Predictors of dysplastic and neoplastic progression for Barrett’s esophagus Saleh Alnasser Supervisors : Dr.Lorenzo Ferri , Dr.Eduardo Franco Experimental surgery McGill University, Montreal Quebec, Canada June 2012 A thesis submitted to McGill University in partial fulfillment of the requirements of the degree of master in experimental surgery. Copyright © Saleh Alnasser MD, 2012.
1. Table of Contents 1. Table of Contents ................................................................................................... 2 1.1 List of Tables: ................................................................................................................... 4 1.2 List of Figures: .................................................................................................................. 5
2. Abbreviations: ........................................................................................................ 6 3. Abstract .................................................................................................................. 7 3.1 Abstract (English) ............................................................................................... 7 3.2 Abstract (French) ................................................................................................ 9 4. Acknowledgements .............................................................................................. 11 5. Introduction: ........................................................................................................ 12 5.1. History: .......................................................................................................................... 12 5.2. BE Definition: ................................................................................................................ 14 5.3. Epidemiology: ............................................................................................................ 16 5.4. Pathogenesis: ................................................................................................................ 17 5.5. Risk Factors: ............................................................................................................. 18 5.5.1. GERD: .......................................................................................................... 18 5.5.2. Demographic factors: ................................................................................ 20 5.5.3. Obesity: ..................................................................................................... 20 5.5.4. Diet: ........................................................................................................... 20 5.5.5. Family history: ........................................................................................... 21 5.5.6. Helicobacter pylori (H. pylori): .................................................................. 21 5.5.7. NSAID and Aspirin: ..................................................................................... 21 5.5.8. Smoking and alcohol: ................................................................................ 22 5.6. Neoplastic and dysplastic progression: ....................................................................... 22 5.7. Surveillance: ............................................................................................................ 23 5.8. Management: ........................................................................................................... 27 5.8.1. Surgery: ............................................................................................. 29 5.8.2. Endoscopic intervention: ............................................................... 30
6. Rationale: .......................................................................................................... 35 7. Review of literature: .......................................................................................... 36 8. Objective: .......................................................................................................... 38 9. Hypothesis ......................................................................................................... 38 10. Methods ............................................................................................................ 40
2
10.1 Study Population ....................................................................................................... 40 10.2 Inclusion Criteria: ....................................................................................................... 40 10.3 Exclusion Criteria: ...................................................................................................... 41 10.4 Study Procedure: ....................................................................................................... 41 10.5 Ethics: ........................................................................................................................ 47 10.6 Statistical analysis: ..................................................................................................... 48
11. Results: .............................................................................................................. 50 11.1 Progression from NDB to any dysplastic or neoplastic grades: ................................... 56 11.2 Progression from NDB to HGD/EAC: ........................................................................... 64
12. Discussion: ......................................................................................................... 71 13. Conclusion: ......................................................................................................... 78 14. Appendices: ...………………………………………………………..………………….…….……………….79 15. References: ........................................................................................................ 81
3
1.1 List of Tables:
Table 1, Risk factors of Barrett's esophagus development .................................................... 19 Table 2, Summary of the surveillance guidelines for BE patients. ...................................... 25 Table 3: Summary of the relevant clinical studies and their outcome .............................. 39 Table 4, Summary of the total and missing endoscopy reports. .......................................... 46 Table 5: Total patients according to their baseline, eligibility and progression status ....................................................................................................................................................................... 52 Table 6, Demographic characteristics for all eligible patients. ............................................ 53 Table 7, Baseline characteristics of patients with BE based on their baseline status. 55 Table 8, Follow up (in months) median and mean for the patients with NDB baseline ....................................................................................................................................................................... 56 Table 9, Dysplastic and neoplastic progression rate in patients with NDB baseline according to the demographic and endoscopic variables ...................................................... 58 Table 10, Analysis of the potential risk factors predicting progression from NDB to dysplastic or neoplastic grades (Cox regression) ....................................................................... 62 Table 12, Analysis of potential risk factors for predicting progression From NDB to dysplastic or neoplastic grades (GEE) ............................................................................................. 63 Table 13, HGD/EAC progression rate in patients with NDB baseline according to the demographic and endoscopic variables ........................................................................................ 65 Table 14, Analysis of potential risk factors for predicting progression from NDB to HGD/EAC (Cox regression) ................................................................................................................. 69 Table 15, Analysis of the potential factors of predicting progression from NDB to HGD/EAC (Cox regression) .................................................................................................................. 70 Table 16, Analysis of the potential factors predicting the progression form NDB to HGD/EAC (GEE) ........................................................................................................................................ 70
4
1.2 List of Figures: Figure 1, Neoplastic progression of Barrett's esophagus ............................................................... 16 Figure 2, Examples of patient treatment timelines to elaborate the presentation of BE grade at the baseline. ................................................................................................................................... 42 Figure 3, Candidate Independent variables for this study. ............................................................ 44 Figure 4, Flow chart for the total number of patients and the process of inclusion ……….51 Figure 5, Box plot shows the median age, interquartile range for all 3 groups. ................... 53 Figure 6, Cumulative incidence of progression to dysplasia or cancer by gender……………59 Figure 7, Cumulative incidence of dysplastic or neoplastic progression by age. ................... 59 Figure 8, Cumulative incidence of dysplastic or neoplastic grades by BE length. ................. 60 Figure 9, Cumulative incidence of HGD/EAC by age. ........................................................................ 66 Figure 10, cumulative incidence of HGD/EAC by gender. ............................................................... 66 Figure 11, Cumulative incidence of HGD/EAC by presence of mucosal irregularities. ........ 67 Figure 12, Cumulative incidence of HGD/EAC by presence of ulcer. .......................................... 67
5
2. Abbreviations: ARS Acid reducing surgery PDT Photodynamic therapy APC Argon plasma coagulation PPIs Proton pump inhibitors BE Barrett’s esophagus PR Progression rate CBE Complete Barrett’s excision RFA Radiofrequency ablation CI Confidence interval RCT Randomized control trial COX2 Cyclooxygenase 2 SIM Specialized intestinal epithelium EAC Esophageal adenocarcinoma SSBE Short segment Barrett’s esophagus EMR Endoscopic mucosal resection U.K. United kingdom GEE Generalized estimation equation GERD Gastroesophageal reflux disease IND Indefinite for dysplasia HGD High grade dysplasia HR Hazard ratio HH Hiatus hernia H.Pylori Helicobacter pylori LGD Low grade dysplasia LSBE Long segment Barrett’s esophagus MUHC McGill University Health Center NDBE Non-‐dysplastic Barrett’s esophagus NSAID Non steroidal antinflammatory drug OR Odds ratio
6
3. Abstract
3.1 Abstract (English) 3.1.1 Introduction: Barrett’s esophagus (BE) is a premalignant condition that may progress through a stepwise process to esophageal adenocarcinoma (EAC). It is unknown why some BE patients progress to EAC rapidly while others do so more slowly, or not at all. The aim of this study is to identify the demographic and endoscopic factors that can be used as predictors of dysplastic and neoplastic progression in BE patients, and thereby be used to stratify BE patients into different surveillance protocols according to the risk of neoplastic progression. 3.1.2 Method: All BE patients at McGill University Health Center between January 2000 and December 2010 were identified from a pathology database. 1054 patients with intestinal metaplasia on esophageal biopsies were identified. Only those with endoscopic findings of columnar lined mucosa were included in this study, (N=518). Data for demographic variables (e.g., age and gender) and endoscopic variables (presence of hiatus hernia, esophagitis, ulcers, mucosal irregularities and strictures) were collected. Neoplastic and dysplastic progression was examined by time to event analysis. Cox proportional hazard regression modeling and Generalized Estimating Equations (GEE) methods were used to identify the variables that are most predictive of neoplastic progression.
7
3.1.3 Result: From a non-‐dysplastic BE (NDB) at baseline state, 35(6.7%) patients progressed to dysplastic or neoplastic grades and 10(1.9%) patients progressed to high-‐grade dysplasia (HGD) or EAC. BE length (HR=1.2, 95%CI=1.1-‐1.3) and advanced age (HR=3.5, 95%CI=1.7-‐ 7.4) were independent predictors of progression from NDB to dysplastic or neoplastic grades. However, mucosal irregularities (HR=8.6, 95%CI=2.4-‐30.4) and advanced age (HR=5.1, 95%CI=1.7-‐16.6) were the independent predictors of progression from NDB to HGD/EAC. 3.1.4 Conclusion: Advanced age, a long BE segment and presence of mucosal irregularities are associated with increased risk of dysplastic and neoplastic progression in BE patients. In addition to the presence of dysplasia, these factors may help to stratify BE patients according to their risk of neoplastic progression and therefore can be used to individualize BE surveillance.
8
3.2 Abstract (French) 3.2.1 Introduction: L'oesophage de Barrett (OB) est un état pré-‐cancéreux qui peut mener à l'adénocarcinome invasif (ACI) de l'oesophage. Les causes de la progression chez certains patients ayant un OB à l'ACI demeurent inconnues. Le but de cette étude est d'identifier les facteurs démographiques et endoscopiques comme prédicteurs de la progression dysplasique et néoplasique chez le patient atteint de l'OB et ainsi permettre la stratification de ceux-‐ci dans différents protocoles de surveillance selon le risque de la progression néoplasique. 3.2.2 Méthodes Tous les patients atteints de l'OB au centre hospitalier du centre universitaire de santé McGill entre janvier 2000 et décembre 2010 ont été revus à partir d'une base de données en pathologie. Les patients présentant une métaplasie intestinale dans les biopsies oesophagiennes ont été identifiés. Seuls ceux ayant un épithélium métaplasique cynlindrique confirmé à la pathologie ont été inclus. Les données suivantes ont été enregistrées : les variables démographiques (l'âge et le sexe) ainsi que les variables endoscopiques (présence de hernie hiatus, esophagites, ulcères, irrégularités au niveau de l'épithélium et sténoses). La progression dysplasique et néoplasique a été examinée par une analyse au temps de l'inclusion à l'évènement. La régression de Cox et la méthode GEE (Generalized Estimating Equations) ont été utilisées pour identifier les variables prédisant la progression néoplasique.
9
3.2.3 Résultats: À partir de 1054 patients atteints de l'OB dont 518 présentant une métaplasie intestinale à la biopsie et ne présentant aucune dysplasie (OBND) à l'entrée de l'étude, 35 (6.7%) d'entre eux ont progressé à un grade de dysplasie ou de néoplasie et 10 (1.9%) d'entre eux ont progressé à une dysplasie de haut grade (DHG) ou à un ACI. Un OB long (HR=1.2, 95%IC=1.1-‐1.3) et l'âge (HR=3.5, 95%IC=1.7-‐7.4) sont des prédicteurs de la progression d'un OBND à un grade de dysplasie. Cependant, une irrégularité de l'épithélium (HR=8.6, 95%CI=2.4-‐30.4) ainsi que l'âge (HR=5.1, 95%CI=1.7-‐16.6) sont des prédicteurs de la progression d'un état OBND à une DHG ou à l'ACI. 3.2.4 Conclusion L'âge, un long segment de OB et la présence d'irrégularités au niveau de l'épithélium ont démontré un risque accru de dysplasie et de progression néoplasique chez les patients atteints de l'OB. En plus de la présence de dysplasies, ces facteurs peuvent aider à stratifier les patients selon leur risque de progression néoplasique et peuvent donc être utilisés pour individualiser la surveillance de l'OB.
10
4. Acknowledgements First of all, it is a pleasure to thank those who made this thesis possible. I am heartily thankful to my supervisor, Dr. Lorenzo Ferri, whose encouragement, guidance and support from the initial to the final level enabled me to develop an understanding of the subject. I also gratefully acknowledge Dr. Eduardo Franco for his advice, supervision, and crucial contribution, which made him a backbone of this research and so to this thesis. My special thanks go to Dr. Agnihotram Ramana-‐kumar for his extra care on the statistical analysis of data and organization of the results. I also want to thank Ms. Myriam Martel for her great work on the data entry, translation and her work to get the IRB approval. To Dr. Serge Mayrand, thank you for your great help in assessment of objectives. Where would I be without my family? My mom and dad deserve a special mention for their inseparable support and prayers. Words fail me to express my appreciation to my wife Lujain whose dedication, love and persistent confidence in me, has taken the load off my shoulders. I owe her for unselfishly letting her intelligence, passions, and ambitions collide with mine, and for her editorial work on the thesis. Finally, many thanks to the heartland of science, McGill University, which paved the way for acquiring knowledge and removes the obstacles so that students can make valuable contributions to the community and refine their skills.
11
5. Introduction: Barrett’s esophagus (BE) is a premalignant condition that may lead to the development of esophageal adenocarcinoma (EAC). EAC develops in a multistep process from non-‐dysplastic BE (NDB) to low-‐grade dysplasia (LGD), then to high-‐grade dysplasia (HGD) that eventually may progresses to invasive adenocarcinoma. The incidence of this malignancy has increased dramatically during the past 3 decades, with a poor prognosis overall and a high mortality. Because of this malignant potential in BE, periodic surveillance via endoscopic biopsies is recommended to detect the cancer at an early stage, suitable for curable treatment. In the last few years, a revolution in the BE field has led to the development of new detection and ablative endoscopic techniques in order to diagnose and treat this condition and reduce the mortality rate from EAC. 5.1. History: More than 200 years ago, Schmidt described the presence of ectopic columnar mucosa in the distal esophagus
(1)
. However, the current concept of the columnar
epithelium goes back to 1906, when the pathologist Wilder Tileston noticed several patients with esophageal ulcers in which the surrounding mucous membrane resembled that in the stomach (2). There were many other early reports describing the presence of columnar epithelium in the esophagus, although few reports tried to establish its pathological significance (3). Esophageal columnar epithelium got more attention in 1950, when Norman Barrett published his first report about the esophageal columnar epithelium (4). He
12
hypothesized that the presence of the columnar epithelium in the distal esophagus was due to congenital shortening of the esophagus with upward extension of the stomach. In 1953, Allison and Johnstone disagreed with Barrett’s contention about the congenital esophageal shortening. They described a case series of reflux esophagitis and concomitant “distal esophagus lined with gastric mucosa”(5). They noted that the columnar-‐lined organ lacked a peritoneal covering, contained few islands of squamous epithelium and had submucosal glands and a muscularis propria. Initially, Barrett argued against this concept but in 1957, he eventually accepted Allison and Johnstone’s arguments, and suggested that the condition be called “lower esophagus lined by columnar epithelium”(6). For the next 2 decades, the histological features of the esophageal columnar epithelium varied and remained a controversial issue. In 1976, Paull et al described the histologic spectrum of BE (7). This spectrum includes one or a combination of 3 types of columnar epithelium—a gastric fundic-‐type, a junctional type and a distinctive type of intestinal metaplasia that the investigators called "specialized intestinal metaplasia" (SIM). Fundic and junctional epithelial types are very similar to the columnar epithelia normally found in the stomach. However, SIM with its prominent goblet cells can be distinguished from normal gastric mucosa. By the 1970’s, the association between gastroesophageal reflux disease (GERD) and BE (8,9) as well as that between EAC and BE were established (10-‐12). By the late 1980s, it was clear that SIM was the columnar epithelial subtype that had the greatest potential to develop dysplasia and EAC (13,14).
13
5.2. BE Definition: The American Gastroenterology Association defines BE as a change in the distal esophageal epithelium of any length that can be recognized as columnar-‐type mucosa with endoscopy and confirmed to have intestinal metaplasia by biopsy (15). The concept behind this definition lies in that a majority of EAC developed in a BE mucosa exhibiting intestinal metaplasia (11,16-‐19). This definition has also been proposed by other medical societies including the French Society of Digestive Endoscopy and the German Society of Pathology in which the histological identification of intestinal metaplasia is mandatory for a diagnosis of BE in addition to endoscopic evidence of columnar lined mucosa (20). The diagnosis of BE by this definition is achieved in patients who have endoscopic and histological Barrett’s findings. However, histologic diagnosis can sometimes be difficult when the biopsies are taken from the gastroesophageal junction. In this situation, the differential diagnosis between short-‐segment BE and intestinal metaplasia of the cardia has to be considered. This differentiation has serious implications because intestinal metaplasia in BE is a precancerous condition and requires regular endoscopic follow-‐up examinations, whereas no follow-‐up is indicated in intestinal metaplasia of the cardia, whose malignant potential is considered to be low (21-‐ 22)
. To avoid overdiagnosis of BE, its presence is considered to be established only if
endoscopic examination reveals a displacement of the squamocolumnar junction that is proximal to the gastroesophageal junction and if SIM is detected by histological examination (22-‐23). On the other hand, the British society of Gastroenterology (24) and the Japanese
14
Esophageal Society (25) do not consider SIM as a prerequisite in their definition of BE. They consider it as less important for a diagnosis of BE than the presence of a proper esophageal gland, squamous island, and/or double muscularis mucosae
(26-‐27)
. Their
rationale behind excluding SIM as criterion is based on the need for extensive biopsies for detection or exclusion of the condition plus the high rate of sampling errors. By using this definition, there will be an increase in the incidence of BE, which in turn will mandate increased surveillance of these patients to monitor cancer occurrence. However, recent data (28) suggest that cardia-‐type epithelium in the esophagus may predispose also to malignancy development but the magnitude of this risk remains unclear. Cognizant of the fact that some cases of esophageal cancer arise from non-‐ intestinalized columnar epithelium, Kelty et al (29), tried to identify the risk of EAC in patients who had a columnar epithelium with and without SIM. They found that there was no statistical difference in EAC risk between the two groups. Another review of 141 cases by Takubo et al
(30)
, demonstrated that more than 70% of primary esophageal
adenocarcinomas were adjacent to cardiac and/or fundic areas rather than intestinal type mucosa with goblet cells. However, the validity and accuracy of these observations need to be proven by larger prospective studies. Because there are insufficient data to make recommendations regarding management of patients who have solely cardia-‐type epithelium in the esophagus, the American Gastroenterology Association do not recommend use of the term “Barrett's esophagus” for those patients and not to be included in the endoscopic surveillance.
15
Figure 1, Neoplastic progression of Barrett's esophagus. A. Salmon colored tongues represent Barrett's epithelium in endoscopy. B. Progression of Barrett’s epithelium to EAC for the same patient 5 years later.
5.3. Epidemiology: Esophageal cancer is the eighth most common cancer and the sixth leading cause of cancer death worldwide (31-‐32). Its overall incidence rates vary greatly throughout the world. In North America, the epidemiology of esophageal cancer has changed dramatically over the past 3 decades. In the 1970s, most esophageal cancers were of squamous cell type. Since then, there has been a dramatic increase in the incidence of EAC (33) making it the fastest rising incidence of a solid cancer in the Western world (34). However, most cases of EAC arise on a background of BE. The reported prevalence of BE varies considerably. There are numerous studies that analyzed the prevalence of BE demonstrating a wide range of results. In a general population, the prevalence of BE is estimated to be between 1.6 and 6.8% (35). Many studies from different populations around the world have shown an increase in the diagnosis of BE during recent decades (36-‐
16
38)
. This may be partially due to more access to esophageal endoscopy and biopsy, but this
does not appear to account for the generalized increase in this condition. BE is a known premalignant condition associated with the greatest risk of developing EAC (39). This risk of EAC is about 30 to 125 times higher for patients with BE than for those without BE. EAC has been estimated to develop in about 0.5%–0.6% of patients with BE annually (40-‐45). The risk of EAC development will be significantly higher for those who have dysplasia. However, esophageal adenocarcinoma is an uncommon cause of death in people with BE (46-‐49) as most patients with BE die from causes other than EAC. 5.4. Pathogenesis: In general, the pathogenesis of BE is poorly understood. It is well known that BE develops through a process of metaplastic change from squamous to columnar epithelium. BE development requires an initial esophageal mucosal injury and the presence of a pathologic environment, allowing for abnormal healing (23), which makes gastroesophageal reflux the strongest risk factor for BE development. This metaplastic columnar epithelium occurs in response to oxidative damage and inflammation of the distal esophageal epithelial mucosa from contact not only with the acidic gastric contents but also with bile. The acids and pepsin enzymes from the stomach weaken cell junctions and widen the intercellular gaps that will allow the acid to access the nerve endings. Because of the low pH environment, bile will be deposited and this will lead to cell disruption and damage (50). As a result of this damage, one theory suggests that the intestinal metaplasia develops as part of the protective mechanism against chronic acid reflux (51). Another theory suggests abnormal differentiation of pluripotential cells in the
17
basal esophageal epithelium (52). This abnormal differentiation results from exposure of these cells to the refluxed gastric juice after damage to the squamous epithelium (53-‐55). Recent studies suggest that, these pluripotential stem cells may contribute to esophageal lesion regeneration and the initiation of metaplasia (56). 5.5. Risk Factors: Different risk factors are associated with BE, some of which are very strong like GERD while others are controversial and require more clinical studies to confirm their association. Here will mention the most important risk factors. 5.5.1. GERD: GERD is considered the strongest and the best understood risk factor for BE development (15,25). Mechanisms that lead to BE development include esophageal mucosal injury from acid reflux that acts as the trigger for the development of Barrett’s esophagus (57-‐58)
. The presence of columnar metaplasia proximal to the gastric conduit in
esophagectomy specimens has confirmed the relationship between GERD and columnar metaplasia
(59)
. Clinically, BE is associated with long-‐standing, symptomatic
gastroesophageal reflux disease (GERD). Studies show that there is a direct proportional relationship between the duration, severity and frequency of GERD and the development of BE and EAC (60). For example, more frequent symptoms of GERD and reflux episodes lasting longer than 5 minutes are associated with BE (61).
18
Table 1, Risk factors of Barrett's esophagus development Risk Factor
Higher risk
Lower risk
Older age
Younger age
Gender
Male
Female
Race
White
Others
GERD
Long duration
Less duration
Nature of reflux
Bile and acid reflux
Only Acid reflux
Overweight and obese
Normal weight
Age
Weight Diet
Less vegetable and fruit
More vegetable and fruit
Smoking
Smoking
No smoking
H. pylori
Absent
Present
The nature of the reflux content also is a relevant factor in BE development as patients who have both gastric acid and bile reflux have been found to have a higher prevalence of BE than patients who have only gastric acid reflux (62). This has been confirmed in an animal model where epithelial damage will regenerate to columnar epithelium only in the presence of both bile and gastric reflux (63).
19
5.5.2. Demographic factors: The epidemiological data show that the BE exhibits a 2-‐3:1 male predominance (55) with the typical age of diagnosis in the 50–59 years age group. There is also a higher prevalence of BE in whites compared to blacks, Asians or Hispanics (56). 5.5.3. Obesity: Obesity is an independent factor for BE development (64). Obese patients have frequent GERD symptoms due to an increase in an intra-‐abdominal pressure that will cause transient lower esophageal sphincter relaxation resulting in more reflux (65). Central obesity has been strongly associated with GERD and BE in several studies (66-‐68). Recent studies have emphasized the importance of central obesity pattern rather than body mass index in the evolution of the BE (67,68). High levels of several adipocytokines including leptin that are associated with central obesity have been proposed as contributors to BE pathogenesis and its progression (69-‐70). 5.5.4. Diet: Vegetable and fruit, together with vitamin C, lower the risk of BE and EAC due to their antioxidant effects (71-‐72). Ingestion of leafy and cruciferous vegetable, carbohydrates, and fiber display an inverse relationship with BE and EAC. Patients at higher risk for BE may benefit from reducing their intake of red meat and processed food items (73), although some studies showed no relationship between this and the incidence of BE (74,75). Iron may be implicated in BE pathogenesis. Experimental studies showed that the rate of BE and EAC in rats after surgical induction of reflux increased in those who
20
received iron supplements (76). However, a reduced iron level in premenopausal women with BE might have a role in delaying the onset of BE and EAC (77). Nitric oxide (NO) has been shown to play a role in the pathophysiology of BE. Nitrates in the saliva and diet will be reduced to nitrites by oral bacteria and these will become an oxidative compounds with mutagenic potential at the gastroesophageal junction in an acidic environment
(78-‐79)
. This makes NO a potential risk factor for
metaplasia development and carcinogenesis initiation (80). 5.5.5. Family history: The development of BE among first degree relatives of patients with BE, with or without GERD symptoms, has been suggested (81). Some epidemiologic data show that there is a familial contribution to BE (82,83). However, it is not known whether BE is a hereditary condition, and no gene has been implicated to date (35,84). 5.5.6. Helicobacter pylori (H. pylori): H. pylori exerts a protective role in BE by reducing acid production secondary to gastritis that creates acholrohydria. An inverse association between H. pylori infection and the presence of BE has been confirmed in many studies (85-‐86). A recent meta-‐analysis supports this association with a reported risk reduction of BE with H. pylori infection (87). 5.5.7. Non-‐steroidal anti-‐inflammatory drugs and Aspirin: Non-‐steroidal anti-‐inflammatory drugs (NSAIDs) and aspirin are considered as protective factors. BE epithelium increases the expression of cyclooxygenase 2 (COX 2)
21
and this expression increases with higher degrees of dysplasia (88). Although unconfirmed, some feel that inhibition of COX 2 by aspirin or NSAIDs can slow the progression of BE. 5.5.8. Smoking and alcohol: Some studies (89-‐90) show an increase in the risk of BE and EAC with smoking, but no relation has been proven to the quantity or duration of smoking. However, it has been suggested that there is a synergistic relationship between smoking and reflux that will increase the risk of BE development (90). The relationship between alcohol consumption and BE is not well established to date. 5.6. Neoplastic and dysplastic progression: Malignant transformation from BE to EAC is believed to occur through the histopathologic stages classified as no dysplasia, LGD and HGD (91). The accurate annual incidence of EAC, HGD or LGD in BE patients remains unclear, as studies have shown considerable variation in the incidence rates. These studies were based primarily on patients referred to tertiary centers, whose cancer risk may exceed that for patients managed by non-‐referral centers. Moreover, published data predominantly come from small retrospective cohort studies with relatively short follow-‐up, showing higher cancer incidence than may be observed in larger surveillance studies. Consequently, evidence of publication bias in these surveillance studies has been reported. The estimated annual risk of EAC development in BE patients is about 0.5%–1.0% of patients with BE (40-‐43). Meta-‐analyses estimate the incidence of EAC among individuals
22
with BE to be 6–7 per 1000 person-‐years (44-‐45). A systematic review by Skiemma et al (92) showed that the annual incidence of EAC in BE patients is about 0.63 %. 51 studies have been included in this systematic review where no evidence of publication bias was found. However, a recent population based study by Hvid-‐Jensen et al (93), showed that the annual risk of EAC in BE patients is about 0.12%. Dysplasia is the first morphologically recognizable lesion in the neoplastic progression of BE to adenocarcinoma, and it is the most common basis of risk stratification in affected patients (94). This makes the risk of EAC development in BE patients higher for those who have dysplasia. However, the estimated risk of progression to EA among patients with dysplasia varies widely among different studies (95-‐98). In patients with low-‐grade dysplasia, incidence rates for esophageal adenocarcinoma range from 0.6% to 1.6% per year (99-‐100). In contrast, the risk of the development of EAC is high among patients with HGD with an estimated incidence of 6.6% per year in a recent meta-‐ analysis (101).
5.7. Surveillance: The key to the prevention and early treatment of EAC in Barrett’s patients is the detection and eradication of cancer or HGD at an early stage. Identification of BE patients with dysplasia or early stages of cancer will reduce the mortality associated with EAC (102-‐ 104)
. If cancer is detected at an early stage, the 5-‐year survival rate is 83%-‐90%, compared
with the 10%-‐20% 5-‐year survival rate at more advanced stages. The goal of endoscopic surveillance of patients with BE is the diagnosis of HGD or
23
EAC at an earlier curable stage with a lower rate of lymph node metastasis. There is some evidence that EAC is detected at an earlier stage in patients with BE who are undergoing surveillance (105), and this improves survival rates in general (106-‐107). It is generally accepted that patients with BE should be enrolled in a surveillance program (15). Patients should also be started on proton pump inhibitors (PPIs) at initial endoscopy to treat GERD and the associated esophagitis in order to improve the endoscopic vision of Barrett’s epithelium and to reduce the inflammation that might give false-‐positive biopsy results for dysplasia. A recent guideline from the American College of Gastroenterology
(15)
suggests
surveillance endoscopic biopsies every 3 years for patients with NDB when 2 endoscopic surveillance biopsies within a year confirm absence of the dysplasia. Biopsy protocols should include 4 quadrant biopsies every 2 cm of Barrett’s mucosa. However, absence of the dysplasia in the biopsy does not completely rule out the presence of dysplasia or even cancer. The presence of LGD mandates follow-‐up endoscopic biopsies in 6 months to rule out the presence of HGD. To reduce false positive biopsy results, an expert pathologist should confirm pathologic diagnosis of LGD. If only LGD is found then endoscopic biopsies should be performed every year until no dysplasia is noted in 2 consecutive years. For patients with HGD, a secondary review with an expert in esophageal pathology is required and if so, endoscopic biopsies should be repeated within 3 months. HGD is generally considered as a threshold to either surgical or endoscopic intervention or extensive surveillance. However, these options should be tailored according to the patient’s overall condition.
24
Table 2, Summary of the surveillance guidelines for BE patients. Dysplasia No dysplasia
Intervention Surveillance endoscopic biopsies every 3 years. When 2 biopsies within a year confirm the absence of dysplasia.
LGD
Repeated endoscopic biopsies within 6 months, once confirmed Surveillance every year.During yearly follow up If two consecutive endoscopic biopsies show no dysplasia, surveillance interval can be shifted to the standard surveillance program.
HGD
Once confirmed by an expert pathologist, endoscopic biopsies need to be repeated in 3 months and should be read by a second pathologist. If confirmed, endoscopic or surgical intervention vs. extensive surveillance.
Invasive E arly stage (T1a) can be treated with Endoscopic mucosal Carcinoma
resection (EMR) (>T1a) can be treated with minimal invasive or open esophagectomy based on the size of the tumor and node status
Endoscopic surveillance of BE patients is considered as a very low-‐risk procedure. Upper endoscopies have a reported morbidity of 0.5% and mortality of 0.05% (108). However, challenges of the endoscopic surveillance include the inconsistency (109-‐110) and the reproducibility of the diagnosis of Barrett’s esophagus, with or without dysplasia (111-‐ 113)
due to sampling error, plus the intra-‐ and inter-‐observer variations. However, these
25
considerations bring into questioning the clinical aspects and cost-‐effectiveness of the surveillance program for patients with NDB and till now the relative benefits remain unproven (114-‐115). Some observational studies on endoscopic surveillance suggest its efficacy, as surveillance-‐diagnosed EAC has a much better prognosis compared those with a symptomatic presentation (15). However, some patients enrolled in the surveillance program might drop out of it (114) or die from causes unrelated to EAC (47,116). These studies are not definitive because they are highly susceptible to different types of biases that can inflate the benefits of surveillance. A randomized trial is required to establish the efficacy of surveillance, which would need large numbers of patients and a long follow-‐up time. However, the results of such a study are unlikely to be available in the near future. Another challenge to this surveillance program is the degree of adherence to these guidelines by clinicians, which has been described in the literature to be poor (117). An endoscopic surveillance practice at one institute revealed that appropriate four-‐quadrant biopsies were taken in only 23% of patients with BE and the mean interval between endoscopies was only 12 months, indicating an increased endoscopy rates per patient (118). This may lead to a waste of clinical and economic resources, as the benefits of surveillance might be reduced by poor adherence to the surveillance protocol plus the cost of the endoscopy in general. In addition, the quality of life for BE patients may be impaired by anxiety concerning the possibility of cancer and the discomfort attributable to endoscopy.
26
5.8. Management: In general, the management of BE is guided by the extent of SIM and grade of dysplasia or cancer, if present. Any BE patients without evidence of dysplasia or cancer should follow a conservative and non-‐invasive management. The primary goal for the management of BE is prevention of the gastroesophageal reflux that may reduce the risk of development or progression to dysplasia or cancer. This involves symptom control and endoscopic surveillance biopsies to exclude progression. Conservative treatment begins with dietary and lifestyle modifications, together with pharmacological acid suppression therapy in symptomatic patients (119). Lifestyle modifications are usually helpful to increase esophageal acid clearance, and decrease the incidence of reflux events. Increase in the body mass index (BMI) is associated with an increase in the incidence of GERD
(120)
. Fatty food, caffeinated
beverages, chocolate and mint may also increase gastric reflux (121). There is physiologic evidence that exposure to tobacco, alcohol, chocolate and high-‐fat meals decreases lower esophageal sphincter pressure. Evidence for the efficacy of dietary modifications in controlling reflux symptoms is still lacking (122). At present, there is no definitive proof supporting an improvement in GERD symptoms after alcohol or smoking cessation (122). Acid-‐suppressing medications are the mainstay treatment for reflux symptoms. Either PPIs or histamine-‐2 receptor antagonists (H2RAs) can be used to decrease gastric acid secretion. However, PPIs provide more complete esophagitis healing and heartburn relief compared to H2RAs and this occurs nearly twice as fast (123). The risk of EAC in patients with GERD not taking PPIs is two to four times higher than the risk of EAC in
27
patients who are on acid suppression therapy (124-‐125). Data on PPIs as chemoprevention agents are controversial. Giving high-‐dose PPIs to prevent progression of Barrett esophagus to dysplasia and cancer is a common practice but there is no strong evidence to support this (125). However, two retrospective studies demonstrated a reduced risk of dysplasia when patients with BE used PPIs (125-‐126). In a double-‐blind randomized control trial comparing the amount of BE regression in patients on H2RAs with those on PPI therapy, a decrease in surface area of 8% was seen in the PPI group and no significant change was seen in the H2RAs (127). Also, in two retrospective observational studies, a decrease in the rate of progression from non-‐dysplastic BE to HGD/EAC was found with PPI use (124,128). Several epidemiologic studies have found aspirin and NSAID are associated with a decreased risk of progression to esophageal carcinoma (129-‐132). The ongoing AspECT trial in the U.K. is now studying the combined effect of PPIs and aspirin in preventing the progression of BE to EAC. Treatment with a cyclooxygenase 2 (COX 2) inhibitor has also been shown to decrease the risk of progression to EAC in an animal model of Barrett’s esophagus (133). Heath et al (134) conducted a multicenter, prospective trial did not find a statistically significant difference between the celecoxib (COX 2 inhibitor) and the placebo groups. However, it has been shown that 50% of patients with symptoms controlled on PPIs and up to 70% of patients who do not respond to PPI therapy show evidence of biliary reflux (135). This finding has led some to propose anti-‐reflux surgery (ARS) as a treatment of BE as it serves to decrease both gastric acid and bile reflux, a benefit that
28
cannot be achieved by medical therapy. In a retrospective review of 312 patients following laparoscopic Nissen fundoplication, 92% reported improvement in symptoms and 70% did not use anti-‐reflux medications at a median follow-‐up of 11 years
(139)
. For the
effectiveness of ARS in controlling the reflux, results from the LOTUS trial (a large multicenter randomized European study), with a three-‐year follow-‐up, showed that the esophageal pH was better controlled in patients with BE who underwent anti-‐reflux surgery compared with patients treated with a PPI (140). However, the symptoms were the same in both groups. The role of ARS in preventing BE progression is highly controversial and there is significant variance in the literature. A case–control study showed that 63.2% of patients on high-‐dose PPIs had regression of LGD, whereas 93.8% of surgical patients were found to have regression of dysplasia at 12 months (141). A systematic review showed that ARS was superior to PPIs at preventing cancer development (142), but when only prospective studies are considered, the benefit for ARS for EAC prevention could not be confirmed. Additionally, when cancer occurs in a BE patient following ARS, it is more likely that the anti-‐reflux barrier of the fundoplication has failed (143). However, a meta-‐analysis by Li et al showed that neither pharmacologic nor surgical anti-‐reflux interventions achieved complete regression of BE, nor eliminated the risk of EAC development (144). 5.8.1. Surgery: In Barrett’s patients with HGD, there is an increased risk of occult malignancy, which makes the surgical or endoscopic resection the traditional standard treatment in these patients. Several studies have shown the presence of malignancy in approximately 35-‐40% of resected samples in HGD patients (145-‐150).
29
Prior to the recent advent of endoscopic therapies, esophagectomy was the only modality of treatment available for HGD and EAC. However, there is considerable postsurgical morbidity and mortality. The mortality rate associated with esophagectomy is inversely related to the number of esophagectomies performed, making the higher volume centers attaining lower mortality rate (151). It has been suggested that there should be at least 20 esophagectomies done a year at an institution in order to decrease operative mortality rates to 5% or less (152). Esophagectomy is a very invasive procedure with a complication rate of approximately 40-‐50% and a hospital stay of 10 to 14 days. The immediate complications include pneumonia, anastomotic leak, wound infection, arrhythmia and heart failure. Long-‐term complications include dysphagia due to stricture formation, weight loss, gastroesophageal reflux and dumping syndrome, which may impair the health-‐related quality of life (153). 5.8.2. Endoscopic intervention: Endoscopic treatment has been designed as an alternative to surgical resection for patients with high-‐grade dysplasia or early stage EAC and is focused on localized resection or ablation of the existing metaplastic tissue using special methods and techniques that eliminate the columnar mucosa. A neosquamous epithelial layer replaces this ablated metaplastic epithelium. Multiple ablative modalities are now used in clinical practice, including argon plasma coagulation (APC), photodynamic therapy (PDT), and more recently radiofrequency ablation (RFA) and cryoablation. The low risk of complications and availability of good treatment techniques in the
30
setting of low rates of lymphatic or hematogenous dissemination have made endoscopic therapy for BE patients with HGD and early EAC a very practical option. One of the major concerns of ablative modalities is the absence of a pathologic specimen for histologic examination. Indeed, there is a possibility of progression of buried Barrett’s metaplasia or dysplasia under the neosquamous epithelial layer, as has been described in the literature (154)
. Multiple techniques have been developed for the endoscopic ablation and
treatment of BE. These techniques might be used either alone or in combination with other ablative techniques in treating Barrett’s patients with HGD or early stage EAC. 5.8.2.1. Endoscopic mucosal resection (EMR): EMR is an endoscopic procedure that was first used in Japan for superficial stomach cancers. Because Barrett’s dysplasia and early cancer is present in more flat lesions, EMR has been applied to the esophagus. This intervention involves an excision of a larger and deeper specimen of Barrett’s mucosa than is afforded by standard biopsy forceps. It has become increasingly important in recent years and it can be used as a diagnostic or a therapeutic tool. One of the advantages of EMR over other ablative therapies is that it provides tissue specimens. So, resecting a large suspicious area using EMR will provide more tissue for the pathological analysis. This might detect a focus of cancer that is more likely to be missed by conventional biopsies as the depth of these specimens can extend down to the middle of the submucosa. Another advantage of EMR is its potential for therapy. This is usually limited to HGD and early stage cancer with negligible risk of lymph node metastases. For the disease
31
confined to the mucosa with negative margins, complete resection can be considered. However, those with submucosal invasion are usually referred for surgery and oncology therapies. Several studies have demonstrated that EMR is safe and very effective for the treatment of superficial lesions (155-‐157). Numerous case series which have been published on using EMR for endoscopic therapy of HGD and superficial cancer showing outstanding results
(158)
. A recent study
(159)
has now provided excellent long-‐term results for
endoscopic resection in 100 consecutive patients with low-‐risk mucosal Barrett’s cancer. EMR techniques that have been described include snare resection with prior submucosal injection to elevate the mucosa and allow easy grasping with either a cap or a ligation device. EMR has also been employed as a method to remove all Barrett’s mucosa (complete Barrett’s excision (CBE)). However, this technique is not more effective than the ablative modalities and confers a higher complication rate (160-‐164). The complication profile of EMR includes stricture formation, with an incidence rate that approaches 50%, bleeding and perforation. Of note, most esophageal strictures and bleeding are amenable to endoscopic treatment (164-‐167). 5.8.2.2
Photodynamic therapy (PDT):
Photodynamic therapy involves the intravenous administration of a photosensitizer, which accumulates in the target mucosa followed by activation of the photosensitizer that is achieved by endoscopically applied laser directly to the malignant tumor. This results in free radical formation leading to the death of the tumor cells. PDT has been successfully used to treat early neoplasia in BE for more than one decade (168-‐169). In patients with HGD, a randomized trial of PDT with PPI versus PPI alone,
32
complete ablation of HGD in 77% of patients in the phototherapy plus PPI group versus 39% of patients in the PPI only group (170). More importantly, PDT was found to reduce the progression of Barrett’s HGD to cancer by approximately 50% compared to patients treated with PPIs alone. The high stricture rate,chest pain, progression of the disease despite treatment and cutaneous photosensitivity plus the high cost are the major drawbacks associated with this treatment modality (171-‐172). Accordingly, this technique has largely been replaced by other modalities including RFA and cryotherapy. 5.8.2.3. Radiofrequency ablation (RFA): Radiofrequency ablation (RFA) is a non-‐invasive technique that uses thermal energy to destroy abnormal cells. This is a newer technique in which a cylindrical balloon with circular electrodes is inflated so the electrodes come into contact with mucosa to which thermal energy is applied. Ablation can be applied to the whole circumference of the esophagus (HALO-‐360) or to smaller areas using a focal device (HALO-‐90). Tissue penetration depth of the RF energy is about 0.5mm (173), which has been demonstrated as sufficient for successful ablation of esophageal epithelium with no injury to the deeper tissues. What make this technique unique are the precise and controlled delivery and the automated delivery of a preset amount of standardized radiofrequency energy. Furthermore, RFA is simple and less time-‐consuming than other treatments. RFA has been shown to be successful with high rates of complete eradication in the treatment of dysplastic
(174)
and NDB
(175-‐177)
. A recent landmark randomized sham-‐
controlled trial showed that RFA was effective in the complete eradication of LGD
33
compared to controls (90.5 vs. 22.7%, P < 0.001)(174) Similar results were seen in the treatment of patients with HGD (77.4 vs. 2.3%, P < 0.001) (174). Although a follow up study has shown this effect to be durable to 3 years (178), information on progression to cancer will not be forthcoming, as patients in the sham group were offered cross-‐over RFA after 1 year. Given that the depth of effect with RFA (0.5-‐1mm) is significantly less than PDT (2-‐ 5mm), the reduction in cancer progression is unlikely to be better than that seen with PDT (approximately 50%) (175). The most common complications associated with RFA include non-‐cardiac chest pain, superficial lacerations, and stricture formation (although a much lower rates than EMR/CBE and PDT). At present, RFA is the preferred modality to be used as an adjunct to endoscopic resection for ablation of the remaining non-‐dysplastic Barrett’s epithelium after successful resection of all irregular mucosa suggestive of HGD or EAC. In patients with visible lesions in the setting of HGD, a combination of EMR and RFA has recently been studied. Pouw and colleagues
(179)
have reported on the
performance of EMR for visible lesions with subsequent ablation of the remaining segment. Complete histological eradication of all dysplasia and SIM was achieved in 43 patients (98%). 5.8.2.3
Cryotherapy:
Cryotherapy is the latest endoscopic ablative technique. Through freeze-‐thaw cycles using liquid nitrogen or carbon dioxide, intracellular disruption and ischemia are produced. These cycles are said to cause intracellular damage while preserving the extracellular matrix and thereby promoting less fibrosis (180-‐181).
34
Two case series have shown good results, but long-‐term follow up is lacking (182-‐183). Furthermore, there are no RCT comparing its efficacy with other modalities of therapy. The fogging of the scope lens, risk of perforation and the prolonged duration of the therapy are some of the limitations of cryotherapy.
6. Rationale: The incidence of EAC has been rapidly increasing in the past 3 decades and only a minority of BE patients will progress to HGD or EAC (184-‐185). Unfortunately, most EAC are detected in the late stage and the associated mortality exceeds 90% (186). For that reason, endoscopic surveillance of patients with BE is recommended because it detects cancer at an early, curable stage (187-‐188) that may help to reduce the associated mortality. The cost-‐effectivness of this surveillance program is dependent on the risk of progression of BE to EAC. However, there is a wide variation in the annual risk of progression from BE to EAC. Recently, this has been estimated at about 0.5% per year (40-‐ 43)
. As the majority of BE patients will never progress, most of them will die due to
unrelated diseases. Because of the relatively large number of patients with BE and because the absolute risk of neoplastic progression in BE that is low, the majority of patients with BE will not benefit from an endoscopic surveillance program, making the clinical and cost-‐ effectiveness of this program in general questionable. However, at present it is unknown which subgroup of patients with BE will be more likely to rapidly progress to HGD and EAC than others. This makes risk stratification in identifying BE patients at the highest risk of developing neoplastic progression important and it may also make the surveillance
35
program more cost-‐effective allowing for individualization of intensive surveillance and follow up for targeted subgroups. So far, only the presence of dysplasia is used as a tool to determine the timing of surveillance intervals and management of patients with BE. However, additional progression predictors are needed for appropriate risk stratification to detect those patients at higher risk of neoplastic progression. Easily attainable demographic and endoscopic features could be used to predict the neoplastic progression and hence provide an ideal non-‐invasive method for risk stratification. When these predictors are used in conjunction with endoscopic biopsy results as determinants of surveillance intervals, may improve the efficiency of the program. A detailed assessment of all demographic and endoscopic factors and their role in progression should be performed in order to know what are the factors that can be used as a predictor of progression. This study will attempt to determine what are the most predictive demographic and endoscopic factors for neoplastic progression in BE patients.
7. Review of literature: Identification of demographic, clinical, endoscopic and pathologic factors that predict the rapid progression to any dysplastic grades and cancer will aid in risk stratification among BE patients. This has been reviewed by several studies, but presently only the presence of dysplasia has been found to be the major determinant that is used clinically for BE management and surveillance intervals. Previous studies have shown some factors that are implicated in the neoplastic progression of BE condition. A 1993
36
study by Weston et al (189) was one of the earliest studies that tried to determine the endoscopic and pathologic predictors of progression. It was a prospective study in which 108 patients were included. It showed that the length of Barrett's segment, presence of hiatal hernia and dysplasia at initial diagnosis or during follow-‐up were found to be predictive of development of HGD or EAC. A retrospective study done by Anandasabapathy et al (190) searched for the clinical, endoscopic and pathologic predictors of HGD or EAC. With a total of 109 patients, they found that male gender, longstanding GERD, hiatal hernia size and BE segment length were strongly associated with higher grades of dysplasia at index diagnosis. However, other studies have shown that hiatal hernia, a long BE segment and LGD are also associated with EAC development (191-‐196). In addition, factors such as male gender, advanced age and Caucasian ethnicity are associated with GERD, BE and EAC development. A recent prospective study by Sikkema et al (197) included 712 patients with a baseline diagnosis of NDB or LGD. It revealed that the risk of developing HGD or EAC is predominantly determined by the presence of LGD, a known duration of BE ≥ 10 years, longer BE segment and the presence of esophagitis. One or a combination of these risk factors identify patients with a low or high risk of neoplastic progression and could therefore be used to individualize surveillance intervals in BE. Pieter et al (198) performed a retrospective study in which 42,207 patients with BE were included. Male gender, older age and LGD in the initial presentation were the independent predictors of malignant progression in this largest cohort study. Bhat et al
37
(199)
did a retrospective study of 8,522 patients. SIM, LGD and male gender where found to
increase the risk of EAC development in BE patients. However, there are some limitations of these studies. Low power due to small sample size is one of the limitations for the first three studies. For the two larger retrospective studies
(198-‐199)
, there was lack of availability of complete data most
important of which is the absence of endoscopic data as the studies were based on administrative pathology databases. This is an important point, as it is difficult to determine which patients have true Barrett’s esophagus (SIM with columnar lined esophageal mucosa on endoscopy) rather than intestinal metaplasia of the cardia. Also there were no standardized biopsy protocol and pathological reporting system used in Irish study (199). Table 3 summarizes the most important studies of this topic.
8. Objective: To identify the factors that predict the progression of non-‐dysplastic Barrett (NDB) to any dysplastic or neoplastic grades.
9. Hypothesis Does the presence of the listed predictors increase the risk of dysplastic and neoplastic progression in BE patients?
38
Table 3: Summary of the relevant clinical studies and their outcome Sikkema et Anandasabapathy Weston et al Bhat et al. a al et al
Pieter et al. a
Year
March 2011
Jan 2007
Jan 2007
May 2011
March 2010
Total no. Patients
713
108
109
8522
42 207
Author
Type
Prospective Prospective
Developed HGD/EAC
26
35
Duration
4 years
101 months
LGD Longer duration of BE Predictors of Longer progression length Esophagitis
Dysplasia HH BE length
Retrospective Retrospective Retrospective 35
36/79
161/505
Jan 2002 to Sept 1993 to 2005 1991 to 2006 2005 Male gender Longstanding GERD HH size Length of BE
Age LGD SIM
Age Male LGD
a = follow the British definition of BE
39
10. Methods Between January 2000 and December 2010, we performed a retrospective study where patients with a diagnosis of BE were evaluated at the McGill University Health Center (MUHC), including the Montreal General and the Royal Victoria Hospitals, in Montreal, Quebec, Canada. Demographic and endoscopic data were reviewed retrospectively for each patient confirmed to have Barrett’s esophagus. 10.1 Study Population Patients who underwent an upper endoscopy at MUHC in the last 10 years and were confirmed to have BE in their biopsy, were identified. Criteria for the diagnosis of BE in this project, following the American society of gastroenterology (ASGE) definition of BE, where endoscopic identification of the squamocolumnar junction proximal to the gastroesophageal junction and targeted biopsies with pathology revealing columnar epithelium with goblet cells. For that we included any patient who had both a pathologic diagnosis of BE together with an endoscopic diagnosis of BE at any time as of recruitment. 10.2 Inclusion Criteria:
Patients in MUHC
From January 2000 till December 2010.
18 years and older.
Diagnosis of BE
Made any time since recruitment. Made based on both pathological and endoscopic findings (not necessary to have both in the same visit).
40
2 or more endoscopic surveillance follow up visits in order to assess the progression status. 10.3 Exclusion Criteria:
Absence of BE in the endoscopy reports even if it is present in pathology.
Absence of BE in the pathology report even if it is present in endoscopy.
Single surveillance follow up.
Any history of esophageal resection or esophageal squamous cancer.
Age less than 18 years.
10.4 Study Procedure: This study began by identifying the patients who had BE. First, the pathology database of the MUHC (LIS PROD) was searched to identify those with a BE in their biopsy. In this search, we used different search terms (Barrett esophagus, Barrett's esophagus, Barrett’s epithelium) to widen our search base and to identify all possible cases that might be missed due to misspelling of the word “Barrett”. Once a histologic diagnosis of BE was found, the endoscopy database software “Endoworks” was searched to locate the related gastroscopy reports.
41
Year 2007 2009 2009 Pathologic diagnosis NA IND HGD Endoscopic diagnosis BE NA NBE Patient 1 Year 2002 2003 2005 2007 2010 Pathologic diagnosis BE BE BE BE BE Endoscopic diagnosis NA NA BE BE BE Patient 2 Year 2006 2006 2007 Pathologic diagnosis BE BE EAC Endoscopic diagnosis BE BE BE Patient 3: Year 2007 2008 2009 Pathologic diagnosis NA BE EAC Endoscopic diagnosis BE BE BE Patient 3:
Figure 2, Examples of patient treatment timelines to elaborate the presentation of BE grade at the baseline.
42
In order to collect the data, demographic and endoscopic case report forms (see Appendices A and B) prepared specifically for this project were used. These forms contain all the possible risk factors that might be implicated in dysplastic or neoplastic progression of BE. These have been filled for each endoscopy report to determine the presence or absence of these factors in the BE population. Because the Endowork software was only installed at the MUHC in 2004, the medical charts of patients prior to this date were requested for those who had BE in pathology to get their endoscopy reports. All biopsy samples from BE patients were examined by an expert esophageal pathologist in the MUHC hospital where the BE patient was identified or followed up. Data collected for this project included patient demographics and findings on the endoscopic and pathology reports. The independent variables included are those that might be associated with an increased risk of Barrett dysplasia or adenocarcinoma development. Independent demographic variables include age and gender. Patients were divided into two categories based on the median age, those less than 60 years old and those 60 years old or older. Endoscopic independent variables include endoscopy date, information on hiatal hernia presence and size (small, large), presence and grade of esophagitis (mild, moderate, severe), presence of ulcer, Barrett segment length (PRAG M), presence of mucosal irregularities, strictures. Selection of these independent variables to be examined was made based on the review of literature that showed associations between BE development and progression to dysplasia or EAC. We also considered all the possible risk factors that can be gathered from the endoscopy or pathology reports.
43
Endoscopic Variables *Hiatus hernia (size) *Esophagitis (severity)
Demographic variables *Age *Gender
* Ulcer *mucosal irregularities *BE length *Stricture
Progression Factors
Figure 3, Candidate Independent variables for this study. Hiatus hernia size was divided into small or large. Hiatus hernias of 1-‐2 cm were determined to be small, while the large hiatus hernias group included those of 3cm or more. Esophagitis severity was evaluated as mild, moderate or severe. Because the classes of Los Angeles (LA) classification of esophagitis severity have been used in some, but not all, endoscopy reports, these were converted into our scale as the following: mild represents class A, moderate represents class B and severe represents classes C and D.
44
The Barrett’s segment length was identified based on the reported PRAG M measurement. It was divided into 2 categories: short segment Barrett’s (SSB) if the length is less than 3 cm and long segment Barrett’s (LSB) if 3 cm or more. Mucosal irregularities included the presence of any nodular mucosa, nodules or polyps in the distal esophagus. Pathologic variables included pathology report dates, presence or absence of dysplasia (including indefinite for dysplasia (IND), LGD and HGD) or cancer. The primary clinical outcome assessed in this project was the progression to dysplasia or cancer. In this project, we wanted to assess the dysplastic and neoplastic progression in this cohort. Two types of progression were investigated: Progression from NDB to any dysplastic or neoplastic grades (including IND/LGD or HGD/EAC). Progression from NDB to HGD/EAC. Based on the outcome, we attempted to identify the predictors that speed up progression to dysplasia or cancer. Due to the lack of standardized endoscopy reporting system within the division of gastroenterology at the McGill University Health Centre, there exist significant variations in the reporting of findings by the numerous endoscopists. Although there is no synoptic reporting, there is a general policy in writing the endoscopy reports among the endoscopists in which only the positive findings are described. Therefore, for the purposes of this study, we made the assumption that any of the possible endoscopic findings that were not mentioned in the endoscopic reports are considered as evidence that the report
45
was negative for such items. This policy also applied to pathologic reporting for the presence and absence of BE, dysplasia or cancer. For endoscopic variables, this is applied to the presence of Hiatus hernia, esophagitis, ulcer, strictures and mucosal irregularities. In regards to the size of hiatus hernia, the severity of esophagitis and the length of the BE segment, we did not follow this policy as not reporting them does not necessarily provide proof of their absence.
Table 4, Summary of the total and missing endoscopy reports. Condition Total endoscopy reports entry
Total number 1961
Total missing endoscopy reports
311
Total missing baseline endoscopy reports
166
Because of the retrospective nature of the study, the availability of the endoscopy reports was an issue. Some of the patients had missing endoscopy reports that could not be found in either the hospital charts or the Endoworks electronic database. These missing reports could be either for the baseline visit or any subsequent visit. However, all the patients included in the final analysis must have had at least 1 visit with a scope report that confirmed the endoscopy diagnosis of BE. This situation also occurred in pathologic data, for different reasons. Unavailability of some pathologic reports in this cohort is mainly due to patient-‐related reasons. These included either refusal of biopsy by the patient or cases where the patient’s medical condition did not
46
allow for biopsies, for example the risk of excessive bleeding for a patient on anticoagulation medication. As the presence or absence of dysplasia or cancer in the baseline pathology status, is important to determine the progression. In the absence of the baseline pathology status, in some cases we considered the subsequent visit pathology result as a baseline status for these patients. We feel that this is an acceptable concession to make as it is unlikely that the BE would be different in between endoscopies of the same patient. In general, the endoscopic surveillance program for BE patients at the MUHC is designed to follow the guidelines of the American society of gastroenterology (ASGE). However, the compliance to this protocol is admittedly low for the reasons mentioned above. Dysplasia associated with BE was graded according to the consensus criteria of 1988, with adjustments as proposed in 2001(200-‐201). The latter comprises IND, LGD, HGD and EAC (202). Patients in this cohort were classified into 3 groups based on their pathologic baseline status into: 1) NDB, 2) IND/LGD, or 3) HGD/EAC. The NDB group included patients who presented to the MUHC with NDB diagnosis (baseline diagnosis), whether it progressed later on or not. The IND/LGD group included those with IND or LGD in the baseline diagnosis. HGD/EAC included those with HGD or EAC in their presentation to the MUHC. 10.5 Ethics: The Medical Ethics Committee of McGill University Health Center (MUHC) approved the study protocol.
47
10.6 Statistical analysis: The primary outcome of this analysis is to identify the predictors of progression from non-‐dysplastic Barrett’s epithelium to dysplasia or cancer. Demographic and endoscopic characteristics were analyzed using the low risk category as the referent group. Categorical endoscopic variables were age (< 60 years old, ≥ 60 years old), gender, esophagitis and grade of esophagitis, hiatus hernia and its size, ulcer, BE length (< 3 cm, ≥ 3cm), presence of mucosal irregularities or strictures. We used the presence or absence of dysplasia or cancer in the pathology report as the outcome to determine the progression status. To know how many patients have progressed compared to the entire cohort, the person-‐year incidence rate of the progression for each type being investigated was calculated. This has been expressed per 1000 person-‐year. The rate of progression as a function of each independent variable was also calculated. Progression from NDB to pre-‐malignant and malignant lesions was initially examined via time-‐to-‐event survival analysis methods. Survival curves were constructed using the Kaplan-‐Meier method. This was used as a descriptive statistical method to analyze the data while controlling for censored observations, where not all patients have an outcome (not all patients progressed). A Kaplan-‐Meier curves were constructed for some variables for which comparisons were made between high risk and low risk categories for a given variable. In order to measure the significance of the differences between the Kaplan-‐Meier curves, the Log-‐rank test was also used. After having a preliminary analysis of the results using the survival curves, a full
48
analysis was performed in order to identify the factors that are most predictive in the progression to dysplasia or cancer. For this, the Cox proportional hazard model was chosen. It was used to calculate the Hazard Ratio (HR) with corresponding 95% confidence intervals (CIs) of progression for each demographic and endoscopic variable using the low risk category as the referent group. This was also calculated after age adjustment. In the analysis, Cox proportional hazards regression modeling (Empirical estimate) is applied in which all the independent variables were examined first to assess their effect individually, then their combined influence, to identify the variables that are most predictive of lesion progression. Two-‐sided p-‐values < 0.05 were considered to be statistically significant. Due to the nature of the data, containing multiple observations per patient, we used the generalized estimating equations (GEE) model to estimate the odds ratios (OR) of progression at the next visit according to each variable. This statistical procedure takes into account the correlation between the variables for each individual. The advantage of using the GEE model is to be able to use multiple pairs of visits from each patient to assess all covariates that can be predictive of progression in the short term. Here again, all the independent variables were examined to estimate their combined influence in order to identify the variables that are most predictive of Barrett’s progression. For statistical analysis, we used the “STATA 10” statistical software program for all analyses (Copyright 2002–2010 StataCorp LP, 4905 Lakeway Drive, College Station, TX 77845 USA).
49
11. Results: From January 2000 to December 2010, the total number of patients identified with BE in the pathology database “LIST PRODE” was 1052; 751 of whom were found to have pathologic and endoscopic diagnosis of BE. 303 patients were excluded due to our inability to find the endoscopy report in either “Endowork” or the medical chart, no documentation of BE in the endoscopy reports, or due to poor documentation of BE in the scope reports like pink mucosa or abnormal mucosa. Of these 751 patients, 518 patients had 2 or more surveillance visits that made them eligible for the final progression analysis. Figure 4 shows the progress of the inclusion process up until the final analysis. During this surveillance and based on the baseline status, a total of 57 patients progressed. From the baseline of NDB (N=458), 35 patients (7.6%) progressed to any dysplastic or neoplastic grades, 7 patients progressed from an IND baseline (7/13= 53.8%), 10 patients progressed from LGD baseline (10/20=50%), and 5 patients progressed from HGD baseline (5/11=45.5%). Table 5 shows the total number of patients according to their baseline, eligibility and progression status.
50
1054 patients had BE by Pathology
120 No BE in EGD
13 not well documented
170 No available EGD
751 patients had BE by pathology and endoscopy
233 had single visits
518 patients entered the final analysis Figure 4: Flow chart for the total number of patients and the process of inclusion.
51
Table 5: Total patients according to their baseline, eligibility and progression status Baseline
Total patients no.
Eligible
Progressed
NDB
681
458
35
IND
14
13
7
LGD
24
20
10
HGD
14
11
5
Cancer
18
16
Total
751
518
-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐ 57
Eligible patients were divided based on their baseline status into NDB, IND/LGD and HGD/EAC. These groups include patients who have presented with the same condition and either progressed to higher grades or remained in the same status. Out of 518, 458 were included in the NDB group, 33 in IND/LGD and 27 in the HGD/EAC group. The demographic characteristics of all eligible patients (n=518) are shown in table 6. For all eligible patients, the median age was 60 years old. Figure 5 shows the median age for the 3 groups of patients. For those in IND/LGD and HGD/EAC groups, a higher percentage is above the median, 54.5% and 70.3 % respectively. Males (n=372) represent the majority of the entire cohort of patients (71.8%).
52
100
100 80
80 age 60
Age (years) 60 40
40 20
20 0 NDB
1 IND/LGD
2 HGD/EAC
Figure 5, Box plot shows the median age, interquartile range for all 3 groups.
Table 6, Demographic characteristics for all eligible patients.
All
NDB
IND/LGD
HGD/EAC
N
N
N
N
518
458
33
27
< 60
263
240
15
8
≥ 60
255
218
18
19
Male
372
319
28
25
Female
146
139
5
2
No. of patients
53
Due to the retrospective nature of the study, endoscopic reports of 166 patients were missing in the baseline visit. However, evaluation of the endoscopic variables revealed that 42% had hiatus hernia, 12.16% had esophagitis, 4.7% had ulcers, 6.9% had mucosal irregularities, and 3% had strictures. Due to poor documentation in the endoscopy reports, more data were missing to document the hiatus hernia size, esophagitis grade and length of BE. Table 7 shows all the endoscopic variables for the whole cohort and all 3 groups, including the missing reports.
54
Table 7, Baseline characteristics of patients with BE based on their baseline status.
All
NDB
IND/LGD
HGD/EAC
N (%)
N (%)
N (%)
N (%)
No. of patients
518
458
33
27
Hiatus hernia (HH) No
209 (40.3)
187(40.8)
11(33.3)
11(40.7)
Yes
143 (27.6)
124(27.1)
11(33.3)
8(29.6)
Missing
166 (32.1)
147(32.1)
11(33.3)
8(29.6)
HH size Small
42 (8.1)
36(7.9)
5 (15.2)
1(3.7)
Large
59(11.4)
50 (10.9)
4(12.1)
5(18.5)
Missing
417(80.5)
372(81.2)
24(72.7)
21(77.8)
Esophagitis No
269 (52)
234 (51.1)
17 (51.5)
18 (66.7)
Yes
83 (16)
77 (16.8)
5 (15.2)
1 (3.7)
Missing
166 (32)
147 (32.1)
11 (33.3)
8 (29.6)
Severity Mild
14 (2.7)
13 (2.8)
1 (3)
0
Moderate
28 (5.4)
26 (5.7)
2 (6.1)
0
Severe
13 (2.5)
11 (2.4)
1 (3)
1 (3.7)
Missing
463 (89.4)
97 (21.2)
29 (87.9)
26 (96.3)
Ulcer No
318 (61.4)
281 (61.4)
18 (54.5)
19 (70.4)
Yes
34 (6.6)
30 (6.6)
4 (12.1)
0
Missing
166 (32)
147 (32.1)
11 (33.3)
8 (29.6)
BE Length < 3 cm
49 (9.5)
40 (8.7)
5 (15.2)
4 (14.8)
≥ 3 cm
98 (18.9)
81 (17.7)
8 (24.2)
9 (33.3)
Missing
371 (71.6)
337 (73.6)
20 (60.6)
14 (51.9)
Mucosal irreg. No
338 (65.3)
305 (66.7)
20 (60.6)
13 (48.1)
Yes
14 (2.7)
6 (1.3)
2 (6.1)
6 (22.2)
Missing
166 (32)
147 (32.1)
11 (33.3)
8 (29.6)
Stricture No
341 (65.8)
301 (65.7)
21 (63.6)
19 (70.4)
Yes
11 (2.1)
10 (2.2)
1 (3)
0
Missing
166 (32)
147 (32.1)
11 (33.3)
8 (29.6)
Total EGD missing
166 (32)
147 (32.1)
11 (33.3)
8 (29.6)
55
Table 8, Follow up (in months) median and mean for the patients with NDB baseline Status
Mean
95% CI
Median
95% CI
All patients
40.1
36.1-‐45.2
55.2
51.9-‐59.7
Remained NDB
44.0
41.9-‐46.4
60.5
57.7-‐63.3
NDB to IND/LGD
20.2
16.9-‐24.1
30.2
17.9-‐38.3
NDB to HGD/EAC
21.2
19.2-‐23.5
16.7
16.7-‐28.4
The surveillance follow up time for patients who remained in NDB status had a mean of 3.7 years and median of 5 years (44 months, 60.5 months respectively). For those who progressed from NDB to IND/LGD and to HGD/EAC, the follow-‐up was shorter. This is because the presence of LGD or HGD increases the risk of progression to EAC and is either treated or the patient dies so a shorter follow up is expected. Table 8 shows the mean and the median for each category. In order to identify the predictors of neoplastic progression, two different stages of progression have been analyzed. They are: Progression from NDB to any dysplastic or neoplastic grade (including IND/LGD or HGD/EAC). Progression from NDB to HGD/EAC. 11.1 Progression from NDB to any dysplastic or neoplastic grades: 35 patients progressed from NDB to a dysplastic grade or cancer, including IND/LGD or HGD/EAC. The progression rate and patient-‐year follow up of these patients, based on
56
their demographic and endoscopic variables are shown in table 8. The incidence of this type of progression in this cohort was 19.8 cases per 1000 person-‐years (95% CI 14.2-‐ 27.6). The total patient year follow up was 1767.6 person-‐years for all 35 patients. The annual risk of progression to any dysplastic or neoplastic grade from NDB baseline is 1.98% per year (95% CI 1.4-‐2.8). Table 9 shows the progression rate per 1000 person-‐years for each variable examined in this project. Of those, 54.3% were older than 60 years old with a progression rate of 26.6 (95% CI 16.9-‐41.7). Males represented 80% with a progression rate of 22.5 (95% CI 15.6-‐32.6). Evaluation of the endoscopic variables shows that the presence of hiatus hernia had a higher progression rate of 23.7 (95% CI 16.2-‐34.5). Progression rates also increased with an increase of its size (19.8, 21.3 respectively). Surprisingly, esophagitis had a lower progression rate (16.6 vs. 21.2) compared to those who did not have esophagitis. Presence of an ulcer or mucosal irregularities had high progression rates, 21.0 (95% CI 7.9-‐55.9) and 32.2 (95% CI 12.1-‐85.9) respectively. The longer the BE segment, the higher the risk of progression, with progression rate of 29.9 (95% CI 18.9 – 47.5) for a long BE segment. Presence of stricture was also associated with a higher progression rate of 34.2 (95% CI 11.0-‐105.9). The 10-‐year cumulative incidence of any dysplastic or neoplastic grades by age, gender and BE length are shown in figures 6, 7 and 8. Those older than 60 years, male and those who had BE length ≥ 3 cm are at higher risk of progression to any dysplastic or neoplastic grades comparing to their lower risk counterparts. However, only BE length ≥ 3 cm showed a significant difference (p=0.031).
57
Table 9, Dysplastic and neoplastic progression rate in patients with NDB baseline according to the demographic and endoscopic variables Variable
Category
Events
Person year Progression P r=ate/ 0.22 1000 person year
95% CI
Age
< 60
16
1052.8
15.2
9.3-‐24.8
≥ 60
19
714.7
26.6
16.9-‐41.7
Gender
Male
28
1242.4
22.5
15.6-‐32.6
Female
7
525.2
13.3
6.4-‐27.9
HH
No
8
626.5
12.8
6.4-‐25.5
Yes
27
1141.1
23.7
16.2-‐34.5
HH size
Small
8
402.9
17.3
10.0-‐29.8
Large
12
564.3
21.3
12.1-‐37.4
Esophagitis
No
26
1225.1
21.2
14.5-‐31.2
Yes
9
542.5
16.6
8.6-‐31.9
Eso. Severity
Mild
1
168.9
5.9
0.8-‐42.0
Moderate
3
145.7
20.6
6.6-‐63.8
Severe
0
81.6
0
Ulcer
No
31
1576.8
19.7
13.8-‐27.9
Yes
4
190.7
20.9
7.9-‐55.9
BE length
< 3 cm
10
772.6
12.9
6.9-‐24.1
≥ 3 cm
18
600.9
29.9
18.9-‐47.5
Mucosal irreg.
No
31
1643.5
18.9
13.3-‐26.8
Yes
4
124.1
32.2
12.1-‐85.9
Stricture
No
32
1679.8
19.1
13.5-‐26.9
Yes
3
87.8
34.2
11.0-‐105.9
Total
35
1767.6
19.8
14.2-‐27.6
58
0.40 0.30 0.20
Male
0.00
0.10
Female
0
24
48
72
96
120
Follow-up (months) Figure 6, Cumulative incidence of progression to dysplasia or cancer by gender.
0.40
P = 0.098
0.20
0.30
60+ years
0.00
0.10
= 3 cm
0.00
0.10
0.20