observed in eight subjects with the distal pH sensor and in three of these subjects also by the proximal sensor. In no subjects was an acid pocket detected only by the proximal pH sensor. The eight acid pockets detected by the distal sensor had a median length of 1.5 cm (range 1–5), and the three acid pockets detected by the proximal sensor had a median length of 3 cm (range 1–4).

Of the 11 acid pockets detected by either electrode in eight healthy subjects, the lowest intragastric pH detected by the respective pH sensor was within the acid pocket. The median postprandial intragastric pH nadir was 1.1 (range 0.7–2.0) and was located 1.3 cm (range 0–4.5) distal to the proximal margin of the gastric folds. The proximal extent of the acid pocket was 1.1 cm distal to the proximal margin of the gastric folds (range −1.7 to 4.5 cm vs 0 cm; p = 0.07).

Seven of the eight healthy subjects with acid pockets had adequate manometry to localise the acid pocket relative to the lower border of the HPZ. Ten acid pockets from the proximal (n = 3) and distal sensors (n = 7) combined exhibited a median distance of 0 cm (range −2 to 3) between their proximal extent and the lower border of the HPZ. The postprandial acid pocket was thus occupying the fasting location of the distal HPZ that opened following the meal (fig 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Location of the fasting pH step-up and postprandial acid pocket relative to manometric landmarks in 11 studies with acid pockets in healthy subjects. All values are medians of the results obtained in the 11 studies in healthy subjects that exhibited an acid pocket. *Significant shortening of the high pressure zone seen after the meal (p<0.05). The respiratory inversion point was a median 1 cm distal to the proximal border of the high pressure zone in both the fasting and postprandial state. The high pressure zone is the region filled by oblique lines.

Reflux patients—preprandial and postprandial anatomy

In the reflux patients, the postprandial position of the proximal border of the HPZ was 41.5 cm (range 34.5–44.5) from the nares, which was similar to its fasting position (table 1). The HPZ extended 3 cm (range 1–6) distal from its proximal border after the meal. There was a strong trend for the HPZ to shorten after the meal (median difference 1 cm; range −1 to 3; p = 0.06) due to the loss of its distal portion. Postprandially, the proximal margin of the gastric folds was 1.6 cm (range −0.4 to 4.5) distal to the proximal border of the HPZ, which was similar to its fasting position. The diaphragmatic clips were 4.7 cm (range 1.9–7.3) distal to the proximal border of the HPZ after the meal, which was similar to their fasting position. The length of the hiatus hernia, as determined by the distance between the clips at the proximal margin of the gastric folds and the diaphragmatic impression, was 2.3 cm (range 0.6–4.7) after the meal, which was similar to its fasting length (1.6 cm; range 0.9–4.1; p = 0.8).

Postprandial acid pockets in reflux patients versus controls

In the 16 reflux patients the fasting maximal intragastric pH was 1.6 (range 0.9–6.9) and was similar to that seen in healthy subjects (table 1). Following the meal the maximal intragastric pH was increased to 4.8 (range 1.1–7.8) which, again, was similar to that of the healthy subjects (table 1).

An unbuffered acid pocket (a region of high acidity of pH ⩽2 distal to the clips at the proximal gastric folds) was observed by the distal pH sensor in 13 of 16 reflux patients. Ten of the 16 reflux patients also had an acid pocket detected by the proximal sensor. In no patient was an acid pocket detected by only the proximal pH sensor. There was a significantly higher frequency of acid pockets in the reflux patients (23/32 studies) than in the healthy subjects (11/24) (p<0.05). This difference in frequency was predominantly due to a higher incidence of acid pockets detected by the proximal sensor in the reflux patients (10/16) compared with healthy volunteers (3/12) (p = 0.049). The frequency of acid pockets detected by the distal sensor was similar in the reflux patients (13/16) and healthy volunteers (8/12) (p = 0.378). This difference between the reflux patients and controls was evident in the earlier postprandial period when the proximal sensor was passing through the gastro-oesophageal region. A typical tracing of an acid pocket in a reflux patient is shown in fig 2.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 Example of a pH and manometric study obtained from a reflux patient during station pull-through at 1 cm intervals each minute. The tracing at the top is from a proximal pH sensor demonstrating a region of low acidity at 43 and 42 cm from the nares. The middle three tracings are from the radially positioned manometry sensors, demonstrating the position of the high pressure zone. The bottom tracing is from the distal pH sensor, again demonstrating a large acid pocket extending from 44 to 40 cm.

The median length distal to the proximal margin of the gastric folds of the 23 acid pockets from both sensors combined in the reflux group was 3 cm (range 1–15) and this was significantly longer than the median length of the 11 acid pockets seen by both sensors in the healthy subjects (median length 2 cm; range 1–5; p<0.05). The length of the acidified zone (pH <2), that extended distal to the proximal margin of the gastric folds, for all sensors for individual pH probes is given in fig 3.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3 The length of the acid pocket (pH ⩽2) distal to the clips at the proximal margin of the gastric folds in 12 healthy subjects and 15 reflux patients. One reflux patient is not included as it was not possible to visualise the radio-opaque clips adequately. Shaded boxes indicate interquartile ranges. All studies in which there was no acid pocket distal to the clips at the proximal margin of the gastric folds were plotted as a value of 0 cm.

Of the 23 acid pockets detected by either electrode, the lowest postprandial intragastric pH was within the acid pocket in 21 of the 23 recordings. The median postprandial intragastric pH nadir within the acid pockets was 1.1 (range 0.7–1.8) and was located a median 1.3 cm (range 0.3–7.3) distal to the proximal margin of the gastric folds. Both the acid pocket pH nadir and the location of the nadir within the acid pocket was similar to that seen in healthy subjects.

The distance between the proximal margin of the gastric folds and the proximal extent of the intragastric acid pocket was calculated for both sensors in each subject group. If the acid pocket extended to or above the level of the proximal gastric folds, a value of 0 cm was recorded. Combined analysis of acid pockets from both sensors revealed the proximal extent of 11 acid pockets (found in eight healthy subjects) to be 1.1 cm (range 0–4.5) distal to the proximal margin of the gastric folds, whereas 23 acid pockets (from 13 reflux patients) had a median distance of 0 cm (range 0–1.5) between the proximal extent of the acid pocket and the proximal margin of the gastric folds (p = 0.005) (fig 4).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4 Distance between the proximal extent of the postprandial acid pocket distal to the proximal margin of the gastric folds in healthy subjects and reflux patients. Combined studies from proximal and distal pH sensors are shown, and were only included if an acid pocket was demonstrated distal to the proximal margin of the gastric folds. Eleven studies in the healthy subject group and 23 studies in the reflux group were analysed. Shaded boxes indicate interquartile ranges. If the proximal edge of the acid pocket was located above the proximal margin of the gastric folds, then a value of 0 cm was reported.

The length of the extension of the acid pocket distal to the diaphragmatic impression was similar in the reflux patients (median 2 cm; range −2 to 9 cm) and healthy subjects (median 2 cm; range 1–5 cm). The longer acid pocket seen in the reflux patients thus corresponded to the proximal migration of their proximal gastric folds above the diaphragmatic hiatus—that is, their hiatus hernia. This location of the extended acid pocket in the reflux patients is seen when the median pH at each catheter station is plotted against the position of the anatomical and manometric landmarks (fig 5). The correlation of the individual postprandial hiatal hernia length and length of the acid pocket measured by the distal pH sensor was not found to be statistically significant (R = 0.212).

• Download figure
• Open in new tab
• Download powerpoint

Figure 5 The median of mean pH measurements by the proximal sensor at each catheter position relative to the nares in 12 healthy subjects and 16 reflux patients. The position of the high pressure zone (HPZ) for each subject group is shown by the boxed area and represents the median positions of the proximal and distal borders of the high pressure zone. The median position of clips at the proximal margin of gastric folds for each group is represented by an asterisk. The vertical dotted line represents the median position of clips at the diaphragmatic impression in the reflux group. The respiratory inversion point is represented by a downward pointing arrow.

Radiological studies

Comparison of fasting and postprandial barium swallows revealed no radiological alteration in the morphology of the hiatal sac in response to the meal. In addition, there was no evidence of food debris seen within the hiatal sac after the meal.

DISCUSSION

This study confirms the frequent presence of a region of unbuffered acidity at the gastro-oesophageal junction in healthy subjects following a meal. Previous work by Fletcher et al5 first described the presence of an unbuffered postprandial acid pocket, defined as a minimum pH <2, using a station pull-through technique. Using the more strict criterion of a mean pH ⩽2, we identified an acid pocket in 46% (11/24) of studies in healthy subjects. The term unbuffered is justified as the pH was similar to fasting pH at a time when the pH in the main body of the stomach was substantially higher due to the effects of the meal. The frequency of acid pockets detected was higher with the distal pH sensor than with the proximal sensor. This may reflect elongation/evolution of the acid pocket with time as the distal sensor traversed the proximal stomach several minutes after the passage of the proximal sensor.

In the present study, we were able to document the location of the unbuffered region relative to the HPZ in our healthy subjects. The acid pocket extended proximally as far as the distal end of the postprandial HPZ. We observed that the HPZ was 1.5 cm shorter after the meal than when fasted and that this was due to loss of its distal segment, as previously observed by Manning et al.9 The median length of the acid pockets was 2 cm in healthy subjects and thus they were located mainly within the abdominal portion of the HPZ that “opens” after ingesting a meal (fig 1).

In the original report of the unbuffered acid pocket in healthy subjects, Fletcher et al5 observed that it extended across the squamocolumnar junction onto the distal oesophagus. In the current study, we also studied the proximal extent of the acid pockets relative to the radio-opaque clips applied to the proximal margin of the gastric folds and which corresponded to the squamocolumnar junction in our healthy subjects. We observed that the median proximal margin of acid pockets was 1.1 cm distal to the clips, and only two of 11 extended proximal to the clips. We believe the discrepancy between the current study and the earlier report of Fletcher et al5 can be explained by the movement with respiration of the clips attached at the squamocolumnar junction relative to the pH probe. We observed in our current study that the clips moved a median of 2.5 cm during the respiratory cycle and therefore we recorded the pH and pressure relative to the clips at the mid-respiratory cycle. In the earlier study by Fletcher et al,5 the clip position was measured only on inspiration, which would tend to overestimate the proximal extent of the acid pocket relative to the squamocolumnar junction. In addition, in the study by Fletcher et al,5 the minimum pH at each site was analysed rather than the mean pH as in the current study. The proximal movement of the squamocolumnar junction with expiration relative to the pH probe will result in it being intermittently acidified even when at a mean position 1 cm above the squamocolumnar junction, and the effect will be much more apparent when employing minimal pH rather than mean pH at each station.

Our observation that the acid pocket rarely extends across the squamocolumnar junction in healthy subjects is consistent with the findings of Fletcher et al10 when a pH electrode was clipped within the HPZ 5 mm proximal to the squamocolumnar junction. At that location, intermittent acidification was observed and interpreted as short segment reflux, but there was no evidence of prolonged acidification following a meal as would be expected from the presence of an acid pocket extending proximal to the squamocolumnar junction.

The location of the unbuffered acid pocket relative to the HPZ in the healthy subjects provides information regarding its nature and origin. If the region of unbuffered pH had been confined to the distal region of the sphincter, it could have been explained by the pH probe being encircled by acid-secreting mucosa and thus recording mucosal surface pH. However, the fact that the unbuffered region extended >1 cm distal to the HPZ supports the existence of a finite amount of unbuffered gastric secretions.

In the present study we investigated whether healthy subjects and patients with severe reflux disease differed with respect to proximal postprandial unbuffered acid pockets. The great majority of the reflux patients had evidence of hiatus hernia, representing a well-recognised association.11^–15 Reflux patients on regular acid inhibitory therapy discontinued it 1 week prior to the studies, a time point at which we have previously shown that gastric acid secretion is similar to pretreatment levels.16 The reflux patients and healthy subjects were also confirmed to be similar with respect to maximal fasting and postprandial intragastric pH.

Three differences were noted with respect to the postprandial unbuffered acid pockets between the reflux patients and healthy subjects. First, the reflux patients had a significantly higher incidence of unbuffered acid pockets detected distal to the proximal margin of the gastric folds than the healthy subjects. Second, the acid pockets in the reflux patients extended a greater distance distal to the proximal gastric folds than those in the healthy subjects. Comparing acid pockets seen in reflux patients with those in the healthy subjects showed that the median length of postprandial pH ⩽2 extending distal to the proximal margin of the gastric folds was 3 cm compared with 2 cm, respectively (p <0.05). The third difference between the reflux patients and healthy subjects was that the acid pockets extended closer to the proximal margin of the gastric folds in the former. In the healthy subjects, the proximal extent of the acid pocket was 1.1 cm distal to the proximal gastric folds, whereas it was 0 cm in the reflux patients.

The current studies do not allow us to explain the aetiology of the extended acid pocket in the reflux patients, but it may be related to their abnormal anatomy in the region of the gastro-oesophageal junction. In the reflux patients, the proximal margin of the gastric folds was displaced proximally due to herniation of the proximal stomach into the chest. The location and median length of the extended region of unbuffered acidity both corresponded to those of the hiatus hernia. We could not demonstrate a significant correlation between the length of the hiatus hernia and length of the acid pocket in the reflux patients, but that may have been due to the fact that the hiatus hernias were of similar size. A hiatus hernia would provide a physiologically plausible unbuffered acid pocket provided it is lined by healthy acid-secreting gastric mucosa and provided food does not reside within it. Our postprandial barium studies showed no evidence of food within the hernias. There is little information on the parietal cell density of the mucosa lining hiatus hernias, but it is a topic worthy of study.

Mittal et al demonstrated in 1987 that gastro-oesophageal refluxate in patients with a hiatus hernia originates from the hernial sac.17 They proposed that the hernial sac provided a reservoir of gastric juice readily available for refluxing into the oesophagus. Our current studies indicate that the gastric juice at this anatomical site following a meal is highly acidic due to escaping the buffering effect of the meal, and potentially highly damaging to the oesophageal mucosa.

In summary, our study demonstrates that postprandial unbuffered acid pockets are larger, more frequent and closer to the gastro-oesophageal junction after meals in reflux patients than in healthy subjects. The extended acid pocket in reflux patients corresponds to the extension of the gastric mucosa above the diaphragmatic hiatus and may be explained by it. This region of unbuffered postprandial acidic juice just below the gastro-oesophageal junction may contribute to severe reflux disease.