Abstract
Overweight and obesity, indicated as increased body mass index, are associated with the risk of some cancers. We carried out a meta-analysis on published cohort and case–control studies to assess the strength of association between body mass index and gastric cancer.
Relevant studies were identified through PubMed, Web of Science and Medline electronic databases. Adjusted relative risks (odds ratios) with 95% confidence interval were used to assess the strength of association between body mass index and gastric cancer.
Sixteen eligible studies were included in this meta-analysis. Overall, obesity (body mass index ≥30 kg/m2) was associated with an increased risk of gastric cancer (odds ratio = 1.13, 95% confidence interval = 1.03–1.24) compared with normal weight (body mass index = 18.5 to <25 kg/m2), while overweight (body mass index = 18.5 to <30 kg/m2) showed no association (odds ratio = 1.04, 95% confidence interval = 0.96–1.12). Specifically, a stratified analysis showed there were associations between obesity and the increased risk of gastric cancer for males (odds ratio = 1.27, 95% confidence interval = 1.09–1.48), non-Asians (odds ratio = 1.14, 95% confidence interval = 1.02–1.28) and both cohort studies (odds ratio = 1.10, 95% confidence interval = 1.00–1.22) and case–control studies (odds ratio = 1.29, 95% confidence interval = 1.03–1.60). Both overweight (odds ratio = 1.22, 95% confidence interval = 1.05–1.42) and obesity (odds ratio = 1.61, 95% confidence interval = 1.15–2.24) were associated with the increased risk of gastric cardia cancer.
The results indicated that obesity was associated with the risk of gastric cancer, especially for males and among non-Asians. Both overweight and obesity were associated with the risk of gastric cardia cancer.
INTRODUCTION
Gastric cancer, a cancer of the digestive tract, can be traced throughout the development of national economy and modern civilization. It is associated with high incidence rate in Asia, as is the case with mortality rate (1). As reported in the latest global cancer statistics, the incidence and mortality rate of gastric cancer are higher in developing countries and in males. Gastric cancer is the fourth frequently diagnosed cancer and the third cause of cancer death among males and it is the fifth leading cause of cancer incidence and death among females (2).
In 2007, the report of World Cancer Research Fund (WCRF) used a more standardized approach to review the evidence, and concluded that the evidence that body fatnesses associated with increased risk of esophageal adenocarcinoma, and with cancers of the pancreas, colorectum, post-menopausal breast, endometrium and kidney was convincing, and that a probably association existed between body fatness and risk of gallbladder cancer (3). In addition, obesity and overweight were linked to increased risk for liver and biliary tract cancer (4) and papillary thyroid cancer (5). However, the abundant epidemiological studies conducted to explore the association between excess body weight and the risk of gastric cancer have shown inconclusive results. Some researchers found that the excess body weight was related to increased risk of gastric cancer (6,7), but others found no association (8–11). It is possible that the different results are caused by differences in exposure definition, outcome judging, gastric cancer site, study designing (cohort or case–control) and the limitations of region and race.
Therefore, to overcome these limitations, we conducted a meta-analysis of 16 eligible cohort and case–control studies to clarify a more accurate estimation of the association between excess body weight and risk of gastric cancer.
PATIENTS AND METHODS
Search Strategy
We conducted a comprehensive search strategy towards the electronic databases including PubMed, Web of Science and Medline using the key words ‘gastric cancer or gastric tumor or gastric carcinoma’, ‘stomach cancer or stomach tumor or stomach carcinoma’, ‘BMI or body mass index’ and ‘obesity or adiposity or overweight’. We reviewed the title and abstract of all citations and retrieved literatures. At the same time, the retrieved reference lists were also reviewed to additional relevant studies manually.
Selection Criteria
The included studies must meet the following criteria: (i) original study, (ii) cohort or case–control study in which gastric cancer incidence or mortality was taken as outcome, (iii) the diagnosis of gastric cancer was based on histological, (iv) having clear description of normal weight, overweight and obesity defined by BMI, (v) relative risk (RR), hazard ratio(HR) or odds ratio(OR) with its corresponding 95% confidence interval (95% CI) for BMI and incidence or mortality of gastric cancer were included, (vi) the RR, HR or OR with its corresponding 95% CI at least adjusted for age, (vii) when multiple publications reported on the same or overlapping data, we used the most recent or largest population and (viii) publication language was confined to English.
Data Extraction
According to pre-specified selection criteria, data was extracted from each study by two reviewers independently (L.X.J. and L.S.). Decisions were made and disagreements about study selection were resolved by consensus or by involving a third reviewer (Y.K.K.). The following information was extracted from the publications: first author, publishing year, country where the study was carried out, study design, duration of data collection, gender of participants, anatomic site, sample size, cutoff of BMI (kg/m2), study population and confounder adjustments. When incidence and mortality were reported as outcomes at the same time, we chose the incidence. When more than one risk estimates were reported, we extracted the ones adjusted for the largest number of potential confounders.
Quality Assessment for Individual Studies
Two reviewers (L.X.J. and L.S.) assessed the quality of each included study using the Newcastle–Ottawa scale (NOS). The ‘star system’ of NOS has been developed in which a study was judged on three broad perspectives: the selection of the study groups, the comparability of the groups and the ascertainment of either the exposure or outcome of interest for case–control or cohort studies, respectively. The NOS assigned a maximum of 4 points for selection, 2 points for comparability and 3 points for outcome. A maximum of 9 points reflected the highest quality. Any disagreements were resolved by a joint reevaluation with a third reviewer (Y.K.K.). A total of 6 points or greater indicated the study was high quality (12).
Exposure Definition
We defined body mass categories using the following BMI (BMI = weight in kilograms/height in meter2) according to World Health Organization (WHO): overweight (BMI = 25 to <30 kg/m2) and obesity (BMI ≥ 30 kg/m2) compared with normal weight (BMI = 18.5 to <25 kg/m2). The current standards based on BMI criteria for overweight and obesity recommended by the WHO are widely accepted and supported by other advisory committees and expert panels to federal agencies (13). But the Chinese people are relatively lean, a population-based investigation conducted by Zhou (14) in China suggested that the cutoff points of BMI for underweight, overweight and obesity of the Chinese people were 18.5, 24 and 28 kg/m2. We also accepted the criteria for Chinese in our study.
Statistical Analysis
Adjusted RRs (HRs or ORs) with 95% CI were used to assess the strength of association between BMI and risk of gastric cancer. Heterogeneity assumption was checked by χ2 test based on Q-test and I2 test (significance level of Phet < 0.10). The pooled RRs with 95% CI were calculated by the random effects model (DerSimonian and Laird) (15). Possible publication bias was tested by Begg's funnel plot and Egger's test (significance level of P < 0.05) (16,17). Subgroup analyses were conducted based on gender (males and females), site of gastric cancer (cardia and non-cardia), study population (Asian and non-Asian) and study design (cohort and case–control). All statistical analyses were performed with STATA version 12.0 software packages (STATA, College Station, TX).
RESULTS
Search Results and Study Characteristics
A total of 1512 articles were relevant to the key words. After excluding for duplication, review, case report, letter, comment, not relevant to the association of gastric cancer and BMI, mechanism study and surgery study and after selecting seriously in accordance with the inclusive criteria, 23 studies were considered of potential value and being evaluated for details. Seven of these 23 articles did not use the WHO categories of BMI exposure. Sixteen finally eligible studies (6–8,11,18–29) were identified, including 13 cohort studies and 3 case–control studies. Among these studies, 4 were carried out in USA, 3 in UK, 3 in Sweden, 1 in the Netherlands, 1 in Norway, 1 in Austria, 1 in China, 1 in Korea and 1 in Japan. The main characteristics of the included studies were summarized in Table 1.
Main characteristics of the 16 studies included in the meta-analysis
| First author, year | Country | Study period | Gender | Anatomic site | Cases/cohort size (controls) | Cutoff of BMI, kg/m2 | Study population | NOS score | Adjustments |
|---|---|---|---|---|---|---|---|---|---|
| Cohort study | |||||||||
| Calle (2003) | USA | 1982–98 | M + F | Stomach | 1453/900 053 | 18.5–24.9 25.0–29.9 30.0–34.9 35.0–39.9 | Cancer Prevention Study II | 7 | Age, education, smoking status, alcohol, physical activity, diet, race, marital status, aspirin use |
| Rapp (2005) | Austria | 1985–2001 | M + F | Stomach | 264/145 931 | 18.5–24.9 25.0–29.9 30.0–34.9 ≥35.0 | VHM&PP in Vorarlberg | 8 | Age, smoking status, occupational group |
| Kuriyama S (2005) | Japan | 1984–92 | M + F | Stomach | 440/27 539 | 18.5–24.9 25.0–27.4 27.5–29.9 ≥30.0 | Residents in 3 municipalities of Miyagi Prefecture | 7 | Age, smoking status, alcohol drinking status, diet, type of health insurance |
| Batty G.D (2005) | UK | 1967–1970 | M | Stomach | 190/18 403 | 18.5–24.9 25.0–29.9 ≥30.0 | Non-industrial London-based male government employees | 7 | Age, plus employment grade, physical activity, smoking habit, marital status, disease at entry, others |
| Samanic (2006) | Sweden | 1971–92 | M | Cardia Non-cardia | 1281/362 552 | 18.5–24.9 25.0–29.9 ≥30.0 | Workers in the construction industry | 6 | Age, calendar years, smoking status |
| Lukanova (2006) | Sweden | 1985–2003 | M + F | Stomach | 72/68 786 | 18.5–24.9 25.0–29.9 ≥30.0 | The Northern Sweden Health and Disease Cohort | 8 | Age, calendar years, smoking, all subjects |
| Reeves (2007) | UK | 1996–2001 | F | Stomach | 521/1 222 630 | <22.5 22.5–24.9 25–27.4 27.5–29.5 ≥30.0 | For screening for breast cancer throughout England and Scotland | 7 | Age, geographical region, socioeconomic status, reproductive history, smoking, alcohol intake, physical activity, hormone therapy |
| Merry (2007) | The Netherlands | 1986–99 | M + F | Cardia Non-cardia | 603/120 852 | <20.0 20.0–24.9 25–29.9 ≥30.0 | The Netherlands Cohort Study | 8 | Age, sex, current smoking, education |
| Abnet (2008) | USA | 1995–96 | M + F | Cardia Non-cardia | 622/480 475 | <18.5 18.5–25.0 25.0–<30.0 30.0–<35.0 ≥35.0 | American Association of Retired Persons members | 7 | Age, sex, cigarette smoking, alcohol consumption, education, physical activity |
| Jee (2008) | Korea | 1992–95 | M + F | Stomach | 18 684/1 213 829 | <20.0 20.0–22.9 23.0–24.9 25.0–29.9 ≥30.0 | National Health Insurance Corporation enrollees | 8 | Age, smoking status |
| Sjödahl (2008) | Norway | 1984–2002 | M + F | Cardia Non-cardia | 313/73 133 | 18.5–24.9 25.0–29.9 ≥30.0 | Inhabitants participated in the Nord-Trondelag Health Study | 8 | Age, recreational physical activity level, smoking, alcohol drinking, salt intake, occupation |
| O'Doherty (2012) | USA | 1995–2006 | M + F | Cardia Non-cardia | 316/218 854 | <18.5 18.5–25.0 25.0–<30.0 30.0–<35.0 ≥35.0 | The NIH-AARP Diet and Health study | 7 | Age, sex, total energy, antacid use, aspirin use, non-steroidal anti-inflammatory drugs, marital status, diabetes, cigarette smoking, education, ethnicity, alcohol, physical activity, diet |
| Lindkvist (2013) | Sweden | 2006 | M + F | Stomach | 1210/578 700 | <18.5 18.5–25.0 25.0–<30.0 ≥30.0 | The Metabolic Syndrome and Cancer Project | 7 | Age, smoking status |
| Case–control study | |||||||||
| Lindblad (2005) | UK | 1994–2001 | M + F | Cardia Non-cardia | 522/10 000 | <20.0 20.0–24.0 25.0–29.0 ≥30.0 | General Practitioners Research Database members | 7 | Age, sex, calendar year, smoking, alcohol consumption, reflux |
| Corley (2008) | USA | 1964–73 | M + F | Cardia | 105/2800 | <18.5 18.5–24.9 25.0–29.9 ≥30.0 | Kaiser Permanent health plan members | 6 | age, sex, year of heath checkup, ethnicity |
| Zhang (2003) | China | 1995–2002 | M + F | Cardia | 300/258 | <18.5 18.5–24.0 24.0–28.0 ≥28 | Patients and controls at Beijing Cancer Hospital | 7 | Age, sex |
| First author, year | Country | Study period | Gender | Anatomic site | Cases/cohort size (controls) | Cutoff of BMI, kg/m2 | Study population | NOS score | Adjustments |
|---|---|---|---|---|---|---|---|---|---|
| Cohort study | |||||||||
| Calle (2003) | USA | 1982–98 | M + F | Stomach | 1453/900 053 | 18.5–24.9 25.0–29.9 30.0–34.9 35.0–39.9 | Cancer Prevention Study II | 7 | Age, education, smoking status, alcohol, physical activity, diet, race, marital status, aspirin use |
| Rapp (2005) | Austria | 1985–2001 | M + F | Stomach | 264/145 931 | 18.5–24.9 25.0–29.9 30.0–34.9 ≥35.0 | VHM&PP in Vorarlberg | 8 | Age, smoking status, occupational group |
| Kuriyama S (2005) | Japan | 1984–92 | M + F | Stomach | 440/27 539 | 18.5–24.9 25.0–27.4 27.5–29.9 ≥30.0 | Residents in 3 municipalities of Miyagi Prefecture | 7 | Age, smoking status, alcohol drinking status, diet, type of health insurance |
| Batty G.D (2005) | UK | 1967–1970 | M | Stomach | 190/18 403 | 18.5–24.9 25.0–29.9 ≥30.0 | Non-industrial London-based male government employees | 7 | Age, plus employment grade, physical activity, smoking habit, marital status, disease at entry, others |
| Samanic (2006) | Sweden | 1971–92 | M | Cardia Non-cardia | 1281/362 552 | 18.5–24.9 25.0–29.9 ≥30.0 | Workers in the construction industry | 6 | Age, calendar years, smoking status |
| Lukanova (2006) | Sweden | 1985–2003 | M + F | Stomach | 72/68 786 | 18.5–24.9 25.0–29.9 ≥30.0 | The Northern Sweden Health and Disease Cohort | 8 | Age, calendar years, smoking, all subjects |
| Reeves (2007) | UK | 1996–2001 | F | Stomach | 521/1 222 630 | <22.5 22.5–24.9 25–27.4 27.5–29.5 ≥30.0 | For screening for breast cancer throughout England and Scotland | 7 | Age, geographical region, socioeconomic status, reproductive history, smoking, alcohol intake, physical activity, hormone therapy |
| Merry (2007) | The Netherlands | 1986–99 | M + F | Cardia Non-cardia | 603/120 852 | <20.0 20.0–24.9 25–29.9 ≥30.0 | The Netherlands Cohort Study | 8 | Age, sex, current smoking, education |
| Abnet (2008) | USA | 1995–96 | M + F | Cardia Non-cardia | 622/480 475 | <18.5 18.5–25.0 25.0–<30.0 30.0–<35.0 ≥35.0 | American Association of Retired Persons members | 7 | Age, sex, cigarette smoking, alcohol consumption, education, physical activity |
| Jee (2008) | Korea | 1992–95 | M + F | Stomach | 18 684/1 213 829 | <20.0 20.0–22.9 23.0–24.9 25.0–29.9 ≥30.0 | National Health Insurance Corporation enrollees | 8 | Age, smoking status |
| Sjödahl (2008) | Norway | 1984–2002 | M + F | Cardia Non-cardia | 313/73 133 | 18.5–24.9 25.0–29.9 ≥30.0 | Inhabitants participated in the Nord-Trondelag Health Study | 8 | Age, recreational physical activity level, smoking, alcohol drinking, salt intake, occupation |
| O'Doherty (2012) | USA | 1995–2006 | M + F | Cardia Non-cardia | 316/218 854 | <18.5 18.5–25.0 25.0–<30.0 30.0–<35.0 ≥35.0 | The NIH-AARP Diet and Health study | 7 | Age, sex, total energy, antacid use, aspirin use, non-steroidal anti-inflammatory drugs, marital status, diabetes, cigarette smoking, education, ethnicity, alcohol, physical activity, diet |
| Lindkvist (2013) | Sweden | 2006 | M + F | Stomach | 1210/578 700 | <18.5 18.5–25.0 25.0–<30.0 ≥30.0 | The Metabolic Syndrome and Cancer Project | 7 | Age, smoking status |
| Case–control study | |||||||||
| Lindblad (2005) | UK | 1994–2001 | M + F | Cardia Non-cardia | 522/10 000 | <20.0 20.0–24.0 25.0–29.0 ≥30.0 | General Practitioners Research Database members | 7 | Age, sex, calendar year, smoking, alcohol consumption, reflux |
| Corley (2008) | USA | 1964–73 | M + F | Cardia | 105/2800 | <18.5 18.5–24.9 25.0–29.9 ≥30.0 | Kaiser Permanent health plan members | 6 | age, sex, year of heath checkup, ethnicity |
| Zhang (2003) | China | 1995–2002 | M + F | Cardia | 300/258 | <18.5 18.5–24.0 24.0–28.0 ≥28 | Patients and controls at Beijing Cancer Hospital | 7 | Age, sex |
M, male; F, female. VHM & PP was Vorarlberg Health Monitoring and Promotion Program. NIH-AARP Diet and Health Study was developed at the National Cancer Institute of the National Institutes of Health to improve our understanding of the relationship between diet and health.
Main characteristics of the 16 studies included in the meta-analysis
| First author, year | Country | Study period | Gender | Anatomic site | Cases/cohort size (controls) | Cutoff of BMI, kg/m2 | Study population | NOS score | Adjustments |
|---|---|---|---|---|---|---|---|---|---|
| Cohort study | |||||||||
| Calle (2003) | USA | 1982–98 | M + F | Stomach | 1453/900 053 | 18.5–24.9 25.0–29.9 30.0–34.9 35.0–39.9 | Cancer Prevention Study II | 7 | Age, education, smoking status, alcohol, physical activity, diet, race, marital status, aspirin use |
| Rapp (2005) | Austria | 1985–2001 | M + F | Stomach | 264/145 931 | 18.5–24.9 25.0–29.9 30.0–34.9 ≥35.0 | VHM&PP in Vorarlberg | 8 | Age, smoking status, occupational group |
| Kuriyama S (2005) | Japan | 1984–92 | M + F | Stomach | 440/27 539 | 18.5–24.9 25.0–27.4 27.5–29.9 ≥30.0 | Residents in 3 municipalities of Miyagi Prefecture | 7 | Age, smoking status, alcohol drinking status, diet, type of health insurance |
| Batty G.D (2005) | UK | 1967–1970 | M | Stomach | 190/18 403 | 18.5–24.9 25.0–29.9 ≥30.0 | Non-industrial London-based male government employees | 7 | Age, plus employment grade, physical activity, smoking habit, marital status, disease at entry, others |
| Samanic (2006) | Sweden | 1971–92 | M | Cardia Non-cardia | 1281/362 552 | 18.5–24.9 25.0–29.9 ≥30.0 | Workers in the construction industry | 6 | Age, calendar years, smoking status |
| Lukanova (2006) | Sweden | 1985–2003 | M + F | Stomach | 72/68 786 | 18.5–24.9 25.0–29.9 ≥30.0 | The Northern Sweden Health and Disease Cohort | 8 | Age, calendar years, smoking, all subjects |
| Reeves (2007) | UK | 1996–2001 | F | Stomach | 521/1 222 630 | <22.5 22.5–24.9 25–27.4 27.5–29.5 ≥30.0 | For screening for breast cancer throughout England and Scotland | 7 | Age, geographical region, socioeconomic status, reproductive history, smoking, alcohol intake, physical activity, hormone therapy |
| Merry (2007) | The Netherlands | 1986–99 | M + F | Cardia Non-cardia | 603/120 852 | <20.0 20.0–24.9 25–29.9 ≥30.0 | The Netherlands Cohort Study | 8 | Age, sex, current smoking, education |
| Abnet (2008) | USA | 1995–96 | M + F | Cardia Non-cardia | 622/480 475 | <18.5 18.5–25.0 25.0–<30.0 30.0–<35.0 ≥35.0 | American Association of Retired Persons members | 7 | Age, sex, cigarette smoking, alcohol consumption, education, physical activity |
| Jee (2008) | Korea | 1992–95 | M + F | Stomach | 18 684/1 213 829 | <20.0 20.0–22.9 23.0–24.9 25.0–29.9 ≥30.0 | National Health Insurance Corporation enrollees | 8 | Age, smoking status |
| Sjödahl (2008) | Norway | 1984–2002 | M + F | Cardia Non-cardia | 313/73 133 | 18.5–24.9 25.0–29.9 ≥30.0 | Inhabitants participated in the Nord-Trondelag Health Study | 8 | Age, recreational physical activity level, smoking, alcohol drinking, salt intake, occupation |
| O'Doherty (2012) | USA | 1995–2006 | M + F | Cardia Non-cardia | 316/218 854 | <18.5 18.5–25.0 25.0–<30.0 30.0–<35.0 ≥35.0 | The NIH-AARP Diet and Health study | 7 | Age, sex, total energy, antacid use, aspirin use, non-steroidal anti-inflammatory drugs, marital status, diabetes, cigarette smoking, education, ethnicity, alcohol, physical activity, diet |
| Lindkvist (2013) | Sweden | 2006 | M + F | Stomach | 1210/578 700 | <18.5 18.5–25.0 25.0–<30.0 ≥30.0 | The Metabolic Syndrome and Cancer Project | 7 | Age, smoking status |
| Case–control study | |||||||||
| Lindblad (2005) | UK | 1994–2001 | M + F | Cardia Non-cardia | 522/10 000 | <20.0 20.0–24.0 25.0–29.0 ≥30.0 | General Practitioners Research Database members | 7 | Age, sex, calendar year, smoking, alcohol consumption, reflux |
| Corley (2008) | USA | 1964–73 | M + F | Cardia | 105/2800 | <18.5 18.5–24.9 25.0–29.9 ≥30.0 | Kaiser Permanent health plan members | 6 | age, sex, year of heath checkup, ethnicity |
| Zhang (2003) | China | 1995–2002 | M + F | Cardia | 300/258 | <18.5 18.5–24.0 24.0–28.0 ≥28 | Patients and controls at Beijing Cancer Hospital | 7 | Age, sex |
| First author, year | Country | Study period | Gender | Anatomic site | Cases/cohort size (controls) | Cutoff of BMI, kg/m2 | Study population | NOS score | Adjustments |
|---|---|---|---|---|---|---|---|---|---|
| Cohort study | |||||||||
| Calle (2003) | USA | 1982–98 | M + F | Stomach | 1453/900 053 | 18.5–24.9 25.0–29.9 30.0–34.9 35.0–39.9 | Cancer Prevention Study II | 7 | Age, education, smoking status, alcohol, physical activity, diet, race, marital status, aspirin use |
| Rapp (2005) | Austria | 1985–2001 | M + F | Stomach | 264/145 931 | 18.5–24.9 25.0–29.9 30.0–34.9 ≥35.0 | VHM&PP in Vorarlberg | 8 | Age, smoking status, occupational group |
| Kuriyama S (2005) | Japan | 1984–92 | M + F | Stomach | 440/27 539 | 18.5–24.9 25.0–27.4 27.5–29.9 ≥30.0 | Residents in 3 municipalities of Miyagi Prefecture | 7 | Age, smoking status, alcohol drinking status, diet, type of health insurance |
| Batty G.D (2005) | UK | 1967–1970 | M | Stomach | 190/18 403 | 18.5–24.9 25.0–29.9 ≥30.0 | Non-industrial London-based male government employees | 7 | Age, plus employment grade, physical activity, smoking habit, marital status, disease at entry, others |
| Samanic (2006) | Sweden | 1971–92 | M | Cardia Non-cardia | 1281/362 552 | 18.5–24.9 25.0–29.9 ≥30.0 | Workers in the construction industry | 6 | Age, calendar years, smoking status |
| Lukanova (2006) | Sweden | 1985–2003 | M + F | Stomach | 72/68 786 | 18.5–24.9 25.0–29.9 ≥30.0 | The Northern Sweden Health and Disease Cohort | 8 | Age, calendar years, smoking, all subjects |
| Reeves (2007) | UK | 1996–2001 | F | Stomach | 521/1 222 630 | <22.5 22.5–24.9 25–27.4 27.5–29.5 ≥30.0 | For screening for breast cancer throughout England and Scotland | 7 | Age, geographical region, socioeconomic status, reproductive history, smoking, alcohol intake, physical activity, hormone therapy |
| Merry (2007) | The Netherlands | 1986–99 | M + F | Cardia Non-cardia | 603/120 852 | <20.0 20.0–24.9 25–29.9 ≥30.0 | The Netherlands Cohort Study | 8 | Age, sex, current smoking, education |
| Abnet (2008) | USA | 1995–96 | M + F | Cardia Non-cardia | 622/480 475 | <18.5 18.5–25.0 25.0–<30.0 30.0–<35.0 ≥35.0 | American Association of Retired Persons members | 7 | Age, sex, cigarette smoking, alcohol consumption, education, physical activity |
| Jee (2008) | Korea | 1992–95 | M + F | Stomach | 18 684/1 213 829 | <20.0 20.0–22.9 23.0–24.9 25.0–29.9 ≥30.0 | National Health Insurance Corporation enrollees | 8 | Age, smoking status |
| Sjödahl (2008) | Norway | 1984–2002 | M + F | Cardia Non-cardia | 313/73 133 | 18.5–24.9 25.0–29.9 ≥30.0 | Inhabitants participated in the Nord-Trondelag Health Study | 8 | Age, recreational physical activity level, smoking, alcohol drinking, salt intake, occupation |
| O'Doherty (2012) | USA | 1995–2006 | M + F | Cardia Non-cardia | 316/218 854 | <18.5 18.5–25.0 25.0–<30.0 30.0–<35.0 ≥35.0 | The NIH-AARP Diet and Health study | 7 | Age, sex, total energy, antacid use, aspirin use, non-steroidal anti-inflammatory drugs, marital status, diabetes, cigarette smoking, education, ethnicity, alcohol, physical activity, diet |
| Lindkvist (2013) | Sweden | 2006 | M + F | Stomach | 1210/578 700 | <18.5 18.5–25.0 25.0–<30.0 ≥30.0 | The Metabolic Syndrome and Cancer Project | 7 | Age, smoking status |
| Case–control study | |||||||||
| Lindblad (2005) | UK | 1994–2001 | M + F | Cardia Non-cardia | 522/10 000 | <20.0 20.0–24.0 25.0–29.0 ≥30.0 | General Practitioners Research Database members | 7 | Age, sex, calendar year, smoking, alcohol consumption, reflux |
| Corley (2008) | USA | 1964–73 | M + F | Cardia | 105/2800 | <18.5 18.5–24.9 25.0–29.9 ≥30.0 | Kaiser Permanent health plan members | 6 | age, sex, year of heath checkup, ethnicity |
| Zhang (2003) | China | 1995–2002 | M + F | Cardia | 300/258 | <18.5 18.5–24.0 24.0–28.0 ≥28 | Patients and controls at Beijing Cancer Hospital | 7 | Age, sex |
M, male; F, female. VHM & PP was Vorarlberg Health Monitoring and Promotion Program. NIH-AARP Diet and Health Study was developed at the National Cancer Institute of the National Institutes of Health to improve our understanding of the relationship between diet and health.
Quantitative Synthesis
In obesity group, the overall analysis revealed a positive association between obesity (BMI ≥ 30 kg/m2) and gastric cancer risk (OR = 1.13, 95% CI = 1.03–1.24) compared with normal weight (BMI = 18.5 to <25 kg/m2). No significant heterogeneity was observed (I2 = 7.7%, P = 0.368) (Fig. 1A). For gender, there was an association between obesity and gastric cancer risk for males (OR = 1.27, 95% CI = 1.09–1.48), but no association for females (OR = 1.04, 95% CI = 0.79–1.39) (Table 2). For the site of gastric cancer, there was an association between obesity and gastric cardia cancer risk (GCC) (OR = 1.61, 95% CI = 1.15–2.24). But we found no significant association between obesity and gastric non-cardia cancer risk (GNCC) (OR = 0.83, 95% CI = 0.68–1.01) (Table 2). For race, among non-Asians, there was an association between obesity and gastric cancer risk (OR = 1.14, 95% CI = 1.02–1.28). Nonetheless, no association between obesity and gastric cancer risk was observed among Asians (OR = 1.11, 95% CI = 0.82–1.50) (Table 2). For study design, there was associations between obesity and gastric cancer risk for cohort studies (OR = 1.10, 95% CI = 1.00–1.22) and case–control studies (OR = 1.29, 95% CI = 1.03–1.60) (Table 2). Meanwhile, it was worth mentioning that no significant heterogeneity was observed in the above subgroups (Table 2).
Stratified meta-analyses of the association between BMI and gastric cancer
| Study groups | Obesity (BMI ≥ 30 kg/m2) | Overweight (25 ≤ BMI < 30 kg/m2) | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| N1 | OR(95% CI) | I2 | Phet | E− T2 | N1 | OR(95% CI) | I2 | Phet | E− T2 | |
| All | 14 | 1.13 (1.03–1.24) | 7.7% | 0.368 | 0.520 | 15 | 1.04 (0.96–1.12) | 45.8% | 0.027 | 0.080 |
| Gender | ||||||||||
| Male | 5 | 1.27 (1.09–1.48) | 0.0% | 0.408 | 0.440 | 5 | 1.12 (0.96–1.29) | 55.2% | 0.063 | 0.375 |
| Female | 3 | 1.04 (0.79–1.39) | 0.0% | 0.790 | 0.139 | 3 | 0.87 (0.71–1.05) | 0.0% | 0.501 | 0.495 |
| Cancer site | ||||||||||
| Cardia | 5 | 1.61 (1.15–2.24) | 38% | 0.168 | 0.543 | 7 | 1.22 (1.05–1.42) | 19.4% | 0.282 | 0.494 |
| Non-cardia | 4 | 0.83 (0.68–1.01) | 0.0% | 0.508 | 0.627 | 6 | 0.94 (0.81–1.10) | 46.1% | 0.098 | 0.026 |
| Race | ||||||||||
| Asia | 3 | 1.11 (0.82–1.50) | 0.0% | 0.771 | 0.835 | 3 | 1.41 (0.93–2.14) | 82.3% | 0.003 | 0.011 |
| Non-Asia | 11 | 1.14 (1.02–1.28) | 26.1% | 0.195 | 0.490 | 12 | 0.98 (0.92–1.04) | 0.0% | 0.559 | 0.432 |
| Design | ||||||||||
| Cohort | 11 | 1.10 (1.00–1.22) | 7.0% | 0.337 | 0.999 | 12 | 1.00 (0.93–1.08) | 29.8% | 0.154 | 0.178 |
| Case–control | 3 | 1.29 (1.03–1.60) | 0.0% | 0.408 | 0.352 | 3 | 1.24 (0.83–1.85) | 70.7% | 0.033 | 0.745 |
| Study groups | Obesity (BMI ≥ 30 kg/m2) | Overweight (25 ≤ BMI < 30 kg/m2) | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| N1 | OR(95% CI) | I2 | Phet | E− T2 | N1 | OR(95% CI) | I2 | Phet | E− T2 | |
| All | 14 | 1.13 (1.03–1.24) | 7.7% | 0.368 | 0.520 | 15 | 1.04 (0.96–1.12) | 45.8% | 0.027 | 0.080 |
| Gender | ||||||||||
| Male | 5 | 1.27 (1.09–1.48) | 0.0% | 0.408 | 0.440 | 5 | 1.12 (0.96–1.29) | 55.2% | 0.063 | 0.375 |
| Female | 3 | 1.04 (0.79–1.39) | 0.0% | 0.790 | 0.139 | 3 | 0.87 (0.71–1.05) | 0.0% | 0.501 | 0.495 |
| Cancer site | ||||||||||
| Cardia | 5 | 1.61 (1.15–2.24) | 38% | 0.168 | 0.543 | 7 | 1.22 (1.05–1.42) | 19.4% | 0.282 | 0.494 |
| Non-cardia | 4 | 0.83 (0.68–1.01) | 0.0% | 0.508 | 0.627 | 6 | 0.94 (0.81–1.10) | 46.1% | 0.098 | 0.026 |
| Race | ||||||||||
| Asia | 3 | 1.11 (0.82–1.50) | 0.0% | 0.771 | 0.835 | 3 | 1.41 (0.93–2.14) | 82.3% | 0.003 | 0.011 |
| Non-Asia | 11 | 1.14 (1.02–1.28) | 26.1% | 0.195 | 0.490 | 12 | 0.98 (0.92–1.04) | 0.0% | 0.559 | 0.432 |
| Design | ||||||||||
| Cohort | 11 | 1.10 (1.00–1.22) | 7.0% | 0.337 | 0.999 | 12 | 1.00 (0.93–1.08) | 29.8% | 0.154 | 0.178 |
| Case–control | 3 | 1.29 (1.03–1.60) | 0.0% | 0.408 | 0.352 | 3 | 1.24 (0.83–1.85) | 70.7% | 0.033 | 0.745 |
All the groups were compared with normal weight (BMI = 18.5 to <25 kg/m2) as the reference category. Heterogeneity assumption was checked by I2and Phet. N1, number of studies; E − T2, P value of Egger's test.
Stratified meta-analyses of the association between BMI and gastric cancer
| Study groups | Obesity (BMI ≥ 30 kg/m2) | Overweight (25 ≤ BMI < 30 kg/m2) | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| N1 | OR(95% CI) | I2 | Phet | E− T2 | N1 | OR(95% CI) | I2 | Phet | E− T2 | |
| All | 14 | 1.13 (1.03–1.24) | 7.7% | 0.368 | 0.520 | 15 | 1.04 (0.96–1.12) | 45.8% | 0.027 | 0.080 |
| Gender | ||||||||||
| Male | 5 | 1.27 (1.09–1.48) | 0.0% | 0.408 | 0.440 | 5 | 1.12 (0.96–1.29) | 55.2% | 0.063 | 0.375 |
| Female | 3 | 1.04 (0.79–1.39) | 0.0% | 0.790 | 0.139 | 3 | 0.87 (0.71–1.05) | 0.0% | 0.501 | 0.495 |
| Cancer site | ||||||||||
| Cardia | 5 | 1.61 (1.15–2.24) | 38% | 0.168 | 0.543 | 7 | 1.22 (1.05–1.42) | 19.4% | 0.282 | 0.494 |
| Non-cardia | 4 | 0.83 (0.68–1.01) | 0.0% | 0.508 | 0.627 | 6 | 0.94 (0.81–1.10) | 46.1% | 0.098 | 0.026 |
| Race | ||||||||||
| Asia | 3 | 1.11 (0.82–1.50) | 0.0% | 0.771 | 0.835 | 3 | 1.41 (0.93–2.14) | 82.3% | 0.003 | 0.011 |
| Non-Asia | 11 | 1.14 (1.02–1.28) | 26.1% | 0.195 | 0.490 | 12 | 0.98 (0.92–1.04) | 0.0% | 0.559 | 0.432 |
| Design | ||||||||||
| Cohort | 11 | 1.10 (1.00–1.22) | 7.0% | 0.337 | 0.999 | 12 | 1.00 (0.93–1.08) | 29.8% | 0.154 | 0.178 |
| Case–control | 3 | 1.29 (1.03–1.60) | 0.0% | 0.408 | 0.352 | 3 | 1.24 (0.83–1.85) | 70.7% | 0.033 | 0.745 |
| Study groups | Obesity (BMI ≥ 30 kg/m2) | Overweight (25 ≤ BMI < 30 kg/m2) | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| N1 | OR(95% CI) | I2 | Phet | E− T2 | N1 | OR(95% CI) | I2 | Phet | E− T2 | |
| All | 14 | 1.13 (1.03–1.24) | 7.7% | 0.368 | 0.520 | 15 | 1.04 (0.96–1.12) | 45.8% | 0.027 | 0.080 |
| Gender | ||||||||||
| Male | 5 | 1.27 (1.09–1.48) | 0.0% | 0.408 | 0.440 | 5 | 1.12 (0.96–1.29) | 55.2% | 0.063 | 0.375 |
| Female | 3 | 1.04 (0.79–1.39) | 0.0% | 0.790 | 0.139 | 3 | 0.87 (0.71–1.05) | 0.0% | 0.501 | 0.495 |
| Cancer site | ||||||||||
| Cardia | 5 | 1.61 (1.15–2.24) | 38% | 0.168 | 0.543 | 7 | 1.22 (1.05–1.42) | 19.4% | 0.282 | 0.494 |
| Non-cardia | 4 | 0.83 (0.68–1.01) | 0.0% | 0.508 | 0.627 | 6 | 0.94 (0.81–1.10) | 46.1% | 0.098 | 0.026 |
| Race | ||||||||||
| Asia | 3 | 1.11 (0.82–1.50) | 0.0% | 0.771 | 0.835 | 3 | 1.41 (0.93–2.14) | 82.3% | 0.003 | 0.011 |
| Non-Asia | 11 | 1.14 (1.02–1.28) | 26.1% | 0.195 | 0.490 | 12 | 0.98 (0.92–1.04) | 0.0% | 0.559 | 0.432 |
| Design | ||||||||||
| Cohort | 11 | 1.10 (1.00–1.22) | 7.0% | 0.337 | 0.999 | 12 | 1.00 (0.93–1.08) | 29.8% | 0.154 | 0.178 |
| Case–control | 3 | 1.29 (1.03–1.60) | 0.0% | 0.408 | 0.352 | 3 | 1.24 (0.83–1.85) | 70.7% | 0.033 | 0.745 |
All the groups were compared with normal weight (BMI = 18.5 to <25 kg/m2) as the reference category. Heterogeneity assumption was checked by I2and Phet. N1, number of studies; E − T2, P value of Egger's test.
Forest plot of meta-analysis for association between BMI (A for obesity and B for overweight) and gastric cancer risk in all studies.
Forest plot of meta-analysis for association between BMI (A for obesity and B for overweight) and gastric cancer risk in all studies.
In overweight group, compared with normal weight (BMI = 18.5 to <25 kg/m2), the overall analysis revealed no association between overweight (BMI = 25 to <30 kg/m2) and gastric cancer risk (OR = 1.04, 95% CI = 0.96–1.12) (Fig. 1B). But, as shown in Table 2, there was an association between overweight and GCC risk (OR = 1.22, 95% CI = 1.05–1.42). Then, no statistically significant link between overweight and gastric cancer risk was observed for males, females, GNCC, Asians, non-Asians, cohort studies and case–control studies (Table 2).
As shown in Fig. 2, the shape of Begg's funnel plots for overweight and obesity group revealed no obvious asymmetry. Then the results were confirmed by Egger's test (for obesity group: P = 0.520; for overweight group: P = 0.080). Neither Begg's funnel plots nor Egger's test detected any obvious evidence of publication bias of literatures (Table 2).
Funnel plot of meta-analysis for association between BMI (A for obesity and B for overweight) and gastric cancer risk in all studies.
Funnel plot of meta-analysis for association between BMI (A for obesity and B for overweight) and gastric cancer risk in all studies.
When the component studies were restricted to cohort studies, we can also find obesity was associated with an increased risk of gastric cancer (OR = 1.10, 95% CI = 1.00–1.22), while overweight showed no association (OR = 1.00, 95% CI = 0.93–1.08). For the site of gastric cancer, obesity was associated with the risk of gastric cardia cancer (OR = 1.68, 95% CI = 1.14–2.47), but not with gastric non-cardia cancer (OR = 0.82, 95% CI = 0.66–1.02). The above conclusions were not altered. But we no longer found an association between overweight and risk of gastric cardia cancer (OR = 1.16, 95% CI = 0.99–1.35), it might be caused by the few samples.
Sensitive Analysis
In the sensitivity analysis, when one study was removed and the rest were analyzed sequentially by meta-analysis. Any study in overweight or obesity group was omitted, the pooled RRs were not materially altered with the overall pooled RRs, indicating that our results were statistically robust (Fig. 3).
The plot of sensitive analysis for association between BMI (A for obesity and B for overweight) and gastric cancer risk in all studies.
The plot of sensitive analysis for association between BMI (A for obesity and B for overweight) and gastric cancer risk in all studies.
DISCUSSION
Previous meta-analysis studies have revealed that higher BMI was associated with the risk of various cancers (30,31). Yang et al. (32) found that excess body weight (BMI ≥ 25 kg/m2) was associated with an increased risk of gastric cancer and GCC but not with GNCC by a meta-analysis of 10 cohort studies. In this study, the BMI references were not unified, such as 18.5–25, <20.3, <23 and <19.9, which will lead the imprecise conclusion. Moreover, they did not explain the reason for the great heterogeneity (like I2 = 81%, P < 0.00001), also they did not take confounders into account. Not coincidentally, Chen et al. (33) found increasing BMI was not a clear risk factor for total gastric cancer, and increased BMI was positively associated with risk of GCC but not with GNCC. Although 24 prospective studies were included in this meta-analysis, but the various cutoff of BMI may lead conclusions to a lack of accuracy.
In our meta-analysis, we included 13 cohort and 3 case–control studies, which used RR, HR and OR to assess the risk of gastric cancer. We strictly specified the normal BMI reference as 18.5 to <25 kg/m2, so that the conclusion may be more meaningful. The pooled results showed that obesity was associated with an increased risk of gastric cancer, especially for males, GCC and among non-Asians.
In this meta-analysis, the association between obesity and gastric cancer was statistically significant for males (based on eight studies), but not for females (based on six studies). The dietary preferences of habit, food types and eating speed may be the reasons for the difference. What is more, the possible reason reflected the gender difference might relate to female sex hormones. Many epidemiologic studies have suggested hormonal factors associated with greater exposure to estrogen and/or progesterone may be associated with decreased risk for gastric cancer (34,35).
After stratifying by the site of gastric cancer, we found in overweight, obesity group, there was a significant link between BMI and GCC risk, the strength of the association increased with the increase of BMI, while no association for GNCC. Nowadays, the trend of GCC was rising, no matter in the developed countries like USA (36) or in developing countries like India (37). Many epidemiological studies have supported that obesity was a risk factor of GCC (27,38). There was also a meta-analysis indicating that a high BMI was weakly associated with the risk of GCC (39). Meanwhile, a new study reported that GNCC was associated with short stature but not with heavy body weight or obesity (40). Although the real mechanism remains unclear, among all the hypotheses of the association between obesity and GCC, the reflux theory was widely accepted (41,42). Obesity can promote gastroesophageal reflux disease by increasing intra-abdominal pressure. Moreover, gastroesophageal reflux predisposes to Barrett's esophagus, which is a metaplastic precursor state for GCC (43).
In this meta-analysis, we found a positive association between obesity and gastric cancer risk among non-Asians, but not among Asians. Similarly, the previous stratified meta-analysis also found that excess body weight (BMI ≥ 25) was associated with an increased risk of gastric cancer among non-Asians (32). Obesity-induced inflammation was thought to promote development of GCA via tumor necrosis factor-α (TNF-α), interleukin-6 (IL-6), and monocyte chemoattractant protein-1 (MCP-1). In vitro and in vivo studies have shown that TNF-a, IL-6, IL-17 and MCP-1 stimulate proliferation and inhibit apoptosis of human gastric cancer cell lines (44). The different conclusions among Asians and non-Asians might be caused by the ethnicity or geographical regions. Besides, only three studies were available in Asia with negative results, which might be caused by few samples.
Some limitations may have been affecting the objectivity of the conclusion of this meta-analysis and should be acknowledged. Although all the cohort and case–control studies included in this meta-analysis have high quality, the potentially unmeasured confounders may influence the results of observational studies. Observational studies are more susceptible to bias risk.
In conclusion, compared with normal weight (BMI = 18.5 to <25 kg/m2), obesity was positively associated with gastric cancer risk for total studies. We found a positive association between obesity and gastric cancer risk especially for males and among non-Asians. Both overweight and obesity were associated with the increased risk of GCC. Considering that some confounders may affect the conclusions more or less, some adjustments for age, ethnicity, education, Helicobacter pylori infection, fruit, vegetable, smoking, alcohol and processed meat are necessary in future studies.
Funding
This study was supported by Jilin Provincial Science and Technology Development Project (grant no. 201205008), Science and Technology Project of the Department of Health of Jilin Province (grant no. 2013Z034), Jilin province postdoctoral scientific research project (grant no. RB201343) and Youth Research Fund Project of Medical Research Support Program of Norman Bethune Health Science Center of Jilin University (grant no. 2013202013).
Conflict of interest statement
None declared.
References
Author notes
Xue-Jun Lin, Chun-Peng Wang and Xiao-Dong Liu contributed equally to this work and shared the co-first authors.
