Tobacco Smoking Increases the Risk of High-Grade Dysplasia and Cancer Among Patients With Barrett s Esophagus

Transcription

1 GASTROENTEROLOGY 2012;142: CLINCAL ALIMENTARY TRACT Tobacco Smoking Increases the Risk of High-Grade Dysplasia and Cancer Among Patients With Barrett s Esophagus HELEN G. COLEMAN,* SHIVARAM BHAT,* BRIAN T. JOHNSTON, DAMIAN MCMANUS, ANNA T. GAVIN,*, and LIAM J. MURRAY* *Cancer Epidemiology and Health Services Research Group, Centre for Public Health, Queen s University Belfast, Northern Ireland; Belfast Health and Social Care Trust, Belfast, Northern Ireland; Northern Ireland Cancer Registry, Belfast, Northern Ireland See Covering the Cover synopsis on page 191. BACKGROUND & AIMS: Esophageal adenocarcinoma arises from Barrett s esophagus (BE); patients with this cancer have a poor prognosis. Identification of modifiable lifestyle factors that affect the risk of progression from BE to esophageal adenocarcinoma might prevent its development. We investigated associations among body size, smoking, and alcohol use with progression of BE to neoplasia. METHODS: We analyzed data from patients with BE identified from the population-based Northern Ireland BE register, diagnosed between 1993 and 2005 with specialized intestinal metaplasia (n 3167). Data on clinical, demographic, and lifestyle factors related to diagnosis of BE were collected from hospital case notes. We used the Northern Ireland Cancer Registry to identify which of these patients later developed esophageal adenocarcinoma, adenocarcinomas of the gastric cardia, or esophageal high-grade dysplasia. Cox proportional hazards models were used to associate lifestyle factors with risk of progression. RESULTS: By December 31, 2008, 117 of the patients with BE developed esophageal high-grade dysplasia or adenocarcinomas of the esophagus or gastric cardia. Current tobacco smoking was significantly associated with an increased risk of progression (hazard ratio 2.03; 95% confidence interval, ) compared with never smoking, and across all strata of smoking intensity. Alcohol consumption was not related to risk of progression. Measures of body size were infrequently reported in endoscopy reports, and body size was not associated with risk of progression. CONCLUSIONS: Smoking tobacco increases the risk of progression to cancer or high-grade dysplasia 2-fold among patients with BE, compared with patients with BE that have never smoked. Smoking cessation strategies should be considered for patients with BE. Keywords: Esophageal Cancer; Tumor Progression; Risk Factor; Cigarette. Esophageal adenocarcinoma (EAC) has received much attention recently because of the rising incidence rates in developed countries and its poor prognosis. 1 3 EAC originates from the premalignant condition Barrett s esophagus (BE), presumably progressing through lowand high-grade dysplasia (HGD). 4,5 However, progression along this pathway is uncommon, and the vast majority of BE patients never develop EAC or HGD. 6 Factors distinguishing the small minority of BE patients that do progress from nonprogressors remain largely elusive. Much research has concentrated on biomarkers for progression in order to stratify patients for surveillance and/or treatment as appropriate. 7 9 Identification of modifiable lifestyle factors that influence neoplastic progression may provide an additional cost-effective method of alleviating future cancer burden in this patient group. Several population-based case-control studies and a few prospective cohort studies have identified associations between lifestyle factors and risk of EAC in the general population. Findings from such studies have highlighted excess weight as one of the strongest risk factors for EAC. 18 The Barrett s Esophagus and Esophageal Adenocarcinoma (BEACON) Consortium have produced pooled analyses on cigarette smoking and observed a significant 2-fold increased risk of EAC in ever smokers compared with never smokers. 19 Conversely, alcohol intake has been consistently refuted as a risk factor for EAC, in contrast to its known direct association with esophageal squamous cell carcinoma. 20,21 Only 2 prospective studies published to date have investigated lifestyle factors and risk of progression from BE to EAC, 15,22 both of which have been limited by relatively small case numbers or inability to adjust for known confounders in their analyses. Therefore, further evidence is required to inform clinicians of appropriate lifestyle advice to offer to BE patients. The aim of this study was to identify modifiable lifestyle factors associated with risk of progression to EAC or HGD in one of the largest population-based cohorts of BE patients worldwide. Abbreviation used in this paper: BE, Barrett esophagus; BMI, body mass index; CI, confidence interval; EAC, esophageal adenocarcinoma; HGD, high-grade dysplasia; HR, hazard ratio; SIM, specialized intestinal metaplasia by the AGA Institute /$36.00 doi: /j.gastro

2 234 COLEMAN ET AL GASTROENTEROLOGY Vol. 142, No. 2 Methods Data Collection Patients were identified from the Northern Ireland BE register, which is a population-based register of all adults diagnosed with columnar-lined epithelium of the esophagus between 1993 and 2005 in Northern Ireland. 23,24 Patients within the Northern Ireland BE register were purposefully targeted for inclusion in the current study if they had ever had a diagnosis of specialized intestinal metaplasia (SIM). A total of 4717 BE patients have been diagnosed with SIM in the entire Northern Ireland BE register, and the 3167 BE patients included in this study did not differ from overall BE patients with SIM in terms of age or sex. The hospital notes of the BE patients (including those who had subsequently died) were reviewed by 1 of 3 trained data abstractors from the Northern Ireland Cancer Registry, who used a standardized electronic proforma (Microsoft Access; Microsoft, Redmond, WA) to extract data from the records corresponding to the patient s initial (index) diagnosis of BE. Data were extracted on a variety of clinical and demographic aspects, such as age, sex, Barrett s segment length, symptoms, and treatment regimen. Reports were retrospectively reviewed between March 2005 and March Tumour Verification Officers were instructed to record information on body weight, height, smoking status, and alcohol intake from the hospital report relating to the index diagnosis of BE. If both weight and height were recorded, body mass index (BMI) was calculated by dividing an individual s weight in kg by their height in m 2. Information collected on smoking status included current, former, unknown or never smoking of cigarettes, pipe tobacco, or cigars. For cigar and pipe smoking, patients reported as unknown were reclassified as nonsmokers because these are relatively rare exposures. Where available, information was also extracted on the number of cigarettes smoked and the duration of smoking. For former smokers, this was calculated as the difference between the ages at which an individual had started and stopped smoking; for current smokers, smoking duration was calculated as age at index BE diagnosis minus the age at smoking initiation. Information was also extracted on alcohol use relating to the number of units consumed per week and alcohol type, if available. Confirmation of Outcome Diagnosis Information on the linkage and verification of EAC and HGD cases in this cohort of BE patients have been published elsewhere. 23 Briefly, the Northern Ireland BE register was linked to the Northern Ireland Cancer Registry to identify patients that progressed to incident adenocarcinomas or unspecified malignancies of the esophagus and gastric cardia adenocarcinomas diagnosed up to December 31, Gastric cardia adenocarcinomas were included as outcomes because of the acknowledged difficulty classifying tumors of the esophagogastric junction and, because these tumors are arising in BE patients, they are most likely to be EACs. 25,26 Indeed, in the latest version of TNM classification, these would now be classified as esophageal in origin. 27 Squamous cell carcinomas of the esophagus were not included as outcomes. Cases of HGD were retrieved using electronic and manual review of all esophageal biopsy reports within Northern Ireland up to December 31, Cancers or HGD cases diagnosed within 6 months of their BE diagnosis were excluded because these were likely to be prevalent cases at the time of index biopsy. Deaths up until December 31, 2008 were identified by matching patient details with information from the General Registrar s Office in Northern Ireland. Statistical Analysis Data was cleaned and analyzed using Intercooled STATA Version 11 (StataCorp, College Station, TX). Continuous and categorical variables were compared between BE patients who progressed to cancer or HGD and BE patients who did not progress using independent t tests and 2 tests, respectively. Cox proportional hazards models were applied to assess risk of progression from BE to cancer or HGD according to body size, smoking, or alcohol variables. Assumptions for Cox proportional hazards models were checked by visual inspection of Kaplan Meier plots. Both unadjusted and adjusted models were applied, using a censor date as the date of death, date of cancer or HGD diagnosis or December 31, In adjusted models, included confounders were age, sex, presence of low-grade dysplasia, socioeconomic status (reflected by area-based income deprivation quintiles), reflux symptoms, and Barrett s segment length, because these are hypothesized risk factors for EAC and may impact lifestyle factors. Tests for trends across categories of lifestyle risk factors were performed after exclusion of individuals with unknown status. Stratified analyses were conducted for cancer-only outcomes by low-grade dysplasia status at BE diagnosis, for long-segment BE patients, and by sex categories based on an a priori hypotheses that lifestyle factors may interact with these factors to differentially influence risk of progression. Tests for interactions were conducted using the likelihood ratio test. In addition, sensitivity analyses excluding outcomes diagnosed within 1 year of follow-up were conducted to minimize the likelihood of reverse causation, whereby subclinical disease may have altered an individual s body size, smoking, or alcohol habits. Results During a total of 23,692 person-years of follow-up and a mean follow-up period of 7.5 years (range, years), 117 of 3167 BE patients included in this study had progressed to develop HGD or cancer of the esophagus or gastric cardia (70 adenocarcinomas or unspecified cancers of the esophagus, 10 adenocarcinomas of the gastric cardia, and 37 esophageal HGD diagnoses). The mean ( SD) lag time from BE diagnosis to cancer or HGD diagnosis was 4.5 ( 3.2) years. Descriptive characteristics of BE patients at their index diagnosis are displayed in Table 1. Progressors (BE patients who progressed to develop cancer or HGD) did not differ from nonprogressors (BE patients who did not progress) with respect to mean age at BE diagnosis, although progressors were less likely to be aged 70 years old. Progressors were more likely to be male, have longsegment BE, and have indefinite or low-grade dysplasia at their index diagnosis than nonprogressors, although segment length was not recorded in hospital case notes for approximately half of the cohort. Neither socioeconomic status nor the recording of the presence of reflux symptoms differed between progressors and nonprogressors. Smoking status was commonly reported, with information available for 90% of BE patients (Table 2), although smoking intensity and duration were less frequently noted. A positive association was evident for cigarette

3 February 2012 LIFESTYLE AND BARRETT NEOPLASTIC PROGRESSION 235 Table 1. Descriptive Characteristics of Patients at Their Index Barrett s Esophagus Diagnosis Characteristics (n 3050) Progressors (n 117) P value Age at BE diagnosis (y), mean SD Age group (y), n (%) (6.4) 4 (3.4) (14.3) 12 (10.2) (22.6) 27 (23.1) (24.9) 47 (40.2) (21.2) 22 (18.8) (10.6) 5 (4.3).003 Sex, n (%).001 Male 1898 (62.2) 91 (77.8) Female 1152 (37.8) 26 (22.2) Barrett segment length,.006 n (%) Short, 3cm 276 (9.0) 1 (0.8) Long, 3cm 1274 (41.8) 58 (49.6) Unknown 1500 (49.2) 58 (49.6) Presence of dysplasia, a.001 n (%) Yes 184 (6.0) 32 (27.4) No 2626 (86.1) 72 (61.5) Unknown 240 (7.9) 13 (11.1) Income deprivation.50 quintile, n (%) I (Most deprived) 644 (21.1) 25 (21.4) II 603 (19.8) 30 (25.6) III 536 (17.6) 19 (16.2) IV 522 (17.1) 16 (13.7) V (Least deprived) 466 (15.3) 20 (17.1) Unknown 279 (9.1) 7 (6.0) Presence of reflux.75 symptoms, n (%) Yes 1641 (53.8) 65 (55.6) No 637 (20.9) 26 (22.2) Unknown 772 (25.3) 26 (22.2) a Low grade or indefinite dysplasia. smoking status and risk of progression, with elevated risks for former smokers (hazard ratio [HR] 1.53; 95% confidence interval [CI]: ) and for current smokers (HR 1.83; 95% CI: ), after adjustment for confounders. Smoking intensity, measured as the number of cigarettes smoked per day, was known for just over half of smokers in this cohort, and findings illustrated no additional increase in progression risk for heavy ( 20 cigarettes/day) compared with lighter smokers ( 20 cigarettes/day). As shown in Table 2, smoking a pipe was also associated with an increased risk of progression after adjustment for potential confounders (HR 2.18; 95% CI: ). Cigar smoking did not influence progression risk but was an extremely rare exposure in these BE patients, with 1% of patients reported as cigar smokers (data not shown). Overall, current smoking of tobacco in any format was associated with a significantly doubled risk of progression to cancer or HGD compared with never smokers (HR 2.07; 95% CI: ) (Figure 1). As expected, mean ( SD) smoking duration was longer in current smokers ( years) compared with former smokers ( years) (P.001). However, duration of smoking was not significantly associated with the risk of progression, although the number of years smoked was known for only 239 individuals, representing only 17% of smokers. Male BE patients in this cohort were more likely to have ever smoked tobacco compared with females (59.2% vs 32.6%, respectively; P.001); however, the test for interaction between sex and smoking status and disease progression risk was not significant (P.39). Although data on alcohol intake were not available for up to 30% of BE patients, there were no indications of differential under-reporting between progressors and nonprogressors. As shown in Table 2, drinking 10 units of alcohol per week was not associated with the risk of progression compared with abstaining from alcohol (HR 1.04; 95% CI: ), nor was lighter consumption of alcohol. In addition, intake of different alcohol types, including wine, beer, and spirits, was not associated with progression risk (data not shown), although this information was only available for approximately 15% of BE patients. These null associations did not change across strata of sex (data not shown). Few patients had both their height and weight recorded, and reporting was differential, with progressors more likely to have had their height and weight recorded compared with nonprogressors (30.8% vs 21.9%; P.03). An exploratory analysis was conducted in which neither height nor weight was associated with neoplastic progression in BE patients (Table 2), nor were any relationships observed with height and weight in sex-specific analyses (data not shown). For BE patients whose BMI could be calculated, 75.0% of progressors and 68.0% of nonprogressors were classified as overweight or obese (P.38). Further stratified analysis was conducted including BE patients who progressed to cancer, excluding HGD outcomes. Results from this analysis were largely similar to those observed for all progressors, and indeed direct associations with smoking exposures were strengthened (Table 2). We also sought to explore the association between tobacco smoking and progression risk for BE patients diagnosed with indefinite or low-grade dysplasia at the outset of the study (Table 3). In this analysis, associations were strengthened and remained significant for dysplastic BE patients; however, they were attenuated in analysis restricted to nondysplastic BE patients only. As shown in Table 4, associations between tobacco smoking and neoplastic progression risk also remained in stratified analyses of long-segment BE patients. However, formal tests for interactions were not significant between smoking status and dysplasia status (P.45) or BE length (P.51). Additional sensitivity analyses were performed restricting included patients to those with at least 1 year of follow-up. This reduced the number of progressors and nonprogressors to 102 and 3007, respectively, who contributed to a total 23,647 person-years of follow-up. All previously observed null or significant associations between risk of progression

4 Table 2. Lifestyle Factors and Risk of Neoplastic Progression in Barrett s Esophagus Risk factors (n 3050) All progressors (n 117) Unadjusted HR (95% CI) P trend Adjusted a HR (95% CI) P trend Cancer progressors (n 80) Adjusted a HR (95% CI) Height, cm b ( ) ( ) ( ).68 per 10-cm increase per 10-cm increase per 10-cm increase Weight, kg c ( ) per 5-kg increase ( ) per 5-kg increase ( ) per 5-kg increase Cigarette smoking status Never 1420 (46.6) 41 (35.0) (33.8) 1.01 Former 657 (21.5) 34 (29.1) 1.85 ( ) 1.53 ( ) 25 (31.2) 1.73 ( ) Current 666 (21.8) 35 (29.9) 1.85 ( ) ( ) (28.8) 2.02 ( ) Unknown 307 (10.1) 7 (6.0) 0.83 ( ) 0.82 ( ) 5 (6.2) 0.89 ( ) No. of cigarettes smoked Never smoker 1420 (46.5) 41 (35.0) (33.8) /day 301 (9.9) 17 (14.5) 2.05 ( ) 2.12 ( ) 12 (15.0) 2.45 ( ) 20/day 591 (19.4) 32 (27.4) 1.91 ( ) ( ) (30.0) 1.96 ( ) Unknown 738 (24.2) 27 (23.1) 1.31 ( ) 1.21 ( ) 17 (21.2) 1.16 ( ) Pipe smoker.005 No/unknown 2989 (98.0) 108 (92.3) (90.0) 1 Yes 61 (2.0) 9 (7.7) 3.36 ( ) ( ).02 8 (10.0) 2.95 ( ) Any smoking status d Never 1409 (46.2) 40 (34.2) (32.5) Former 625 (20.5) 30 (25.6) 1.76 ( ) 1.45 ( ) 22 (27.5) 1.66 ( ) Current 718 (23.5) 43 (36.8) 2.09 ( ) ( ) (36.2) 2.29 ( ) Unknown 298 (9.8) 4 (3.2) 0.51 ( ) 0.53 ( ) 3 (3.8) 0.62 ( ) Years smoked e ( ) ( ) ( ).38 per 5 years per 5 years per 5 years Alcohol intake.19 None 899 (29.5) 35 (29.9) (32.5) 1 10 units/week 708 (23.2) 28 (23.9) 0.95 ( ) 0.94 ( ) 17 (21.2) 0.81 ( ) 10 units/week 531 (17.4) 25 (21.4) 1.17 ( ) ( ) (17.5) 0.82 ( ) Unknown 912 (29.9) 29 (24.8) 0.85 ( ) 0.89 ( ) 23 (28.8) 1.01 ( ) NOTE. Descriptive data are presented as n (%) for categorical variables and mean SD for continuous variables. CI, confidence interval; HR, hazard ratio. a Adjustments: age groups (16 40, 40 50, 50 60, 60 70, 70 80, and 80 years), sex (male, female), presence of low-grade dysplasia (yes, no, unknown), income deprivation quintile (I, II, III, IV, V, and unknown), Barrett segment length (long/short or unknown) and presence of reflux symptoms (yes, no, unknown). b Information on height available for 28% of BE patients. c Information on weight available for 58% of BE patients. d Includes cigarette, pipe, and cigar smoking status combined. e Known for 224 nonprogressors and 15 HGD and cancer progressors that comprised of 101 former and 138 current smokers. P trend 236 COLEMAN ET AL GASTROENTEROLOGY Vol. 142, No. 2

5 February 2012 LIFESTYLE AND BARRETT NEOPLASTIC PROGRESSION 237 Figure 1. Kaplan Meier plot showing the proportion of BE patients who progressed to HGD or EAC by smoking status. and body size, tobacco smoking, and alcohol consumption remained unchanged in these analyses. Discussion In this population-based study, which sought to investigate the association between lifestyle exposures among BE patients and their subsequent risk of esophageal or gastric cardia adenocarcinomas or esophageal HGD, tobacco smoking emerged as the strongest risk factor for progression. Alcohol consumption and body size were not related to neoplastic progression risk in this cohort of BE patients. BE patients who currently smoke tobacco had a significant 2-fold increased risk of progressing to esophageal or gastric cardia adenocarcinoma or esophageal HGD. The magnitude of this association is similar to that seen in case-control studies of EAC risk. 10,14,19 Recent reports from smaller prospective studies have shown conflicting results in relation to smoking and progression risk. 15,22 A study from Olmsted County, Minnesota, USA, that included 19 progressors from BE to EAC, illustrated a direct association with progression risk only for former, and not current smokers, in unadjusted analyses. 15 Contrastingly, Sikkema et al 22 did not identify any significant associations between current or former smoking and progression risk in 713 BE patients in The Netherlands, among whom 26 progressed to HGD or EAC. In the current study, former smokers risk of progression remained considerably higher than that of never smokers, although it should be noted that average duration of smoking for former smokers was still in excess of 30 years. In addition, the risk of progression remained elevated regardless of smoking intensity; suggesting that merely reducing the number of cigarettes smoked per day may not consequently reduce the risk of progression in BE patients. The significant associations between tobacco smoking and neoplastic progression remained in stratified analyses of long-segment BE patients, indicating that these findings are not a result of misclassification of patients with SIM Table 3. Lifestyle Factors and Risk of Neoplastic Progression in Barrett s Esophagus Stratified by Dysplasia Status Dysplastic (indefinite/low-grade) BE only Nondysplastic BE only Risk factors (n 184) Progressors (n 32) Adjusted a HR (95% CI) P trend (n 2626) Progressors (n 72) Adjusted a HR (95% CI) Cigarette smoking status Never 93 (50.6) 9 (28.1) (46.8) 28 (38.9) 1.00 Former 33 (17.9) 10 (31.3) 3.79 ( ) 553 (21.0) 20 (27.8) 1.23 ( ) Current 42 (22.8) 13 (40.6) 3.23 ( ) 577 (22.0) 19 (26.4) 1.37 ( ) Unknown 16 (8.7) 0 (0.0) 268 (10.2) 5 (6.9) 0.81 ( ) No. of cigarettes smoked Never smoker 93 (50.5) 9 (28.1) (46.8) 28 (38.9) /day 13 (17.9) 7 (21.9) 5.63 ( ) 263 (10.0) 7 (9.7) 1.14 ( ) 20/day 41 (22.3) 12 (37.5) 2.72 ( ) 498 (19.0) 20 (27.8) 1.46 ( ) Unknown 37 (20.1) 4 (12.5) 1.34 ( ) 637 (24.2) 17 (23.6) 1.04 ( ) Pipe smoker No/unknown 179 (97.3) 31 (96.9) (98.1) 65 (90.3) 1.00 Yes 5 (2.7) 1 (3.1) 2.12 ( ) 51 (1.9) 7 (9.7) 3.36 ( ) Any smoking status b Never 92 (50.0) 9 (28.1) (46.4) 27 (37.5) 1.00 Former 33 (17.9) 10 (31.3) 3.69 ( ) 524 (20.0) 16 (22.2) 1.10 ( ) Current 45 (24.5) 13 (40.6) 2.91 ( ) 623 (23.7) 26 (36.1) 1.69 ( ) Unknown 14 (7.6) 0 (0.0) 261 (9.9) 3 (4.2) 0.52 ( ) NOTE. Descriptive data are presented as n (%) for categorical variables. CI, confidence interval; HR, hazard ratio. a Adjustments: age groups (16 40, 40 50, 50 60, 60 70, 70 80, and 80 years), sex (male, female), income deprivation quintile (I, II, III, IV, V, and unknown), Barrett segment length (long/short or unknown), and presence of reflux symptoms (yes, no, unknown). b Includes cigarette, pipe, and cigar smoking status combined. P trend

6 238 COLEMAN ET AL GASTROENTEROLOGY Vol. 142, No. 2 Table 4. Lifestyle Factors and Risk of Neoplastic Progression in Barrett Esophagus Stratified by Barrett s Esophagus Segment Length Long-segment BE only Risk factors (n 1274) Progressors (n 58) Adjusted a HR (95% CI) P trend Cigarette smoking status.01 Never 610 (47.9) 18 (31.0) 1.00 Former 283 (22.2) 20 (34.5) 2.04 ( ) Current 268 (21.0) 17 (29.3) 2.38 ( ) Unknown 113 (8.9) 3 (5.2) 0.77 ( ) No. of cigarettes smoked.02 Never smoker 610 (47.9) 18 (31.0) /day 133 (10.4) 9 (15.5) 2.37 ( ) 20/day 228 (17.9) 15 (25.9) 2.06 ( ) Unknown 303 (23.8) 16 (27.6) 1.60 ( ) Pipe smoker.35 No/unknown 1,252 (98.3) 55 (94.8) 1.00 Yes 22 (1.7) 3 (5.2) 1.80 ( ) Any smoking status b.009 Never 604 (47.4) 18 (31.0) 1.00 Former 273 (21.4) 17 (29.3) 1.76 ( ) Current 290 (22.8) 20 (34.5) 2.42 ( ) Unknown 107 (8.4) 3 (5.2) 0.91 ( ) NOTE. Descriptive data are presented as n (%) for categorical variables. CI, confidence interval; HR, hazard ratio. a Adjustments: age groups (16 40, 40 50, 50 60, 60 70, 70 80, and 80 years), sex (male, female), presence of low-grade dysplasia (yes, no, unknown), income deprivation quintile (I, II, III, IV, V, and unknown) and presence of reflux symptoms (yes, no, unknown). b Includes cigarette, pipe, and cigar smoking status combined. at the gastroesophageal junction. In further analysis restricted to nondysplastic BE patients, however, many of the previously observed significant associations became attenuated. Associations remained significant and appeared stronger in analysis of patients diagnosed with indefinite or low-grade dysplasia at their BE diagnosis, although this analysis had limited statistical power. Nevertheless, these results suggest that tobacco smoking promotes neoplastic progression in BE patients with lowgrade dysplasia rather than earlier in the pathway. Tobacco smoking has been long established as highly carcinogenic 28 and these findings provide additional evidence that it remains the top modifiable lifestyle factor for cancer risk. Smoking has also been specifically demonstrated to inflict DNA damage on Barrett s mucosa that may explain subsequent progression to EAC. 29 There are also suggestions that smoking is associated with an increased number of reflux episodes, and that nicotine may reduce lower esophageal sphincter pressure, 30 which would contribute to excess acid/bile exposure that may consequently increase neoplastic progression risk in BE. Further research into the effects of smoking in Barrett s mucosa would prove useful insights into the mechanisms involved. Body size measures were not associated with risk of progression in this cohort of BE patients, although the majority of BE patients in the current study for whom data were available were classified as overweight or obese. This contrasts with previous population-based studies that have repeatedly illustrated body fat to be directly associated with EAC risk, 21 although other prospective studies were also unable to identify an association between BMI and progression risk in BE. 15,22 The lack of association between BMI and risk of neoplastic progression in BE should be interpreted with caution, because height and weight were infrequently and differentially recorded in the endoscopy reports reviewed in our study, with BE patients who progressed to develop cancer or HGD more likely to have had both their height and weight reported than nonprogressors. Our findings can therefore be considered exploratory only. However, an alternative explanation for our null results could be that overweight and obesity do not increase risk of progression from BE to EAC, but instead act earlier in the disease pathway. This is supported by findings from a systematic review of adiposity and BE risk that concluded the increased risk to be an indirect consequence of increased gastroesophageal reflux disease risk, and not an etiological determinant of BE. 31 Consumption of alcohol was not significantly associated with the risk of progression from BE to esophageal or gastric cardia adenocarcinoma or HGD in the current study, even among heavy drinkers. The results from the current study are in line with previous findings 15,20,22 and suggest that alcohol is not a causative agent in EAC development. Unlike previous research, we did not show any associations, protective or otherwise, for specific alcohol types, such as beer and wine, 20,32 although our analysis on alcohol types had limited statistical power. Therefore, we can only recommend that BE patients follow the

7 February 2012 LIFESTYLE AND BARRETT NEOPLASTIC PROGRESSION 239 advice of current public health guidelines for cancer prevention to consume alcohol in moderation, if taken. 21 Strengths of this study include its large size and population-based setting in the established Northern Ireland BE register, which is one of the largest and most robust population-based registers of BE patients worldwide. Although information was collected retrospectively, it was based on the hospital case notes from a patients initial diagnosis of BE. Therefore, we were able to investigate lifestyle exposures in BE patients before their subsequent diagnosis of EAC or HGD. This is the largest study to date that has achieved such temporality in the investigation of lifestyle factors and progression risk. In addition, we were able to control for a number of known confounders in our analysis. Limitations of this investigation lie primarily with the retrospective collection of data from hospital case notes; considerable proportions of patients had to be recorded as unknown with regard to body size and alcohol consumption in particular because of a lack of reporting by clinicians. In addition, we do not have any data on subsequent body size measures or alcohol and tobacco exposures in these BE patients. We also had limited statistical power in our stratified analyses. In conclusion, BE patients who smoke tobacco are at a 2-fold increased risk of progressing to EAC or HGD. Although these findings need to be confirmed in future studies, we would suggest that tobacco smoking be discouraged and smoking-cessation strategies considered in BE patients in order to reduce future cancer risk. References 1. Lofdahl HE, Lu Y, Lagergren J. Sex-specific risk factor profile in oesophageal adenocarcinoma. Br J Cancer 2008;99: Vial M, Grande L, Pera M. Epidemiology of adenocarcinoma of the esophagus, gastric cardia, and upper gastric third. Recent Results Cancer Res 2010;182: Devesa SS, Blot WJ, Fraumeni JF Jr. Changing patterns in the incidence of esophageal and gastric carcinoma in the United States. Cancer 1998;83: Gatenby P, Ramus J, Caygill C, et al. Routinely diagnosed lowgrade dysplasia in Barrett s oesophagus: a population-based study of natural history. Histopathology 2009;54: Miros M, Kerlin P, Walker N. Only patients with dysplasia progress to adenocarcinoma in Barrett s oesophagus. Gut 1991;32: Yousef F, Cardwell C, Cantwell MM, et al. The incidence of esophageal cancer and high-grade dysplasia in Barrett s esophagus: a systematic review and meta-analysis. Am J Epidemiol 2008;168: Ong CA, Lao-Sirieix P, Fitzgerald RC. Biomarkers in Barrett s esophagus and esophageal adenocarcinoma: predictors of progression and prognosis. World J Gastroenterol 2010;16: Murray L, Sedo A, Scott M, et al. TP53 and progression from Barrett s metaplasia to oesophageal adenocarcinoma in a UK population cohort. Gut 2006;55: Prasad GA, Bansal A, Sharma P, Wang KK. Predictors of progression in Barrett s esophagus: current knowledge and future directions. Am J Gastroenterol 2010;105: Anderson LA, Watson RG, Murphy SJ, et al. Risk factors for Barrett s oesophagus and oesophageal adenocarcinoma: results from the FIN- BAR study. World J Gastroenterol 2007;13: Lagergren J, Bergstrom R, Lindgren A, et al. The role of tobacco, snuff and alcohol use in the aetiology of cancer of the oesophagus and gastric cardia. Int J Cancer 2000;85: Brown LM, Swanson CA, Gridley G, et al. Adenocarcinoma of the esophagus: role of obesity and diet. J Natl Cancer Inst 1995;87: Chen H, Ward MH, Graubard BI, et al. Dietary patterns and adenocarcinoma of the esophagus and distal stomach. Am J Clin Nutr 2002;75: Whiteman DC, Sadeghi S, Pandeya N, et al. Combined effects of obesity, acid reflux and smoking on the risk of adenocarcinomas of the oesophagus. Gut 2008;57: Jung KW, Talley NJ, Romero Y, et al. Epidemiology and Natural History of Intestinal Metaplasia of the Gastroesophageal Junction and Barrett s Esophagus: a population-based study. Am J Gastroenterol 2011;106: ; quiz Freedman ND, Abnet CC, Leitzmann MF, et al. A prospective study of tobacco, alcohol, and the risk of esophageal and gastric cancer subtypes. Am J Epidemiol 2007;165: Corley DA, Kubo A, Zhao W. Abdominal obesity and the risk of esophageal and gastric cardia carcinomas. Cancer Epidemiol Biomarkers Prev 2008;17: Murray L, Romero Y. Role of obesity in Barrett s esophagus and cancer. Surg Oncol Clin N Am 2009;18: Cook MB, Kamangar F, Whiteman DC, et al. Cigarette smoking and adenocarcinomas of the esophagus and esophagogastric junction: a pooled analysis from the international BEACON consortium. J Natl Cancer Inst 2010;102: Freedman ND, Murray LJ, Kamangar F, et al. Alcohol intake and risk of oesophageal adenocarcinoma: a pooled analysis from the BEACON Consortium. Gut 2011;60: World Cancer Research Fund/American Institute for Cancer Research. Food, nutrition, physical activity, and the prevention of cancer: a global perspective. Washington DC: American Institute for Cancer Research, Sikkema M, Looman CW, Steyerberg EW, et al. Predictors for Neoplastic Progression in Patients with Barrett s Esophagus: a prospective cohort study. Am J Gastroenterol 2011;106: Bhat S, Coleman HG, Yousef F, et al. Risk of Malignant Progression in Barrett s Esophagus Patients: results from a large population-based study. J Natl Cancer Inst 2011;103: Coleman HG, Bhat S, Murray LJ, et al. Increasing incidence of Barrett s oesophagus: a population-based study. Eur J Epidemiol 2011;26: Chandrasoma P, Wickramasinghe K, Ma Y, et al. Adenocarcinomas of the distal esophagus and gastric cardia are predominantly esophageal carcinomas. Am J Surg Pathol 2007;31: Marsman WA, Tytgat GN, ten Kate FJ, et al. Differences and similarities of adenocarcinomas of the esophagus and esophagogastric junction. J Surg Oncol 2005;92: Sobin LH, Gospodarowicz MK, Wittekind C, eds. TNM classification of malignant tumours. 7th ed. Hoboken, NJ: Wiley-Blackwell, International Association for Research on Cancer. International Association for Research on Cancer monographs. Volume 83: Tobacco smoke and involuntary smoking. Geneva: World Health Organization, Olliver JR, Hardie LJ, Gong Y, et al. Risk factors, DNA damage, and disease progression in Barrett s esophagus. Cancer Epidemiol Biomarkers Prev 2005;14: Pandolfino JE, Kahrilas PJ. Smoking and gastro-oesophageal reflux disease. Eur J Gastroenterol Hepatol 2000;12: Cook MB, Greenwood DC, Hardie LJ, et al. A systematic review and meta-analysis of the risk of increasing adiposity on Barrett s esophagus. Am J Gastroenterol 2008;103: Anderson LA, Cantwell MM, Watson RG, et al. The association between alcohol and reflux esophagitis, Barrett s esophagus, and esophageal adenocarcinoma. Gastroenterology 2009;136:

8 240 COLEMAN ET AL GASTROENTEROLOGY Vol. 142, No. 2 Received June 29, Accepted October 24, Reprint requests Address requests for reprints to: Helen G. Coleman, PhD, BSc (Hons), Cancer Epidemiology and Health Services Research Group, Centre for Public Health, Queen s University Belfast, ICS-B Building, RVH Site, Grosvenor Road, Belfast, BT12 6BA, Northern Ireland. h.mulholland@qub.ac.uk; fax: 0044 (0) Acknowledgments We would like to acknowledge the contribution of the Tumour Verification Officers in the Northern Ireland Cancer Registry, namely Mrs Kate Donnelly, Mrs Rosemary Ward and Ms Olwyn Dawson, who collected the data presented in this report. We are also grateful for the input of all staff in the Centre for Public Health at Queen s University Belfast, past and present, who have contributed to the development of the Northern Ireland BE register. The Northern Ireland Cancer Registry is funded by the Public Health Agency for Northern Ireland. Conflicts of interest The authors disclose no conflicts. Funding This work was supported by funding from the Ulster Cancer Foundation and the Health and Social Care Research and Development Office, Northern Ireland.