Prevalence and predictive factors of computed tomography-detected pancreatopathy in a cohort of adult patients hospitalized for newly diagnosed or uncontrolled non-auto-immune diabetes

Abstract

Objective

To assess the prevalence of pancreatic lesions in patients hospitalized for newly diagnosed or uncontrolled diabetes mellitus and identify potential predictive factors for pancreatic pathology.

Methods

We conducted a retrospective, single-center observational study at Brest University Hospital between February 2016 and February 2022. Adult patients hospitalized for newly diagnosed diabetes or uncontrolled diabetes mellitus who underwent a computed tomography (CT) scan within six months of admission were included. Patients with type 1 diabetes, prior pancreatectomy, or a known history of pancreatic adenocarcinoma were excluded. Clinical, biological, and imaging data were analyzed.

Results

Among 412 patients analyzed, 53 (12.9 %) presented pancreatic abnormalities (PA), including 11 cases (2.7 %) of malignant pancreatic lesion. Predictive factors for PA (P < 0.05) included (odds ratio [95 % confidence interval]): age ≥ 65 years (2.00 [1.52;5.27]); body mass index ≤ 26.4 kg/m² (2.65 [1.1;4.98]); LDL/HDL ratio ≤ 3.0 (4.75 [2.03;11.10]); and presence of at least one clinical warning sign (alcoholic use, steatorrhea, abdominal pain) (2.29 [1.21;4.33]). Using all four criteria together, 68 of 412 CT scans would have been avoided, with no missed cases.

Conclusions

The prevalence of pancreatic lesions in patients hospitalized for diabetes-related glycemic imbalance was significant although malignancy remained low. Age ≥ 65 years, low body mass index, and altered lipid profile may help identify patients requiring pancreatic imaging. Future prospective studies should refine these criteria to develop screening strategies for early pancreatic cancer detection in high-risk selected diabetic populations.