Association between physical activity and early-stage gastric cancer: A cross-sectional study of the Wuwei cohort

Abstract

BACKGROUND

Gastric cancer remains a major global burden, especially in China. While early detection improves prognosis, prevention requires identification of modifiable risk factors. Physical activity (PA) is protective against several cancers, but its association with early gastric cancer (EGC) is unclear. This study investigates this relationship using data from the Wuwei cohort.

AIM

To assess the association between PA and EGC risk.

METHODS

This cross-sectional study included 1829 adults (35-70 years) who underwent endoscopy with pathological diagnosis from the Wuwei cohort (2013-2016). PA was assessed using the International PA Questionnaire-Short Form and categorized into tertiles of total metabolic equivalent (MET)-minutes/week. High-grade dysplasia (HGD) and EGC were combined as the outcome. Multivariable logistic regression and restricted cubic spline models were applied to estimate adjusted odds ratios (ORs) and dose-response relationships.

RESULTS

Among 1829 participants, 170 had HGD/EGC (74 HGD and 96 EGC). Moderate PA (1368.5-3213 MET-minutes/week) was associated with lower odds of HGD/EGC compared with low activity [model 3: OR = 0.64, 95% confidence interval (CI): 0.43-0.95], corresponding to a risk difference of 3.1% (8.7% vs 11.8%). A significant trend across tertiles was observed (P = 0.047). Restricted cubic spline analysis demonstrated a non-linear inverse association, with the lowest risk at approximately 4200 MET-minutes/week. Stronger inverse associations were observed among males (high PA: OR = 0.58, 95%CI: 0.34-0.99), Helicobacter pylori-positive individuals (OR = 0.40, 95%CI: 0.19-0.83), and cardia lesions (OR = 0.27, 95%CI: 0.10-0.68).

CONCLUSION

Moderate PA is associated with lower odds of early gastric neoplasia, particularly among Helicobacter pylori-positive individuals, supporting a potential role in EGC prevention.

Key Words: Early gastric cancer; Stomach neoplasms; Physical activity; Exercise; Risk factors

Core Tip: This cross-sectional analysis of 1829 adults from a high-risk Chinese population demonstrates that moderate physical activity (PA), assessed by the International PA Questionnaire-Short Form, is associated with lower odds of high-grade dysplasia and early gastric cancer. We observed a significant dose-response trend and a non-linear inverse association, with the lowest risk at intermediate activity levels. The inverse association is more pronounced among Helicobacter pylori-positive individuals and patients with cardia lesions. These findings suggest that moderate PA may play a role in early gastric carcinogenesis and prevention strategies.

INTRODUCTION

Gastric cancer is the fifth leading cause of morbidity and the fourth leading cause of mortality in the world. According to statistics, there were more than 1 million new gastric cancer patients in 2020, and 769000 people died of gastric cancer. In China, gastric cancer incidence and mortality rank third among various malignant tumors[1].

The mortality rate of early gastric cancer (EGC) is significantly lower than that of advanced gastric cancer; early detection is important for longer survival. While early detection is a strategy for the secondary prevention of cancer, it is important to identify and mitigate potential modifiable risk factors for both cancer prevention and treatment. Studies by many scholars have found that Helicobacter pylori (H. pylori) infection, smoking, and alcohol consumption are risk factors for gastric cancer[2]. H. pylori infection was clearly listed as a class of carcinogens by the World Health Organization as early as 2017, which is closely related to the incidence of gastric cancer, and it has become a consensus that eradication of H. pylori can reduce the occurrence of gastric cancer. In terms of living habits, eating more fresh vegetables and fruits and reducing the intake of pickled foods can somewhat reduce the occurrence of stomach cancer. Studies have shown that physical activity (PA) can also reduce cancer incidence; PA is a protective factor against a variety of tumors, including bladder cancer, breast cancer, and prostate cancer[3-5]. The number of epidemiological studies on the association between exercise and gastric cancer is relatively small, and current results suggest that exercise is a protective factor against gastric cancer[6,7]. However, the patients with gastric cancer included in the relevant studies had advanced gastric cancer, and no studies on the effect of PA on EGC have been reported. Therefore, this study aims to explore the association between PA and the risk of early-stage gastric cancer by conducting a cross-sectional analysis of the Wuwei cohort.

MATERIALS AND METHODS

Study population

A total of 23346 participants aged 35-70 years were identified from the Wuwei cohort between 2013 and 2016. The Wuwei cohort is a general, population-based prospective cohort study established in a high-risk region for gastric cancer between March 2013 and April 2016. Detailed descriptions of the Wuwei cohort have been published elsewhere[8]. The cohort study conforms to the guidelines of the Declaration of Helsinki, approved by the Ethics Committee of The First Hospital of Lanzhou University, No. LDYYLL2012001. All participants provided written informed consent.

Participants who underwent gastroscopic examination and had gastric pathological diagnoses of normal gastric mucosa, chronic non-atrophic gastritis, high-grade dysplasia (HGD), and EGC were included. Participants with incomplete information on PA were excluded. This resulted in a total of 1829 individuals in our analysis (7.8% of the original cohort) (Figure 1).

Figure 1 Flow chart of the study population. This figure depicts the study population screening process, exclusion criteria and inclusion criteria. no-CAG: Chronic non-atrophic gastritis; HGD: High-grade dysplasia; EGC: Early gastric cancer.

To assess potential selection bias, baseline characteristics were compared between included and excluded participants (Supplementary Table 1). Significant differences were observed in age, sex, education, occupation, H. pylori infection, dietary behaviors, medical history, total energy intake, and PA grade (all P < 0.05), whereas body mass index (BMI) and marital status were comparable. These findings suggest that selection bias cannot be excluded.

Assessment of PA

The International PA Questionnaire-Short Form (IPAQ-SF) was used to measure PA behaviors. The IPAQ has been translated into Chinese with good reliability and validity[9]. However, the IPAQ-SF only captures PA over the past 7 days and relies on self-report, which may not fully reflect long-term habitual activity relevant to gastric carcinogenesis. Measurement error and recall bias are therefore possible, particularly at higher activity levels. The IPAQ-SF consists of seven items assessing PA over the past 7 days: (1) How many days did you do vigorous PA, such as lifting, digging, aerobics, or fast cycling; (2) How much time do you usually spend in vigorous PA on one of these days; (3) In the last 7 days, have you done moderate/moderate PA on several days (such as playing golf, lifting light objects, cycling at a normal pace, playing badminton), please do not include walking; (4) How much time do you usually spend on moderate PA on each of these days; (5) On which days did you walk for at least 10 minutes at a time; (6) How much time do you usually spend walking on one of these days; and (7) How much of the workday did you sit?[10]. According to the IPAQ-SF protocol, PA is classified into three intensity categories based on metabolic equivalent (MET) values: Low-intensity activity = 3.3 METs, moderate-intensity activity = 4.0 METs, and vigorous-intensity activity = 8.0 METs. The total PA score was calculated using standardized IPAQ-SF scoring protocols to yield total MET minutes per week of PA. For categorical analysis of total PA, the total MET minutes of every individual were categorized into tertiles, where the first tertile was defined as the low PA group, the second tertile as the moderate PA group, and the third tertile as the high PA group. In this cross-sectional study, PA assessment was conducted at the same visit as, or immediately prior to, endoscopy. No lag period was applied. Given the cross-sectional design, reverse causation cannot be excluded, and causal inference is not possible.

Dietary assessment

Dietary intake was assessed using a semi-quantitative food frequency questionnaire adapted from the 2010 China National Nutrition and Health Survey[11] and modified for local dietary habits. Foods were grouped into ten categories. Participants reported frequency of intake over the previous 12 months (nine response options) and usual portion size (small, medium, large). Average daily intake (g/day) was calculated by multiplying frequency by portion size. Total energy intake (kcal/day) was estimated using the China Food Composition Tables (2019 editions)[12] and included as a covariate in model 3.

Assessment of covariables

Covariates included age, sex, marital status, education, occupation, household income, smoking, drinking, diet habits (eating hot food, eating quickly, eating fried food), medical history (gastritis, peptic ulcers, gallbladder diseases and gastric polyps) and family history of gastric cancer, and were derived from standardized and structured questionnaires, which were completed by trained interviewers through in-person interviews with participants. Height and weight measurements were performed by clinicians after the completion of the in-person interview. BMI was calculated as weight (kg) divided by height (m) squared. Infection with H. pylori was determined using the 14C-urea breath test (urea-14C breath test, Shenzhen Zhonghe Headway BIO-SCI and TECH, Shenzhen, China).

Statistical analysis

Student’s t-test, χ² test, or Fisher’s exact test were performed to compare means for continuous variables and proportions for categorical variables between individuals by their category of total MET minutes per week of PA. HGD and EGC were combined as a composite outcome (HGD/EGC), consistent with the Vienna classification of early gastric neoplasia[13]. Although biological differences exist, combining these outcomes improved statistical power while maintaining clinical relevance.

To flexibly model the dose-response relationship between PA and HGD/EGC risk, we performed restricted cubic spline regression within the logistic regression framework. The number and placement of knots were determined based on Akaike Information Criterion and Bayesian Information Criterion minimization. Specifically, we compared models with 3, 4, and 5 knots placed at percentiles recommended by Harrell (5^th, 35^th, 65^th, 95^th percentiles for 4 knots, etc.). The model with 4 knots yielded the lowest Akaike Information Criterion and Bayesian Information Criterion values and was therefore selected as the final model. The reference value for odds ratio (OR) estimation was set at the 10^th percentile of the total MET-minutes/week distribution to avoid instability at extremely low values. The restricted cubic spline analysis adjusted for the same set of covariates as in model 3, and nonlinearity was tested by examining the null hypothesis that the coefficient of the second spline transformation equaled zero.

Adjusted ORs with 95% confidence intervals (CIs) for the associations between different levels of PA and HGD/EGC were calculated using multivariable binary logistic regression. Participants diagnosed with normal gastric mucosa or non-atrophic gastritis served as controls. Restricted cubic spline analysis was used to examine the relationship between total MET minutes per week and the risk of HGD/EGC. Potential confounders were identified based on established risk factors in the literature, clinical judgment, and a conceptual framework linking covariates to both PA and HGD/EGC. The OR estimates were calculated for three models: Model 1 was unadjusted; model 2 was adjusted for age, sex, marital status, education, occupation, smoking, drinking, diet habits (eating hot food, eating quickly, eating fried food), household income, H. pylori infection status, BMI, and medical histories (gastritis, peptic ulcers, gallbladder diseases and gastric polyps) and family history of gastric cancer; model 3 was additionally adjusted for total energy intake (derived from the food frequency questionnaire as described in the Dietary Assessment section). Subgroup analyses were conducted after stratification by sex, H. pylori infection status, and tumor location.

Multicollinearity among independent variables was assessed using the variance inflation factor, with all variance inflation factor values below 2, indicating no significant multicollinearity. P values < 0.05 were considered statistically significant. All statistical analyses were performed using Stata (version 15.0, StataCorp LLC, College Station, TX, United States).

RESULTS

Comparison between included and excluded participants

Compared with excluded individuals, included participants were slightly older (51.2 years vs 50.4 years, P < 0.001), more often male (52% vs 47%, P < 0.001), less likely to be farmers, and had a lower prevalence of H. pylori infection (45% vs 54%, P < 0.001). They also differed in total energy intake, dietary behaviors, and certain medical history variables (all P < 0.05), whereas BMI and marital status were comparable. Detailed data are shown in Supplementary Table 1.

A total of 1829 participants were included, comprising 210 normal gastric mucosa, 1449 chronic non-atrophic gastritis, 74 HGD, and 96 EGC. PA was categorized into tertiles: Low n = 628, moderate (1368.5-3213 MET-minutes/week; n = 776), and high > 3213 MET-minutes/week; n = 425). Participants in the low PA group were generally older and had lower educational levels (Table 1).

Table 1 Patient characteristics based on physical activity level, n (%)/mean ± SD/median (interquartile range).

Variables	Low (n = 628)	Moderate (n = 776)	High (n = 425)	P value
Age	52 ± 8	51 ± 8	51 ± 7	0.018
Male	334 (53.2)	387 (49.9)	233 (54.8)	0.210
Married	594 (94.6)	748 (96.4)	410 (96.5)	0.180
Illiteracy	112 (17.8)	97 (12.5)	61 (14.4)	0.019
Farmers	570 (90.8)	671 (86.5)	365 (85.9)	0.019
Household income1	2.0 (1.0-3.0)	2.0 (1.0-3.75)	2.0 (1.0-3.0)	0.130
Body mass index	23.9 ± 3.1	23.8 ± 3.0	23.9 ± 2.9	0.850
H. pylori infection status	294 (47.0)	330 (42.9)	193 (45.5)	0.310
Missing	1 (0.1)	7 (0.9)	2 (0.4)
Smoking				0.037
Never	392 (62.5)	520 (67.1)	253 (59.5)
Current	198 (31.6)	228 (29.4)	149 (35.1)
Ever	37 (5.9)	27 (3.5)	23 (5.4)
Missing	1 (0.1)	1 (0.1)
Drinking2	25 (4.0)	25 (3.2)	21 (4.9)	0.330
Diet habits
Eating hot food	327 (52.1)	425 (54.8)	261 (61.4)	0.010
Eating quickly	157 (25.0)	129 (16.6)	92 (21.6)	< 0.001
Eating fried foods3	11 (1.8)	11 (1.4)	5 (1.2)	0.740
Average daily energy intake (kcal)	1325.8 (793.2-2015.2)	1319.8 (839.8-2053.9)	1700.4 (1129.5-2439.8)	< 0.001
Medical history
Gastritis	87 (13.9)	114 (14.7)	63 (14.8)	0.880
Peptic ulcers	9 (1.4)	9 (1.2)	12 (2.8)	0.084
Gallbladder diseases	73 (11.6)	94 (12.1)	54 (12.7)	0.870
Gastric polyps	14 (2.2)	8 (1.0)	16 (3.8)	0.006
Family history of gastric cancer	0 (0.0)	4 (0.5)	1 (0.2)	0.180

¹Annual family income in ten thousand Yuan.

²At least 1 time per week, each medium amount.

³At least 4 times a week.

H. pylori: Helicobacter pylori.

Association between PA and HGD/EGC risk

Compared with low PA, moderate PA was associated with significantly lower odds of HGD/EGC after multivariable adjustment (model 3: OR = 0.64, 95%CI: 0.43-0.95). High PA showed a similar inverse direction but did not reach statistical significance (Table 2). A significant dose-response trend across PA tertiles was observed in the fully adjusted model (P for trend = 0.047). Restricted cubic spline analysis revealed an inverse dose-response relationship between total MET-minutes per week and HGD/EGC risk, with the lowest observed risk occurring at approximately 4200 MET-minutes/week (Figure 2). When analyzed separately (Supplementary Table 2), moderate PA was associated with lower odds of HGD (OR = 0.53, 95%CI: 0.29-0.97), whereas for EGC, high PA showed a borderline inverse association (OR = 0.59, 95%CI: 0.32-1.07). A significant trend across PA tertiles was observed for EGC (P for trend = 0.008), but not for HGD.

Figure 2 Association between total metabolic equivalent-minutes/week and high-grade dysplasia/early gastric cancer risk. MET: Metabolic equivalent.

Table 2 Association between physical activity and high-grade dysplasia/early gastric cancer risk.

Physical activity level	Model 13			Model 2			Model 3
	Case/control	OR (95%CI)	P value1	Case/control	OR (95%CI)	P value1	Case/control	OR (95%CI)	P value1
Low	75/553	Reference		75/550	Reference		75/550	Reference
Moderate	56/720	0.57 (0.40-0.82)	0.003	56/712	0.66 (0.45-0.98)	0.041	56/712	0.64 (0.43-0.95)	0.027
High	39/386	0.74 (0.50-1.12)	0.158	39/385	0.74 (0.47-1.15)	0.180	39/385	0.67 (0.43-1.05)	0.081
P for trend2									0.047

¹P value for comparison with low physical activity within each model (logistic regression).

²P for trend represents the P value for linear trend across physical activity tertiles in model 3.

³Odds ratios and 95% confidence intervals were estimated using logistic regression. Model 1 was unadjusted. Model 2 was adjusted for age, sex, marital status, education, occupation, household income, smoking, alcohol consumption, dietary habits (hot food, fast eating, fried food), body mass index, Helicobacter pylori infection, medical history (gastritis, peptic ulcer, gallbladder disease, gastric polyps), and family history of gastric cancer. Model 3 was additionally adjusted for total energy intake.

OR: Odds ratios; CI: Confidence interval.

Subgroup analyses of PA and HGD/EGC risk

Among males, both moderate and high PA were associated with lower odds of HGD/EGC. We also observed a significant inverse association among H. pylori-positive individuals, whereas there was no association among H. pylori-negative participants. For tumor location, inverse associations were evident for cardia lesions but not for non-cardia lesions (Table 3). Formal interaction tests indicated effect modification by H. pylori status (P = 0.029) and tumor location (P < 0.001), but not by sex (P = 0.110). Given the limited number of cases in certain strata and the absence of correction for multiple comparisons, these subgroup findings should be considered exploratory.

Table 3 Subgroup analyses of the association between physical activity and high-grade dysplasia/early gastric cancer risk.

Physical activity level	Gender3				H. pylori status				Tumor location
		Case/control	OR (95%CI)	P value1		Case/control	OR (95%CI)	P value1		Case/control	OR (95%CI)	P value1
	Males				Positive				Cardia
Low		61/272	Reference			41/253	Reference			30/1647	Reference
Moderate		40/341	0.59 (0.37-0.95)	0.031		24/305	0.65 (0.35-1.18)	0.157		17/1647	0.44 (0.22-0.86)	0.016
High		28/204	0.58 (0.34-0.99)	0.047		14/179	0.40 (0.19-0.83)	0.015		6/1647	0.27 (0.10-0.68)	0.006
P for interaction2				0.110				0.029				< 0.001
	Females				Negative				Non-cardia
Low		14/278	Reference			34/297	Reference			45/1647	Reference
Moderate		16/371	0.87 (0.40-1.90)	0.725		32/407	0.73 (0.42-1.27)	0.264		39/1647	0.74 (0.46-1.18)	0.203
High		11/181	1.14 (0.48-2.75)	0.764		25/206	1.26 (0.68-2.33)	0.459		33/1647	0.93 (0.56-1.54)	0.774

¹P value for comparison with low physical activity within each stratum.

²P value for multiplicative interaction.

³Odds ratios and 95% confidence intervals were estimated using multivariable logistic regression. Models were adjusted for age, sex (except in sex-stratified analyses), marital status, education, occupation, household income, smoking, alcohol consumption, dietary habits, body mass index, Helicobacter pylori infection (except in Helicobacter pylori-stratified analyses), medical history, family history of gastric cancer, and total energy intake.

H. pylori: Helicobacter pylori; OR: Odds ratios; CI: Confidence interval.

DISCUSSION

This cross-sectional study with a large sample size of 1829 participants indicates that PA, especially moderate PA was associated with HGD and EGC. Stratified analysis revealed that the association between PA and HGD/EGC was pronounced among men, cardia HGD/EGC, and H. pylori-positive patients. However, given the cross-sectional design, these findings should be interpreted cautiously.

HGD and EGC represent early stages of gastric carcinogenesis and are classified as non-invasive neoplasia under the Vienna system[13]. Because these lesions are potentially curable with endoscopic resection, identifying modifiable risk factors at this stage has important clinical implications. Most prior epidemiological studies have focused on advanced gastric cancer, and data on early gastric lesions remain limited.

When endpoints were analyzed separately, the inverse association appeared more consistent for EGC than for HGD. This difference may reflect limited statistical power or potential biological heterogeneity across stages of carcinogenesis. However, the relatively small number of events and wide CIs warrant cautious interpretation.

Previous studies have generally reported inverse associations between PA and gastric cancer risk[14-17]. The Continuous Update Project of the World Cancer Research Fund/American Institute for Cancer Research concluded that higher PA levels are likely associated with reduced stomach cancer risk, although the evidence remains limited and suggestive. A recent umbrella review further strengthened this evidence, showing that moderate-to-high PA is associated with significantly lower gastric cancer risk[18]. Consistently, Ma et al[19] reported in a real-world evidence synthesis that higher PA was associated with reduced gastric cancer incidence and mortality. Our findings regarding moderate activity are broadly consistent with these reports.

However, discrepancies across studies may reflect several methodological differences. First, most prior investigations examined advanced gastric cancer, whereas our study focused on early neoplastic lesions (HGD and EGC), which may represent distinct stages of carcinogenesis. Second, variations in PA assessment (self-reported questionnaires vs objective measures), exposure categorization, and adjustment for confounders could contribute to inconsistent findings. Third, population characteristics, including background H. pylori prevalence, dietary patterns, and obesity distribution, differ substantially across regions, potentially modifying the observed associations.

Guidelines recommend 150-300 minutes/week moderate activity for cancer prevention[20], though gastric cancer-specific recommendations vary (600-3000 MET-minutes/week)[21,22]. In this cohort, individuals whose activity levels fell within this broad range demonstrated significantly lower odds of HGD/EGC compared with those in the lowest tertile. The significant trend across PA tertiles suggests a possible dose-response relationship. However, the cross-sectional design precludes causal inference, and reverse causation cannot be excluded. Individuals with early symptoms or undiagnosed disease may have reduced their PA prior to diagnosis, potentially biasing the association. In addition, the limited number of events and extensive covariate adjustment may introduce statistical instability. Therefore, the observed range should not be interpreted as a clinical recommendation.

Notably, the inverse association with HGD/EGC appeared confined to moderate PA, with no comparable effect observed at high intensity. Several explanations merit consideration. One reason is the small case count in the high-intensity group, which lowered statistical power. Self-reported activity data are prone to misclassification, particularly at upper intensity ranges where recall becomes less reliable. Second, inaccurate self-reporting may also distort the true association. Third, moderate activity may be more strongly associated with lower odds in a potentially non-linear manner.

Subgroup analyses indicated a significant inverse association among H. pylori-positive individuals. Chronic H. pylori-induced mucosal inflammation has been implicated in gastric carcinogenesis[23,24], and PA has been reported to modulate inflammatory responses[25]. Our data revealed a significant association for cardia HGD/EGC (OR = 0.27, 95%CI: 0.10-0.68), but not for non-cardia lesions. Only 53 cardia cases were included in this analysis. The wide CIs suggest limited statistical precision. Given the distinct etiologic backgrounds of cardia and non-cardia gastric cancer reported in previous studies[26-30], PA may have divergent effects on reflux: Some studies suggest benefit, while others report worsening with intense exercise[31,32]. However, we did not assess relevant clinical indicators, including reflux symptoms and metabolic parameters, limiting further interpretation.

The interactions with H. pylori status and tumor location suggest that the protective effect of PA may vary by these factors. These analyses were exploratory, with small event numbers in some strata and no correction for multiple comparisons, so chance findings cannot be excluded. The stronger inverse association in H. pylori-positive individuals supports the idea that PA may reduce infection-related chronic inflammation. The site-specific effect (cardia vs non-cardia) may reflect different etiologies, though residual confounding is possible. These findings need to be replicated in independent cohorts.

A number of biological mechanisms may underlie the inverse association of PA against cancer development. These include improved insulin sensitivity, decreased hyperinsulinemia[33], altered adipokine profiles, and enhanced antitumor immune function[34]. However, metabolic and inflammatory biomarkers were not assessed in the present study; thus, mechanistic inferences remain speculative[35]. Future prospective studies that include objective biomarker measurements will be crucial for clarifying the precise mechanisms.

This study has several strengths. It is among the first to examine PA in relation to early gastric neoplastic lesions. The Wuwei cohort provides a population-based sample, enhancing generalizability. We also adjusted for multiple important confounders, including BMI, H. pylori infection, smoking, alcohol consumption, and dietary factors.

Several limitations should be acknowledged. PA was assessed using the IPAQ-SF, and endoscopic findings were assessed at the same visit without a defined temporal sequence, which may introduce recall bias and measurement error. The cross-sectional design precludes establishing temporality and raises the possibility of reverse causation. Selection bias is also a concern, as only 7.8% of the original cohort met the inclusion criteria. Included participants differed from excluded individuals in several baseline characteristics; thus, external validity may be limited. However, key variables such as BMI and marital status were comparable, and most differing covariates were adjusted for in multivariable models. In addition, more detailed dietary patterns, occupational PA, broader socioeconomic indicators, and comorbidities were not comprehensively assessed, and residual confounding from these factors cannot be excluded. Furthermore, although covariates were pre-specified based on established evidence, the number of adjusted variables relative to the limited outcome events may introduce statistical instability and potential model overfitting. Residual dietary confounding remains possible despite adjustment for total energy intake. Finally, multiple subgroup analyses increase the risk of type I error. Larger prospective studies are needed to confirm these findings.

CONCLUSION

This study demonstrates that moderate PA is inversely associated with HGD/EGC, particularly in men, individuals with H. pylori infection, and those with cardia HGD/EGC. However, as this study is of a cross-sectional design, it cannot establish causality. The observed associations should be regarded as hypothesis-generating rather than causal inferences. Future studies are warranted to confirm these findings and investigate potential mechanisms.

ACKNOWLEDGEMENTS

The participants of the study are warmly acknowledged for their commitment to the study. We thank the interviewers from the Wuwei Center for Disease Control and Prevention and Lanzhou University School of Public Health, as well as the participants in the Wuwei cohort.