Original Paper Received: July 19, 2013 Accepted: December 6, 2013 Published online: July 29, 2014
Digestion 2014;90:1–9 DOI: 10.1159/000357776
Association between the Body Mass Index and the Risk of Barrett’s Esophagus in Japan Hirohiko Shinkai a Katsunori Iijima a Tomoyuki Koike a Yasuhiko Abe e Naohiro Dairaku f Yoshifumi Inomata b Syoichi Kayaba g Fumitake Ishiyama g Tomoyuki Oikawa h Motoki Ohyauchi i Hirotaka Ito i Sho Asonuma j Tatsuya Hoshi k Katsuaki Kato c Shuichi Ohara d Tooru Shimosegawa a
a
Division of Gastroenterology, Tohoku University Graduate School of Medicine, b Department of Gastroenterology, JR Sendai Hospital, c Cancer Detection Center, Miyagi Cancer Society, and d Department of Gastroenterology, Tohoku Rosai Hospital, Sendai, e Department of Gastroenterology, Yamagata University Faculty of Medicine, Yamagata, f Department of Gastroenterology, Iwaki Kyoritsu General Hospital, Iwaki, g Department of Gastroenterology, Iwate Prefectural Isawa Hospital, Isawa, h Department of Gastroenterology, Hachinohe City Hospital, Hachinohe, i Department of Gastroenterology, Osaki Citizen Hospital, Osaki, j Department of Gastroenterology, South Miyagi Medical Center, Ogawara, and k Department of Gastroenterology, Kesennuma City Hospital, Kesennuma, Japan
Key Words Barrett’s esophagus · Body mass index · Obesity
Abstract Background: We investigated the association between long-segment Barrett’s esophagus and obesity in the Japanese population in a multicenter case-control trial. Methods: One hundred thirteen patients with endoscopically detected Barrett’s esophagus with a length of more than 2 cm and the same number of sex- and age-matched controls were prospectively enrolled. Barrett’s esophagus was diagnosed based on the Prague C and M criteria. The body mass index (BMI) of the subjects was categorized into the following groups: normal, BMI 25.0. To determine the associa-
© 2014 S. Karger AG, Basel 0012–2823/14/0901–0001$39.50/0 E-Mail [email protected] www.karger.com/dig
tion between BMI and the risk of Barrett’s esophagus, multivariate logistic regression analyses were performed. Results: The basically adjusted regression model adjusted for smoking and alcohol consumption revealed that overweight and obesity were significantly associated with an elevated risk of Barrett’s esophagus (OR 2.4, 95% CI 1.2–4.7, and OR 2.5, 95% CI 1.3–4.6, respectively). The intensity of the association was not attenuated even after adjustment for gastroesophageal reflux disease-related parameters. Conclusions: An increased BMI was associated with an increased risk for Barrett’s esophagus through a gastroesophageal reflux-independent mechanism in the Japanese population. Further, unlike in Caucasian populations, being even slightly overweight with a BMI of 23.0–24.9 was an independent risk factor in the Japanese population. © 2014 S. Karger AG, Basel
Iijima Katsunori, MD Division of Gastroenterology Tohoku University Graduate School of Medicine Seiryo-machi, Aobaku, Sendai 980-8574 (Japan) E-Mail kijima @ med.tohoku.ac.jp
Introduction
Barrett’s esophagus (BE) is a metaplastic change in which normal squamous epithelium is replaced with columnar epithelium. It is considered a premalignant condition associated with an increased risk of developing esophageal adenocarcinoma [1]. Long-segment BE, defined as a columnar mucosa extending 2–3 cm above the gastroesophageal (GE) junction [2–4], is considered to be primarily important as a premalignant condition compared to short-segment BE [5, 6]. The identification of putative risk factors for the increasing prevalence of BE is crucial for an early diagnosis and potential intervention. In addition to the established risk factors for BE, such as GE reflux diseases (GERD), hiatal hernia, being an elderly male, and Caucasian race [7], obesity has been frequently identified as a predisposing factor for BE [7–9]. However, almost all of the studies on risk factors for BE thus far have originated from Western countries, largely because the incidence of BE, and in particular long-segment BE, is much lower in Asian countries compared to Western countries [10, 11]. Since some studies have suggested that the risk factors differ among different ethnic groups [12, 13], the risk factors for Asian subjects with BE remain to be clarified. Nonetheless, there are a few studies from Japan that have investigated potential associations between BE and obesity, but with inconsistent results (a positive association or no association) [10, 14]. In this regard, correlations between reflux esophagitis and obesity in the Japanese population have also been somewhat conflicting [15–18], although a positive association has been observed in previous studies [15–17]. Meanwhile, the vast majority of subjects in these studies have had short-segment BE, and in particular ultra-short (2 cm [34]. Statistics Values regarding the length of BE are expressed as medians (IQR), and other values are expressed as means (SD). The distributions of the demographic characteristics and related factors were compared between BE patients and controls using Student’s paired t test for continuous variables and a sign test for categorical variables. All p values were two-sided, and p < 0.05 was considered statistically significant. In examining the association of BMI and other parameters with the risk of BE, conditional logistic regression analysis was used to control for potential confounders using the SAS PHREG procedure in the SAS version 8.2 statistical analysis software package (SAS Institute, Cary, N.C., USA), and OR with 95% CI were estimated. To test for trends, categorical variables were included in the model as continuous data. In our basic multivariable model, BMI (categorical variable), current smoking (yes or no), and current alcohol consumption (yes or no) were included as independent variables. Then, in order to estimate the potential contribution of GE reflux to the association between BMI and the risk of BE, various GERD-related parameters [heartburn, hiatal hernia, erosive esophagitis, and intake of proton pump inhibitors (PPI)] were included together in the basic model. We also undertook stratified analyses separately for males and females for all analyses.
Results
A total of 113 cases of BE >2 cm in length and the same number of sex- and age-matched controls were prospectively enrolled into this study. None of the subjects who were asked to participate refused to enroll into the study. Of the enrolled BE patients, 55 had been under surveillance. The mean age of the BE cases was BE and BMI in Japan
63.1 (12.7) years and the majority (93/113 = 82%) were male. Overall, the median length of BE according to the Prague C and M criteria was 3.0 (3.0) cm for the M value and 2.0 (2.0) for the C value, respectively. BE cases with lengths >3 cm accounted for 72% (81/113) of the enrolled subjects, with 4.0 (3.0) cm for the M value and 2.0 (3.0) cm for the C value (table 1). In 33 subjects, the C value was 3.0 cm or more. None of the BE cases showed mucosal irregularities such as a concavo-convex shape or had histologically proven dysplasia in the CLE. Of the 89 BE patients who allowed biopsies to be taken from the CLE, SIM was confirmed in 54 (61%) cases. Background Characteristics of BE Although height and body weight were not significantly different between the BE cases and the controls, the BMI was significantly higher in BE cases than in controls [24.4 (3.8) vs. 23.4 (3.0), p = 0.04]. The proportions of normal weight, overweight, and obesity in the controls according to WPRO criteria were 53, 20, and 27%, respectively, which is consistent with a previous report regarding BMI classification defined by the same criteria in normal Japanese individuals [29]. Meanwhile, the proportions of normal weight, overweight, and obesity in the BE patients were 32, 27, and 41%, respectively, and the distribution of subjects in the BMI categories was significantly different between the two groups (p < 0.01) (table 1). Various GERD-related parameters were more prevalent in BE patients than in controls, as expected. The prevalence of both erosive reflux esophagitis and hiatal hernia was significantly higher in BE patients than in controls (p < 0.01 for both), and especially hiatal hernia turned out to be highly prevalent (90%) in BE. Similarly, the prevalence of GERD symptoms such as heartburn and acid regurgitation was also significantly higher in BE patients than in controls (p < 0.01 for both). Concerning the drug intakes that we examined in the present study, PPI was administered to half of the BE patients, mainly to alleviate GERD symptoms, and such drug use was significantly more prevalent in BE patients than in controls (table 1). When the analysis was confined to subjects whose BE was longer than 3 cm, the above patients’ backgrounds were similar to those of the entire cohort. Namely, the mean BMI was significantly higher in BE patients than in controls [24.3 (4.0) vs. 23.2 (3.0), p = 0.04], and the distribution of subjects in the BMI categories was significantly different between BE patients and controls (p < 0.01) (table 1). Digestion 2014;90:1–9 DOI: 10.1159/000357776
3
Table 1. Characteristics of the subjects with BE with a length of more than 2 or 3 cm
Parameters
Male/female ratio Age, years Height, cm Weight, kg BMI BMI 25.0 Length of BE Median M value Median C value Current smokers Current drinkers Erosive esophagitis Hiatal hernia GERD symptom Heartburn Acid regurgitation Dysphagia Medication PPI H2 blockers Aspirin NSAIDs
Overall subjects (more than 2 cm of BE), n (%)
p value
More than 3 cm of BE, n (%)
p value
BE cases (n = 113)
controls (n = 113)
BE cases (n = 81)
controls (n = 81)
93/20 63.1 (12.7) 163.0 (9.6) 65.2 (14.2) 24.4 (3.8)
93/20 63.4 (13.2) 163.4 (9.9) 62.6 (10.0) 23.4 (3.0)
68/13 64.6 (12.2) 162.9 (9.8) 65.0 (14.7) 24.3 (4.0)
68/13 64.7 (12.9) 163.2 (10.1) 61.7 (9.6) 23.2 (3.0)
36 (31.9) 31 (27.4) 46 (40.7)
60 (53.1) 22 (19.5) 31 (27.4)
28 (34.6) 21 (25.9) 32 (39.5)
46 (56.8) 15 (18.5) 20 (25.0)
3.0 (3.0) 2.0 (2.0) 14 (12.4) 53 (46.9) 32 (28.3) 102 (90.3)
19 (16.8) 54 (47.8) 5 (4.4) 30 (26.5)
NS NS
Digestion 2014;90:1–9 DOI: 10.1159/000357776
Association between the Body Mass Index and the Risk of Barrett’s Esophagus in Japan Hirohiko Shinkai a Katsunori Iijima a Tomoyuki Koike a Yasuhiko Abe e Naohiro Dairaku f Yoshifumi Inomata b Syoichi Kayaba g Fumitake Ishiyama g Tomoyuki Oikawa h Motoki Ohyauchi i Hirotaka Ito i Sho Asonuma j Tatsuya Hoshi k Katsuaki Kato c Shuichi Ohara d Tooru Shimosegawa a
a
Division of Gastroenterology, Tohoku University Graduate School of Medicine, b Department of Gastroenterology, JR Sendai Hospital, c Cancer Detection Center, Miyagi Cancer Society, and d Department of Gastroenterology, Tohoku Rosai Hospital, Sendai, e Department of Gastroenterology, Yamagata University Faculty of Medicine, Yamagata, f Department of Gastroenterology, Iwaki Kyoritsu General Hospital, Iwaki, g Department of Gastroenterology, Iwate Prefectural Isawa Hospital, Isawa, h Department of Gastroenterology, Hachinohe City Hospital, Hachinohe, i Department of Gastroenterology, Osaki Citizen Hospital, Osaki, j Department of Gastroenterology, South Miyagi Medical Center, Ogawara, and k Department of Gastroenterology, Kesennuma City Hospital, Kesennuma, Japan
Key Words Barrett’s esophagus · Body mass index · Obesity
Abstract Background: We investigated the association between long-segment Barrett’s esophagus and obesity in the Japanese population in a multicenter case-control trial. Methods: One hundred thirteen patients with endoscopically detected Barrett’s esophagus with a length of more than 2 cm and the same number of sex- and age-matched controls were prospectively enrolled. Barrett’s esophagus was diagnosed based on the Prague C and M criteria. The body mass index (BMI) of the subjects was categorized into the following groups: normal, BMI 25.0. To determine the associa-
© 2014 S. Karger AG, Basel 0012–2823/14/0901–0001$39.50/0 E-Mail [email protected] www.karger.com/dig
tion between BMI and the risk of Barrett’s esophagus, multivariate logistic regression analyses were performed. Results: The basically adjusted regression model adjusted for smoking and alcohol consumption revealed that overweight and obesity were significantly associated with an elevated risk of Barrett’s esophagus (OR 2.4, 95% CI 1.2–4.7, and OR 2.5, 95% CI 1.3–4.6, respectively). The intensity of the association was not attenuated even after adjustment for gastroesophageal reflux disease-related parameters. Conclusions: An increased BMI was associated with an increased risk for Barrett’s esophagus through a gastroesophageal reflux-independent mechanism in the Japanese population. Further, unlike in Caucasian populations, being even slightly overweight with a BMI of 23.0–24.9 was an independent risk factor in the Japanese population. © 2014 S. Karger AG, Basel
Iijima Katsunori, MD Division of Gastroenterology Tohoku University Graduate School of Medicine Seiryo-machi, Aobaku, Sendai 980-8574 (Japan) E-Mail kijima @ med.tohoku.ac.jp
Introduction
Barrett’s esophagus (BE) is a metaplastic change in which normal squamous epithelium is replaced with columnar epithelium. It is considered a premalignant condition associated with an increased risk of developing esophageal adenocarcinoma [1]. Long-segment BE, defined as a columnar mucosa extending 2–3 cm above the gastroesophageal (GE) junction [2–4], is considered to be primarily important as a premalignant condition compared to short-segment BE [5, 6]. The identification of putative risk factors for the increasing prevalence of BE is crucial for an early diagnosis and potential intervention. In addition to the established risk factors for BE, such as GE reflux diseases (GERD), hiatal hernia, being an elderly male, and Caucasian race [7], obesity has been frequently identified as a predisposing factor for BE [7–9]. However, almost all of the studies on risk factors for BE thus far have originated from Western countries, largely because the incidence of BE, and in particular long-segment BE, is much lower in Asian countries compared to Western countries [10, 11]. Since some studies have suggested that the risk factors differ among different ethnic groups [12, 13], the risk factors for Asian subjects with BE remain to be clarified. Nonetheless, there are a few studies from Japan that have investigated potential associations between BE and obesity, but with inconsistent results (a positive association or no association) [10, 14]. In this regard, correlations between reflux esophagitis and obesity in the Japanese population have also been somewhat conflicting [15–18], although a positive association has been observed in previous studies [15–17]. Meanwhile, the vast majority of subjects in these studies have had short-segment BE, and in particular ultra-short (2 cm [34]. Statistics Values regarding the length of BE are expressed as medians (IQR), and other values are expressed as means (SD). The distributions of the demographic characteristics and related factors were compared between BE patients and controls using Student’s paired t test for continuous variables and a sign test for categorical variables. All p values were two-sided, and p < 0.05 was considered statistically significant. In examining the association of BMI and other parameters with the risk of BE, conditional logistic regression analysis was used to control for potential confounders using the SAS PHREG procedure in the SAS version 8.2 statistical analysis software package (SAS Institute, Cary, N.C., USA), and OR with 95% CI were estimated. To test for trends, categorical variables were included in the model as continuous data. In our basic multivariable model, BMI (categorical variable), current smoking (yes or no), and current alcohol consumption (yes or no) were included as independent variables. Then, in order to estimate the potential contribution of GE reflux to the association between BMI and the risk of BE, various GERD-related parameters [heartburn, hiatal hernia, erosive esophagitis, and intake of proton pump inhibitors (PPI)] were included together in the basic model. We also undertook stratified analyses separately for males and females for all analyses.
Results
A total of 113 cases of BE >2 cm in length and the same number of sex- and age-matched controls were prospectively enrolled into this study. None of the subjects who were asked to participate refused to enroll into the study. Of the enrolled BE patients, 55 had been under surveillance. The mean age of the BE cases was BE and BMI in Japan
63.1 (12.7) years and the majority (93/113 = 82%) were male. Overall, the median length of BE according to the Prague C and M criteria was 3.0 (3.0) cm for the M value and 2.0 (2.0) for the C value, respectively. BE cases with lengths >3 cm accounted for 72% (81/113) of the enrolled subjects, with 4.0 (3.0) cm for the M value and 2.0 (3.0) cm for the C value (table 1). In 33 subjects, the C value was 3.0 cm or more. None of the BE cases showed mucosal irregularities such as a concavo-convex shape or had histologically proven dysplasia in the CLE. Of the 89 BE patients who allowed biopsies to be taken from the CLE, SIM was confirmed in 54 (61%) cases. Background Characteristics of BE Although height and body weight were not significantly different between the BE cases and the controls, the BMI was significantly higher in BE cases than in controls [24.4 (3.8) vs. 23.4 (3.0), p = 0.04]. The proportions of normal weight, overweight, and obesity in the controls according to WPRO criteria were 53, 20, and 27%, respectively, which is consistent with a previous report regarding BMI classification defined by the same criteria in normal Japanese individuals [29]. Meanwhile, the proportions of normal weight, overweight, and obesity in the BE patients were 32, 27, and 41%, respectively, and the distribution of subjects in the BMI categories was significantly different between the two groups (p < 0.01) (table 1). Various GERD-related parameters were more prevalent in BE patients than in controls, as expected. The prevalence of both erosive reflux esophagitis and hiatal hernia was significantly higher in BE patients than in controls (p < 0.01 for both), and especially hiatal hernia turned out to be highly prevalent (90%) in BE. Similarly, the prevalence of GERD symptoms such as heartburn and acid regurgitation was also significantly higher in BE patients than in controls (p < 0.01 for both). Concerning the drug intakes that we examined in the present study, PPI was administered to half of the BE patients, mainly to alleviate GERD symptoms, and such drug use was significantly more prevalent in BE patients than in controls (table 1). When the analysis was confined to subjects whose BE was longer than 3 cm, the above patients’ backgrounds were similar to those of the entire cohort. Namely, the mean BMI was significantly higher in BE patients than in controls [24.3 (4.0) vs. 23.2 (3.0), p = 0.04], and the distribution of subjects in the BMI categories was significantly different between BE patients and controls (p < 0.01) (table 1). Digestion 2014;90:1–9 DOI: 10.1159/000357776
3
Table 1. Characteristics of the subjects with BE with a length of more than 2 or 3 cm
Parameters
Male/female ratio Age, years Height, cm Weight, kg BMI BMI 25.0 Length of BE Median M value Median C value Current smokers Current drinkers Erosive esophagitis Hiatal hernia GERD symptom Heartburn Acid regurgitation Dysphagia Medication PPI H2 blockers Aspirin NSAIDs
Overall subjects (more than 2 cm of BE), n (%)
p value
More than 3 cm of BE, n (%)
p value
BE cases (n = 113)
controls (n = 113)
BE cases (n = 81)
controls (n = 81)
93/20 63.1 (12.7) 163.0 (9.6) 65.2 (14.2) 24.4 (3.8)
93/20 63.4 (13.2) 163.4 (9.9) 62.6 (10.0) 23.4 (3.0)
68/13 64.6 (12.2) 162.9 (9.8) 65.0 (14.7) 24.3 (4.0)
68/13 64.7 (12.9) 163.2 (10.1) 61.7 (9.6) 23.2 (3.0)
36 (31.9) 31 (27.4) 46 (40.7)
60 (53.1) 22 (19.5) 31 (27.4)
28 (34.6) 21 (25.9) 32 (39.5)
46 (56.8) 15 (18.5) 20 (25.0)
3.0 (3.0) 2.0 (2.0) 14 (12.4) 53 (46.9) 32 (28.3) 102 (90.3)
19 (16.8) 54 (47.8) 5 (4.4) 30 (26.5)
NS NS
