Abstract: Background/Objectives: Acute pancreatitis (AP) is increasingly recognized as a risk factor for diabetes mellitus (DM) even after a single episode of AP. Observational studies in childhood AP are lacking but represent the ideal forum to develop risk models. While pancreatitis risk genes are associated with pancreatitis, it is not known if they affect risk of DM. Methods: This was an observational cohort study that enrolled subjects <21 years old from their index AP and followed them to 12 months after. Subjects with prior pancreas surgery or pre-existing DM were not eligible. Pancreatitis risk genes (CASR, CEL, CFTR, CLDN2, CPA1, CTRC, PRSS1, SBDS, SPINK1, and UBR1) were tested by next generation sequencing. Endocrine insufficiency was defined according to standards from the American Diabetes Association for pre-DM and DM. Results: Total 120 index AP subjects were genotyped. Sixty-three subjects (52.5%) had at least 1 variant affected. Forty-six (38.3%) had 1 variant, 12 (10%) had 2 variants, 3 (2.5%) had 3 variants, and 2 (1.7%) had 4 variants. Among the 10-gene panel, 49 distinct variants were detected across 9 genes, with a total of 87 positive results, variants were observed in multiple subjects. The most common genes affected were CFTR (47 variants), SPINK1 (10), and PRSS1 (8), Figure 1.
Of the 120 with gene testing and related diabetes data, 4 were excluded (3 with total pancreatectomy, and 1 death), resulting in 116 with diabetes data included in analysis. At 12 months 95 had normal endocrine testing and, 21 (18%) had endocrine insufficiency (6 DM, 15 Pre-DM).
Those with versus without endocrine insufficiency were similar in age, sex, race, ethnicity, body mass index, and AP recurrence. Severe AP (57.1% vs 20.6%; p=0.002), requiring insulin during AP (14.3% vs 1.0%; p=0.02), and at least one gene affected (71.4% vs 47.3%, p=0.046) were enriched among the endocrine insufficient group compared to the normal group. At the gene level, SBDS had higher prevalence in the endocrine insufficient group (9% vs 0%; p=0.03). We observed a higher prevalence of the CFTR 5T allele among subjects who developed DM/pre-DM, with each copy of the 5T allele associated with an OR = 3.49 (1.14, 10.7), p=0.029.
A genetic risk score was derived from all genes with univariable p<0.2, combining the number of variants across SBDS, UBR1, CTRC, and the CFTR 5T allele. Using stepwise model-building, we identified a predictive logistic regression model for preDM/DM (AUC=0.76) that includes AP severity (OR=4.873, p=0.0006), insulin use during AP (OR=12.385, p=0.052), and genetic risk score (OR=3.954, p=0.005), Figure 2.

Conclusion: Pre-DM and DM were found in 18% in our cohort of children 12 months after a single episode of AP. Pancreatitis risk genes and AP disease severity, insulin use in AP, can predict risk to diabetes development post AP.

Figure 1. Waterfall plot of 89 variants across 9 genes clustered according to PreDM/DM status and AP attack severity generated through CbioPortal's Oncoprinter.

Figure 2. A.Venn diagram of co-occurrence of clinical and genetic variables included in the logistic regression model for PreDM and DM. B. Forest plot of odds ratios and 95% confidence limits of variables included in the predictive model.

Jessica Woo

Appearances