Our study population was derived from the Kailuan Cohort (registration number:ChiCTR-TNC-11001489; https://www.chictr.org.cn/showprojEN.html?proj=48316), which consists of employees (including retirees) of the Kailuan Group. Detailed information about the Kailuan Cohort has been previously reported (16). In summary, the Kailuan Cohort was established in 2006 by recruiting 101,510 employees (aged 18–98 years) from the Kailuan Group. Participants underwent a series of physical examinations, biochemical tests, and questionnaires at 11 affiliated hospitals of the Kailuan Group. Subsequently, these assessments were repeated every two years, resulting in eight waves of data collection to date. Participants were excluded if they met any of the following criteria: (1) did not complete three consecutive health examinations between 2006 and 2010; (2) had missing data for any of the CVAI-related variables, including age, BMI, WC, triglyceride (TG), and high-density lipoprotein-cholesterol (HDL-C) in the three examinations; (3) had a history of cancer at baseline in 2006 or developed cancer between 2006 and 2010; (4) had missing data for any of the following covariates: age, sex, reported income, education level, marital status, smoking status, drinking status, physical activity, family history of malignancy, total cholesterol (TC), high-sensitivity C-reactive protein (hs-CRP), and low-density lipoprotein-cholesterol (LDL-C). Finally, a total of 44,213 participants were included in this study ( Figure 1 ).

Participants underwent physical examinations conducted by trained medical staff, including measurements of height, weight, and WC. BMI was calculated as weight (kg) divided by height squared (m²). After fasting for at least 8 hours, each participant underwent an ultrasound examination of the liver, gallbladder, and other organs by a trained sonographer. Diagnosis of cirrhosis, fatty liver, gallstones, and gallbladder polyps was made according to pre-established clinical criteria (17, 18).

Blood samples were collected from all participants in the morning after an overnight fast. A total of 5 mL of peripheral venous blood was drawn by trained medical staff and analyzed on a Hitachi 747 automatic analyzer (Hitachi, Tokyo, Japan) to determine serum concentrations of TG, TC, HDL-C, LDL-C, and hs-CRP. Serum TC and TG were measured using enzymatic colorimetric methods, while hs-CRP was measured using a high-sensitivity turbidimetric assay.

Participants completed a standardized questionnaire to collect data on age, gender, reported household income, education level, marital status, smoking status, drinking status, physical activity, and family history of malignancy. Smokers were defined as those who reported smoking at least one cigarette per day, on average, for at least 6 months in the past year. Drinkers were defined as those who reported consuming at least 100 ml of alcohol per day, on average, for at least 6 months in the past year. Physical activity was defined as engaging in at least 30 minutes of exercise per session, at least 3 times per week.

The CVAI calculation formula is (13):

The cumCVAI calculation formula is (19):[(CVAI₂₀₀₆+CVAI₂₀₀₈)/2×time_1-2]+ [(CVAI₂₀₀₈+CVAI₂₀₁₀)/2×time_2-3], That is the weighted sum of the mean CVAI obtained at each examination, where CVAI₂₀₀₆, CVAI₂₀₀₈, and CVAI₂₀₁₀ represent the CVAI obtained at the examinations in 2006, 2008, and 2010, respectively, and time_1–2 and time_2–3 represent the time intervals between consecutive examinations.

The follow-up period started from the physical examination in 2010 and ended with the occurrence of new-onset GI cancers. For participants who did not develop GI cancers, the follow-up ended on December 31, 2021. For those who died without developing GI cancers, the follow-up ended at the time of death. According to the International Classification of Diseases, Tenth Revision (ICD-10), this study included the following GI cancers: liver cancer (C22), gallbladder or extrahepatic bile duct cancer (C23 and C24), stomach cancer (C16), pancreatic cancer (C25), small bowel cancer (C17), esophageal cancer (C15), and colorectal cancer (C18-C21). Diagnosis was confirmed through the following methods: (1) conducting a physical examination and questionnaire survey for participants every two years, with a cutoff date of December 31, 2021; (2) annually querying relevant medical records from the municipal healthcare system and the Social Security Information System of Kailuan General Hospital; (3) reviewing death certificates from the provincial vital statistics office to obtain additional missing information. These three methods maximize the collection of all cancer cases and minimize omissions. All cancer cases required a definitive pathological diagnosis for re-confirmation. In the absence of a pathological diagnosis, the cases were further evaluated by two or more specialized oncologists. Only when the physicians reached a consensus on the cancer diagnosis would the case be confirmed and included in the cohort.

Data analysis was performed using SAS 9.4 statistical software, with a two-sided P-value of <0.05 considered statistically significant. Normally distributed continuous data are presented as mean ± standard deviation, and comparisons between groups were conducted using analysis of variance (ANOVA). Skewed continuous data are presented as median and interquartile range, with comparisons between groups performed using the non-parametric Kruskal-Wallis test. Categorical data are expressed as frequencies and percentages, with intergroup comparisons conducted using the chi-square (χ²) test.

The incidence of GI cancers is expressed per 1,000 person-years, and the cumulative incidence of GI cancers was estimated using Kaplan-Meier curves and log-rank tests. A multivariable-adjusted restricted cubic spline (RCS) analysis was conducted to examine the dose-response relationship between CVAI and the risk of GI cancer occurrence. A multivariable Cox proportional hazards regression analysis was conducted to calculate the hazard ratios (HR) and 95% confidence intervals (CI) for the association between cumCVAI quartiles and the incidence of GI cancers. Model 1 represents the univariate analysis; Model 2 adjusts for age and gender based on Model 1; Model 3 further adjusts for TC, hs-CRP, BMI, LDL-C, physical activity, reported income, educational background, marital status, smoking status, drinking status, and family. In multivariate analysis of specific cancer sites, adjustments were made for fatty liver and liver cirrhosis in liver cancer cases, and for gallstones and gallbladder polyps in gallbladder or extrahepatic bile duct cancer cases. Subgroup analyses were conducted stratified by age (<60 years vs. ≥60 years), sex (male vs. female), BMI (<24 kg/m² vs. ≥24 kg/m²), as well as smoking and alcohol consumption status. Finally, several sensitivity analyses were performed to validate the robustness of the study results. We excluded individuals diagnosed with GI cancers within one year of follow-up to eliminate the influence of reverse causation on the findings. Additionally, we excluded participants who were taking statin medications to mitigate the effects of drugs on CVAI levels.

This study included a total of 44,213 participants, comprising 34,880 males and 9,333 females, with a mean age of 49.00 ± 11.70 years. Participants were categorized into four groups based on their cumCVAI quartiles. The baseline characteristics of the four groups are presented in Table 1 , which reveals significant differences in age, sex, TG, TC, HDL-C, LDL-C, WC, BMI, marital status, current drinking status, current smoking status, reported income, educational background (high school graduation or above), sedentary lifestyle, physical exercise, salt intake, non-alcoholic fatty liver disease (NAFLD), liver cirrhosis, and gallstone disease (P <0.05). However, no significant differences were observed in the prevalence of gallbladder polyps (P >0.05).

The median follow-up time in this study was 10.57 ± 1.72 years, during which 760 new cases of GI cancers were identified. These included 74 cases of esophageal cancer, 156 of gastric cancer, 20 of small intestine cancer, 281 of colorectal cancer, 158 of liver cancer, 21 of gallbladder or extrahepatic bile duct cancer, and 50 of pancreatic cancer. Kaplan-Meier curves showed a progressive increase in GI cancers incidence from Q1 to Q4, with statistically significant differences in cumulative incidence across the groups (P < 0.001) ( Figure 2 ). Additionally, the restricted cubic spline (RCS) curve indicated a nonlinear relationship between cumCVAI and GI cancers events (P for overall trend < 0.001; P for nonlinear trend < 0.001) ( Figure 3 ).

The multivariable analysis of the association between cumCVAI quartiles and the risk of GI cancers, along with incidence rates, is presented in Table 2 . The incidence rates for GI cancers in the Q1 to Q4 groups were 1.00, 1.45, 1.62, and 2.11 per 1,000 person-years. Compared to the Q1 group, the risk of GI cancers was significantly elevated in the Q2 group (HR 1.26; 95% CI 1.01, 1.58), Q3 group (HR 1.31; 95% CI 1.05, 1.64), and Q4 group (HR 1.48; 95% CI 1.19, 1.85) in Model 3. In the multivariable analysis for specific cancer sites, the risk of cancer was significantly elevated in the gastric cancer Q4 group (HR 1.86; 95% CI 1.16, 2.99), colorectal cancer Q2 group (HR 1.29; 95% CI 1.09, 1.66), Q3 group (HR 1.34; 95% CI 1.02, 1.96), and Q4 group (HR 1.44; 95% CI 1.07, 2.13), as well as in the liver cancer Q3 group (HR 1.49; 95% CI 1.01, 2.51) and Q4 group (HR 1.61; 95% CI 1.05, 2.72). However, no significant associations were found for esophageal cancer, small intestine cancer, gallbladder or extrahepatic bile duct cancer, or pancreatic cancer ( Supplementary Table S1 ).

Based on baseline CVAI quartiles, participants were divided into four groups. The Kaplan-Meier curve indicated that the incidence of GI cancers progressively increased from the Q1 to Q4 groups, with statistically significant differences in cumulative incidence rates between the groups (p < 0.001) ( Supplementary Figure S1 ). After adjusting for confounding factors, the risk of GI cancers was significantly higher in the Q2 group (HR 1.29; 95% CI 1.03, 1.62), Q3 group (HR 1.36; 95% CI 1.09, 1.70), and Q4 group (HR 1.59; 95% CI 1.28, 1.97) compared to the Q1 group ( Supplementary Table S2 ).

We conducted stratification based on participants’ age, gender, BMI, smoking status, and drinking status ( Figure 4 ), and found no significant interactions with cumCVAI (p > 0.05). Except for younger individuals, current smokers, and current drinkers, all other subgroups demonstrated a significant increase in the risk of GI cancers with higher quartiles of cumCVAI. In sensitivity analyses, we excluded 117 participants who developed GI cancers within one year of follow-up ( Supplementary Table S3 ) and 372 participants taking lipid-lowering medications ( Supplementary Table S4 ). The results showed no significant changes, and the association between cumCVAI and the risk of GI cancers remained significant.