Lifestyle-Related Risk Factors for Pancreatic Ductal Adenocarcinoma: A Longitudinal Analysis of 1,120,377 Individuals From the NHISS Cohort

Abstract

Objective: Utilizing data from the National Health Insurance Sharing Service database, this study explored significant risk factors for pancreatic cancer in a cohort of 1,120,377 South Korean individuals over a 10-year period (2009-2019).

Methods: Propensity score matching was employed to ensure comparability between 3535 pancreatic cancer patients and a control group with a common cold diagnosis. The study analyzed various lifestyle factors and biochemical markers, including smoking status, alcohol consumption, fasting blood glucose (FBS) levels, liver enzyme levels, and Charlson comorbidity index (CCI) scores.

Results: The findings revealed that current smoking, frequent alcohol consumption, and elevated levels of FBS and liver enzymes were associated with an increased risk of pancreatic cancer. Conversely, engaging in high-intensity exercise (≥ 20 min, twice weekly) was correlated with a 20% reduction in pancreatic cancer risk (p < 0.05). Additionally, optimal thresholds for total cholesterol (179.50 mg/dL), GGT (29.50 U/L), low-density lipoprotein cholesterol (104.50 mg/dL), and CCI score (2.50) were identified, which may facilitate early diagnosis and intervention.

Conclusions: These findings underscore the importance of modifiable lifestyle factors in managing pancreatic cancer risk and highlight the potential of personalized, evidence-based interventions-such as high-intensity exercise programs-in improving prevention and treatment outcomes.

Keywords: exercise; exercise program development; pancreatic cancer; receiver operating characteristic curve analysis; risk factors.

FIGURE 1

Study flowchart detailing the data origin and analysis parameters for pancreatic cancer patients. NHISS data were used in this study. Out of the 1,120,377 registered patients, the records of 5204 pancreatic cancer patients and 700,336 common cold patients were extracted. After eliminating missing covariates, 1:1 PSM was performed based on sex, age, region, and type of healthcare insurance. Ultimately, 3535 pancreatic cancer patients and 3535 common cold patients were selected for the analysis. The 21 parameters under evaluation were subjected to t‐test, logistic regression, and ROC curve analyses. NHISS DB, National Health Insurance Sharing Service database; PSM, propensity score matching; ROC, Receiver operating characteristic.

FIGURE 2

ROC curve analysis of the GGT and CCI score results. For the ROC curve analysis, variables with an AUC of 0.5 or greater, along with sensitivity and specificity greater than 0.5, were selected. The optimal threshold values of TC (AUC, 0.554), GGT (AUC, 0.546), LDL (AUC, 0.555), and CCI score (AUC, 0.637) were determined to be 179.50 (mg/dL), 29.50 (U/L), 104.50 (mg/dL), and 2.50, respectively. AUC, area under the curve; CCI, Charlson comorbidity index; GGT, gamma‐glutamyl transferase; LDL, low‐density lipoprotein; PSM, propensity score matching; ROC, Receiver operating characteristic; TC, total cholesterol.