Risk prediction model for high-grade dysplasia in resected pancreatic mucinous cystic neoplasms: a nomogram-based approach

Abstract

Background: Mucinous cystic neoplasms (MCN) are precancerous pancreatic lesions. Their management is challenged by the lack of robust predictors of malignancy. This study aimed to identify preoperative predictors of high-grade dysplasia (HGD) or invasive carcinoma and to develop a predictive nomogram.

Methods: This multicenter retrospective study included patients who underwent resection for MCN between 2008 and 2021 across 13 centers. Surgical decisions were based on European guideline criteria. Diagnosis required ovarian-type stroma. Preoperative CT/MRI were centrally reviewed by an expert radiologist blinded to pathology. Independent predictors of ≥HGD were identified using multivariable logistic regression.

Results: Among 198 patients, 32 (16 %) had ≥ HGD, including 14 (7 %) with invasive carcinoma. Independent predictors were older age (median 52.5 vs 45 years, p = 0.01), symptomatic presentation (72% vs 47%, OR 2.88, p = 0.01), larger cyst size (85 vs 46.5 mm, p < 0.001), and mural nodules (56% vs 15 %, OR 7.25, p < 0.001). The nomogram achieved an accuracy of 0.89 and an AUC of 0.86; a 10% risk threshold provided 85 % sensitivity. An interactive web-based calculator is publicly available.

Conclusion: Age, symptoms, cyst size, and mural nodules independently predict malignancy in resected MCNs. A validated nomogram and online calculator may support individualized surgical versus surveillance strategies.

Keywords: Mucinous cystic neoplasm; Pancreatic cancer; Pancreatic cyst; Pancreatic surgery; Precancerous lesion.