Abstract
Purpose
Despite extensive research on gastric cancer (GC), efforts to consolidate the numerous associations between possible factors and GC risk remain lacking. This systematic review aimed to provide an overview of potential GC-associated pairs.
Materials and Methods
We systematically searched PubMed, Embase, and Cochrane databases, from their inception to April 23, 2022, for eligible systematic reviews and meta-analyses to investigate the association between any possible factors and GC risk. After the inclusion of 75 systematic reviews and meta-analyses, 117 association pairs were examined. We reanalyzed the included meta-analyses and produced effect estimates using uniform analytical models. The certainty of the evidence for each association pair was evaluated using the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) criteria.
Results
Iatrogenic factors, including antibacterial drugs, were associated with an increased risk of GC. Epstein-Barr virus and Helicobacter pylori infections were also associated with an increased risk of GC, while human T-lymphotropic virus type 1 (HTLV-1) infections were associated with a reduced risk. Dietary habit was a major factor influencing moderate to high GRADE associations. Positive associations were observed for heavy alcohol consumption (relative risk [RR], 1.13; 95% confidence interval [CI], 1.06–1.12), refined grain consumption (RR, 1.36; 95% CI, 1.21–1.53), and habitual salt intake (RR, 1.41; 95% CI, 1.04–1.91).
Conclusions
The associations between GC risk and dietary and nutritional factors were considerably heterogeneous, whereas other factors, such as lifestyle and iatrogenic and environmental exposures, were consistent across regions. Therefore, dietary interventions for GC prevention should be tailored specific to regions.
Trial Registration
PROSPERO Identifier: CRD42020209817
Keywords: Umbrella review, Meta-analysis, Gastric cancer, Stomach cancer, Risk factors
INTRODUCTION
Gastric cancer (GC) is the fifth most commonly diagnosed cancer type worldwide and the third leading cause of cancer-related death, accounting for 1,089,103 new cases and 768,793 deaths in 2020 [1]. Despite a global decline in GC incidence driven by improved screening programs and lifestyle modification, the burden of GC among young adults continues to rise in many countries [1].
The burden of GC is generally high in Eastern Asia [1], likely due to ethnic factors and widespread Helicobacter pylori infections [2]. Other regional factors may also influence GC risk in Asia and Western regions. However, no quantitative analysis has compared the regional variations in the risk factors for GC. In this study, we performed extensive statistical analyses to compare the regional differences in 117 risk and protective factors for GC. We pooled effect estimates for the association between each factor and GC in Asian and non-Asian regions by classifying 1,900 observational studies and randomized control trials (RCTs) investigating GC risk-altering factors according to regions.
Numerous systematic reviews, meta-analyses, and RCTs have assessed the environmental, lifestyle, and nutritional factors associated with the risk of GC [3]. However, these studies have often focused on investigating individual associations (i.e., vitamin C and GC), and the efforts remain fragmented, particularly given the large number of factors influencing GC risk. To consolidate the evidence, we conducted an umbrella review, which provided the highest level of evidence through a comprehensive quantitative synthesis of effect estimates and a thorough evaluation of the evidence level for each association [4]. We performed a statistical replication and re-evaluated 75 systematic reviews and meta-analyses. We used a uniform analytic framework to enable a systematic evidence-level evaluation of 117 risk and protective factors associated with GC using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) framework [5,6]. This systematic evaluation will provide a comprehensive overview and be beneficial in developing screening programs and lifestyle modification strategies.
MATERIALS AND METHODS
Literature search and selection criteria
We systematically searched PubMed/Medical Literature Analysis and Retrieval System Online, Embase, and Cochrane databases for systematic reviews, meta-analyses, and pooled analyses to investigate potential factors related to GC development. The databases were searched from their inception to April 23, 2022. The primary outcome of interest was a risk of developing GC. No restrictions were imposed on the publication date, nationality, race, or age of participants, and only articles published in English were selected. Furthermore, the reference lists of all eligible articles were reviewed. The study was registered in the Prospective Register of Systematic Reviews (CRD42020209817) and adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) guidelines [7].
Search strategy
Systematic reviews, meta-analyses, and pooled analyses were searched using a predefined search strategy. The search strategy is outlined in Appendix 1.
Inclusion and exclusion criteria
We included systematic reviews and meta-analyses of retrospective cohort studies, prospective cohort studies, and RCTs that investigated the association between various factors and GC risk. Systematic reviews and meta-analyses that evaluated outcomes other than GC risk, such as mortality rates, were excluded. Additionally, studies focused on specific populations that did not provide sufficient statistical data to enable a pooled analysis were excluded. In cases where multiple reviews addressed the same association, we excluded reviews that were published more than 5 years before the latest review.
Data extraction
Two authors (S.W.K. and M. S. K.) independently conducted the literature search, and the search results were screened for duplicates. The titles, abstracts, and keywords of each study were independently reviewed by 2 researchers to identify eligible studies. Fig. 1 presents the PRISMA flowchart of the search and selection process [8].
Fig. 1. Flow diagram of the search and selection processes.
A predefined template was used for data collection. The following data were extracted from the included systematic reviews, meta-analyses, and pooled analyses: exposure, publication year, number of studies, type of study, number of participants, metrics, the presence or absence of a dose-response relationship, and summary effect sizes (ESs), along with the corresponding 95% confidence interval (CI) for each study (Supplementary Table 1). Additionally, we obtained the number of included studies and their respective metrics (odds ratio [OR], risk ratio [RR], and hazard ratio [HR]). Considering that metrics such as OR, RR, and HR are inherently linked to specific study designs, the use of a single standardized metric across diverse studies was not feasible. To address this, we retained the metrics reported in the original meta-analyses and included a metric-specific column in the figures to present various metrics, as previously described [9,10,11,12]. We also reviewed the nationality of the participants to facilitate subgroup analyses by region.
Data analysis
To standardize the 75 meta-analyses and enhance their comparability, we reanalyzed all included meta-analyses and generated the effect estimates using generic inverse variance fixed- and random-effects methods. The following metrics were used to assess bias: the I2 statistic to measure heterogeneity among individual studies [13], Egger’s tests to evaluate the presence of publication bias and small study effects (significance threshold: P<0.10) [14], P-curve test [15], and 95% prediction intervals (which represents a range of values that are likely to contain the value of a single new observation with 95% certainty) [16]. We used the “meta” package of R (version 4.0.2; R Foundation, Vienna, Austria) software to reanalyze the extracted data.
Subgroup analysis
Subgroup analyses were performed according to sex (female and male), geographic region (Asian and non-Asian), study design (case-control, hospital-based case-control, and population-based case-control), and anatomical location (gastric cardia cancer and gastric non-cardia cancer). We conducted an extensive reanalysis of geographic regions at the individual study level (rather than at the systematic review level). Studies were categorized into Asian and non-Asian regions based on the country where the participants were enrolled. Given the relatively low incidence of GC in certain Western Asian populations, such as those in Syria and Turkey, these countries were excluded from the Asian region [17]. The non-Asian region was defined as all countries that are not included in the Asian region. For each region (Asian and non-Asian), the effect estimates along with their 95% CIs were pooled using fixed- and random-effects models. The following metrics are also presented in Supplementary Table 2: number of included studies, study design, and metrics. The reanalyzed data also encompassed P-values, heterogeneity measurement using the I2 statistic, Egger’s P-value, and 95% prediction intervals. A P-value of less than 0.05 was considered significant.
Evaluation of the certainty of evidence
To assess potential bias and the quality of evidence for each association, the GRADE criteria were applied [6]. The study design, risk of bias, inconsistency, indirectness, imprecision, publication bias, large magnitude of effect, and dose-response associations were assessed.
RESULTS
Literature review
A total of 1,928 studies were identified after searching the PubMed, Embase, and Cochrane databases. After removing 739 duplicates, 271 and 671 studies were further excluded based on the pre-specified exclusion criteria following a review of titles and abstracts, respectively. Subsequently, 247 systematic reviews, meta-analyses, and pooled analyses were assessed for eligibility through a full-text review. Of these, 172 reviews were excluded due to the following reasons: 12 were outdated, 58 were irrelevant, 40 were identical to the selected articles, 14 focused only on overly specific populations, and 48 lacked sufficient data. Finally, 75 systematic reviews and meta-analyses were analyzed. The search and selection processes are illustrated in Fig. 1.
Meta-analyses of observational studies
We reanalyzed the association between GC and 117 factors (Supplementary Data 1-5). According to the random-effects model, 33 factors were identified as increasing the risk of GC, while 39 factors were found to be protective against GC. Given the large number of risk factors associated with GC, we categorized them into environmental factors (n=8), genetic factors (n=4), drug exposure-related factors (n=9), lifestyle factors (n=20), infection-related factors (n=3), health-related factors (n=5), socioeconomic factors (n=1), dietary factors (n=50), and nutritional factors (n=17) (Figs. 2 and 3). Factors related to chemical exposure were grouped under “environmental factors,” which included exposure to 2,4-dichlorophenoxyacetic acid, cement, and occupational exposure to crystalline silica. A family history of GC (having a first-degree relative with GC) and blood type were categorized under “family history and blood type.” Long-term use of proton pump inhibitors, antibacterial drugs, H. pylori eradication therapy, aspirin, non-aspirin nonsteroidal anti-inflammatory drugs (NSAIDs), statins, acid-suppressive drugs, and hormone replacement therapy were grouped under “drug exposure-related factors.” Twenty factors, including sedentary behavior, night-shift work, alcohol consumption, coffee consumption, obesity, physical activity, and smoking, were classified under “lifestyle factors.” The “infection-related factors” encompassed human T-lymphotropic virus type 1 (HTLV-1), Epstein-Barr virus (EBV), and H. pylori infections. Factors such as glycemic index, glycemic load, and those related to parity were grouped under “health-related factors.” Education level was categorized under “socioeconomic factors.” The “dietary factors” category comprised a variety of vegetables, fruits, dairy products, grains, rice, meat, dietary fats, nuts, salt, and pickled foods. Dietary supplements like vitamins, carotenoids, and flavonoids were included in the “nutritional factors” category.
Fig. 2. Environmental, genetic, drug, lifestyle, and social factors associated with the risk of gastric cancer in overall (A), Asian (B), and non-Asian (C) regions. The results are based on random-effects models. The certainty of evidence underlying each association between risk/protective factors and gastric cancer was evaluated using the GRADE framework. Subgroup analyses of geographical regions were conducted at the individual-study level. This approach was referred to as “reanalysis” as it generated separate results by region for factors not originally assessed by region in the meta-analyses.
GRADE = Grading of Recommendations Assessment, Development, and Evaluation; ES = effect size; RR = risk ratio or relative risk; OR = odds ratio; CI = confidence interval; PPI = proton pump inhibitor; NSAID = non-steroidal anti-inflammatory drugs; HTLV-1 = human T-lymphotropic virus.
*Helicobacter pylori eradication for asymptomatic patient; †Total, occupational, or recreational physical activity.
Fig. 3. Nutritional and dietary factors associated with the risk of gastric cancer in overall (A), Asian (B), and non-Asian regions (C). The results are based on random-effect models. The certainty of evidence underlying each association between risk/protective factors and gastric cancer was evaluated following the GRADE framework. A subgroup analysis of geographical regions was conducted at the individual-study level. This approach was referred to as “reanalysis” as it generated separate results by region for factors not originally assessed by region in the meta-analyses.
GRADE = Grading of Recommendations Assessment, Development, and Evaluation; ES = effect size; RR = risk ratio or relative risk; OR = odds ratio; CI = confidence interval.
*High intake of vegetables, fruits, fish, low-fat milk, and whole grains; †High intake of all types of red and/or processed meats, refined grains, sweets, high-fat dairy products, and high-fat gravy.
Environmental exposures
Two factors showed notable associations in the random-effects results: crystalline silica and selenium exposure (Fig. 2). The overall summary ES of occupational crystalline silica exposure was 1.25, with a 95% CI of 1.18–1.32. Specifically, the ESs of crystalline silica exposure during construction, foundry, and mining were 1.18 (95% CI, 1.07–1.30), 1.32 (95% CI, 1.22–1.42), and 1.35 (95% CI, 1.23–1.49), respectively. Conversely, the RR of selenium exposure was 0.86 (95% CI, 0.77–0.96), indicating a reduced risk of GC.
Family history and blood type
Family history and blood type, including having a first-degree relative with GC and blood groups A and AB, were strongly associated with GC. The RR of having a first-degree relative with GC was 2.35 (95% CI, 1.96–2.81). Blood groups A and AB were correlated with an increased risk of GC compared with blood group O. The OR of blood group A was 1.19 (95% CI, 1.13–1.2), while that of blood group AB was 1.09 (95% CI, 1.02–1.16).
Drug exposure
Exposure to antibacterial and acid-suppressive drugs was also associated with an increased risk of GC. The overall RR of antibacterial drug exposure was 1.12 (95% CI, 1.08–1.16), while that of acid-suppressive drug exposure was 1.44 (95% CI, 1.18–1.75). H. pylori eradication therapy, aspirin, NSAIDs, and hormone replacement therapy were found to considerably reduce the risk.
Lifestyle
Seven of the 30 lifestyle factors significantly increased the risk of developing GC. The overall RR of the highest category of alcohol consumption was 1.25 (95% CI, 1.15–1.36) compared with the lowest category. Compared with non-drinkers, heavy alcohol drinkers had a significantly increased risk (RR, 1.13; 95% CI, 1.06–1.21), whereas light and moderate drinkers showed no significant correlations. The RR of the highest category of waist circumference was 1.48 (95% CI, 1.24–1.77) compared with the lowest category, while the RR of the highest category of waist-to-hip ratio was 1.33 (95% CI, 1.04–1.70) compared with the lowest category. The RR of current smokers was 1.60 (95% CI, 1.35–1.90) compared with that of never smokers. Two lifestyle factors were associated with a reduced risk; compared with no physical activity, the overall OR of performing any type of physical activity was 0.81 (95% CI, 0.73–0.89), while the OR of performing recreational physical activities was 0.83 (95% CI, 0.73–0.95).
Infection
The overall RR of an HTLV-1 infection status was 0.45 (95% CI, 0.28–0.72). Conversely, EBV and H. pylori infection statuses significantly increased GC risk. The overall OR of an EBV infection status was 18.12 (95% CI, 15.21–21.59), while that of an H. pylori infection status was 1.86 (95% CI, 1.47–2.35).
Socioeconomic status
A higher education level correlated with a decreased risk of GC. The OR of the highest education level category was 0.60 (95% CI, 0.43–0.84).
Dietary intake
The RR of allium vegetable intake was 0.78 (95% CI, 0.66–0.91), while that of Chinese chive intake was 0.44 (95% CI, 0.26–0.74). The OR of carrot consumption was 0.43 (0.26–0.73), while that of garlic intake was 0.46 (95% CI, 0.30–0.71). Compared with no whole grain consumption, the OR for overall whole grain consumption was 0.85 (95% CI, 0.76–0.95), with no significant heterogeneity (I2=6.60%). Frequent consumption of whole grains (more than thrice/week) was associated with a reduced risk of GC (OR, 0.5; 95% CI, 0.43–0.70) compared with no consumption. However, the overall RR of refined grain consumption was 1.36 (95% CI, 1.21–1.53), while that of a moderate amount of refined grain consumption (1–2 times/week) was 1.24 (95% CI, 1.10–1.40) compared with no consumption. The RR of the highest category of red meat consumption was 1.41 (95% CI, 1.21–1.66) compared with the lowest category, while that of the highest category of processed meat consumption was 1.57 (95% CI, 1.37–1.81) compared with the lowest category. Habitual salt (RR, 1.41) and pickled food (RR, 1.22) intake increased the risk of GC.
Nutritional factors
The overall RR of the highest category of vitamin intake was 0.77 (95% CI, 0.72–0.83) compared with the lowest category. Compared with the lowest category, the highest categories of vitamin C, vitamin E, and beta-carotene intake showed a reduced risk of GC.
Level of evidence
The GRADE framework (Supplementary Table 3) was used to evaluate the level of evidence derived from observational studies. In all included studies, the association between heavy alcohol consumption and GC was supported by a high level of certainty (GRADE), as described in the evidence map presented in Table 1 (Supplementary Table 4). The AMSTAR 2 tool was used to evaluate the quality of all included meta-analyses, and 52 of the 75 meta-analyses and systematic reviews were rated as having a moderate or high level of evidence.
Table 1. Certainty of evidence for protective and risk factors associated with the risk of gastric cancer.
| Certainty of evidence (GRADE*) | Category of the factors involved | Health protective (reduced risk)† | No benefit (no increase or decrease in risk) | Health risk (increased risk)† |
|---|---|---|---|---|
| High | Lifestyle | - None | - None | - Alcohol consumption |
| Moderate | Lifestyle | - None | - Alcohol consumption | - Current smoking |
| - Waterpipe tobacco smoking | ||||
| Infections | - None | - None | - Epstein-Barr virus infection | |
| Dietary intake | - Chinese chive intake | - None | - Refined grain consumption | |
| - Garlic intake | - Red meat consumption | |||
| - Intake of citrus fruits | - Habitual salt intake | |||
| - Whole grain consumption | - Pickled food intake | |||
| - Soy product intake | - Salted fish intake | |||
| Risk differs by region (Asian regions and Non-Asian regions)‡ | Family history and blood type | - Blood group AB (non-Asian) | - Blood group AB (Asian) | |
| Exposure to drugs | - Helicobacter pylori eradication therapy (Asian) | - Helicobacter pylori eradication therapy (non-Asian) | ||
| - Hormone replacement therapy (non-Asian) | - Hormone replacement therapy (Asian) | |||
| Lifestyle | - Occupational physical activity (non-Asian) | - Occupational physical activity (Asian) | ||
| - Recreational physical activity (non-Asian) | - Recreational physical activity (Asian) | |||
| - Alcohol consumption (Asian) | - Alcohol consumption (non-Asian) | |||
| Health status | - Years of fertility (non-Asian) | - Years of fertility (Asian) | ||
| Socioeconomic status | - Higher education level (non-Asian) | - Higher education level (Asian) | ||
| Dietary intake | - Onion intake (non-Asian) | - Onion intake (Asian) | ||
| - Refrigerator use (Asian) | - Refrigerator use (non-Asian) | |||
| - Dietary carbohydrate intake (non-Asian) | - Dietary carbohydrate intake (Asian) | |||
| - Dietary fat intake (Asian) | - Dietary fat intake (non-Asian) | |||
| - Dietary saturated fat intake (Asian) | - Dietary saturated fat intake (non-Asian) | |||
| - Dietary polyunsaturated fat intake (non-Asian) | - Dietary polyunsaturated fat intake (Asian) | |||
| - Garlic intake (non-Asian) | - Garlic intake (Asian) | |||
| - Red meat consumption (Asian) | - Red meat consumption (non-Asian) | |||
| - Habitual salt intake (non-Asian) | - Habitual salt intake (Asian) | |||
| - Pickled food intake (non-Asian) | - Pickled food intake (Asian) | |||
| Nutritional | - Vitamin intake (non-Asian) | - Vitamin intake (Asian) | ||
| - Folate intake (non-Asian) | - Folate intake (Asian) | |||
| - Alpha-carotene intake (Asian) | - Alpha-carotene intake (non-Asian) |
Of the 139 factors, 84 factors demonstrated very low (GRADE) evidence and are presented in the Supplementary Table 4.
GRADE = Grading of Recommendations, Assessment, Development, and Evaluation.
*Definitions of the GRADE: High: The evidence provides a strong indication of the likely effect, with a low likelihood of substantial variation; Moderate: The evidence provides a good indication of the likely effect, with a moderate likelihood of substantial variation; Low: The evidence provides some indication of the likely effect, but substantial variation (enough to influence decision-making) is highly likely; Very low: The evidence does not provide a reliable indication of the likely effect, with a very high likelihood of substantial variation (enough to influence decision-making).
†Factors significantly associated with reduced or increased risk of gastric cancer.
‡Factors with heterogeneous effects across regions (benefit/no benefit/risk), supported by moderate or high GRADE evidence. All factors, including those supported by very low or low GRADE evidence, are provided in Supplementary Table 4.
Subgroup analysis by region
We evaluated the association between GC risk and 112 and 134 factors in the Asian and non-Asian populations, respectively (Supplementary Table 2). Of these, 21 factors showed regional differences (Asian compared with non-Asian) in their association with GC risk (Table 1). These factors are: H. pylori eradication in the “infection” category; occupational physical activity, recreational physical activity, and alcohol consumption in the “lifestyle” category; hormone replacement therapy in the “drug exposure” category, higher education level in the “socioeconomic status” category; onion, garlic, dietary carbohydrate, dietary fat, dietary saturated fat, dietary polyunsaturated fat, habitual salt, and pickled food intake, as well as red meat consumption, in the “dietary intake” category; folate, total vitamin, and alpha-carotene intake in the “vitamin intake” category; and blood group AB in the “family history and blood type” category. Compared with the lowest category of dietary fat intake, the RR of the highest category in the Asian population was 0.68 (95% CI, 0.54–0.85), whereas that in the non-Asian population was 1.27 (95% CI, 1.08–1.51). The subgroup analysis results for sex, study design, and anatomical position of GC are provided in the Appendix 2 (Supplementary Tables 5 and 6).
DISCUSSION
This comprehensive consolidation of systematic reviews provides a critical assessment of the dietary, lifestyle, environmental, iatrogenic, and genetic factors associated with GC. Recent umbrella reviews have explored various factors associated with GC risk. Previous studies, such as those by Bouras et al. (2022) [18], have primarily focused on investigating the relationship between dietary factors and GC risk. In contrast, our study offers a broader perspective by assessing not only dietary factors, but also environmental exposure, family history and blood types, drug exposure, lifestyle factors, infection statuses, socioeconomic status, and nutritional intake. Nine risk factors and seven protective factors showed a moderate to high level of certainty (Table 1). Certain factors were more strongly associated with GC in Asian regions than in non-Asian regions, indicating differences in genetic susceptibility and environmental exposure. H. pylori infection is a well-known risk factor for GC, and the eradication of H. pylori is recognized as a potential preventive measure. However, the efficacy of H. pylori eradication therapy varies across regions. In Asian countries, this therapy has shown a protective effect against GC; however, it has not demonstrated the same level of efficacy as observed in non-Asian countries. The associations between dietary and nutritional factors and the risk of GC were heterogeneous in Asian and non-Asian regions, whereas other factors, including lifestyle, environmental/drug exposure, and infections, showed more consistent associations across regions. A major factor affecting the certainty of evidence was imprecision, likely due to small sample sizes, wide CIs, or lower statistical powers (Supplementary Table 7).
Dietary factors frequently showed a high level of evidence (Table 1), implying that dietary habits and food components substantially influenced the development of GC. This is unsurprising, given that the gastric mucosa remains in contact with food. However, the specific foods identified as beneficial or as risk factors for GC remained unclear and varied widely. We consolidated all relevant meta-analyses into a single umbrella analysis (with reanalysis), which identified a clear pattern suggesting that salty or spicy foods are predominantly associated with negative outcomes (Fig. 3), likely due to their potential to damage the gastric epithelium and trigger inflammatory responses [19,20]. This pattern is consistent with the beneficial effects of vegetable and fruit consumption, which exert antioxidant and anti-inflammatory effects [21].
The association between dietary factors and GC varies between Asian and non-Asian regions. The associations that shifted from protective or risk factors in the overall pooled results to null findings when stratified by region are likely due to reduced statistical power; therefore, we did not place a significant emphasis on interpreting these associations. However, null associations in the overall analysis becoming significant (indicating a protective or risk relationship) after stratifying by region (even with a loss of statistical power), suggested meaningful regional differences. For example, dietary carbohydrate intake became a significant factor when the analysis was restricted to Asian regions. Additionally, high fat consumption, another main food ingredient, was protective against GC in Asian regions, whereas an opposite effect was observed in non-Asian regions. Salt intake was a risk factor in Asian regions but not in non-Asian regions. These differing responses and susceptibilities to major food components and actual variations in dietary habits between Asian and non-Asian regions may partially explain the higher GC burden in Asian regions [22]. Therefore, dietary recommendations and policies for GC prevention should be tailored to specific regions rather than extrapolated.
Modifiable factors warrant more attention from a public health perspective. Our analysis showed that alcohol consumption should be avoided to reduce the risk of GC; however, the association was dose dependent and significant only with heavy drinking (Fig. 2). Moreover, higher obesity measures, such as waist circumference and waist-to-hip ratio, were associated with an increased GC risk; these results align with the findings from a recent Mendelian randomization study that elucidated the causal role of adiposity in GC [23]. Furthermore, increased physical activity was associated with a reduced risk of GC, potentially due to its effects on reducing obesity. Investigating the effect of each drug on the risk of GC could provide valuable insights for managing the disease. NSAIDs were protective against GC, possibly due to their anti-inflammatory effect by inhibiting cyclooxygenase and anticancer effects through the NSAID-induced apoptosis of cancerous cells [24]. Meanwhile, antibiotics and acid-suppressing agents were associated with an increased risk of GC. Promoting healthy lifestyles and appropriate drug use represents practical and effective public health strategies. This comprehensive analysis provided a list of modifiable factors, along with a systematic evaluation of the certainty of evidence behind each association, which can be useful for individuals at risk of GC and policymakers alike.
Despite the higher incidence of GC in Asian regions, a notable disparity exists in the number of studies conducted in Asian regions compared with non-Asian regions. Among the 117 factors explored, 81 factors from 507 studies were investigated in the Asian population, whereas 108 factors from 1,226 studies were examined in non-Asian populations. This discrepancy influenced the credibility of the findings, as the non-Asian subgroup generally exhibited narrower CIs and stronger correlations with higher statistical power compared with the Asian subgroup. This disproportionality underscores the need for further GC research focused on the Asian population to strengthen the confidence in the findings and provide more robust results for the Asian populations most affected by GC [25,26].
This study has several limitations. First, we only reviewed previous meta-analyses, which may have excluded some risk or protective factors that were not explored at the meta-analysis level. Second, some analyses may have been subject to type II errors due to the small sample size. To address this issue, we assessed the combination of sample size, confidence interval width, and statistical power of each association using the “imprecision” criteria in the GRADE framework and downgraded the evidence level when such an error-prone risk was detected. Third, the regional subgroup estimates were more conservative and had wider CIs than the overall results. This was partly due to the reduced statistical power caused by the small number of studies included in each subgroup analysis. This limitation complicates the interpretation of null findings as it may be difficult to distinguish between a true absence of association and false-negative findings caused by low statistical power. Therefore, the subgroup associations identified in the present study should be interpreted with caution.
We highlighted the considerable heterogeneity in the association between GC risk and dietary and nutritional factors across regions. However, the association between GC and other factors such as lifestyle, environmental/drug exposure, and infections was consistent across regions. This pattern suggests that regional differences in GC risk are attributable to the varying responses and susceptibilities to major food components and differences in dietary habits between Asian and non-Asian regions. Moreover, GC was more strongly correlated with modifiable factors than with non-modifiable factors, underscoring the importance of public health efforts aimed at promoting favorable lifestyle modifications. Finally, further studies in Asian regions are warranted to resolve the knowledge gap and improve confidence in the findings between Asian and non-Asian regions.
Appendix 1
Search strategy for systematic review and meta-analysis
PubMed
(gastric cancer* OR (cancers,gastric) OR "stomach cancer*" OR "gastric neoplasm*" OR (neoplasms, gastric) OR "stomach neoplasm*" OR "gastric malignanc*" OR "stomach malignanc*" OR "gastric tumor*" OR "stomach tumor*" OR "Stomach Neoplasms"[Mesh]) AND (meta-analysis[ptyp] OR meta[ti] OR meta-analy*[ti] OR pooled[ti] OR mendelian[tiab] OR "Meta-Analysis"[publication type])AND (risk*[tiab] OR risk *[tiab] OR association*[tiab] OR associat*[tiab] OR "Risk"[Mesh] OR "Risk "[Mesh] OR "Association"[Mesh]) NOT gastrectom*[ti] NOT surviv*[ti] NOT prognos*[ti] NOT protocol*[ti] NOT comment*[ti] NOT polymorphism*[ti] NOT kid*[ti] NOT child*[ti] NOT adolesc*[ti] NOT pediatric*[ti]
Embase
(stomac*:ti,ab,kw OR gastri*:ti,ab,kw)AND (cancer*:ti,ab,kw OR tumor*:ti,ab,kw OR neoplasm*:ti,ab,kw) AND (meta:ti OR 'meta analy*':ti OR pooled:ti) AND (risk*:ti,ab,kw OR incidenc*:ti,ab,kw OR associat*:ti,ab,kw) NOT (gastrectom*:ti OR surviv*:ti OR prognos*:ti OR protocol*:ti OR comment*:ti OR polymorphism*:ti OR kid*:ti OR child*:ti OR adolesc*:ti OR pediatric*:ti) NOT ('conference abstract':it OR 'conference paper':it OR 'conference review':it OR editorial:it OR note:it OR letter:it OR 'short survey':it)
Cochrane database of systematic reviews
(stomach* OR gastri*) AND (cancer* OR neoplasm* OR tumor*)
Appendix 2
Results
Male vs. female
We re-analyzed 7 factors: total diary consumption, dietary inflammatory index, aspirin use, performing any type of physical activity, habitual salt intake, current smoker, and former smoker. A heterogeneous direction was observed between the male and female groups (Supplementary Table 8). The relative risk (RR) of performing any type of physical activity compared with that of not performing physical activity in males was 0.90 (95% confidence interval [CI], 0.85-0.95), whereas female populations revealed no clear association (RR, 0.77; 95% CI, 0.53-1.12). The RR of former smokers vs. never smokers was 1.23 (95% CI, 0.77-1.97) in females; however, the RR in males was 1.37 (95% CI, 1.06-1.77).
Stratification by study design
We stratified all systematic reviews and meta-analyses into 4 groups; case-control, hospital-based case-control, population-based case-control, and cohort. We identified 23 factors in case-control studies, 5 factors in hospital-based case-control studies, 5 factors in population-based case-control studies, and 29 factors in cohorts studies. We re-evaluated each study group and reanalyzed the data (Supplementary Table 5).
Stratification by anatomical position of gastric cancer
We evaluated all systematic reviews and divided the studies into gastric cardiac cancer (GCA) and gastric non-cardiac cancer (GNCA). Twelve factors evaluated the risk of GCA and 11 factors evaluated the risk of GNCA (Supplementary Table 6). The RR of tooth loss in GCA was 0.95 (95% CI, 0.65-1.38), and in GNCA was 1.71 (95% CI, 1.17-2.50). The effect size (ES) of H. pylori infection in GCA was not significant, but the ES in GNCA was 2.81 (95% CI, 2.14-3.68). Aspirin use also reduced the risk in GNCA, the odds ratio was 0.82 (95% CI, 0.63-1.06), compared with that of no reduction in GCA.
Footnotes
Conflict of Interest: No potential conflict of interest relevant to this article was reported.
- Conceptualization: K.M.S.
- Data curation: K.S.W., K.M.S.
- Formal analysis: K.S.W., K.M.S., K.J.Y.
- Methodology: K.S.W., K.M.S.
- Project administration: K.S.W., K.M.S.
- Software: K.S.W., K.M.S., S.J.I.
- Supervision: P.S.S.
- Validation: K.Y.K.
- Visualization: K.S.W., K.M.S.
- Writing - original draft: K.S.W., K.M.S., K.J.E., C.Y.H.
- Writing - review & editing: K.Y.K., M.J.S., Y.D.K., P.S.S..
SUPPLEMENTARY MATERIALS
Overall population (Figs. 1-138)
Population stratified by sex (Figs. 139-154)
Population stratified by geographic area (Figs. 155-373)
Population stratified by study type (Figs. 374-436)
Outcome stratified by anatomical classification of gastric cancer (Figs. 437-459)
Evaluation of effect size, heterogeneity, small study effect, and evidence of 145 meta-analyses investigating the protective and risk factors of gastric cancer: overall population
Evaluation of effect size, heterogeneity, small study effects, and evidence of level of 10 meta-analyses investigating the protective and risk factors of gastric cancer: populations stratified by geographic area
Quality assessment of 75 included systemic review and meta-analyses with AMSTAR 2
Certainty of evidence for protective and risk factors associated the risk of gastric cancer
Evaluation of effect size, heterogeneity, small study effects, and evidence of level of 23 meta-analyses investigating the protective and risk factors of gastric cancer: populations stratified by study design
Evaluation of effect size, heterogeneity, small study effect, and evidence of 43 meta-analyses investigating the protective and risk factors of gastric cancer: population stratified by anatomical position of gastric cancer
Details of evidence grading for meta-analyses of risk factors for gastric cancer
Evaluation of effect size, heterogeneity, small study effects, and evidence of level of 41 meta-analyses investigating the protective and risk factors of gastric cancer: populations stratified by sex
References
- 1.Sung H, Ferlay J, Siegel RL, Laversanne M, Soerjomataram I, Jemal A, et al. Global Cancer Statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71:209–249. doi: 10.3322/caac.21660. [DOI] [PubMed] [Google Scholar]
- 2.Correa P. Gastric cancer: overview. Gastroenterol Clin North Am. 2013;42:211–217. doi: 10.1016/j.gtc.2013.01.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Poorolajal J, Moradi L, Mohammadi Y, Cheraghi Z, Gohari-Ensaf F. Risk factors for stomach cancer: a systematic review and meta-analysis. Epidemiol Health. 2020;42:e2020004. doi: 10.4178/epih.e2020004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Fusar-Poli P, Radua J. Ten simple rules for conducting umbrella reviews. Evid Based Ment Health. 2018;21:95–100. doi: 10.1136/ebmental-2018-300014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Puhan MA, Schünemann HJ, Murad MH, Li T, Brignardello-Petersen R, Singh JA, et al. A GRADE Working Group approach for rating the quality of treatment effect estimates from network meta-analysis. BMJ. 2014;349:g5630. doi: 10.1136/bmj.g5630. [DOI] [PubMed] [Google Scholar]
- 6.Iorio A, Spencer FA, Falavigna M, Alba C, Lang E, Burnand B, et al. Use of GRADE for assessment of evidence about prognosis: rating confidence in estimates of event rates in broad categories of patients. BMJ. 2015;350:h870. doi: 10.1136/bmj.h870. [DOI] [PubMed] [Google Scholar]
- 7.Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. doi: 10.1136/bmj.n71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Moher D, Liberati A, Tetzlaff J, Altman DG PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. Int J Surg. 2010;8:336–341. doi: 10.1016/j.ijsu.2010.02.007. [DOI] [PubMed] [Google Scholar]
- 9.Okoth K, Chandan JS, Marshall T, Thangaratinam S, Thomas GN, Nirantharakumar K, et al. Association between the reproductive health of young women and cardiovascular disease in later life: umbrella review. BMJ. 2020;371:m3502. doi: 10.1136/bmj.m3502. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Piovani D, Danese S, Peyrin-Biroulet L, Nikolopoulos GK, Lytras T, Bonovas S. Environmental risk factors for inflammatory bowel diseases: an umbrella review of meta-analyses. Gastroenterology. 2019;157:647–659.e4. doi: 10.1053/j.gastro.2019.04.016. [DOI] [PubMed] [Google Scholar]
- 11.He Y, Li X, Gasevic D, Brunt E, McLachlan F, Millenson M, et al. Statins and multiple noncardiovascular outcomes: umbrella review of meta-analyses of observational studies and randomized controlled trials. Ann Intern Med. 2018;169:543–553. doi: 10.7326/M18-0808. [DOI] [PubMed] [Google Scholar]
- 12.Kim MS, Kim WJ, Khera AV, Kim JY, Yon DK, Lee SW, et al. Association between adiposity and cardiovascular outcomes: an umbrella review and meta-analysis of observational and Mendelian randomization studies. Eur Heart J. 2021;42:3388–3403. doi: 10.1093/eurheartj/ehab454. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med. 2002;21:1539–1558. doi: 10.1002/sim.1186. [DOI] [PubMed] [Google Scholar]
- 14.Egger M, Smith GD. Bias in location and selection of studies. BMJ. 1998;316:61–66. doi: 10.1136/bmj.316.7124.61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Simonsohn U, Nelson LD, Simmons JP. p-Curve and effect size: correcting for publication bias using only significant results. Perspect Psychol Sci. 2014;9:666–681. doi: 10.1177/1745691614553988. [DOI] [PubMed] [Google Scholar]
- 16.van Assen MA, van Aert RC, Wicherts JM. Meta-analysis using effect size distributions of only statistically significant studies. Psychol Methods. 2015;20:293–309. doi: 10.1037/met0000025. [DOI] [PubMed] [Google Scholar]
- 17.Wong MCS, Huang J, Chan PSF, Choi P, Lao XQ, Chan SM, et al. Global incidence and mortality of gastric cancer, 1980-2018. JAMA Netw Open. 2021;4:e2118457. doi: 10.1001/jamanetworkopen.2021.18457. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Bouras E, Tsilidis KK, Triggi M, Siargkas A, Chourdakis M, Haidich AB. Diet and risk of gastric cancer: an umbrella review. Nutrients. 2022;14:1764. doi: 10.3390/nu14091764. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Takahashi M, Hasegawa R. Enhancing effects of dietary salt on both initiation and promotion stages of rat gastric carcinogenesis. Princess Takamatsu Symp. 1985;16:169–182. [PubMed] [Google Scholar]
- 20.Wang XQ, Terry PD, Yan H. Review of salt consumption and stomach cancer risk: epidemiological and biological evidence. World J Gastroenterol. 2009;15:2204–2213. doi: 10.3748/wjg.15.2204. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Ricker MA, Haas WC. Anti-inflammatory diet in clinical practice: a review. Nutr Clin Pract. 2017;32:318–325. doi: 10.1177/0884533617700353. [DOI] [PubMed] [Google Scholar]
- 22.Tan WSK, Tan SY, Henry CJ. Ethnic variability in glycemic response to sucrose and isomaltulose. Nutrients. 2017;9:347. doi: 10.3390/nu9040347. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Kim MS, Song M, Kim S, Kim B, Kang W, Kim JY, et al. Causal effect of adiposity on the risk of 19 gastrointestinal diseases: a Mendelian randomization study. Obesity (Silver Spring) 2023;31:1436–1444. doi: 10.1002/oby.23722. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Kazberuk A, Zareba I, Palka J, Surazynski A. A novel plausible mechanism of NSAIDs-induced apoptosis in cancer cells: the implication of proline oxidase and peroxisome proliferator-activated receptor. Pharmacol Rep. 2020;72:1152–1160. doi: 10.1007/s43440-020-00140-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Button KS, Ioannidis JPA, Mokrysz C, Nosek BA, Flint J, Robinson ES, et al. Confidence and precision increase with high statistical power. Nat Rev Neurosci. 2013;14:585–586. doi: 10.1038/nrn3475-c4. [DOI] [PubMed] [Google Scholar]
- 26.Song Y, Liu X, Cheng W, Li H, Zhang D. The global, regional and national burden of stomach cancer and its attributable risk factors from 1990 to 2019. Sci Rep. 2022;12:11542. doi: 10.1038/s41598-022-15839-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Overall population (Figs. 1-138)
Population stratified by sex (Figs. 139-154)
Population stratified by geographic area (Figs. 155-373)
Population stratified by study type (Figs. 374-436)
Outcome stratified by anatomical classification of gastric cancer (Figs. 437-459)
Evaluation of effect size, heterogeneity, small study effect, and evidence of 145 meta-analyses investigating the protective and risk factors of gastric cancer: overall population
Evaluation of effect size, heterogeneity, small study effects, and evidence of level of 10 meta-analyses investigating the protective and risk factors of gastric cancer: populations stratified by geographic area
Quality assessment of 75 included systemic review and meta-analyses with AMSTAR 2
Certainty of evidence for protective and risk factors associated the risk of gastric cancer
Evaluation of effect size, heterogeneity, small study effects, and evidence of level of 23 meta-analyses investigating the protective and risk factors of gastric cancer: populations stratified by study design
Evaluation of effect size, heterogeneity, small study effect, and evidence of 43 meta-analyses investigating the protective and risk factors of gastric cancer: population stratified by anatomical position of gastric cancer
Details of evidence grading for meta-analyses of risk factors for gastric cancer
Evaluation of effect size, heterogeneity, small study effects, and evidence of level of 41 meta-analyses investigating the protective and risk factors of gastric cancer: populations stratified by sex