Risk factors and early prediction of pancreatic cancer among patients with diabetes mellitus: a systematic review and meta-analysis.
[AIM] Diabetes mellitus (DM) increases the risk of pancreatic cancer (PC).
- 95% CI 0.75-0.84
- 연구 설계 meta-analysis
APA
Cong H, Song J, et al. (2025). Risk factors and early prediction of pancreatic cancer among patients with diabetes mellitus: a systematic review and meta-analysis.. Frontiers in endocrinology, 16, 1698850. https://doi.org/10.3389/fendo.2025.1698850
MLA
Cong H, et al.. "Risk factors and early prediction of pancreatic cancer among patients with diabetes mellitus: a systematic review and meta-analysis.." Frontiers in endocrinology, vol. 16, 2025, pp. 1698850.
PMID
41347135
Abstract
[AIM] Diabetes mellitus (DM) increases the risk of pancreatic cancer (PC). This study evaluates risk factors for PC in DM patients and the predictive accuracy of machine learning (ML) models to provide research-backed data for the development and update of intelligent prediction tools.
[METHODS] PubMed, Cochrane, Embase, and Web of Science were systematically retrieved, up to December 1, 2024. The quality of the original studies was assessed through the Newcastle-Ottawa Scale (NOS). A meta-analysis was conducted on the c-index that reflects the comprehensive accuracy of the prediction models.
[RESULTS] 18 studies were included. The rough annual incidence of PC among DM was estimated at 0.4% (95% CI: 0.1% - 0.9%), and the incidence rates of PC for new-onset DM and pre-existing DM were 0.3% (95% CI: 0.1% - 0.5%) and 0.5% (95% CI: 0% - 2.7%), respectively. The possible risk factors included age at DM diagnosis, weight changes, blood sugar, ALP, GI symptoms, pancreatic disease history, and the usage of hypoglycemic drugs. ML models based on risk factors had ROC-AUCs of 0.79 (95% CI: 0.75-0.84) in the training set and 0.79 (95% CI: 0.71-0.87) in the validation set.
[CONCLUSIONS] Risk factors for PC in DM are diverse. Current ML models appear to exhibit favorable predictive accuracy but are built on severely imbalanced data. Future studies with larger, broader populations are needed to address this limitation.
[SYSTEMATIC REVIEW REGISTRATION] https://www.crd.york.ac.uk/PROSPERO/, identifier CRD42025631534.
[METHODS] PubMed, Cochrane, Embase, and Web of Science were systematically retrieved, up to December 1, 2024. The quality of the original studies was assessed through the Newcastle-Ottawa Scale (NOS). A meta-analysis was conducted on the c-index that reflects the comprehensive accuracy of the prediction models.
[RESULTS] 18 studies were included. The rough annual incidence of PC among DM was estimated at 0.4% (95% CI: 0.1% - 0.9%), and the incidence rates of PC for new-onset DM and pre-existing DM were 0.3% (95% CI: 0.1% - 0.5%) and 0.5% (95% CI: 0% - 2.7%), respectively. The possible risk factors included age at DM diagnosis, weight changes, blood sugar, ALP, GI symptoms, pancreatic disease history, and the usage of hypoglycemic drugs. ML models based on risk factors had ROC-AUCs of 0.79 (95% CI: 0.75-0.84) in the training set and 0.79 (95% CI: 0.71-0.87) in the validation set.
[CONCLUSIONS] Risk factors for PC in DM are diverse. Current ML models appear to exhibit favorable predictive accuracy but are built on severely imbalanced data. Future studies with larger, broader populations are needed to address this limitation.
[SYSTEMATIC REVIEW REGISTRATION] https://www.crd.york.ac.uk/PROSPERO/, identifier CRD42025631534.
MeSH Terms
Humans; Pancreatic Neoplasms; Risk Factors; Diabetes Mellitus; Machine Learning; Prognosis