Intestinal Metaplasia in Patients With Columnar Lined Esophagus is Associated With High Levels of Duodenogastroesophageal Reflux

Abstract

Objective

To evaluate the rate of duodenogastroesophageal reflux in patients with columnar lined esophagus compared with patients with gastroesophageal reflux disease without columnar lined esophagus, and to analyze whether it is related to the presence of specialized columnar epithelium in the metaplastic segment.

Summary Background Data

The carcinomatous degeneration of columnar lined esophagus originates from a specialized columnar epithelium. The appearance of this metaplastic phenomenon is clearly related to severe prolonged gastroesophageal reflux, but only some of these patients finally develop columnar lined esophagus. For this reason other factors have been suggested, particularly the role played by the reflux of duodenal contents into the esophagus.

Methods

The authors studied 15 healthy volunteers (control group), 10 patients with reflux symptoms but without endoscopic lesions, 20 patients with reflux esophagitis without columnar lined esophagus, and 35 patients with columnar lined esophagus (complicated with ulcers or stenosis in 8 cases), of whom 22 had intestinal metaplasia. To assess the reflux of duodenal contents into the esophagus, all the patients underwent Bilitec 2000 and 24-hour esophageal pH monitoring.

Results

The presence of bilirubin in the material refluxed into the esophagus was greater in the patients with columnar lined esophagus than in the rest of the groups. Likewise, duodenogastroesophageal reflux was greater in the columnar lined esophagus patients who had intestinal metaplasia.

Conclusions

Duodenogastroesophageal reflux may play a major role in the development of columnar lined esophagus, especially in patients with intestinal metaplasia.

The traditional concept of Barrett’s esophagus involved the existence of a segment of columnar metaplasia in the distal esophagus of 3 cm or more in length. However, because the type of epithelium that really predisposes to the development of adenocarcinoma has been shown to be the intestinal type, ¹ Barrett’s is currently defined as intestinalized columnar mucosa of any length. Patients without intestinalization should be referred to as having columnar lined esophagus (CLE). Why some patients with CLE develop intestinal metaplasia and others do not remains unknown.

The appearance of CLE is clearly related to severe prolonged acid gastroesophageal reflux (GER), although not all patients with high rates of acid GER develop CLE. ² For this reason other factors have been suggested. ³ Among these is the role played by the reflux of duodenal contents into the esophagus. Experimental studies have shown that this reflux can induce the appearance of intestinal metaplasia and adenocarcinoma. ^4,5 Moreover, several clinical studies have suggested that duodenogastroesophageal reflux (DGER) is related to the development of CLE, especially if it is associated with complications such as ulcers, stenosis, or dysplasia. ^6,7 Most of these studies have involved the evaluation of duodenal contents in the gastric aspirate ^8,9 or pH monitoring. ^10,11 However, the reliability of these techniques is questionable: the former does not directly measure biliopancreatic reflux into the esophagus, and studies with pH monitoring may have numerous artifacts. ¹² In 1993 a technique was introduced for evaluating DGER directly using 24-hour outpatient monitoring of the presence of bilirubin in the material refluxed into the esophagus (Bilitec 2000; Synectics, Stockholm, Sweden). ¹³ With this technique, studies have shown that the rate of biliopancreatic reflux into the esophagus is increased in patients with CLE. ^14,15

The present study aims to evaluate the rate of DGER in patients with CLE versus patients with gastroesophageal reflux disease (GERD) without CLE and to analyze the relation of DGER to the presence of specialized columnar epithelium in the metaplastic segment.

METHODS

The present study included 15 healthy volunteers without reflux symptoms and with normal findings on upper gastrointestinal endoscopy (control group) and 65 patients with GERD seen at our esophageal surgery department during the past 2 years. The 65 patients were divided into three groups: patients with symptoms typical of GER and pathologic rates of acid GER revealed by 24-hour esophageal pH monitoring, but with no endoscopic evidence of esophagitis (n = 10); patients with reflux esophagitis but without CLE (n = 20); and patients diagnosed with CLE (n = 35), divided into two subgroups: those with uncomplicated CLE (n = 27) and those with CLE complicated by stenosis or ulcers (n = 8). The patients with CLE were divided depending on the presence (n = 22) or absence (n=13) of intestinal metaplasia in the histologic study. The patients with intestinal metaplasia were in turn separated into two subgroups according to whether they had mild dysplasia (n = 6) or not (n = 16). No patient had high-grade dysplasia or adenocarcinoma. CLE was considered when there was circumferential columnar metaplasia in the distal esophagus of 3 cm or more in length independent of the type of epithelium found, or when there was specialized columnar epithelium in the biopsy samples taken from shorter metaplastic segments. The mean length of the metaplastic segment in the patients with CLE without intestinal metaplasia was 6 ± 2.1 cm (median 5.5, range 3 to 10). In patients with CLE with intestinal metaplasia, the mean length was 6.4 ± 3.4 cm (median 5, range 2 to 15). All patients gave written informed consent, and the procedures were conducted in accord with the ethical standards of the committee on human experimentation of the institution. The age and sex of the different study groups are shown in Table 1.

Table 1. DISTRIBUTION OF PATIENTS BY AGE AND SEX

CLE, columnar lined esophagus; GER, gastroesophageal reflux.

All the patients underwent upper gastrointestinal endoscopy, which in the patients with CLE included multiple biopsies, esophageal manometry, 24-hour pH monitoring, and 24-hour determination of bilirubin in the esophagus. The endoscopy was performed by the same endoscopist, and the biopsies from the CLE patients were taken systematically as follows: the first four (one in each quadrant of the circumference) immediately above the squamocolumnar junction, the next immediately below, and then distally every 1 to 2 cm, according to the length of the metaplastic segment, as far as the cardia. For the 24-hour pH monitoring, a Mark III pH meter (Synectics) was used with a double-canal pH probe and with two electrodes 15 cm apart. The proximal electrode was placed 5 cm above the lower esophageal sphincter, previously located by esophageal manometry, which enabled us to quantify the esophageal pH. The second, or distal, electrode was positioned at the stomach and was used to analyze gastric pH. The readings obtained were evaluated in accord with Johnson and DeMeester’s methodology. ¹⁶

The DGER study was based on colorimetric analysis of the material refluxed into the esophagus using a portable spectrophotometer, Bilitec 2000 (Synectics), which detects the presence of bilirubin. This system consists of a fiberoptic probe, 140 cm long and 3 mm in diameter. At the distal end it has a 9.5-mm-long, 5-mm-in-diameter head with a 2-mm orifice in which there are two diodes that emit light at different wavelengths: 470 nm (bilirubin value) and 565 nm (reference value). The probe is placed beside the esophageal pH electrode, 5 cm above the lower esophageal sphincter, and simple chest radiography is performed to check that the probes are placed correctly. The study is performed as an outpatient routine over 24 hours. Measurements are taken every 8 seconds, and the mean between two measurements is calculated to obtain a total of 5,400 counts in 24 hours. A measurement is considered positive (presence of bilirubin in the esophagus) when absorption is 0.14 units or greater. With values less than this, the measurement may be affected by the presence of mucosity or gastric particles, thus giving false-positive results. At least a week before the study, all medications that might alter motility and gastric secretion (e.g., proton pump inhibitors, histamine antagonists, prokinetics) are suspended. The patient is instructed to eat three meals per day and to avoid foods and medication that might alter the study as a result of their colorimetric or pH characteristics. Alcohol and smoking are also prohibited. The patient is given a diary in which to note the times of meals, situations of lying down and standing, and the appearance of symptoms. The probes are removed after 24 hours, and a software package (Esophogram; Gastrosoft, Irving, TX) is used to extract data and perform calculations to obtain numeric and graphic results.

The mean, standard deviation, median, and range were calculated for all the parameters studied. For statistical comparison of the data, we used Mann–Whitney nonparametric test. All the results are expressed as medians and ranges.

RESULTS

The overall study results are shown in Table 2. The median percentage of time with pH less than 4 was statistically greater (P < .01) in the patients with CLE than in the other groups. There were no differences between the reflux patients without esophagitis and those with esophagitis. There were also no differences between patients with complicated and uncomplicated CLE. Analysis of the time recorded with esophageal pH of more than 7 showed no significant differences between the groups (Table 3); for most of the recording time, pH was 4 to 7 in all the groups. Comparison of the mean percentage of total time with gastric pH less than 4 and of mean gastric pH showed no significant differences between the groups (Table 4).

Table 2. MEDIAN PERCENTAGE OF TOTAL TIME WITH pH < 4 AND BILITEC > 0.14

CLE, columnar lined esophagus; GER, gastroesophageal reflux.

*P < .05 compared with control group.

**P < .01 compared with the rest of the groups.

Table 3. RESULTS OF 24-HOUR pH MONITORING

CLE, columnar lined esophagus; GER, gastroesophageal reflux.

Table 4. RESULTS OF INTRAGASTRIC ACIDITY WITH 24-HOUR pH MONITORING

CLE, columnar lined esophagus; GER, gastroesophageal reflux.

The Bilitec 2000 study revealed that the presence of bilirubin in the refluxed material was greater in the patients with CLE than in the rest of the groups (P < .01), with no significant differences between the patients with GERD without CLE with or without esophagitis (see Table 2 and Fig. 1). As in the pH study, there were no differences between patients with complicated and uncomplicated CLE (27% and 26.7%, respectively).

Figure 1. Percentage of Bilitec readings of more than 0.14 in the study groups.

The median percentage of time with a bilirubin level of 0.14 or more was higher in the CLE patients who had intestinal metaplasia in the histologic study (P < .05). This did not occur when comparing the percentage of total time with pH less than 4 (Table 5). The patients with mild dysplasia had significantly greater (P < .05) bile reflux than those without dysplasia, and there were no differences in acid reflux between the two groups (see Table 5).

Table 5. PERCENTAGE OF CLE PATIENTS WITH pH < 4 AND BILITEC > 0.14

CLE, columnar lined esophagus.

*P < .05.

DISCUSSION

The involvement of bile and pancreatic juice in the metaplastic change of esophageal mucosa and its possible evolution to dysplasia and adenocarcinoma has been studied using various techniques, such as quantification of bile acids or isotopic markers (Tc₉₉-HIDA) in the gastric or esophageal aspirate, ^9,17 or with 24-hour outpatient pH monitoring, with the existence of alkaline reflux considered when the pH was 7 or more. ¹¹ However, the term “alkaline reflux” is erroneous because it may be due to factors such as saliva, food remains, or oral infections; hence, it does not always imply the existence of DGER. ¹¹ Some authors ¹⁸ have found that the presence of bile in the esophagus is correlated principally with the existence of an acid pH (or at least of a pH of less than 7). The validity of Bilitec 2000 for evaluating DGER has also been questioned, ¹⁹ because in acid pH the bile acids precipitate and the Bilitec underestimates the measurement of bilirubin in the esophagus by 30%; when bilirubin precipitates, its wavelength changes from 453 to 400 nm and it cannot be detected by the system. In fact, this is what happens in the stomach: in vitro studies have shown that for bilirubin to precipitate completely, it requires 30 minutes at pH 1.5 or 45 minutes at pH 2.5. However, alkalinization of the sample when duodenogastric material refluxes into the esophagus, reaching a pH of 4 to 7, restores the spectrophotometric characteristics suitable for detecting bilirubin. ^20,21 This is in line with our results: as mentioned in the previous section, esophageal pH ranged from 4 to 7 for most of the recording time.

What is not fully clear is whether bile alone can induce the development of CLE. Whereas some authors report the appearance of CLE in patients undergoing total gastrectomy, ²² others ²³ have found that the incidence of CLE in gastric surgery patients with proven DGER and reduced or absent gastric acid secretion is not significantly different compared with patients with GER and an intact stomach. In other words they, like others, ^14,18 believe that bile needs the concomitant action of acid to bring about the appearance of a metaplastic change.

Like other authors, ^14,15,18 we found that the patients with CLE had a significantly higher rate of DGER than the rest of the study groups. However, unlike Vaezi and Richter, ¹⁵ we did not observe differences between patients with complicated and uncomplicated CLE. This may be due to the small sample size, although some of our patients with complicated CLE did not have bilirubin detected in the material refluxed into the esophagus, which differs from the findings reported by Vaezi and Richter. ¹⁵

The higher rate of DGER encountered in the CLE patients with specialized epithelium seems to confirm the attractive theory that major acid reflux would lead to the development of cardial- or fundal-type epithelium as a measure to protect the mucosa, whereas refluxed material with duodenal contents would give rise to a more bile-resistant epithelium such as specialized columnar epithelium. Likewise, the rate of bile reflux to the esophagus proved to be greater in patients with mild dysplasia, as reported previously by Fein et al. ²⁴ This suggests it is probably involved in the evolution of esophageal lesions, and even in the development of adenocarcinoma. This is in line with the report of Stein et al, ²⁵ who found that patients with Barrett’s and adenocarcinoma had greater biliopancreatic reflux into the esophagus than patients with nonmalignant CLE.

Considering the degenerative potential of intestinal epithelium, it might be asked whether medical treatment with proton pump inhibitors is efficient in these patients: these drugs would not control biliopancreatic reflux, ^8,14 and the alkalinization caused by these drugs would allow the activation of certain components of the biliopancreatic juice, such as trypsin, which according to some authors ²⁶ may be far more harmful than bile. Moreover, with a neutral pH, the bile salts undergo a deconjugation process and become more damaging to the esophageal mucosa.

In conclusion, DGER is increased in patients with CLE, especially those with intestinal-type epithelium, which points to the role of bile and pancreatic juice in the genesis of Barrett’s esophagus and its possible degeneration into adenocarcinoma. We believe that the existence of DGER is an argument in favor of antireflux surgery ^27,28 as the treatment of choice in such patients, especially if they are young and do not represent a major surgical risk.

Acknowledgments

The authors thank Ms. Dolores Navarro for her extraordinary contribution.