The relationship between length of Barrett s oesophagus mucosa and body mass index

Alimentary Pharmacology and Therapeutics The relationship between length of Barrett s oesophagus mucosa and body mass index J. Abdallah*, C. Maradey-Romero*, S. Lewis, A. Perzynski & R. Fass* *Division of Gastroenterology and Hepatology, The Esophageal and Swallowing Center, MetroHealth Medical Center, Case Western Reserve University, Cleveland, OH, USA. Center for Health Care Research and Policy, Case Western Reserve University at MetroHealth Medical Center, Cleveland, OH, USA. Correspondence to: Prof R. Fass, Division of Gastroenterology and Hepatology, Esophageal and Swallowing Center, Case Western Reserve University, MetroHealth Medical Center, 2500 MetroHealth Drive, Cleveland, OH 44109, USA. E-mail: Ronnie.fass@gmail.com Publication data Submitted 27 August 2014 First decision 11 September 2014 Resubmitted 23 September 2014 Accepted 26 September 2014 EV Pub Online 17 October 2014 This article was accepted for publication after full peer-review. SUMMARY Background Length of Barrett s oesophagus (BE) is a risk factor for oesophageal cancer. The underlying mechanisms that determine BE length are not fully known. Aim To determine if there is a correlation between obesity and length of BE. Methods Using a population-based study, 381 patients diagnosed with Barrett s oesophagus between 1999 and 2013 were included. Body mass index (BMI) at the time of BE diagnosis was calculated. Upper endoscopy reports were reviewed to obtain the length of BE. Spearman s correlation coefficient was performed to assess the strength of the relationship between Barrett s length and BMI. A multivariate logistic regression analysis was conducted to further examine the association between BMI and length of BE. Results The adjusted odds ratio for each five-point increase in BMI was 1.5 (95% CI 1.24 1.81, P < 0.001). The mean BMI was significantly higher in patients with long segment BE as compared to patients with short segment BE (32.7 vs. 30.3, P = 0.001). There was also a positive trend in long segment BE as patients entered into higher BMI categories (Z = 4.25, P < 0.001). There was a significant correlation between BMI and length of BE (r = 0.25, P < 0.0001). Conclusion The study demonstrated a correlation between BMI and the length of Barrett s oesophagus mucosa. Thus, increased BMI is associated with longer segment of Barrett s oesophagus. Aliment Pharmacol Ther 2014; 41: 137 144 137 doi:10.1111/apt.12991

J. Abdallah et al. INTRODUCTION Barrett s oesophagus (BE) is the displacement of squamocolumnar junction proximal to the gastroesophageal junction with histological evidence of specialised intestinal metaplasia on biopsy specimens. 1 BE develops in 6% to 12% of patients with chronic gastroesophageal reflux disease (GERD) related symptoms and predisposes patients to the development of oesophageal adenocarcinoma. 2 5 Although the risk of progression from BE without dysplasia to oesophageal adenocarcinoma is approximately 0.1 0.2% per year, patients diagnosed with BE have a 30 40-fold increased risk of developing oesophageal adenocarcinoma overtime as compared to subjects without BE. 6 The incidence and mortality from oesophageal adenocarcinoma have increased by more than five-fold in the past four decades. 6, 7 Caucasians and middle-aged men are more likely to be affected. 6 Despite the increase in frequency of performing upper endoscopies and thus a greater potential for an early detection, the prognosis of patients with oesophageal adenocarcinoma remains poor. A recent report put the 5-year survival rate from oesophageal adenocarcinoma at around 19%. 8 The main risk factors for the development of BE include male sex, Caucasian race, older age, the duration 9, 10 and severity of GERD, peptic stricture and obesity. Based on upper endoscopy, BE is currently classified into short segment BE (SSBE < 3 cm) and long segment BE (LSBE 3 cm). Within the BE population without dysplasia, LSBE carries a greater risk of progression to dysplasia and oesophageal adenocarcinoma than SSBE. 11 Factors linked to increased BE length include a larger hiatal hernia, reduced lower oesophageal sphincter basal pressure, evidence of dysplasia, increased severity and duration of GERD-related symptoms, lack of acid-suppressive therapy prior to initial BE diagnosis and degree, duration and rate of change in acid exposure along the distal oesophagus. 10 14 Overall, it has been demonstrated that BE is associated with obesity. 15 19 Obesity increases the risk of BE by two to three-fold. 19 Obesity has been shown to increase the risk of GERD, BE and oesophageal adenocarcinoma through several different mechanisms. Volume of abdominal fat, which is present either in the subcutaneous or visceral compartments may determine this risk. In particular, visceral fat has been shown to release several proinflammatory cytokines, including interleukin-6 and tumour necrosis factor, which are over expressed in BE. 18 20 Visceral fat also causes direct mechanical pressure on the stomach, increasing the intragastric pressure and consequently leading to increased frequency of transient lower oesophageal sphincter relaxation with subsequent reflux. 20 Obese patients also demonstrate an increase in gastroesophageal pressure gradients, prevalence of hiatal hernia, oesophageal motor and sensory abnormalities, and presence of various comorbidities such as obstructive sleep apnoea, all of which increase the risk of developing GERD and its adverse effects. 21 Thus far, only two studies evaluated the association between obesity and length of BE. Neither study was able to demonstrate a relationship between obesity and Barrett s oesophagus length. 19, 22 However, in both studies the relationship between obesity and Barrett s length was not the main outcome. In addition, one study 19 focused on male veterans only and the other on women only 22 and thus the results could not be generalisable. The aim of our study was to determine if there is a relationship between the length of Barrett s oesophagus and body mass index (BMI). Our hypothesis was that increased BMI is associated with longer segment of BE mucosa. Thus, increased BMI may contribute to the development of oesophageal adenocarcinoma through the generation of longer segment of BE. MATERIALS AND METHODS Patients Using Epic electronic medical records for data collection, we conducted a population-based study using clinical and administrative records of patients diagnosed with BE at MetroHealth Medical Center (MHMC) in Cleveland, Ohio between 1999 and 2013. The study was approved by the Institutional Review Board at MHMC. We searched for patients who were diagnosed with BE (ICD-9 code 530.85) and excluded patients with co-existing oesophageal cancer (ICD-9 code 150) at the time of diagnosis. Patients with oesophageal cancer were excluded because they are commonly diagnosed after developing progressive dysphagia with rapid weight loss. All patients included in this study had to have a diagnosis of GERD (ICD-9 code 530.81), report GERD-related symptoms and receive anti-reflux medications. Data collection We collected patients data regarding gender, race, age, date of BE diagnosis, comorbidities, smoking status and number of pack years, alcohol use, anti-reflux medications, body mass index, presence and size of hiatal hernia, length of BE mucosa at time of diagnosis and presence or absence of dysplasia. Basic demographics 138 Aliment Pharmacol Ther 2014; 41: 137-144

Barrett s oesophagus and BMI and date of diagnosis were made available on the data report and confirmed during chart review. The Charlson Comorbidity Index was calculated using certain comorbidities, such as chronic obstructive pulmonary disease, diabetes and chronic liver disease preceding BE, as well as age at BE diagnosis. Comorbidities were obtained by viewing the medical history, problem list and progress notes from primary care physicians. BMI was calculated using the following equation: BMI = [(weight in kilograms)/(height in metres) 2 ]. BMI was examined as a continuous variable as well as a categorical variable according to World Health Organization (WHO) criteria (<25, 25 29.99, 30 34.99, 35 39.99, >40). Height and weight measured in the out-patient setting were used. The BMI was calculated using averaged weight and height within 3 months of BE diagnosis to account for minor variations. Baseline weight in that time frame documented in progress notes for patients with cirrhosis, congestive heart failure and chronic renal failure were used to calculate the BMI. Tobacco and alcohol use was obtained by viewing patients social history, problem list, past medical history and progress notes. Pack years of smoking was obtained by reviewing progress notes. Data regarding the duration, frequency of administration and type of anti-reflux medications used prior to BE diagnosis, such as proton pump inhibitors (PPIs) and histamine type 2 receptors antagonists (H2RA), were obtained to estimate the duration and severity of GERD. Only medication data within the last month before BE diagnosis was used in this study, since type of anti-reflux medications and frequency of their use were variable. Upper endoscopy reports were used to help confirm the diagnosis of BE, determine BE mucosa length and the size of hiatal hernia (if present). The upper endoscopy report was used to corroborate the diagnosis of BE. All of the patients in the study had displacement of the squamocolumnar junction proximal to the oesophagogastric junction (EGJ). A hiatal hernia was considered present if the EGJ was displaced at least 1 cm proximal to the diaphragmatic hiatus. 12 The distance between the diaphragmatic constriction and the GEJ was measured and considered as the hiatal hernia length. In the pathology report, diagnosis of BE required the presence of intestinal metaplasia on oesophageal mucosal biopsy. Pathology reports were then reviewed by two expert raters to confirm proper diagnosis of BE. Patients with suspected BE on endoscopy with a biopsy showing fundic or cardiac type of columnar epithelium without intestinal metaplasia were excluded. Evidence and grade of dysplasia was also obtained from the pathology report. Patients were then classified into two categories, SSBE (<3 cm) and LSBE ( 3 cm). Statistical analysis Spearman s rank-order correlation coefficient was used to assess the strength of the linear relationship between Barrett s length and BMI. A P-value below 0.05 denotes statistical significance. A simple logistic regression model was used to determine whether a significant relationship exists between Barrett s length (where LSBE was modelled) and BMI. An odds ratio (OR) estimate along with a Wald confidence interval was obtained considering an increase in BMI of five units (consistent with the ranges of WHO BMI classes). A confidence interval excluding 1 indicates significance at the 0.05 level. A C-statistic was also generated for this model. A second logistic regression model was used to assess the relationship between Barrett s length and BMI after adjusting for additional covariates (e.g. gender, race, alcohol use). The OR estimates along with Wald confidence intervals were obtained for each covariate including BMI. The primary focus is the adjusted OR for BMI. A C-statistic was also generated for this model. The likelihood ratio test was used to compare the two nested logistic regression models. A P-value derived from the Chi-square statistic below 0.05 indicates that the logistic regression model with additional covariates was significantly better than the simple logistic regression model with BMI as the sole covariate. The v 2 and Wilcoxon Rank Sum tests were used to formally test for relationships between our dichotomised outcome and categorical and continuous covariates, respectively. The Cochrane-Armitage trend test was used to test for a trend across BMI classifications (in accordance with WHO criteria) for SSBE and LSBE. P-values less than 0.05 were considered statistically significant. RESULTS There were 392 patients initially identified with histologically confirmed BE. Five patients were excluded due to co-existing oesophageal cancer at the time of diagnosis. Two patients had recorded BE lengths of more than 30 cm and were excluded because of likely error in data entry. After a 99% inter-observer agreement by two expert raters, 381 patients were included in this study. There were NO 226 (60%) males and 155 (40%) females. Mean age at the time of diagnosis was 55.4 years (range 21 87). The mean Barrett s length was 3.6 cm (range 0.25 20 cm). Mean BMI was 31.4 kg/m 2 (range 17 56 kg/m 2 ). Of those included in this study, 55% were Aliment Pharmacol Ther 2014; 41: 137-144 139

J. Abdallah et al. smokers and the mean number of pack years (current/ past) was 17.5. Mean size of hiatal hernia was 1.9 cm. There was a significant correlation between the length of BE and BMI using Spearman s correlation coefficient (r = 0.25, P < 0.001). Figure 1 shows the distribution of BMI for both the long and short segments. Here, we see that the distribution appears higher for the long segment. The unadjusted OR for five unit increase in BMI was 1.4 (95% CI 1.16 1.64, P = 0.001). The C-statistic for this analysis (not shown) was 0.636. The adjusted OR for the same level of increase in BMI was 1.5 (95% CI 1.24 1.81, P < 0.001). The adjusted area under the curve (C-statistic, not shown) was 0.739. The difference in C-statistics between the models was significantly different (95% CI 0.042 0.160, P < 0.001). The likelihood ratio test comparing the two models also favoured the adjusted model (69.81, df = 14, P < 0.001). Figure 2 displays ORs for all of the variables in the adjusted model. Women were 1.6 times more likely to have LSBE than males, although with borderline significance (P = 0.053). When compared to Hispanics, Caucasians were three times more likely to have LSBE (OR 0.32, 95% CI 0.12 0.91, P = 0.033). The hiatal hernia was significantly longer in patients with LSBE than in those with SSBE (P = 0.003). Patients with dysplasia (P = 0.004) or tobacco smoking (P < 0.001) were also more likely to have LSBE (Figure 2). There were 200 patients with SSBE and 181 with LSBE. Table 1 summarises the different patient characteristics of the two groups. There were no significant differences BMI 50 40 30 20 Distribution of BMI by Barrett s Length Wilcoxon Rank Sum Test P-value < 0.001 + LSBE (n = 181) Barrett s Length + SSBE (n = 200) Figure 1 Notched box plots representing the distribution of body mass index by BE length (SSBE, short segment Barrett s oesophagus; LSBE, long segment Barrett s oesophagus). The median BMI was significantly higher in patients with LSBE as compared to those with SSBE (P < 0.001). between age, race or gender between both groups. There was a significantly higher mean BMI in the LSBE category as compared with SSBE (32.7 vs. 30.3, P = 0.001). Figure 3 illustrates the relationship between the different BMI categories and the length of BE. The Cochran-Armitage Trend test confirmed a positive trend, suggesting the probability that a patient was in the long segment Barrett s oesophagus group increased for those in higher BMI categories (Z = 4.25, P < 0.001). Hiatal hernia was documented more commonly in patients with LSBE than in those with SSBE (67% vs. 43% of patients, P < 0.001). There was a significant difference between mean length of hiatal hernia of patients with LSBE and those with SSBE (2.4 cm vs. 1.5 cm, P = 0.003). Low grade dysplasia was noted in 11% of the patients with LSBE as compared with 4% of the patients with SSBE at the time of diagnosis (P = 0.02). No patients were reported as having high grade dysplasia at the time of diagnosis. Past or current tobacco use was documented in 62% of the patients with LSBE vs. 49% of the patients with SSBE (P < 0.001). Alcohol use was more common in patients with LSBE than those with SSBE, although not statistically significant (P = 0.188). No significant difference was observed in regards to comorbidities of the two groups (P = 0.624). In addition, no significant difference was observed between the two groups in relation to duration, type or frequency of anti-- reflux medications used prior to the diagnosis of BE. DISCUSSION This population-based study suggests that there is a relationship between BMI and the length of BE mucosa, even when accounting for multiple confounders. It showed that patients with a higher BMI at the time of BE diagnosis were more likely to have both longer BE and LSBE. Furthermore, increase in BMI by five points is associated with a 1.5-fold increase in the odds of having LSBE. Thus, our study suggests that BMI is significantly associated with the length of BE and thus may be an important factor in determining the extent of Barrett s mucosa. Jacobson et al. as well as Stein et al. evaluated the association between BMI and BE and neither of these studies showed a correlation between BMI and the length of Barrett s mucosa. 19, 22 Jacobson et al. studied only women and relied on self-reported height and weight. Stein et al. reported about a small sample of male Caucasian patients. Due to a small sample size a type 2 error could not be excluded in either study. The larger, more diverse sample size in our study and the presence of more rigorous methodology likely explains 140 Aliment Pharmacol Ther 2014; 41: 137-144

Barrett s oesophagus and BMI Barrett s Length Odds Ratios with 95% Wald Confidence Intervals BMI: units = 5 Reference = LSBE GENDER: Female vs. Male RACE: African American vs. Caucasian RACE: Hispanic vs. Caucasian RACE: Other vs. Caucasian ALCOHOL ABUSE: Yes vs. No MEDICATION: PPI vs. H2 Blocker MEDICATION: PPI+H2 Blocker vs. H2 Blocker DYSPLASIA: Undetermined vs. No DYSPLASIA: Yes vs. No OR = 1 0 2 4 Odds Ratio 6 8 10 Figure 2 Odds ratios (with 95% Wald confidence limits) of demographics and clinical variables that are associated with Barrett s oesophagus (BMI, body mass index; PPI, proton pump inhibitor; H2 blocker, histamine 2 blocker). why our study found an association between BMI and the length of BE mucosa when prior studies did not. Furthermore, it is interesting to note that in the multivariate analysis, females were more likely to have an LSBE than men by 1.6-fold, albeit, this result had borderline statistical significance (P = 0.053). In contrast to our study, Falk et al. compared the features of Barrett s oesophagus between men and women and showed that the length of the BE segment in men was greater than that in women. 23 The difference in multivariate models is likely responsible for the difference between the two studies. More studies that compare Barrett s oesophagus characteristics between the two genders are needed. It is also interesting to note that Caucasians were three-fold more likely to have LSBE when compared to Hispanics. These results are similar to a study by Bersentes et al. who demonstrated a two-fold increase in the length of BE in Caucasians as compared to Hispanics. 24 Overall, patients with a higher BMI were more likely to have LSBE. In addition, patients with dysplasia at the time of BE diagnosis were four times more likely to have LSBE. Patients with a history of smoking or the presence of hiatal hernia were also more likely to have LSBE as compared with the SSBE group. These findings are consistent with several previous studies. 10 12 The duration, type, and frequency of anti-reflux medications were used as surrogates to predict the duration and severity of GERD. However, these variables were not shown to significantly impact the length of BE in our analysis. While increased duration and severity of GERD have been shown to increase the risk for LSBE, it is likely that because we used anti-reflux medications consumption as a surrogate marker for GERD severity in our study, no correlation was found. BMI was also categorised based on WHO classification to determine if a trend exists in relation to the length of BE and a higher BMI category. Our study indicates that there is a statistically significant positive relationship between patients categorised into higher BMI groups and the presence of LSBE. There are no Aliment Pharmacol Ther 2014; 41: 137-144 141

J. Abdallah et al. Table 1 Demographical, endoscopic, histological and anthropometric features of patients with short segment Barrett s oesophagus vs. to those with long segment Barrett s oesophagus Length of Barrett s oesophagus Variable SSBE (n = 200, 52%) LSBE (n = 181, 48%) P-value Age in years (mean s.d.) 55.5 13.2 55.2 11.4 0.781 Male, n (%) 124 (62) 102 (56) 0.263 Caucasian, n (%) 146 (73) 145 (80) 0.308 Frequency of anti-reflux medication prior to BE, n (%) Daily 175 (88) 164 (91) 0.333 Twice daily 25 (12) 17 (9) 0.333 Mean duration of anti-reflux medication prior to BE (years) 3.1 2.8 0.321 Type of anti-reflux medication used prior to BE, n (%) PPI 169 (85) 153 (85) 0.643 H2 blocker 8 (4) 10 (5) 0.644 Both 23 (11) 18 (10) 0.663 Smoking, n (%) 98 (49) 109 (60) 0.001 Mean BE length (cm) 1.5 6.0 0.001 Mean hiatal hernia length (cm) 1.5 2.4 0.003 Mean BMI (kg/m 2 ) 30.3 32.7 0.001 WHO BMI classification, n (%) <25 (reference) 32 (16) 11 (6) <0.001 25 29.99 (overweight) 70 (35) 38 (21) <0.001 30 34.99 (moderately obese) 57 (29) 70 (38) <0.001 35 39.99 (severely obese) 22 (11) 43 (24) <0.001 40 (very severely obese) 19 (9) 19 (11) <0.001 LG dysplasia, n (%) 8 (4) 20 (11) 0.026 Charlson comorbidity index 2.0 2.0 0.624 SSBE, short segment Barrett s oesophagus (Barrett s length <3 cm); LSBE, long segment Barrett s oesophagus (Barrett s length 3 cm); BMI, body mass index; LG, low grade. other studies that provided similar insight into the relationship between obesity and Barrett s oesophagus. Thus, these findings suggest that it is possible that the relationship between obesity and oesophageal adenocarcinoma is driven by the generation of longer segments of Barrett s mucosa. This study provides an identifiable and modifiable contributor to the length of BE. As mentioned earlier, 1.0 BMI Group and Barrett s Length LSBE ( 3 cm) SSBE (< 3 cm) 0.8 0.74 0.65 0.66 Proportion 0.6 0.4 0.55 0.45 0.35 0.34 0.5 0.5 0.26 0.2 0.0 BMI < 25 25 BMI < 30 30 BMI < 35 35 BMI < 40 BMI 40 (n = 43) (n = 108) (n = 127) (n = 65) (n = 38) BMI Group Figure 3 Comparison of proportion of patients with LSBE vs. those with SSBE in each BMI category. 142 Aliment Pharmacol Ther 2014; 41: 137-144

Barrett s oesophagus and BMI current evidence suggests that within the BE population without dysplasia, LSBE carries a greater risk of progression to dysplasia and oesophageal adenocarcinoma as compared with SSBE. The critical factors that contribute to the length of BE needs to be identified to prevent BE mucosa from progressing to dysplasia and oesophageal adenocarcinoma. From this study, it can be concluded that one of these critical factors is obesity. As a result, it is important to evaluate in a prospective fashion the effect of weight loss programs in preventing the development of LSBE in obese patients with GERD. A few limitations of this study need to be addressed. First, since this was a population-based study, there was not a standardised method in obtaining information regarding endoscopic findings, patient characteristics or variables. Although each upper endoscopy was performed by a single endoscopist, there were many different endoscopists performing the procedures throughout the 14 year study period. Practitioner variations in measuring the length of BE as well as other endoscopic landmarks were also possible. However, there is no reason to suspect that any bias existed among the different endoscopists. Given the design of this study, we were unable to measure or account for lower oesophageal sphincter basal pressure or level of acid exposure in the distal oesophagus; both of which have been shown to affect the length of BE. In addition, during chart review, it was rather difficult to obtain information regarding the duration and severity of GERD. This was due to lack of both subjective and objective data used to describe GERD in patients history. The duration and severity of GERD were commonly underreported in patients charts and in their upper endoscopy reports prior to the diagnosis of BE. Therefore, anti-reflux medications prescribed prior to BE diagnosis were used as surrogate markers for the duration and severity of GERD. However, patients could have been taking over the counter anti-reflux medications prior to the first prescription or had these medications prescribed initially or changed by providers outside the health system in which data were collected. Finally, given the nature of our study, the only parameter that could be used to classify obesity was BMI. However, BMI is a combined measure of muscle, fat and bone mass. Consequently, BMI may not be an accurate parameter in determining patient s true body fat composition. BMI, for example, does not distinguish between patients with high fat mass and those with high muscle 25, 26 mass. Previous studies suggested that waist-to-hip ratio or waist circumference are more accurate parameters to assess body fat composition. 15, 27 However, it is still unclear which of the proposed parameters (BMI, waist-to-hip ratio, waist circumference and bioelectrical impedance) provides better prediction of BE length. In summary, we demonstrated for the first time a significant relationship between BMI and the length of BE mucosa. This relationship became stronger when adjusting for important selected covariates. This study suggests that increased BMI is associated with longer segments of Barrett s oesophagus. In addition, a modifiable contributor to the length of BE has been identified. This is very important, because both length of Barrett s mucosa and increased BMI have been shown to increase the risk for oesophageal adenocarcinoma. Lastly, the study also suggests that the positive relationship between obesity and oesophageal adenocarcinoma may be driven by longer segments of Barrett s mucosa. Further studies demonstrating a similar relationship between BMI and BE length are needed. AUTHORSHIP Guarantor of the article: Ronnie Fass. Author contributions: Jason Abdallah collected the data; Jason Abdallah, Carla Maradey-Romero and Ronnie Fass analysed the data; Steven Lewis and Adam Perzysnki performed statistical analysis; Jason Abdallah, Carla Maradey-Romero, Steven Lewis and Ronnie Fass wrote the paper; Ronnie Fass designed the research study. All authors approved the final version of the article, including the authorship list. ACKNOWLEDGEMENT Declaration of personal and funding interests: None. REFERENCES 1. Wang KK, Sampliner RE. Updated guidelines 2008 for the diagnosis, surveillance and therapy of Barrett s esophagus. Am J Gastroenterol 2008; 103: 788 97. 2. Cameron AJ, Ott BJ, Payne WS. The incidence of adenocarcinoma in columnar-lined (Barrett s) esophagus. N Engl J Med 1985; 313: 857 9. 3. Winters C Jr, Spurling TJ, Chobanian SJ, et al. Barrett s esophagus. A prevalent, occult complication of gastroesophageal reflux disease. Gastroenterology 1987; 92: 118 24. Aliment Pharmacol Ther 2014; 41: 137-144 143

J. Abdallah et al. 4. Mann NS, Tsai MF, Nair PK. Barrett s esophagus in patients with symptomatic reflux esophagitis. Am J Gastroenterol 1989; 84: 1494 6. 5. Cameron AJ, Kamath PS, Carpenter HA. Prevalence of Barrett s esophagus and intestinal metaplasia at the esophagogastric junction [abstract]. Gastroenterology 1997; 112: AB82. 6. Cook MB, Chow WH, Devesa SS. Oesophageal cancer incidence in the United States by race, sex, and histologic type, 1977-2005. Br J Cancer 2009; 101: 855 9. 7. Steevens J, Botterweck AA, Dirx MJ, van den Brandt PA, Schouten LJ. Trends in incidence of oesophageal and stomach cancer subtypes in Europe. Eur J Gastroenterol Hepatol 2010; 22: 669 78. 8. Siegel R, Naishadham D, Jemal A. Cancer statistics, 2012. CA Cancer J Clin 2012; 62: 10 29. 9. Oberg S, DeMeester TR, Peters JH, et al. The extent of Barrett s esophagus depends on the status of the lower esophageal sphincter and the degree of esophageal acid exposure. J Thorac Cardiovasc Surg 1999; 117: 572 80. 10. Fass R, Hell RW, Garewal HS, et al. Correlation of oesophageal acid exposure with Barrett s oesophagus length. Gut 2001; 48: 310 3. 11. Tharalson EF, Martinez SD, Garewal HS, et al. Relationship between rate of change in acid exposure along the esophagus and length of Barrett s epithelium. Am J Gastroenterol 2002; 97: 851 6. 12. Wakelin DE, Al-Mutawa T, Wendel C, Green C, Garewal HS, Fass R. A predictive model for length of Barrett s esophagus with hiatal hernia length and duration of esophageal acid exposure. Gastrointest Endosc 2003; 58: 350 5. 13. Dickman R, Green C, Chey WD, et al. Clinical predictors of Barrett s esophagus length. Gastrointest Endosc 2005; 62: 675 81. 14. Weston AP, Badr AS, Hassanein RS. Prospective multivariate analysis of clinical, endoscopic, and histological factors predictive of the development of Barrett s multifocal high-grade dysplasia or adenocarcinoma. Am J Gastroenterol 1999; 94: 3413 9. 15. Corley DA, Kubo A, Levin TR, et al. Abdominal obesity and body mass index as risk factors for Barrett s esophagus. Gastroenterology 2007; 133: 34 41; quiz 311. 16. Kamat P, Wen S, Morris J, Anandasabapathy S. Exploring the association between elevated body mass index and Barrett s esophagus: a systematic review and meta-analysis. Ann Thorac Surg 2009; 87: 655 62. 17. Steevens J, Schouten LJ, Driessen AL, et al. A prospective cohort study on overweight, smoking, alcohol consumption, and risk of Barrett s esophagus. Cancer Epidemiol Biomarkers Prev 2011; 20: 345 58. 18. Edelstein ZR, Farrow DC, Bronner MP, Rosen SN, Vaughan TL. Central adiposity and risk of Barrett s esophagus. Gastroenterology 2007; 133: 403 11. 19. Stein DJ, El-Serag HB, Kuczynski J, Kramer JR, Sampliner RE. The association of body mass index with Barrett s oesophagus. Aliment Pharmacol Ther 2005; 22: 1005 10. 20. Ryan AM, Healy LA, Power DG, et al. Barrett esophagus: prevalence of central adiposity, metabolic syndrome, and a proinflammatory state. Ann Surg 2008; 247: 909 15. 21. Fass R. The pathophysiological mechanisms of GERD in the obese patient. Dig Dis Sci 2008; 53: 2300 6. 22. Jacobson BC, Chan AT, Giovannucci EL, Fuchs CS. Body mass index and Barrett s oesophagus in women. Gut 2009; 58: 1460 6. 23. Falk GW, Thota PN, Richter JE, Connor JT, Wachsberger DM. Barrett s esophagus in women: demographic features and progression to high-grade dysplasia and cancer. Clin Gastroenterol Hepatol 2005; 3: 1089 94. 24. Bersentes K, Fass R, Padda S, Johnson C, Sampliner RE. Prevalence of Barrett s esophagus in Hispanics is similar to Caucasians. Dig Dis Sci 1998; 43: 1038 41. 25. Nevill AM, Stewart AD, Olds T, Holder R. Relationship between adiposity and body size reveals limitations of BMI. Am J Phys Anthropol 2006; 129: 151 6. 26. Prentice AM, Jebb SA. Beyond body mass index. Obes Rev 2001; 2: 141 7. 27. Thrift AP, Kramer JR, Alsarraj A, El- Serag HB. Fat mass by bioelectrical impedance analysis is not associated with increased risk of Barrett esophagus. J Clin Gastroenterol 2014; 48: 218 23. 144 Aliment Pharmacol Ther 2014; 41: 137-144