The potential impact of contemporary developments in the management of patients with gastroesophageal reflux disease undergoing an initial gastroscopy

1Division of Gastroenterology; 2Department of Pathology, McGill University Health Centre, McGill University, Montreal, Quebec Correspondence: Dr Alan N Barkun, McGill University Health Centre, Division of Gastroenterology and Clinical Epidemiology, Montreal General Hospital Site, Room D7-148, 1650 Cedar Avenue, Montreal, Quebec H3G 1A4. Telephone 514-934-8233, fax 514-934-8375, e-mail alan.barkun@muhc.mcgill.ca Received for publication March 28, 2008. Accepted November 4, 2008 R developments have the potential to alter the current approach to patients with suspected gastroesophageal reflux disease (GERD). First, there is an increased recognition of atypical presenting symptoms. Second, a controversial revision of the definition of Barrett’s esophagus was recently proposed (1). Finally, recent data suggest that eosinophilic esophagitis (EE) may be more prevalent than previously recognized among patients with GERD-like symptoms (2). The recent consensus paper by Vakil et al (1) proposed the Montreal definition and classification of GERD. This population of patients with GERD is increasingly being recognized as having a variety of symptoms apart from heartburn and regurgitation, including dysphagia, atypical chest pain and hoarseness, which are all now more frequently investigated as part of suspected GERD. Furthermore, the recommendations from this International Consensus Group suggested a significant change original arTicle

to the histological definition of Barrett's esophagus.In brief, the participating experts proposed that the definition of Barrett's esophagus be expanded to include all types of columnar metaplasia noted on histology, and that the presence or absence of intestinal metaplasia (IM) be specified (1).This new 'Montreal definition' of Barrett's esophagus, however, was the most contentious topic among the participants at the meeting and its application remains a controversial issue.
In addition, it has become increasingly recognized that EE may present with GERD-like symptoms (2).In spite of this, it is unclear how frequently EE goes unrecognized in patients with GERD-like symptoms, especially in adults.Appropriate targeted treatment for EE, such as the use of systemic or topical glucocorticoids, leads to symptomatic and histological improvement, while patients incorrectly diagnosed with another condition such as GERD may continue to experience EE-related symptoms with ongoing evidence of esophageal injury (2).
These developments thus bear the potential to alter our current approach to patients presenting with GERD-like symptoms.The purpose of the present study was to assess the impact of such considerations in a general multiphysician, hospital-based, gastroenterology practice, by reviewing a cohort of patients referred for GERD-like symptoms undergoing an initial endoscopy.In particular, the present study aimed to determine the impact of the Montreal definition of Barrett's esophagus and the possible underdiagnosis of EE in a typical patient population managed by gastroenterologists.

Patient population
An audit of all patients undergoing a gastroscopy for a recorded indication of GERD-like symptoms (3), over a one-year period ending November 2006, was undertaken.Patients were excluded if they had undergone a previous gastroscopy for GERD.All medical charts and endoscopic reports were retrospectively reviewed using a standardized data abstraction form by one research associate, while independent prospective reviews of all available endoscopic images (grading of esophagitis and endoscopically suspected endoscopic metaplasia [ESEM]) and esophageal biopsies were performed by a single digestive endoscopist (A Barkun) and two experienced pathologists (Y Kushner and V Marcus).The study adhered to the ethical standards of McGill University, Montreal, Quebec.

historical information
Patients were identified using an institutional electronic endoscopy database, and related charts were reviewed retrospectively.The clinical information included demographic information and the presence of symptoms (including heartburn, acid regurgitation, atypical chest pain, dysphagia, nocturnal symptoms, sleep disturbance, cough, asthma, hoarseness and other complaints).Ongoing intake of medications (proton pump inhibitors, H 2 receptor antagonists, antacids, as well as acetylsalicylic acid and nonsteroidal anti-inflammatory medications) and patient habits (alcohol use and smoking) were also noted.

endoscopic findings
Endoscopic findings were categorized as nonerosive or erosive esophagitis.The presence or absence of ESEM was noted using established criteria (1).Endoscopic esophagitis was independently classified prospectively by an expert endoscopist blinded to all other data, according to the Los Angeles (LA) classification, using available photodocumentation (4).ESEM was similarly independently and prospectively classified as long segment, short segment or ultra-short segment using published definitions (5).The presence or absence of strictures was also noted, as were other findings (eg, hiatal hernia, Schatzki's ring, etc).

histological analysis
A review of every esophageal biopsy specimen was independently undertaken by two digestive pathologists blinded to all other data.The biopsies had been fixed in 10% buffered formalin, processed routinely and stained with hematoxylin and eosin.All fragments from each biopsy site were evaluated.Two slides from each biopsy site were prepared.The type of mucosa was classified as pure squamous, pure glandular (cardiac, oxyntocardiac or oxyntic mucosa), or admixed squamous and glandular mucosa, as previously described (6).The evaluation of the glandular component also included the assessment of the presence or absence of IM, as characterized by the presence of goblet cells.The biopsy fulfilled the Montreal definition of Barrett's esophagus if it contained any cardiac or oxyntocardiac mucosa with or without IM (1,6).All slides were also scanned to identify the area with the highest density of eosinophils.This area was used to count the number of eosinophils per high-power field (hpf; Olympus BX40 microscope [Olympus, Japan]; UPlanFL 40× lens, ocular magnification 10×; area of microscopic field 0.34 mm 2 ).The eosinophil count included intact eosinophils and degranulated eosinophils if the cluster of granules was discretely localized and interpretable as deriving from a single eosinophil.The presence of eosinophil microabscesses (defined as aggregates of four or more contiguous eosinophils) was noted.EE was defined as greater than 20 eosinophils per hpf (7,8).To rule out the presence of possible fungal organisms, Grocott's silver and periodic acid-Schiff with diastase stains were also performed, when indicated.

Statistical analysis
Categorical variables are reported as proportions with the appropriate numerators and denominators, and 95% CIs using the normal approximation of the binomial distribution.Continuous variables are reported as means and SDs.

Pathology results
The esophageal biopsies of 65 patients (18%; 95% CI 14% to 23%) were available for evaluation, including 34 of 41 patients (83%; 95% CI 68% to 93%) with ESEM.Of these, 63 (97%; 95% CI 89% to 100%) had biopsies of the squamocolumnar junction, while six patients (1.7%; 95% CI 0.6% to 3.7%) also had biopsies of the esophagus proximal to the squamocolumnar junction.Two patients (0.57%; 95% CI 0.069% to 2%) had biopsies of the mid esophagus only.EE was suspected in three cases, but a biopsy was taken in only one of these cases.Between one and 11 fragments (mean 3±2.1 fragments) were evaluated from each biopsy site.Two slides were available from each biopsy site, with each slide displaying two to four levels.Pathological findings are shown in Table 3.

diSCuSSioN
Of the 353 patients who presented for gastroscopy during the study period, the predominant symptoms, as expected, were heartburn and acid regurgitation, but patients also presented with nocturnal symptoms (11.3%), dysphagia (9.4%), hoarseness (5.5%) or atypical chest pain (4.2%).Endoscopically diagnosed esophagitis, identified in 26% of patients, may be under-represented considering 56% of patients were already taking a PPI at the time of endoscopy.ESEM was present in 12% of patients who underwent endoscopies.In addition, EE was suspected at endoscopy in three cases, but a biopsy was taken in only one of these patients.An independent, blinded histological review of all biopsy material in 65 patients yielded 10 cases of Barrett's esophagus, using a traditional definition.
Adopting the Montreal definition of Barrett's esophagus proposed by Vakil et al (1), the number of Barrett's cases increased to 49.In addition, five cases of EE were diagnosed histologically.
The strengths of the present study include the broad generalizability of the study population referred for an initial gastroscopy for GERD-like symptoms to a general gastroenterology practice, and the independent reviews of endoscopic images and histology slides.While the present study is useful in answering the clinical aims we set out to address, there are important limitations.First, as a retrospective study, the clinical information for all of the categories studied was not readily available for every patient and depended on what was recorded as part of routine practice.This limitation is especially true for the presence of specific symptoms that may be influenced by detection or recall biases.Second, while the endoscopic images were prospectively reviewed, pictures were absent in a minority of cases.Finally, only 65 patients had esophageal biopsies.
Dysphagia emerged as a frequent presenting symptom of GERD in the present study.Of the 29 patients who complained of dysphagia, this symptom was explained by endoscopic findings in five cases, including four with Schatzki rings and one with a stricture.In addition, 11 of the 29 patients (37.9%) with dysphagia had reflux esophagitis, a figure comparable with previously reported data (11).
The new definition proposed by Vakil et al ( 1) is controversial, and clearly has the potential to dramatically affect the number of patients diagnosed with Barrett's esophagus.Whereas only 10 of 353 patients (2.8%) referred for evaluation met the classic diagnostic criteria for Barrett's esophagus, using the Montreal consensus definition would increase that number to 49 (14%), a fivefold increase.Among patients biopsied, 15% were diagnosed with Barrett's esophagus using traditional criteria, in contrast to 75% of patients if the new criteria were applied.We cannot be certain about the 'true' incidence because not all patients were biopsied; however, biopsies were collected in 34 of 41 patients exhibiting ESEM.In these patients, the proportion of Barrett's diagnoses would rise from 24% to 97% with adoption of the Montreal histological definition.Whether this represents a definition that has greater sensitivity for detecting patients at increased risk of esophageal adenocarcinoma is unclear.Previous research has indicated that esophageal columnar metaplasia with specialized IM is most strongly associated with adenocarcinoma (12).In contrast, the association between esophageal columnar metaplasia (cardiac or oxyntocardiac mucosa) without IM and adenocarcinoma has been more controversial (13,14).The inadequacy of biopsy protocols currently being practiced worldwide results in a poor sensitivity at detecting IM within an ESEM (1,12,15), which is one of the reasons for the proposed change in definition.Additionally, although the risk for adenocarcinoma is indeed highest with columnar metaplasia associated with IM, other forms of columnar metaplasia may also carry an associated risk (16).In fact, studies on patients with IM have shown that up to 35% of biopsies fail to demonstrate IM (12,17), and that at least eight biopsies may be needed for a reasonable diagnostic yield (12).The clinical impact of increased Barrett's diagnostic rates is significant because both long-term daily PPI therapy and inclusion in an endoscopic surveillance program are currently recommended for patients with a diagnosis of Barrett's esophagus (18)(19)(20).
EE was observed in five patients (1.4% of all patients).Among patients biopsied, 7.7% of patients met the criteria for EE.These figures may represent underestimates because this diagnosis was not suspected endoscopically in four of the five patients.In addition, only one of the five patients was diagnosed with EE on histology before our audit.Three of the five cases of EE were convincing, with eosinophil counts of greater than 100/hpf, while the remaining two had borderline eosinophil counts of 21/hpf and 26/hpf, respectively.The histological underdiagnosis may be partly explained by reports suggesting that variable numbers of eosinophils may also be noted in GERD (21).Currently, it is unclear whether the plethora of recent adult EE reports are related to an increase in diagnosis or an increase in the incidence of this condition (2,22).Presently, the most useful predictors appear to be clinical and not endoscopic, including male sex and a history of atopy (2).Despite the small sample size, our results show that instances of EE may have no distinguishing clinical or endoscopic features, suggesting that perhaps routine biopsies should be considered in all patients presenting with esophageal complaints (2).In addition, there appears to be a need for a heightened awareness of EE among clinicians, endoscopists and pathologists.A better correlation between the clinical and/or endoscopic suspicions and pathological findings through closer communication between clinicians and pathologists may help improve detection rates.This is especially relevant because specifically targeted therapy for EE may be required for timely and complete symptom resolution.With all of the above information, we attempted to quantify, in an exploratory fashion, the possible impact of these issues on clinical management, by extrapolating from the documented formal management recommendations recorded in the chart based on the initial consult and endoscopy report.Patients who had had a change in medication or its dose, those diagnosed with EE or those patients who were subsequently diagnosed with Barrett's esophagus, and thus needing subsequent surveillance (irrelevant of any change in medication that may have also occurred), were considered as having undergone a change in management compared with the preassessment status.This classification was independently reviewed by an experienced clinician with particular expertise in esophageal diseases (A Barkun).The clinical management of GERD patients was altered in 95 patients (27%; 95% CI 22% to 32%).Eighty-four of the 95 patients had a change of medications, including starting a PPI in 65 patients or increasing the PPI dose in 19 patients.Ten of the 95 patients were diagnosed with Barrett's esophagus and entered into an endoscopic surveillance program.One was diagnosed as having EE.Using the Montreal consensus definition of Barrett's esophagus, above and beyond these 95 patients, an additional 39 patients with columnar metaplasia but without IM would be considered for surveillance.Finally, an additional four patients were diagnosed with EE after the independent review was initiated as part of the present audit.The breakdown of patients is shown in Figure 2; up to 43 additional patients (12%; 95% CI 8.6% to 16%) would have undergone a change from current management.
The newly proposed Montreal consensus definition of Barrett's esophagus is controversial and, if implemented, would greatly increase the number of patients diagnosed with this condition.In addition, EE is a diagnosis that is probably much more prevalent than currently recognized, and may be missed in patients presenting with GERD-like symptoms.Both of these issues thus have the potential to significantly alter clinical practice.Prospective studies need to be undertaken to better define the implications of these findings.
diSCloSureS: Dr Barkun is a research scholar funded by the Fonds de la Recherche en Santé du Québec.Dr AN Barkun is a consultant for AstraZeneca Inc. Dr Carlo Fallone is a consultant for or has received research support from AstraZeneca Canada Inc, Janssen-Ortho Canada, Nycomed Canada, Negma and Abbott Laboratories Canada Inc. Dr Serge Mayrand is a consultant for AstraZeneca Canada Inc, Janssen-Ortho Canada, Nycomed Canada and Axcan Pharma Inc. Dr Yael Kushner, Dr Victoria Marcus and Dr Suhail Salem have no conflict of interest disclosures or financial affiliations to declare.This study was undertaken with arm's-length support from Abbott Laboratories Canada Inc.

+
Figure 2) Exploratory analysis of the potential impact on routine management of patients with gastroesophageal reflux disease (GERD)-like symptoms undergoing a gastroscopy

Figure 1 )
Figure 1) Proportions of patients with Barrett's esophagus according to the traditional and proposed Montreal consensus definitions