A novel prediction model of mortality in chronic pancreatitis using clinical characteristics and gene promoter hypermethylation status.

Background: Chronic pancreatitis (CP) is an inflammatory disease characterized by pain, functional deficits and increased mortality. The clinical course is unpredictable, and there are no classification systems or biomarkers to predict this. Identifying patients with high mortality risk is crucial for guiding clinical management and improving outcomes. This study presents a novel approach to a prognostic prediction model that combines clinical parameters and promoter hypermethylation (ph) of genes.

Methods: We performed methylation-specific quantitative polymerase chain reaction(qPCR) on a panel of 28 genes, using an accelerated bisulfite treatment protocol. We then developed a prognostic prediction model by backwards stepwise elimination using the methylation status of genes with a ph frequency > 5% and seven clinical factors. Survival was assessed with Kaplan-Meier survival curves and Cox regression.

Results: Ninety-seven patients with CP were included in the study. The final model included: Age, sex, exocrine insufficiency, diabetes, prior history of acute pancreatitis, and the methylation status of MLH1, HIC1, and RASSF1A. The model had an area under the curve (AUC) of 0.84 (95%CI: 0.76-0.92). A risk score was computed, and patients stratified into high and low-risk groups. The high-risk group had a significantly higher hazard ratio (HR) of death of 14.1 (95% CI; 4.3-46.0, p < 0.01).

Conclusions: This study serves as proof-of-concept that clinical factors can be combined with gene methylation status to provide additional prognostic information in patients with chronic pancreatitis. This could potentially aid the clinician in estimating which patients require intense follow-up. However, external validation is required.