The Study of Barrett's Esophagus: What Are the Factors of Progression

Barrett's esophagus (BE) is defined as a change in the esophageal mucosa from normal squamous epithelium to columnar epithelium with intestinal metaplasia (1). There is compelling evidence that BE is a precursor lesion for adenocarcinoma of the esophagus (2). There is also evidence that the prevalence of BE is much higher than was initially suspected, and that the true incidence rate of cancer in patients with BE is much lower than previously estimated. Endoscopic surveillance of all patients with BE is not likely to be cost effective, or save many lives. Therefore, it is important to understand which patients with BE are likely to progress to malignancy, and equally important to determine which patients are not likely to progress.Because progression is an uncommon event, definitive studies of natural history including evaluation of numerous risk factors require large numbers of patients, that can only be accomplished in the context of a multicenter study. This model will be a useful tool leading to a reduction in overall health care costs.

An Opportunity to Generate the Essential Knowledge for Intervention. We have established an initial network of investigators interested in the 'Barrett's Esophagus Study' (a long term, longitudinal, prospective study) and collected preliminary data in patients with BE. To date, 1376 patients from 4 centers have been entered into a central databank. We have presented preliminary data on the incidence and prevalence rates of high-grade dysplasia and cancer at national meetings. We now propose the creation of a formal consortium to perform a cohort study with the ultimate primary aim of defining risk factors for the development of high-grade dysplasia and adenocarcinoma in patients with BE. Such a multi-center, prospective, cohort, multi-disciplinary approach to BE was endorsed by the "NIH PRG on stomach and esophageal cancers" (May 2002).

BACKGROUND \& CLINICAL SIGNIFICANCE Barrett's Esophagus the pre-malignant lesion for esophageal adenocarcinoma - a Condition of Rapidly Increasing Incidence. Patients with BE are at an increased risk for the development of esophageal adenocarcinoma and esophagogastric junction adenocarcinoma. Adenocarcinoma of the esophagus is believed to arise from BE in a step-wise progression from low-grade dysplasia to high-grade dysplasia, finally leading to frank adenocarcinoma (4,5). The cost effectiveness of surveillance endoscopy in patients with BE is very sensitive to the incidence of cancer (11). The exact incidence of adenocarcinoma in patients with BE is not known and has been reported to vary from 1 in 50 patient-years of follow up to 1 in 285 patient-years of follow up (18-22). The published incidence data from existing cohorts of patients with BE have been influenced by the small number of patients being evaluated, short duration of follow-up as well as by the demographics of the population being studied (e.g. older patients, male sex). To date, the maximum number of patients included in a published study determining the incidence of adenocarcinoma has been only 327 patients (23,24). The ultimate goal of the proposed study is to establish a large cohort of patients with BE to address issues of incidence, progression and risk factors influencing progression. Such an approach was recently endorsed by the "NIH 2002 PRG on stomach and esophageal cancers".

Factors predicting the development of dysplasia and cancer in BE patients are unknown. A few possible factors have been associated with esophageal adenocarcinoma, but not studied in BE patients. These include GERD symptoms, obesity, diet, smoking, prior cholecystectomy, drugs and H pylori infection (negative association) (25-36). The presence of hiatal hernia and BE length have been associated with dysplasia and cancer development in BE patients in a preliminary single center study (37). Reflux symptoms have also been reported to be significantly more prevalent among parents and siblings of patients with BE and esophageal adenocarcinoma than spouse control relatives (38). A few uncontrolled reports have provided conflicting results on the role of anti-reflux surgery and the use of acid suppressive medications (proton pump inhibitors, H2 receptor antagonists) and their effect on the neoplastic progression in BE patients (39-43). However, all the studies to date in BE patients have been limited by small sample sizes, univariate analysis, short duration of follow-up, lack of adjustment for confounding variables etc. In this proposed study, risk factors and mechanisms of capturing data at each site and data transfer to the main study database will be discussed and implemented. Sample size calculations for appropriate univariate and multiple logistic regression analysis will be performed.

Proposed Study will Fill Significant Gaps in Knowledge. In the recent "NIH PRG on stomach and esophageal cancers (May 2002)" and a workshop sponsored by the NIH \& The American Digestive Health Foundation on endoscopic priorities (44), experts in the field acknowledged that there was a need for multi-center, longitudinal, studies and there were significant gaps in the knowledge base of BE including:

1. Risk stratification of cancer risk
2. Appropriate method of surveillance and technique for identifying dysplasia
3. Frequency and benefits of surveillance
4. Optimal management of BE such as acid suppression and endoscopic reversal techniques The ultimate goal of the BE Study is to address issues #1 and #3 i.e. risk stratification of Barrett's patients, and by developing a large, multi-site database, will prepare the way for future studies that can address the other key issues.

The information gained will be clinically useful. We will define the natural history of progression of BE patients and to determine factors involved in this progression to high-grade dysplasia and cancer. The potentially unique answers obtained from the study would have major clinical relevance in

• Defining the natural history of dysplasia/cancer in BE patients and to identify risk factors involved in progression to dysplasia/cancer.
• Helping clarify what predisposes only a small subset of BE patients to develop dysplasia/cancer. Rigorous examination of a number of risk factors may define this patient group, thereby focusing our limited health care resources to the patient subset at increased risk for the development of dysplasia/cancer - "reduction in health care costs"
• Potential to change our current costly clinical practice of endoscopic surveillance of all BE patients, by limiting surveillance to the 'high-risk group' - "improve clinical decision making"
• Providing better and accurate information to our BE patients regarding their 'true' risk for dysplasia and cancer.

PRELIMINARY STUDIES At this time, preliminary data have been collected from the 4 participating centers, and merged into one single databank. These preliminary results were presented at the plenary sessions (45,46) of the American Gastroenterological Association (AGA) annual meeting in Atlanta, GA (May 2001) and at the American College of Gastroenterology (ACG) annual meeting in Las Vegas, NV (October 2001), indicating a common interest amongst the investigators and capability of working together.

Distribution of Degree of Dysplasia. From the four participating centers, 1376 patients have met the study criteria and had at least one endoscopy with biopsy revealing intestinal metaplasia, thus confirming the diagnosis of BE. At the initial endoscopy, the distribution of dysplasia was as follows:

Diagnosis Number of pts. Prevalence Low-grade dysplasia (LGD) 101 7.3% High-grade dysplasia (HGD) 42 3% Esophageal adenocarcinoma 91 6.7% Incidence and progression to low-grade dysplasia, high-grade dysplasia and esophageal adenocarcinoma. 1376 patients were enrolled in the prevalence phase of the study. To date, 618 patients (95% Caucasians, 14% Females) have had follow-up endoscopic procedures to examine the incidence of dysplasia/cancer. They have been followed for a total of 2546 patient-years; mean follow up 4.12 patient-years (range: 1-22.5 yrs). Twelve patients developed cancer during follow up for a cancer incidence of 1 in 212 patient-years of follow up or 0.5% per year. The mean time to development of cancer was 5.3 yrs (range: 2.4-11.2 yrs). Of interest, 7 of 12 patients had high-grade dysplasia before cancer development while 2 had only low-grade dysplasia; 3 developed cancer from BE without dysplasia documented at any point. 22 patients developed high-grade dysplasia during follow up for an estimate of high-grade dysplasia incidence of 1 in 116 patient-years of follow up or 0.9% per year. The mean time to high-grade dysplasia development was 3.8 yrs (range: 1.2-7.9 yrs). 11 of 22 patients developed high-grade dysplasia from BE without documented dysplasia, whereas the remaining 11 went from low-grade dysplasia to high-grade dysplasia.

Risk Factors for BE and adenocarcinoma. The epidemiology and natural history of BE has been poorly studied. Some factors have been associated with an increased risk for esophageal adenocarcinoma in the general population (see Table 1), but risk factors for progression in BE patients are not clear. A number of risk factors have been suggested as factors of risk in BE patients including age, gender, ethnicity, GERD symptoms, family history of BE/cancer, alcohol, tobacco, dietary factors (fat intake, cereal fiber, calcium etc.) acid suppressive medications, biomarkers, genetic markers etc. Given this long list of possible risk factors, we propose to convene a panel of experts in a planning committee to serve as jury in order to determine which of the potential risk factors should be measured in a prospective study and how they should be measured. This would narrow and refine the list of risk factors for the prospective study.

There are a lot of purported risk factors to be evaluated. Thus, there is a long list (see below) of potential factors of risk in patients with BE associated with progression to dysplasia and/or cancer:

(1) Age (2) Gender (3) Ethnicity (4) Reflux symptoms (5) Smoking (6) Alcohol (7) Dietary fat (8) Dietary fiber, vegetables, fruits (9) Dietary calcium (10) Obesity (11) Family history of reflux, BE or esophageal cancer (12) Use of aspirin, non-steroidal anti-inflammatory drugs (13) Use of lower esophageal sphincter relaxing medications (14) Use of acid suppressive medications (15) Cholecystectomy (16) Anti-reflux surgery (fundoplication) (17) Length of BE (18) Hiatal hernia (19) Helicobacter pylori infection (20) Low-grade dysplasia on biopsy This proposal will make it feasible to narrow on the factors to be studied in the prospective trial.

Statistical Analysis Prevalence and incidence rates: We expect to increase the baseline-screened population from 1376 to more than 2500 individuals in the final study. The number of patients we expect to enroll each year will be calculated and the expected number of new prevalence cases using the 9.7% prevalence rate with the preliminary study data will be estimated. In the preliminary study to date 618 patients have had at least one follow-up visit. Including the remaining 700 patients from the preliminary study and assuming a 10% rate for loss to follow up, we will have 6100-7450 patient years of follow up. The expected number of new incidence cases using the 1.3% incidence rate with the preliminary study data will be estimated.