Obesity and lifestyle risk factors for gastroesophageal reflux disease, Barrett esophagus and esophageal adenocarcinoma

Summary

The aim of this study was to examine the association of obesity with esophageal adenocarcinoma, and with the precursor lesions Barrett esophagus and gastroesophageal reflux disease (GERD). This case-control study included cases with GERD (n = 142), Barrett esophagus (n = 130), and esophageal adenocarcinoma (n = 57). Controls comprised 102 asymptomatic individuals. Using logistic regression methods, we compared obesity rates between cases and controls adjusting for differences in age, gender, and lifestyle risk factors. Relative to normal weight, obese individuals were at increased risk for esophageal adenocarcinoma (Odds Ratio [OR] 4.67, 95% Confidence Interval [CI] 1.27–17.9). Diets high in vitamin C were associated with a lower risk for GERD (OR 0.40, 95% CI 0.19–0.87), Barrett esophagus (OR 0.44, 95% CI 0.20–0.98), and esophageal adenocarcinoma (OR 0.21, 95% CI 0.06–0.77). For the more established risk factors, we confirmed that smoking was a significant risk factor for esophageal adenocarcinoma, and that increased liquor consumption was associated with GERD and Barrett esophagus. In light of the current obesity epidemic, esophageal adenocarcinoma incidence rates are expected to continue to increase. Successful promotion of healthy body weight and diets high in vitamin C may substantially reduce the incidence of this disease.

Introduction

Over the past three decades, a marked change in the epidemiology of esophageal malignancy in North America and Europe has been reported, with an increasing incidence of esophageal adenocarcinoma.^1–3 The reasons for this are largely unknown, but several lifestyle and dietary risk factors, have been proposed.^4–7 As prevalence rates for obesity have increased dramatically,^8–10 and in parallel with reported epidemiologic changes for cancer,¹⁰ obesity was proposed as a risk factor for esophageal adenocarcinoma, an observation now supported by a number of case-control studies.^11–15 Although a strong statistical association between gastroesophageal reflux disease (GERD) and risk for esophageal adenocarcinoma has also been reported,^16–19 the role of lifestyle risk factors are less well defined and remain controversial.^{11,16,20–24} To date, relatively few studies have evaluated obesity and lifestyle risk factors associated with esophageal premalignancy, or potential biologic mechanisms underlying these epidemiologic observations.

Primary esophageal adenocarcinomas are thought to arise from Barrett esophagus, an acquired condition in which the normal esophageal squamous epithelium is replaced by a specialized metaplastic columnar cell-lined epithelium.^25,26 Progression of Barrett esophagus to invasive adenocarcinoma is reflected histologically by the metaplasia-dysplasia-carcinoma sequence.²⁷ As GERD is a risk factor for Barrett esophagus, there is a plausible link between GERD, Barrett esophagus and esophageal adenocarcinoma.²⁸ The recent identification of molecular markers associated with Barrett metaplasia-dysplasia-adenocarcinoma progression may provide further insight into the molecular pathogenesis of this disease.^27,29–31 However, as GERD and Barrett esophagus are relatively common, and only a fraction of individuals progress to invasive adenocarcinoma,²⁸ it is more likely that obesity and lifestyle risk factors interact to modulate individual susceptibility for progression to esophageal adenocarcinoma.

To examine the relationship between obesity and susceptibility to GERD, Barrett esophagus and esophageal adenocarcinoma, we conducted a two-year prospective case-control study, the Nova Scotia Barrett Esophagus Study.

Materials and Methods

Study population

Between February 2001 and February 2003, a total of 431 individuals were enrolled in a prospective hospital-based case-control study including newly diagnosed cases with GERD (n = 142), Barrett esophagus (n = 130) and esophageal adenocarcinoma (n = 57). Because GERD is common in the general population,^28–32 we selected strictly asymptomatic individuals as controls (n = 102) to alleviate the comparison with GERD cases. These controls were frequency matched to the age and gender distribution of the combined population of GERD, Barrett esophagus and esophageal adenocarcinoma cases. The study was based at the QEII Health Science Center (QEII HSC), Halifax, Nova Scotia, Canada, the largest tertiary referral center in Atlantic Canada. Over this two-year study, the participation rate among cases was 100% for patients with esophageal adenocarcinoma, 93% (140 initially identified and approached to participate, 130 enrolled) for patients with Barrett esophagus, and 88% (162 approached, 142 enrolled) for patients with GERD. The participation rate for controls was 58% (177 eligible, 102 enrolled); the most frequent reason for declining was related to scheduling a convenient time to complete the questionnaire. All participants gave informed, written consent, and approval for this study was granted by the Capital Health Research Ethics Board (QE-2000-277).

For each participant, a 102-point structured questionnaire was administered through a face-to-face interview with a single trained research nurse. This included collection of information on sociodemographic factors, family and medical history, weight history, measured height and weight, and questions on tobacco and alcohol consumption. It also included a 150-item food frequency questionnaire on dietary patterns five years prior to the interview, as recent diet may be affected by conditions under study. Similarly, we considered body weight one year prior to the interview, and at age 20 years to be the most informative. Underweight was defined as a body mass index (BMI) of < 20 kg/m², normal weight as a BMI of ≥ 20 and < 25 kg/m², overweight as a BMI of ≥ 25 and < 30 kg/m², and obesity as a BMI ≥ 30 kg/m².⁹

Strict clinicopathologic criteria³³ were used to define adenocarcinomas of primary esophageal origin, thereby excluding adenocarcinomas of the cardia or proximal stomach. The diagnosis of Barrett esophagus was established by the histologic finding of intestinal metaplasia within an endoscopically identified columnar-lined esophagus, which was confirmed independently by two consultant gastrointestinal histopathologists. Patients were diagnosed with GERD on the basis of dominant symptoms of heartburn and/or regurgitation, previous response to acid suppression therapy, with or without evidence of esophagitis diagnosed endoscopically and histologically from biopsies. Asymptomatic individuals were recruited concurrently from two ambulatory general surgery outpatient clinics. Subjects who attended for consultation regarding unrelated benign conditions, including inguinal hernia and miscellaneous non-malignant ‘lumps and bumps’, were screened for GERD-related symptoms, a history of ‘hiatus hernia’, ‘dyspepsia’, antacid use, previous malignancy or upper gastrointestinal surgery, and if negative, were asked to participate as controls.

Statistical analysis

We used logistic regression to compare controls with each of the cases of GERD, Barrett esophagus and esophageal adenocarcinoma with respect to body weight and to lifestyle risk factors that have previously been suggested in the etiology of esophageal malignancy. These included smoking, consumption of alcohol (beer, wine, liquor), use of multivitamins, and dietary factors including energy intake, percent of calories obtained from dietary fat, number of servings of fruit and vegetables, and intake of vitamin C and fiber. For these dietary factors, we categorized participants into tertiles and we present odds ratios comparing the highest tertile with the reference be the lowest tertile.

Results

Table 1 presents descriptive characteristics for asymptomatic controls, and for cases with GERD, Barrett esophagus and esophageal adenocarcinoma. Age and gender adjusted risks for overweight, obesity and lifestyle risk factors among patients with GERD, Barrett esophagus and esophageal adenocarcinoma, relative to asymptomatic controls, are presented in Table 2. Although obesity was not associated with an increased risk for GERD, obesity was associated with a trend towards increased risk for Barrett esophagus (Odds Ratio [OR] 2.09, 95% Confidence Interval [CI] 0.95–4.58), and was a statistically significant risk factor for esophageal adenocarcinoma (OR 4.52, 95% CI 1.47–13.88). The less severe form, overweight, was also associated with a trend towards increased risk for Barrett esophagus and esophageal adenocarcinoma, although statistical significance was not achieved. Further consideration of the confounding potential of lifestyle risk factors did not substantially alter the risk estimates for overweight and obesity (Table 3). Tables 1, 2 and 3 present body weight one year prior to the inclusion in the study. However, from the perspective of tumor initiation, body weight at age 20 years may be more relevant. The multivariate risk adjustment for esophageal adenocarcinoma was 2.15 (95% CI 0.72–6.46) and 6.97 (95% CI 1.57–31.03) for overweight and obesity at age 20 years, respectively.

Table 1

Characteristics of controls and cases with gastroesophageal reflux disease, Barrett esophagus and esophageal adenocarcinoma.

	Controls (n = 102)	GERD (n = 142)	BE (n = 130)	EADC (n = 57)
Age (mean years)	57	55	59	67
Gender (% female)	33	47	28	11
Underweight (%)	2	3	2	0
Normal weight (%)	23	28	15	16
Overweight (%)	45	38	48	43
Obese (%)	30	31	35	41
Smoking (% ever)	59	66	63	86
Beer (drinks/month)	21	30	42	43
Wine (drinks/month)	8	7	10	15
Liquor (drinks/month)	19	25	44	40
Energy intake (kcal/day)	2100	1990	2113	2078
% calories for fat	32.3	30.7	31.3	32.7
Fruit/vegetables (serve/day)	6.7	6.6	6.3	6.7
Fiber (g/day)	23.3	22.5	22.0	23.5
Vitamin C (mg/day)	145	134	132	132
Multivitamins (% use)	21	16	16	5

BE, Barrett esophagus; EADC, esophageal adenocarcinoma; GERD, gastroesophageal reflux disease.

Table 2

Age and gender adjusted risks of overweight, obesity and lifestyle risk factors for gastroesophageal reflux disease, Barrett esophagus and esophageal adenocarcinoma

	GERD		BE		EADC
	OR	95%CI	OR	95%CI	OR	95%CI
Underweight	0.64	0.10–4.02	2.67	0.37–19.37	n/a
Normal weight	1		1		1
Overweight	0.69	0.36–1.33	1.68	0.81–3.46	1.55	0.56–4.30
Obesity	0.73	0.36–1.48	2.09	0.95–4.58	4.52	1.47–13.88
Smoking
Non-smoking	1		1		1
< 5000 lps	1.95	0.96–3.98	1.07	0.49–2.33	3.64	0.98–13.56
≥ 5000 lps	1.19	0.66–2.15	1.38	0.78–2.45	5.31	2.06–13.67
Beer
< one drink/mo	1		1		1
≥ 1–40 drinks/month	1.07	0.57–1.99	1.08	0.55–2.12	0.49	0.30–1.21
≥ 40 drinks/month	1.21	0.54–2.71	1.43	0.63–3.24	0.99	0.37–2.64
Wine
< one drink/month	1		1		1
≥ 1 & < 40 drinks/month	0.53	0.31–0.93	0.78	0.46–1.35	0.25	0.10–0.59
≥ 40 drinks/month	0.71	0.23–2.20	0.80	0.27–2.34	0.92	0.27–3.13
Liquor
< one drink/month	1		1		1
≥ 1–40 drinks/month	0.93	0.52–1.67	1.08	0.27–1.98	0.59	0.25–1.41
≥ 40 drinks/month	2.38	0.99–5.72	3.16	1.35–7.40	2.96	1.04–8.40
Energy	0.68	0.36–1.30	0.81	0.43–1.54	0.58	0.24–1.44
% calories from fat	0.56	0.28–1.11	0.49	0.24–1.00	0.81	0.20–2.03
Fruit/vegetables	0.83	0.40–1.73	0.49	0.23–1.06	0.37	0.13–1.06
Fiber	1.17	0.51–2.66	0.41	0.19–0.88	0.38	0.12–1.21
Vitamin C	0.46	0.22–0.96	0.41	0.19–0.88	0.19	0.06–0.60
Multivitamin use	0.72	0.37–1.40	0.74	0.37–1.46	0.11	0.03–0.46

BE, Barrett esophagus; EADC, esophageal adenocarcinoma; GERD, gastroesophageal reflux disease; lps, lifetime packs of cigarettes; n/a, not applicable; OR, odds ratio comparing the highest tertile with the reference be the lowest tertile.

Table 3

Multivariate risks of overweight, obesity and lifestyle risk factors for gastroesophageal reflux disease, Barrett esophagus and esophageal adenocarcinoma

	GERD		BE		EADC
	OR	95% CI	OR	95% CI	OR	95% CI
Underweight	0.53	0.08–3.68	2.75	0.35–21.63	n/a
Normal weight	1		1		1
Overweight	0.66	0.33–1.34	1.56	0.73–3.34	1.58	0.48–5.26
Obesity	0.66	0.31–1.39	1.80	0.78–4.14	4.67	1.27–17.19
Smoking
Non-smoking	1		1		1
< 5000 lps	2.33	1.09–4.94	1.00	0.45–2.23	5.29	1.19–23.53
≥ 5000 lps	0.99	0.51–1.90	1.03	0.55–1.95	3.86	1.23–12.10
Liquor
< one drink/month	1		1		1
≥ 1–40 drinks/month	0.97	0.52–1.81	1.20	0.63–2.29	0.61	0.21–1.76
≥ 40 drinks/momth	2.69	1.05–6.92	3.06	1.23–7.62	2.64	0.74–9.39
Vitamin C	0.40	0.19–0.87	0.44	0.20–0.98	0.21	0.06–0.77
Multivitamin use	0.84	0.41–1.73	0.96	0.75–1.13	0.17	0.03–0.90

BE, Barrett esophagus; EADC, esophageal adenocarcinoma; GERD, gastroesophageal reflux disease; lps, lifetime packs of cigarettes; n/a, not applicable; OR, odds ratio comparing the highest tertile with the reference be the lowest tertile.

The risk for esophageal adenocarcinoma increased with age, by 2.65 per 10 years increment, and with male gender (OR 3.84, 95% CI 1.43–10.30). In this study, smoking was associated with a significantly increased risk for esophageal adenocarcinoma (OR 3.86; 95% CI 1.23–12.10). Increased liquor consumption was, however, an associated risk factor for both GERD (OR 2.69, 95% CI 1.05–6.92) and Barrett esophagus (OR 3.06, 95% CI 1.23–7.62), but was not a statistically significant risk factor for esophageal adenocarcinoma (OR 2.64, 95% CI 0.74–9.39). We observed colinearity between smoking and consumption of liquor in their association with risk for developing esophageal adenocarcinoma. In a multivariate adjusted model without consideration of smoking, liquor consumption would provide a statistically significant risk for esophageal adenocarcinoma (OR 3.65, 95% CI 1.08–12.31). In a model without consideration of liquor consumption, the risk for esophageal adenocarcinoma would be of similar magnitude in both smoking categories (OR for < 5000 and > 5000 lifetime packs of cigarettes were 5.26 and 4.66, respectively).

Individuals with high vitamin C diet and the use of multivitamin supplements were associated with diminished risk for GERD (OR 0.40, 95% CI 0.19–0.87), Barrett esophagus (OR 0.44, 95% CI 0.20–0.98), and esophageal adenocarcinoma (OR 0.21, 95% CI 0.06–0.77), as shown in Table 3. The use of multivitamin supplements further decreased the risk for esophageal adenocarcinoma (OR 0.17, 95% CI 0.03–0.90). We also evaluated the potential importance of income and educational attainment; however, these were not found to have substantial or statistically significant effects on the occurrence of GERD, Barrett esophagus and esophageal adenocarcinoma.

Discussion

In this study, we found that relative to persons of normal weight, obese individuals had an almost five-fold increased risk for esophageal adenocarcinoma. This finding is comparable with previously reported case-control studies,^11–16 although precise comparisons across studies are limited as they contrasted different BMI subcategories. The magnitude of the risk for esophageal adenocarcinoma that we observed among individuals who were obese at age 20 was even higher (approximately 6.5-fold). As earlier tumor initiation is likely to result in diagnosis at younger age, this observation seems consistent with the reported higher risk for esophageal adenocarcinoma in obese patients diagnosed before the age of 50 years.¹³ In addition, the prevalence of childhood obesity is particularly high in Atlantic Canada.³⁴ The higher risk for esophageal adenocarcinoma associated with obesity observed in this study may well be explained by a higher proportion of patients who were obese in childhood or early adulthood.

To date, relatively few studies have evaluated the effects of excess weight on the precursor lesion, Barrett esophagus, and clinically symptomatic GERD, a well recognized risk factor for both Barrett esophagus and esophageal adenocarcinoma.^17–19 One recent study reported a two-and-a-half-fold increased risk for Barrett esophagus in overweight males,³⁵ whereas a second study found no association between BMI and Barrett esophagus and selected biomarkers (flow cytometry, allelic loss of chromosome 9p or 17p).³⁶ The latter study, however, did demonstrate increasing risk for cell cycle (aneuploidy) and genetic abnormalities (17p loss) in Barrett esophagus with increasing waist:hip ratio, supporting the notion that distribution of body fat (male-pattern obesity) may be more important than BMI.³⁶ Similarly, the association between BMI and GERD has been controversial, with a number of well performed studies reporting no clear relationship.^37–39 However, more recent reports have demonstrated obesity to be a significant risk factor for GERD,^40–43 with a clear association between increasing BMI and the severity and frequency of GERD symptoms⁴¹ and objective measurement of distal esophageal acid exposure.⁴⁴ That the strongest association between obesity and GERD was reported in females,^42,43 especially premenopausal or with use of hormone replacement therapy, suggests that female sex hormones may play an important role in the pathogenesis of GERD. It was hypothesized that estrogen may mediate nitric oxide-mediated reduction of smooth muscle tone in the lower esophageal sphincter (LES), an important barrier to reflux.⁴³

The precise mechanism(s) underlying the reported association between body mass and GERD or Barrett esophagus, risk factors for progression to esophageal adenocarcinoma,^{17–19,27,28} have not yet been defined. It has been suggested that overweight/obese individuals may have associated esophageal dysmotility,^45,46 resulting in reduced esophageal clearance, or increased intra-abdominal pressure and hiatus hernia, with consequent mechanical disruption of LES function.^47,48 However, in the present study we observed no increased risk for GERD among obese subjects. LES function may also be altered by smoking, various dietary factors including high fat, increased alcohol consumption, and a variety of medications which relax the LES and predispose to GERD.^{4–7,49–51} Indeed, an increased risk for esophageal adenocarcinoma has been reported for individuals taking various LES-relaxing drugs,⁵² including long-term use of theophylline and beta-agonists.⁵³ Finally, several metabolic consequences of obesity may play a role in carcinogenesis,⁵⁴ possibly mediated by leptin,^55,56 insulin and insulin-like growth factor,^57–59 which has been shown to stimulate cell proliferation and inhibit apoptosis, although these would not be specific to esophageal malignancy.

The contribution of smoking, alcohol consumption and diet, well established risk factors for squamous cell carcinoma of the esophagus, to the pathogenesis of esophageal adenocarcinoma is controversial.^4–7 In contrast to studies reporting only a moderate increased risk among smokers,^15,16,23,24 we found a strong association between cigarette smoking and risk for esophageal adenocarcinoma. A recent study has clearly established that smoking is associated with high rates of DNA damage in esophageal squamous (and columnar) epithelium, increasing the likelihood of genetic change in esophageal epithelium, providing a sound biologic basis for malignant progression.⁶⁰ The role of alcohol consumption as a risk factor for esophageal adenocarcinoma is also unclear,^6,7 with only about half of all published studies reporting a positive, but generally weak, association.^{12,23,24,52,61} In this study, we found a strong association between increased consumption of alcohol, specifically liquor, and GERD and Barrett esophagus, suggesting that alcohol may be a significant risk factor at early stages in the molecular pathogenesis of esophageal adenocarcinoma. Although relatively few studies have critically evaluated dietary risk factors, increased dietary fat, low intake of fruit, raw vegetables, fiber and vitamins have been associated with increased risk for esophageal adenocarcinoma.^{4,11,20–22,62} Similar dietary risk factors were reported for GERD⁶³ and Barrett esophagus.⁶⁴ Our observation that diets with high vitamin C content and multivitamin supplementation were associated with reduced risk for GERD, Barrett esophagus, and esophageal adenocarcinoma, may provide a basis for future preventive public health strategies.

Strengths of the present study include that strict clinico-pathologic criteria were used to identify cases with GERD, Barrett esophagus and primary esophageal adenocarcinomas,³³ thereby avoiding misclassification with adenocarcinomas arising from the esophagogastric junction (cardia) or proximal stomach. Furthermore, every effort was made in this study to identify asymptomatic individuals (with regard to GERD) as a control group, although it is recognized that a small proportion of clinically asymptomatic individuals may have occult esophagitis or Barrett esophagus. Potential limitations relate to the use of hospital-based controls, recall bias regarding diet and weight, limited use of ambulatory 24 h pH monitoring to confirm pathologic acid reflux in GERD patients, and the total number of study subjects. With just under one million inhabitants, the population of Nova Scotia has experienced little migration or ethnic diversity. Despite the striking changes reported for the epidemiology of esophageal cancer in North America, adenocarcinomas of the esophagus are still relatively uncommon tumors in this province. As all cancers must legally be reported, data from the Nova Scotia Cancer Registry revealed that 69 adenocarcinomas of the esophagus were registered for the corresponding two-year interval of this study. Therefore, the 57 cancers included in this hospital-based study represent 83% of all esophageal adenocarcinomas seen in the province, and given the rarity of this malignancy, would appear to be representative of the general population of Nova Scotia.

As the prognosis for invasive esophageal cancer is generally poor,²⁵ it therefore seems likely that progress with this malignancy will only be made with early detection, prevention, and a clearer understanding of its etiology and tumor biology. Endoscopic surveillance of patients with Barrett esophagus is a widely accepted and utilized strategy to detect early, and potentially curable esophageal cancer. However, as most patients with Barrett esophagus do not progress to cancer, the value and cost-efficacy of this approach has recently been questioned.²⁸ Similarly, there appears to be little evidence to support routine screening of individuals with GERD. The use of molecular biomarkers to define subsets of patients with Barrett esophagus at increased risk of progression to esophageal adenocarcinoma, has yet to be widely incorporated into clinical practice.²⁹ We recently reported that risk for progression to esophageal adenocarcinoma is associated with polymorphisms of genes regulating the cell cycle (cyclin D1)⁶⁵ and DNA repair.⁶⁶ Here we demonstrated the independent importance of obesity and several lifestyle risk factors; it is therefore plausible that interactions between molecular alterations and such lifestyle risk factors may modulate individual susceptibility for progression from GERD and Barrett esophagus to esophageal adenocarcinoma. Strategies to prevent malignant progression would therefore appear to be a particularly promising approach. Based on the results of this, and other reported studies, prevention of obesity, healthy diet and lifestyle as well as multivitamin supplementation would appear to be most effective at reducing risk.

Acknowledgments

This work was supported by The Nova Scotia Health Research Foundation; the Capital Health Research Fund; and the National Cancer Institute of Canada with funds from the Canadian Cancer Society. AGC is supported by a Senior Clinical Research Scholarship from the Dalhousie University Faculty of Medicine. We thank the following for help with specific aspects of this study: Angela Fitzgerald for expert analysis of nutritional data; Drs Dickran Malatjalian and Heidi Sapp for expert independent histopathological review of esophageal tissue sections; the nursing staff of the Endoscopy Unit at the QE II Health Sciences Center for their enthusiastic support; and Ron Dewar for provincial cancer registry data.

Appendix 1

Collaborators in the Nova Scotia Barrett Esophagus Study

Principal Investigator: Alan G. Casson FRCSC

Study Co-ordinator: Susan Winch RN

Collaborating Gastroenterologists (case accrual): Nina Abraham MD, Bernard Badley MD, Dana Farina MD, Mani Kareemi MD, Des Leddin MD, Jonathan Love MD, Don Macintosh MD, Sunil Patel MD, Kervork Peltekian MD, Ron Tanton MD, Geoff Turnbull MD, Sander van Zanten MD, Noel Williams MD

Collaborating Internists (case accrual): Benedict Cookey MD, Brian O’Brian MD

Collaborating General Surgeons (control accrual): John Curry MD, Bob Stone MD

Collaborating Thoracic Surgeons (case accrual): Drew Bethune MD, Harry Henteleff MD

Pathologists: Laurette Geldenhuys MD, Dickran Malatjalian MD, Heidi Sapp MD

Molecular Pathology Laboratory: Susan Evans MSc, Kim Gallant BSc, Duane Guernsey PhD, Christie Riddell MD, Nadine Vaninetti BSc, Zuoyu Zheng MD

Data Management: John Fris, Sarah Maaten MSc

Statistical Analysis: Angela Fitzgerald MSc, Judy Guernsey PhD, Geoff Porter MD, Paul Veugelers PhD

Administrative Assistance: Joanne Dibblee, Dianne Russell, Lesli Smith

Author notes

*

A copy of the questionnaire may be obtained directly from the corresponding author.