Abstract
Background: To provide a quantitative assessment of the association between body mass index (BMI) and the risk of gastric cancer, we summarized the evidence from prospective studies.
Methods: Eligible studies published up to November 30, 2012, were retrieved via computer searches of MEDLINE and EMBASE as well as manual review of references. Summary relative risks (SRR) with their corresponding 95% confidence intervals (CI) were calculated using a random-effects model.
Results: A total of 24 prospective studies of BMI and gastric cancer risk with 41,791 cases were included in our analysis. Overall, both overweight (BMI, 25–30 kg/m2) and obesity (BMI, ≥30 kg/m2) were not associated with risk of total gastric cancer (overweight: SRR, 1.01; 95% CI, 0.96–1.07; obesity: SRR, 1.06; 95% CI, 0.99–1.12). Furthermore, we found increased BMI was positively associated with the risk of gastric cardia cancer (GCC; SRR = 1.21 for overweight and 1.82 for obesity), but not with gastric non-cardia cancer (GNCC; SRR = 0.93 for overweight and SRR = 1.00 for obesity). Similar results were observed in a linear dose–response analysis.
Conclusion: On the basis of meta-analysis of prospective studies, we find high BMI is positively associated with the risk of GCCs but not with GNCCs.
Impact: (i) On the basis of more definite and quantitative evidence than previously available, we found that increasing BMI was not a clear risk factor for total gastric cancer. (ii) Increased BMI was positively associated with risk of GCC but not with GNCCs. Cancer Epidemiol Biomarkers Prev; 22(8); 1395–408. ©2013 AACR.
Introduction
Gastric cancer is the fourth most common cancer type in the world and the second leading cause of death due to cancer (1). Gastric cancer incidence and mortality rates are different throughout the world, with rates much higher in parts of Asian (such as China, Japan, and Korea), Eastern Europe, and South America than those in the United States, where gastric cancer incidence rates have decreased substantially since the mid-1980s (2). Gastric cancers are classified as gastric non-cardia cancer (GNCC) and gastric cardia cancer (GCC) according to anatomic location of the lesion. In contrast to a significant decrease in the incidence of the GNCCs, a significant increase in the incidence rate of GCCs has been observed in Western countries (3, 4), which may suggest different etiology related to the 2 subtypes (3).
Recently, progress has been made through epidemiologic studies investigating environmental risk factors for gastric cancer. While GCCs may be related to gastroesophageal reflux (GER), white race, male gender, and tobacco smoking (5), the majority of GNCCs are attributable to chronic Helicobacter pylori infection (6), tobacco smoking (7), consumption of salt and salt-preserved foods (8, 9), and alcohol abuse (10).
The prevalence of obesity, as measured by body mass index (BMI, defined as a BMI ≥ 30 kg/m2), has grown markedly over the past 2 decades. A narrative review of the WCRF/AICR 2007 report has linked obesity with an increased risk for several cancers including cancers of the colon, breast, esophagus (adenocarcinoma), endometrium, kidney, as well as possibly additional sites (11). Unfortunately, in this report, no results on the association between obesity and gastric cancer risk were reported (12). A recent meta-analysis on this issue was conducted in 2009, which included 10 prospective studies involving a total number of 9,492 gastric cancer cases, and found that overweight (BMI = 25–29.9 kg/m2) and obesity were associated with an increased risk of gastric cancer [obesity: summary relative risks (SRR), 1.36, 95% confidence interval (CI), 1.21–1.54 and overweight: SRR, 1.21; 95% CI, 1.08–1.36; ref. 13]. Unfortunately, several large cohort studies were not included in this meta-analysis (14–25). Moreover, the previous analysis did not explore the effect of important confounders (such as smoking, alcohol use, and physical activity) on the relationship of BMI and gastric cancer. That analysis also did not explore the source of the observed heterogeneity among studies (P < 0.001). In addition, the exact shape of the dose–risk relationship between BMI and gastric cancer, specifically by GCCs and GNCCs, risk has not been clearly defined.
Therefore, to better characterize this issue, we conducted a comprehensive meta-analysis of the epidemiologic literature using our own methods and criteria in the selection of studies, in the presentation of data, and in our conclusions and interpretation of the evidence.
Materials and Methods
Literature search strategies
Two of us (Y. Chen and L. Liu) identified eligible studies by a comprehensive searching the MEDLINE and EMBASE databases through November 30, 2012. Research articles were selected using the following keywords or Medical Subject Heading (MeSH) terms: “body mass index”, “BMI”, or “obesity”, or “overweight”; “gastric” or “stomach” or “cardia”; and “cancer” or “carcinoma” or “neoplasia” or “adenocarcinoma”; and “risk” or “incidence” or “prevalence”. No language restrictions were imposed. Furthermore, the reference lists of retrieved articles were examined for additional relevant studies. This meta-analysis was planned, conducted, and reported according to the PRISMA statement (26).
Inclusion and exclusion criteria
Three authors (Y. Chen, X. Wang, and L. Liu) independently evaluated all of the studies retrieved according to the prespecified selection criteria. Discrepancies between the 3 reviewers were solved by discussion. To be included, the study had to meet the following criteria: (i) with a prospective design, (ii) reported relative risk (RR) estimates with corresponding 95% CIs for BMI and incidence of or mortality from GCCs, GNCCs, or all types; and (iii) the RRs and corresponding 95% CIs at least adjusted for age. If results based on the same study population were reported in more than one study, we included the one with the largest number of cases. So, four studies were excluded (27–30).
Data extraction
The following information from each included study were extracted using a standardized data collection protocol: the first author's last name, country of origin, publication year, sample size, the source of the study population, ascertainment of exposure and outcome, cutoff of exposure, duration of follow-up, and adjustments for confounders. When several risk estimates were presented, we used the ones adjusted for the largest number of potential confounders. Data abstraction was conducted independently by 2 investigators (L. Liu and Y. Chen) and then cross-checked. When both gastric cancer incidence and mortality data were provided, preference was given to the incidence, unless the necessary information was available only for mortality data.
Quality assessment for individual studies
Two reviewers (Y. Chen and L. Liu) assessed the quality of each selected study using the Newcastle–Ottawa quality assessment scale (NOS; ref. 31). For cohort studies, the NOS consists of 3 parameters of quality: selection, comparability, and outcome. The NOS assigns a maximum of 4 points for selection, 2 points for comparability, and 3 points for outcome. Thus, a maximum of 9 points reflects the highest quality. Any discrepancies were addressed by a joint re-evaluation of the original article with a third reviewer. A total score of 7 or greater was used to indicate high-quality studies and a total score of 6 or smaller indicated low-quality studies.
Statistical analysis
We defined body mass categories using the following BMI categories: normal (BMI = 18.5–24.9 kg/m2), overweight (BMI = 25–29.9 kg/m2), and obesity (BMI ≥ 30 kg/m2). All statistical analyses were conducted using STATA, version 11.0 (STATA). A 2-tailed P < 0.05 was considered statistically significant. SRR estimates with their corresponding 95% CIs were derived by the method of DerSimonian and Laird using the assumptions of a random-effects model, which incorporates between-study variability (32).
For studies that reported RRs for several categories of BMI that fell into the category representing overweight or obesity, we pooled the RRs (weighted by the inverse of their variance) and used the pooled RRs in the meta-analysis. When sex-specific results were presented, respectively, and did not provide information for both genders, we first analyzed together for gastric cancer (weighted by the inverse of their variance) and then separately for males and females. When subtype-specific estimates were available and did not provide information for total gastric cancer (17, 18, 24, 33–36), they were only included in the analysis of cardia and non-cardia cancers. When studies provided both total gastric cancer and subtype-specific estimates (21, 37, 38), they were included in the analysis for both total gastric cancer and subtype-specific estimates, respectively.
We also estimated the RR per 5 kg/m2 increase in BMI by regressing the natural logarithm of the RRs to the corresponding median values of BMI across exposure categories in each study, using the variance-weighted least-squares regression (39, 40). This requires that the number of cases and person-time or non-cases for at least 3 quantitative exposure categories is known. When the highest and the lowest categories of exposure were open, we considered them of the same amplitude as the previous categories.
Heterogeneity among studies was assessed using the Q and I2 statistics, which tested total variation across studies that was attributable to heterogeneity rather than to chance (41). For I2, values of larger than 50% are assigned to high heterogeneity (42). Sources of heterogeneity were explored using subgroup analyses and random-effects meta-regression analysis, according to sex, geographic locations (Asia and non-Asia), duration of follow-up (≥10 vs. <10 years), assessment of BMI (self-reported vs. measured), and outcome (cancer/death registry vs. histologic finding), study quality score (high vs. low), and adjustment for confounding factors including smoking, alcohol use, and physical activity. Sensitivity analysis was conducted to estimate the influence of each individual study on the summary estimates by repeating the random-effects meta-analysis after omitting one study at a time. Funnel plots and statistical tests for funnel plot asymmetry were conducted to test evidence of publication bias. We carried out formal testing using Begg's adjusted rank correlation and Egger's regression asymmetry test (43, 44).
Results
Search results and study characteristics
There were 1,620 articles relevant to the search words, of which 54 were considered of potential value and being retrieved for detailed evaluation. Additional 7 articles were included from reference review. Thirty-seven of these 61 articles were subsequently excluded from the meta-analysis for various reasons (Fig. 1). Thus, a total of 24 prospective studies, which involved 41,791 gastric cancer cases and a total studied population of more than 10 million, were found to match our inclusion criteria. Of these 24 studies, 7 studies were conducted in Asia, 4 in the United States, 11 in Europe, and 2 in Australia. According to the NOS, 17 studies (71%) were of high quality (NOS score > 6) and 7 (29%) of acceptable quality (NOS score ≤ 6). The most common selection bias was underrepresentativeness of that exposed cohort was not from general population in 6 studies (25%). The most common outcome bias was the lack of reporting of completeness of follow-up in 14 studies (58%). Characteristics of studies included in the meta-analysis were presented in Table 1.
Flow diagram of systematic literature search on BMI and risk of gastric cancer.
Flow diagram of systematic literature search on BMI and risk of gastric cancer.
Characteristics of prospective studies of BMI and risk of gastric cancer
| Author, year, country | Study participants, n | Anatomic site | No. of cases | Follow-up, y | Assessment of anthropometry | Case ascertainment | Cutoff of BMI, kg/m2 | Newcastle–Ottawa score | Adjustments |
|---|---|---|---|---|---|---|---|---|---|
| Møller and colleagues, 1994, Denmark (14) | General population total N = 43,968 | Stomach | 30 (M), 43 (F) | 4.8 | Discharge diagnosis | Death certificates | <30 | 6 | Age |
| ≥30 | |||||||||
| Tulinius and colleagues, 1997, Iceland (15) | The Cardiovascular Risk Factor Study. Total N = 22,946 | Stomach | 171 (M), 75 (F) | NA | Self-reported | Cancer registry | Per 1 | 5 | Age |
| Wolk and colleagues, 2001, Sweden (16) | Hospital patients total N = 28,129 | Stomach | 19 (M), 40 (F) | 10.3 | Discharge diagnosis | Cancer registry | >30 (M); >28.6 (F) | 6 | Age, calendar year |
| Calle and colleagues, 2003, USA (20) | Cancer Prevention Study II, Total N = 900,053 | Stomach | 945 (M), 508 (F) | 16 | Self-reported | Death certificates | 18.5–24.9 | 8 | Age, smoking, alcohol, physical activity, diet, education, HRT, marital status, NSAIDs use |
| 25.0–29.9 | |||||||||
| 30.0–34.9 | |||||||||
| 35.0–39.9 | |||||||||
| ≥40.0 | |||||||||
| Samanic and colleagues, 2004, USA (21) | White and black veterans total N = 4,500,700 (M) | Stomach cardia | 7,486 (M) | 12 | Discharge diagnosis | Cancer registry | <30 | 5 | Age, calendar year |
| ≥30 | |||||||||
| Batty and colleagues, 2005, UK (22) | Nonindustrial London-based government employees total N = 18,403 (M) | Stomach | 190 (M) | 35 | Measured | Death certificates | 18.5–24.9 | 6 | Age, smoking, physical activity, socioeconomic status, marital status, other |
| 25.0–29.9 | |||||||||
| ≥30 | |||||||||
| Kuriyama and colleagues, 2005, Japan (58) | Population subjects total N = 27,539 | Stomach | 314 (M), 126 (F) | 7.6 | Self-reported | Cancer registry | 18.5–24.9 | 8 | Age, smoking, alcohol, diet, type of health insurance |
| 25.0–27.4 | |||||||||
| 27.5–29.9 | |||||||||
| ≥30.0 | |||||||||
| Lindblad and colleagues, 2005, UK (37) | General Practitioners Research Database 10000 (controls) | Stomach | 1,023, M + F | 7 | Self-reported | Histologic finding | <20 | 6 | Age, sex, calendar year, smoking, alcohol consumption, reflux |
| Cardia | 20–24 | ||||||||
| Non-cardia | 25–29 | ||||||||
| ≥30.0 | |||||||||
| Oh and colleagues, 2005, Korea (33) | Civil servants and private school workers and their dependants total N = 781,283 (M) | Non-cardia | 5,293, M + F | 9.9 | Measured | Histologic finding | <18.5 | 7 | Age, smoking, alcohol, family History, physical activity, place of birth |
| 18.5–22.9 | |||||||||
| 23.0–24.9 | |||||||||
| 25.0–26.9 | |||||||||
| 27.0–29.9 | |||||||||
| ≥30.0 | |||||||||
| Rapp and colleagues, 2005, Australia (23) | Adult Vorarlberg residents total N = 67,447 (M) | Stomach | 146 (M), 118 (F) | 10 | Measured | Histologic finding | 18.5–24.9 | 8 | Age, smoking, socioeconomic status |
| 25–29.9 | |||||||||
| 30–34.9 | |||||||||
| ≥35 | |||||||||
| Tran and colleagues, 2005, China (17) | The Linxian General Population Trial total N = 29,584 | Cardia | 1,452, M + F | 15 | Measured | Radiology+ pathology | <20 | 8 | Age, smoking |
| Non-cardia | 20—21 | ||||||||
| 21–22 | |||||||||
| >23 | |||||||||
| Lukanova and colleagues, 2006, Sweden (59) | Northern Sweden Health and Disease Cohort total N = 256,000 | Stomach | 46 (M), 26 (F) | 8.3 | Measured | Cancer registry | 18.5–23.4 | 8 | Age, calendar year and smoking |
| 23.5–25.3 | |||||||||
| 25.4–27.6 | |||||||||
| ≥27.7 | |||||||||
| Maclnnis and colleagues, 2006, Australia (34) | Melbourne Collaborative Cohort Study total N = 41,528 | Cardia | 68, M + F | 6 | Measured | Cancer registry | <25 | 9 | Sex, country of birth, highest level of education and physical activity |
| Non-cardia | 25–29 | ||||||||
| >30 | |||||||||
| Samanic and colleagues, 2006, Sweden (35) | Workers of Construction Industry total N = 362,552 (M) | Cardia | 1,052, M + F | 19 | Measured | Cancer registry | 18.5–24.9 | 6 | Age, smoking |
| Non-cardia | 25–29.9 | ||||||||
| >30 | |||||||||
| Merry and colleagues, 2007, Netherlands (36) | The Netherlands Cohort Study total N = 120,852 | Cardia | 603, M + F | 13.3 | Self-reported | Cancer registry | <20.0 | 8 | Age, sex, current smoking, education |
| Non-cardia | 20–24.9 | ||||||||
| 25–29.9 | |||||||||
| >30 | |||||||||
| Reeves and colleagues, 2007, UK (60) | The Million Women Study total N = 1.3 million | Stomach | 521, M + F | 5.4 | Self-reported | Cancer registry | <22.5 | 7 | Age, geographical region, socioeconomic status, reproductive history, smoking status, alcohol intake, physical activity, menopause, HRT |
| 22.5–24.9 | |||||||||
| 25.0–27.4 | |||||||||
| 27.5–29.9 | |||||||||
| ≥30 | |||||||||
| Tanaka and colleagues, 2007, Japan (61) | A community-based cohort of Japanese men and women total N = 28,443 | Stomach | 86 (M), 43 (F) | 7 | Self-reported | Death certificates | <20.3 | 8 | Age, smoking history, alcohol intake, education, physical activity score and marital status. |
| 20.3–22.2 | |||||||||
| >22.2 | |||||||||
| Corley and colleagues, 2008, USA (24) | Multiphasic health checkup cohort consists of Kaiser Permanente health plan members total N = 206,974 | Cardia | 105, M + F | 22.5 | Measured | Cancer registry and individual record review | 18.5–24.9 | 8 | Age, sex, ethnicity and year of heath checkup |
| 25.0–29.9 | |||||||||
| ≥30 | |||||||||
| Jee and colleagues, 2008, Korea (25) | National Health Insurance Corporation total N = 1,213,829 | Stomach | 14,568 (M), 4,116 (F) | 10.8 | Measured | Cancer registry or hospital admission diagnosis | <20.0 | 8 | Age, smoking |
| 20.0–22.9 | |||||||||
| 23.0–24.9 | |||||||||
| 25.0–29.9 | |||||||||
| ≥30.0 | |||||||||
| Persson and colleagues, 2008, Japan (62) | Japan Public Health Center–based Prospective Study total N = 44,453 | Stomach | 368 (F) | 12.2 | Self-reported | Histologic finding | <19.9 | 7 | Age, family history of gastric cancer, and study area |
| 20–24.9 | |||||||||
| >25 | |||||||||
| Sjodahl and colleagues, 2008, Norway (38) | The Nord Trondelag Health Study total N = 73,133 | Stomach | 126 (F), 187 (M) | 15.4 | Measured | Cancer registry | <20.0 | 8 | Age, recreational physical activity level, smoking, alcohol drinking, salt intake, and occupation |
| Non-cardia | 20.0–22.9 | ||||||||
| 23.0–24.9 | |||||||||
| 25.0–29.9 | |||||||||
| ≥30.0 | |||||||||
| Inoue and colleagues, 2009, Japan (63) | Inhabitants total N = 27,724 | Stomach | 233 (M), 138 (F) | 10.2 | Measured | Cancer registries | <25 | 9 | Age, study area, smoking status ethanol intake and total serum cholesterol |
| ≥25 | |||||||||
| O'Doherty and colleagues, 2012, USA (18) | The NIH-AARP Diet and Health study total N = 480,475 | Cardia | 316, M + F | 9 | Self-reported | Cancer registry | 18.5–24.9 | 7 | Age, sex, total energy, antacid use, NSAID use, marital status, diabetes, cigarette smoking, education, ethnicity, alcohol consumption, physical activity, red and white meat intake, and fruit and vegetable intake |
| Non-cardia | 25.0–29.9 | ||||||||
| ≥30 | |||||||||
| Lindkvist and colleagues, 2013, Sweden (19) | The Metabolic Syndrome and Cancer Project total N = 578,700 | Stomach | 828 (M), 382 (F) | NA | Measured | Cancer registry | 21.5 | 7 | Age, sex, study cohort, smoking |
| 23.8 | |||||||||
| 25.4 | |||||||||
| 27.1 | |||||||||
| 30.8 |
| Author, year, country | Study participants, n | Anatomic site | No. of cases | Follow-up, y | Assessment of anthropometry | Case ascertainment | Cutoff of BMI, kg/m2 | Newcastle–Ottawa score | Adjustments |
|---|---|---|---|---|---|---|---|---|---|
| Møller and colleagues, 1994, Denmark (14) | General population total N = 43,968 | Stomach | 30 (M), 43 (F) | 4.8 | Discharge diagnosis | Death certificates | <30 | 6 | Age |
| ≥30 | |||||||||
| Tulinius and colleagues, 1997, Iceland (15) | The Cardiovascular Risk Factor Study. Total N = 22,946 | Stomach | 171 (M), 75 (F) | NA | Self-reported | Cancer registry | Per 1 | 5 | Age |
| Wolk and colleagues, 2001, Sweden (16) | Hospital patients total N = 28,129 | Stomach | 19 (M), 40 (F) | 10.3 | Discharge diagnosis | Cancer registry | >30 (M); >28.6 (F) | 6 | Age, calendar year |
| Calle and colleagues, 2003, USA (20) | Cancer Prevention Study II, Total N = 900,053 | Stomach | 945 (M), 508 (F) | 16 | Self-reported | Death certificates | 18.5–24.9 | 8 | Age, smoking, alcohol, physical activity, diet, education, HRT, marital status, NSAIDs use |
| 25.0–29.9 | |||||||||
| 30.0–34.9 | |||||||||
| 35.0–39.9 | |||||||||
| ≥40.0 | |||||||||
| Samanic and colleagues, 2004, USA (21) | White and black veterans total N = 4,500,700 (M) | Stomach cardia | 7,486 (M) | 12 | Discharge diagnosis | Cancer registry | <30 | 5 | Age, calendar year |
| ≥30 | |||||||||
| Batty and colleagues, 2005, UK (22) | Nonindustrial London-based government employees total N = 18,403 (M) | Stomach | 190 (M) | 35 | Measured | Death certificates | 18.5–24.9 | 6 | Age, smoking, physical activity, socioeconomic status, marital status, other |
| 25.0–29.9 | |||||||||
| ≥30 | |||||||||
| Kuriyama and colleagues, 2005, Japan (58) | Population subjects total N = 27,539 | Stomach | 314 (M), 126 (F) | 7.6 | Self-reported | Cancer registry | 18.5–24.9 | 8 | Age, smoking, alcohol, diet, type of health insurance |
| 25.0–27.4 | |||||||||
| 27.5–29.9 | |||||||||
| ≥30.0 | |||||||||
| Lindblad and colleagues, 2005, UK (37) | General Practitioners Research Database 10000 (controls) | Stomach | 1,023, M + F | 7 | Self-reported | Histologic finding | <20 | 6 | Age, sex, calendar year, smoking, alcohol consumption, reflux |
| Cardia | 20–24 | ||||||||
| Non-cardia | 25–29 | ||||||||
| ≥30.0 | |||||||||
| Oh and colleagues, 2005, Korea (33) | Civil servants and private school workers and their dependants total N = 781,283 (M) | Non-cardia | 5,293, M + F | 9.9 | Measured | Histologic finding | <18.5 | 7 | Age, smoking, alcohol, family History, physical activity, place of birth |
| 18.5–22.9 | |||||||||
| 23.0–24.9 | |||||||||
| 25.0–26.9 | |||||||||
| 27.0–29.9 | |||||||||
| ≥30.0 | |||||||||
| Rapp and colleagues, 2005, Australia (23) | Adult Vorarlberg residents total N = 67,447 (M) | Stomach | 146 (M), 118 (F) | 10 | Measured | Histologic finding | 18.5–24.9 | 8 | Age, smoking, socioeconomic status |
| 25–29.9 | |||||||||
| 30–34.9 | |||||||||
| ≥35 | |||||||||
| Tran and colleagues, 2005, China (17) | The Linxian General Population Trial total N = 29,584 | Cardia | 1,452, M + F | 15 | Measured | Radiology+ pathology | <20 | 8 | Age, smoking |
| Non-cardia | 20—21 | ||||||||
| 21–22 | |||||||||
| >23 | |||||||||
| Lukanova and colleagues, 2006, Sweden (59) | Northern Sweden Health and Disease Cohort total N = 256,000 | Stomach | 46 (M), 26 (F) | 8.3 | Measured | Cancer registry | 18.5–23.4 | 8 | Age, calendar year and smoking |
| 23.5–25.3 | |||||||||
| 25.4–27.6 | |||||||||
| ≥27.7 | |||||||||
| Maclnnis and colleagues, 2006, Australia (34) | Melbourne Collaborative Cohort Study total N = 41,528 | Cardia | 68, M + F | 6 | Measured | Cancer registry | <25 | 9 | Sex, country of birth, highest level of education and physical activity |
| Non-cardia | 25–29 | ||||||||
| >30 | |||||||||
| Samanic and colleagues, 2006, Sweden (35) | Workers of Construction Industry total N = 362,552 (M) | Cardia | 1,052, M + F | 19 | Measured | Cancer registry | 18.5–24.9 | 6 | Age, smoking |
| Non-cardia | 25–29.9 | ||||||||
| >30 | |||||||||
| Merry and colleagues, 2007, Netherlands (36) | The Netherlands Cohort Study total N = 120,852 | Cardia | 603, M + F | 13.3 | Self-reported | Cancer registry | <20.0 | 8 | Age, sex, current smoking, education |
| Non-cardia | 20–24.9 | ||||||||
| 25–29.9 | |||||||||
| >30 | |||||||||
| Reeves and colleagues, 2007, UK (60) | The Million Women Study total N = 1.3 million | Stomach | 521, M + F | 5.4 | Self-reported | Cancer registry | <22.5 | 7 | Age, geographical region, socioeconomic status, reproductive history, smoking status, alcohol intake, physical activity, menopause, HRT |
| 22.5–24.9 | |||||||||
| 25.0–27.4 | |||||||||
| 27.5–29.9 | |||||||||
| ≥30 | |||||||||
| Tanaka and colleagues, 2007, Japan (61) | A community-based cohort of Japanese men and women total N = 28,443 | Stomach | 86 (M), 43 (F) | 7 | Self-reported | Death certificates | <20.3 | 8 | Age, smoking history, alcohol intake, education, physical activity score and marital status. |
| 20.3–22.2 | |||||||||
| >22.2 | |||||||||
| Corley and colleagues, 2008, USA (24) | Multiphasic health checkup cohort consists of Kaiser Permanente health plan members total N = 206,974 | Cardia | 105, M + F | 22.5 | Measured | Cancer registry and individual record review | 18.5–24.9 | 8 | Age, sex, ethnicity and year of heath checkup |
| 25.0–29.9 | |||||||||
| ≥30 | |||||||||
| Jee and colleagues, 2008, Korea (25) | National Health Insurance Corporation total N = 1,213,829 | Stomach | 14,568 (M), 4,116 (F) | 10.8 | Measured | Cancer registry or hospital admission diagnosis | <20.0 | 8 | Age, smoking |
| 20.0–22.9 | |||||||||
| 23.0–24.9 | |||||||||
| 25.0–29.9 | |||||||||
| ≥30.0 | |||||||||
| Persson and colleagues, 2008, Japan (62) | Japan Public Health Center–based Prospective Study total N = 44,453 | Stomach | 368 (F) | 12.2 | Self-reported | Histologic finding | <19.9 | 7 | Age, family history of gastric cancer, and study area |
| 20–24.9 | |||||||||
| >25 | |||||||||
| Sjodahl and colleagues, 2008, Norway (38) | The Nord Trondelag Health Study total N = 73,133 | Stomach | 126 (F), 187 (M) | 15.4 | Measured | Cancer registry | <20.0 | 8 | Age, recreational physical activity level, smoking, alcohol drinking, salt intake, and occupation |
| Non-cardia | 20.0–22.9 | ||||||||
| 23.0–24.9 | |||||||||
| 25.0–29.9 | |||||||||
| ≥30.0 | |||||||||
| Inoue and colleagues, 2009, Japan (63) | Inhabitants total N = 27,724 | Stomach | 233 (M), 138 (F) | 10.2 | Measured | Cancer registries | <25 | 9 | Age, study area, smoking status ethanol intake and total serum cholesterol |
| ≥25 | |||||||||
| O'Doherty and colleagues, 2012, USA (18) | The NIH-AARP Diet and Health study total N = 480,475 | Cardia | 316, M + F | 9 | Self-reported | Cancer registry | 18.5–24.9 | 7 | Age, sex, total energy, antacid use, NSAID use, marital status, diabetes, cigarette smoking, education, ethnicity, alcohol consumption, physical activity, red and white meat intake, and fruit and vegetable intake |
| Non-cardia | 25.0–29.9 | ||||||||
| ≥30 | |||||||||
| Lindkvist and colleagues, 2013, Sweden (19) | The Metabolic Syndrome and Cancer Project total N = 578,700 | Stomach | 828 (M), 382 (F) | NA | Measured | Cancer registry | 21.5 | 7 | Age, sex, study cohort, smoking |
| 23.8 | |||||||||
| 25.4 | |||||||||
| 27.1 | |||||||||
| 30.8 |
Abbreviations: HRT, hormone therapy; NA, data not applicable; NSAID, nonsteroidal anti-inflammatory drugs.
Total gastric cancer
As shown in Fig. 2, the overall RR was 1.01 (95% CI, 0.96–1.07) for overweight (Fig. 2A), and 1.06 (95% CI, 0.99–1.12) for obesity (Fig. 2B). There was no evidence of heterogeneity among studies (overweight: Pheterogeneity = 0.239 and I2 = 22.2%; obesity: Pheterogeneity = 0.490 and I2 = 0).
SRRs of gastric cancer for BMI between 25 and 30 kg/m2 (A) and BMI over 30 kg/m2 (B).
SRRs of gastric cancer for BMI between 25 and 30 kg/m2 (A) and BMI over 30 kg/m2 (B).
Table 2 showed the results of subgroup meta-analyses by gender, geographic locations, ascertainment of exposure and outcome, duration of follow-up, study quality, and adjustments for confounders (smoking, alcohol use, and physical activity). The pooled estimates and 95% CIs were somewhat higher for males than those for females [males: 1.07 (1.01–1.13) for overweight and 1.12 (1.00–1.24) for obesity; females: 0.99 (0.89–1.11) for overweight and 1.04 (0.93–1.16) for obesity]. For the overweight and obese category, results for the majority of the subgroups were not significant, except several subgroups: studies with self-reported anthropometry and adjustments for smoking or physical activity for obesity–gastric cancer link (Table 2).
Stratified meta-analyses of BMI and the risk of total gastric cancer
| Overweight | Obesity | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Characteristics | n | SRR (95% CI) | Phet | Pd | I2, % | n | SRR (95% CI) | Phet | Pd | I2, % |
| All | 10 | 1.01 (0.96–1.07) | 0.239 | 22.2 | 13 | 1.06 (0.99–1.12) | 0.490 | 0 | ||
| Sex | ||||||||||
| Male | 6 | 1.07 (1.01–1.03) | 0.980 | 0 | 10 | 1.12 (1.00–1.24) | 0.204 | 26.0 | ||
| Female | 6 | 0.99 (0.89–1.11) | 0.145 | 0.112 | 39.1 | 9 | 1.04 (0.93–1.16) | 0.416 | 0.374 | 2.3 |
| Geographic location | ||||||||||
| Asia | 2 | 1.06 (1.00–1.12) | 0.394 | 4.6 | 2 | 1.06 (0.99–1.14) | 0.575 | 0 | ||
| Non-Asia | 8 | 0.98 (0.92–1.04) | 0.466 | 0.213 | 0 | 11 | 1.01 (0.84–1.21) | 0.370 | 0.736 | 7.8 |
| BMI ascertainment | ||||||||||
| Self-reported | 4 | 1.03 (0.96–1.11) | 0.486 | 0 | 7 | 1.08 (1.01–1.16) | 0.501 | 0 | ||
| Measured | 6 | 0.99 (0.90–1.08) | 0.112 | 0.421 | 43.9 | 6 | 0.98 (0.87–1.11) | 0.523 | 0.534 | 0 |
| Outcome ascertainment | ||||||||||
| Cancer/death registry | 8 | 1.01 (0.95–1.07) | 0.155 | 34.2 | 11 | 1.05 (0.99–1.12) | 0.504 | 0 | ||
| Histologic finding | 2 | 1.03 (0.88–1.21) | 0.348 | 0 | 0.835 | 2 | 1.06 (0.71–1.56) | 0.174 | 0.893 | 45.8 |
| Duration of follow-up, y | ||||||||||
| <10 | 3 | 1.09 (0.98–1.20) | 0.858 | 0 | 6 | 1.03 (0.93–1.15) | 0.514 | 0 | ||
| ≥10 | 7 | 0.98 (0.90–1.06) | 0.096 | 0.192 | 46.5 | 7 | 1.07 (0.98–1.18) | 0.327 | 0.645 | 13.5 |
| Study quality score | ||||||||||
| High (NOS score > 6) | 7 | 1.04 (0.99–1.08) | 0.800 | 8 | 1.04 (0.94–1.15) | 0.373 | 7.4 | |||
| Low (NOS score ≤ 6) | 3 | 0.99 (0.82–1.19) | 0.082 | 0.142 | 59.9 | 5 | 1.07 (0.99–1.16) | 0.449 | 0.948 | 0 |
| Adjustment for confounders | ||||||||||
| Alcohol use, yes | 6 | 1.04 (0.96–1.11) | 0.746 | 0 | 6 | 1.16 (1.04–1.31) | 0.555 | 0 | ||
| No | 4 | 0.97 (0.87–1.09) | 0.036 | 0.343 | 64.9 | 7 | 1.02 (0.95–1.09) | 0.702 | 0.797 | 0 |
| Smoking, yes | 10 | 1.01 (0.96–1.07) | 0.239 | 22.2 | 10 | 1.07 (0.97–1.18) | 0.314 | 14.0 | ||
| No | — | — | — | — | — | 3 | 1.05 (0.96–1.14) | 0.629 | 0.105 | 0 |
| Physical activity, yes | 4 | 1.01 (0.93–1.10) | 0.733 | 0 | 4 | 1.14 (1.02–1.33) | 0.390 | 0.3 | ||
| No | 6 | 1.01 (0.93–1.10) | 0.069 | 0.896 | 51.1 | 9 | 1.03 (0.96–1.10) | 0.665 | 0.135 | 0 |
| Overweight | Obesity | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Characteristics | n | SRR (95% CI) | Phet | Pd | I2, % | n | SRR (95% CI) | Phet | Pd | I2, % |
| All | 10 | 1.01 (0.96–1.07) | 0.239 | 22.2 | 13 | 1.06 (0.99–1.12) | 0.490 | 0 | ||
| Sex | ||||||||||
| Male | 6 | 1.07 (1.01–1.03) | 0.980 | 0 | 10 | 1.12 (1.00–1.24) | 0.204 | 26.0 | ||
| Female | 6 | 0.99 (0.89–1.11) | 0.145 | 0.112 | 39.1 | 9 | 1.04 (0.93–1.16) | 0.416 | 0.374 | 2.3 |
| Geographic location | ||||||||||
| Asia | 2 | 1.06 (1.00–1.12) | 0.394 | 4.6 | 2 | 1.06 (0.99–1.14) | 0.575 | 0 | ||
| Non-Asia | 8 | 0.98 (0.92–1.04) | 0.466 | 0.213 | 0 | 11 | 1.01 (0.84–1.21) | 0.370 | 0.736 | 7.8 |
| BMI ascertainment | ||||||||||
| Self-reported | 4 | 1.03 (0.96–1.11) | 0.486 | 0 | 7 | 1.08 (1.01–1.16) | 0.501 | 0 | ||
| Measured | 6 | 0.99 (0.90–1.08) | 0.112 | 0.421 | 43.9 | 6 | 0.98 (0.87–1.11) | 0.523 | 0.534 | 0 |
| Outcome ascertainment | ||||||||||
| Cancer/death registry | 8 | 1.01 (0.95–1.07) | 0.155 | 34.2 | 11 | 1.05 (0.99–1.12) | 0.504 | 0 | ||
| Histologic finding | 2 | 1.03 (0.88–1.21) | 0.348 | 0 | 0.835 | 2 | 1.06 (0.71–1.56) | 0.174 | 0.893 | 45.8 |
| Duration of follow-up, y | ||||||||||
| <10 | 3 | 1.09 (0.98–1.20) | 0.858 | 0 | 6 | 1.03 (0.93–1.15) | 0.514 | 0 | ||
| ≥10 | 7 | 0.98 (0.90–1.06) | 0.096 | 0.192 | 46.5 | 7 | 1.07 (0.98–1.18) | 0.327 | 0.645 | 13.5 |
| Study quality score | ||||||||||
| High (NOS score > 6) | 7 | 1.04 (0.99–1.08) | 0.800 | 8 | 1.04 (0.94–1.15) | 0.373 | 7.4 | |||
| Low (NOS score ≤ 6) | 3 | 0.99 (0.82–1.19) | 0.082 | 0.142 | 59.9 | 5 | 1.07 (0.99–1.16) | 0.449 | 0.948 | 0 |
| Adjustment for confounders | ||||||||||
| Alcohol use, yes | 6 | 1.04 (0.96–1.11) | 0.746 | 0 | 6 | 1.16 (1.04–1.31) | 0.555 | 0 | ||
| No | 4 | 0.97 (0.87–1.09) | 0.036 | 0.343 | 64.9 | 7 | 1.02 (0.95–1.09) | 0.702 | 0.797 | 0 |
| Smoking, yes | 10 | 1.01 (0.96–1.07) | 0.239 | 22.2 | 10 | 1.07 (0.97–1.18) | 0.314 | 14.0 | ||
| No | — | — | — | — | — | 3 | 1.05 (0.96–1.14) | 0.629 | 0.105 | 0 |
| Physical activity, yes | 4 | 1.01 (0.93–1.10) | 0.733 | 0 | 4 | 1.14 (1.02–1.33) | 0.390 | 0.3 | ||
| No | 6 | 1.01 (0.93–1.10) | 0.069 | 0.896 | 51.1 | 9 | 1.03 (0.96–1.10) | 0.665 | 0.135 | 0 |
Thirteen studies were included in the dose–response analysis of BMI and total gastric cancer risk. The SRRs and 95% CIs of gastric cancer per 5-unit increment in BMI was 1.00 (95% CI, 0.94–1.05), with evidence of heterogeneity (Q = 46.15, P < 0.001, I2 = 58.8%). The nonsignificant risk relations were observed in both males (SRR, 1.03; 95% CI, 0.95–1.12) and females (SRR, 0.97; 95% CI, 0.90–1.05; data not shown).
Gastric cardia cancer
As shown in Fig. 3, the SRR was 1.21 (95% CI, 1.03–1.42) for overweight (Fig. 3A) and 1.82 (95% CI 1.32–2.49) for obesity (Fig. 3B). There was evidence of heterogeneity among studies for the obese category (Pheterogeneity = 0.009 and I2 = 65.0%), but not for the overweight category (Pheterogeneity = 0.619 and I2 = 0).
SRRs of gastric cardia cancer for BMI between 25 and 30 kg/m2 (A) and BMI over 30 kg/m2 (B).
SRRs of gastric cardia cancer for BMI between 25 and 30 kg/m2 (A) and BMI over 30 kg/m2 (B).
Supplementary Table S1 showed the results of subgroup meta-analyses. For the overweight and obese categories, variables such as the ascertainment of exposure and outcome, duration of follow-up, and adjustments for confounders (smoking, alcohol use, and physical activity) did not significantly alter the summary risk estimates. However, study quality scores significantly modified the obesity–GCC link; summary estimates for high-quality studies were significantly stronger than those for low-quality studies (Pd = 0.011).
Seven studies were included in the dose–response analysis of BMI and GCC risk. The SRR of GCC per 5-unit increment in BMI was 1.32 (95% CI, 1.07–1.64), with evidence of heterogeneity (Q = 33.08, P < 0.001, I2 = 81.9%; data not shown).
Gastric non-cardia cancer
Figure 4 showed the SRRs of GNCCs for 2 levels of BMI. The overall SRR was 0.93 (95% CI, 0.82–1.05) for overweight (Fig. 4A) and 1.00 (95% CI, 0.87–1.15) for obesity (Fig. 4B). There was evidence of heterogeneity for the overweight category (Pheterogeneity = 0.042 and I2 = 54.1%) but not for the obesity category (Pheterogeneity = 0.196 and I2 = 29.1%).
SRRs of gastric non-cardia cancer for BMI between 25 and 30 kg/m2 (A) and BMI over 30 kg/m2 (B).
SRRs of gastric non-cardia cancer for BMI between 25 and 30 kg/m2 (A) and BMI over 30 kg/m2 (B).
As shown in Supplementary Table S2, results from all the subgroup analyses were not statistically significant, and there was no difference in this association between strata in geographic locations, ascertainment of exposure and outcome, duration of follow-up, study quality, and adjustment for confounders (smoking, alcohol consumption, and physical activity).
Eight studies were included in the dose–response analysis of BMI and total GNCC risk. The SRR of GNCC per 5-unit increment in BMI was 0.92 (95% CI, 0.85–1.01), with evidence of heterogeneity (Q = 14.30, P = 0.046, I2 = 51.0%; data not shown).
Sensitivity analysis, meta-regression analysis
The evaluation of influential studies of the BMI–total gastric cancer link showed a single study (35); exclusion of this provided more homogeneous results and a somewhat stronger association between overweight and gastric cancer (OR, 1.04; 95% CI, 1.00–1.09; P = 0.897 and I2 = 0). The evaluation of influential studies of the BMI–GCC and BMI–GNCC links did not show any single influential study, although 2 studies found an inverse association between BMI and the risk of GNCCs (17, 35). In univariate meta-regression analysis, none of the above variables were found to be a significant factor for the links of BMI–GC, overweight–GCC, and BMI–GNCC. Study quality score was found to significantly modify the obesity–GCC link; the between-study variance was reduced from 0.09492 to 0.00804, and the heterogeneity explained by study quality was 91.6%.
Publication bias
Both the rank correlation and regression asymmetry tests did not suggest the presence of publication bias in the summary estimates for total gastric cancer (overweight: PBegg's = 1.00 and PEgger's = 0.648; obesity: PBegg's = 0.760 and PEgger's = 0.281), GCCs (overweight: PBegg's = 0.452 and PEgger's = 0.999; obesity: PBegg's = 0.174 and PEgger's = 0.480), and GNCCs (overweight: PBegg's = 0.764 and PEgger's = 0.182; obesity: PBegg's = 0.711 and PEgger's = 0.997). A review of funnel plots also did not show patterns suggestive of publication bias (Supplementary Fig. S1).
Discussion
The association between BMI and gastric cancer risk has been explored in many previous studies, and recently a meta-analysis reported that overweight and obesity were positively associated with risk of GCCs and overall gastric cancer but not with risk of GNCCs (13). However, the number of included studies (n = 10) and cases (n = 9,492) was relatively smaller, and many a few important studies were not included. Moreover, the previous analysis did not explore the important confounders and the source of the observed heterogeneity, although there was great heterogeneity among studies (P < 0.001). Two additional meta-analyses which also assessed the association between BMI and GCCs showed a positive relationship between BMI and GCCs (45, 46); one was based on 7 case–control and 7 cohort studies (46) and the other was based on 5 case–control and 1 cohort studies (45).
In the current meta-analysis, we investigated the association between BMI and the risk of gastric cancer from 24 large cohort studies comprising more than 10 million subjects, involving 41,791 cases of gastric cancer. The stratification by cancer site in this study suggested that BMI was not a clear risk factor in the studies that combined GNCCs and GCCs together or GNCCs alone; however, a positive association was shown when analyses were limited to studies of GCCs alone. The current evidence extends our knowledge to the association between BMI and gastric cancer (including GNCCs and GCCs), which was not discussed in the body of WCRF 2007 report (11). Importantly, subgroup analyses showed obesity was positively associated with GCCs observed in high-quality studies with no evidence of heterogeneity. Moreover, meta-regression analyses showed study quality might account for 91.6% of the high heterogeneity among studies on obesity–GCC link. We also conducted dose–response analyses and similar results were observed.
In this meta-analysis, the association with overall gastric cancer was significant for males (based on 10 studies), but not for females (based on 9 studies), although the difference is not significant. Reasons for the disparity in the association of gender and overall gastric cancer are not clear. However, it may be inferred that menopausal status and reproductive hormone would have an effect on the BMI–gastric link.
The exact biologic mechanisms underlying the association between increased BMI and risk of GCCs remain unclear. Obesity promotes GER and its transition to Barrett esophagus and thus esophageal adenocarcinoma and GCCs (47, 48). In addition, accumulation of adipose tissue may upregulate the production of endogenous hormones, including sex steroids, insulin, and insulin growth factor-1, which may increase cell proliferation and impair apoptosis, and consequently favor preneoplastic and neoplastic cell growth (49). Obesity is a recognized proinflammatory state which leads to elevated levels of the proinflammatory cytokine, such as TNF and interleukin (IL)-6, which favor the development of cancer (50, 51). In the current study, we also observe higher BMI is not associated with GNCC risk, which is most likely because persons at increased risk of this type of gastric cancer, that is, persons using high levels of tobacco and alcohol, are often malnourished. Moreover, the inverse association between higher BMI and prevalence of Helicobacter pylori infection, an important risk factor for GNCCs, is likely to be another mechanism (52).
The strengths of this study include as follows: (i) our analysis includes all the prospective studies investigating the association between BMI and gastric cancer, involving a total of more than 40,000 cases. This allowed the definition of comparable exposure categories and the investigation of the association of interest in different subgroups, such as gender, geographic locations, confounding factors, and subsite of cancer (only GCCs or only GNCCs). (ii) We restricted our analyses to cohort and nested case–control studies and excluded traditional case–control studies, which are prone to the possibility of recall or selection bias. (iii) All the included studies evaluated multiple potential confounders and the relationships between BMI and gastric cancer risk in each study were derived from regression. (iv) There was no publication bias among these included studies, which may add strength to the conclusions drawn.
As with any meta-analysis of observational studies, our study has limitations. First, most studies in this meta-analysis relied on self-reported anthropometric measures and thus subject to information bias. It is reported that weight tends to be systematically underestimated (more so by heavier subjects), and height is typically overestimated, especially by shorter subjects (53, 54). BMIs calculated from self-reported data will therefore tend to be lower than those from more objective measures (55). However, the results did not materially change when we included in the analysis exclusively the studies in which had measured weight and height. In addition, misclassification of the cancer site may have occurred in some studies. If some GNCCs were incorrectly considered as GCCs, the association between BMI and GCCs could have been diluted (56). Similarly, misclassification of GCCs and esophageal adenocarcinoma could not be fully excluded, which may overestimate the risk relationship of GCC–BMI (because the association was somewhat stronger for esophageal adenocarcinoma than for GCCs). In addition, we must acknowledge that one study suggested higher BMI a convincing risk factor for esophageal adenocarcinoma and GCCs combined (34).
Second, it is possible that the observed relation between increasing BMI and gastric cancer (including GCCs and GNCCs) risk is due to the unmeasured or residual confounding, as inadequate adjustments for confounders may bias the results toward exaggeration or underestimation of risk estimates. Obesity tends to be associated with unhealthy behaviors linked to increased risk of gastric cancer, such as heavy alcohol consumption and physical inactivity. However, a risk association between obesity and gastric cancer persisted when we restricted the meta-analysis of studies that adjusted for both confounders. Overweight was not associated with GCC risk when we limited studies with adjustments for the 2 variables, respectively, which may be related to the low statistical power (only 2 studies were included). Furthermore, smoking is positively related to GCC risk and inversely related to BMI. The majority of studies included in our meta-analysis were adjusted for smoking habits. When conducting meta-analysis of these studies, we obtained similar risk estimates. These results suggested that the associations between obesity and gastric cancer (including GCCs and GNCCs) are not explained by smoking status.
Helicobacter pylori infection is a cause of the GNCCs, which may not be associated with GCC development (57). Moreover, because of acid reflux and consequent lesions on cardiac epithelia, the declined prevalence of Helicobacter pylori infection has been associated with increased obesity (52). However, none of the included studies were adjusted for Helicobacter pylori infection, the unmeasured or residual confounding can therefore not be fully excluded.
Third, high heterogeneity across studies was presented for the obesity–GCC and overweight–GNCC links, which would throw some doubt on the reliability of the SRR estimates for both relationships. On the basis of meta-regression analysis, we found that study quality might account for 91.6% of the obesity–GCC link. Moreover, the considerable variation in duration of follow-up, the ascertainment of diagnosis of GCCs or GNCCs (cancer registry, death certificates, and histologic findings) and the sets of confounders used for adjustment, may contribute to the high degree of heterogeneity presented for the obesity–GCC and overweight–GNCC links.
Finally, as in any meta-analysis, it is possible that an observed association is the result of publication bias, which may have led to an overestimation of the true association. However, the results obtained from funnel plot analysis and formal statistical tests did not provide evidence for such bias.
In summary, this meta-analysis provides more definite and quantitative evidence than previously available that increasing BMI was not a clear risk factor in the studies that combined non-cardia and cardia cancers together or in non-cardia gastric cancer alone; however, a positive association with GCCs was observed. Given the increasing rates of obesity and GCCs worldwide, the contribution of obesity to the development of GCCs might constitute a significant proportion of the global burden of GCCs. Future studies may focus on the effect of weight reduction in obese populations at risk of GCCs.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: Y. Chen, L. Xiu, J. Wang, J. Cheng, G. Gong
Development of methodology: Y. Chen, L. Xiu, J. Wang, J. Cheng, G. Li
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Chen, L. Xiu, Z. Yan, J. Cheng, G. Gong
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): L. Xiu, X. Wang, J. Cheng, G. Li
Writing, review, and/or revision of the manuscript: Y. Chen, L. Xiu, X. Wang, J. Wang, Z. Yan, J. Cheng, G. Gong, G. Li
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): X. Wang, J. Cheng
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
