Background: Solid pseudopapillary neoplasm of the pancreas (SPN) represents a rare form of low-grade malignant pancreatic cystic neoplasm. This study seeks to construct a predictive nomograph model for SPN recurrence.
Methods: Data was collected from patients with SPN from the Surveillance, Epidemiology, and End Results (SEER) database in the United States from the years 2010 to 2021 as the training cohort. We collected data from our two hospitals as an external validation cohort from the years 2011 to 2021. Logistic regression analysis was used to analyze the related factors of recurrence, and a predictive nomograph model was established and verified. The calibration curve was plotted by the Bootstrap method, and the clinical efficacy of the model was evaluated by the decision curve analysis.
Results: The SEER database included 455 patients. Five of them (1.10%) experienced recurrence, and the liver is the main recurrence site. There is a significant difference in tumor size (P = 0.001) between recurrent patients and the non-recurrent. Tumor size (P = 0.012) and regional nodes positive (P = 0.007) were independent predictors of relapse. We constructed a nomograph model based on them, the C-index 0.782 with a p-value 0.001. The C-index of the model in the external validation queue was 0.865 with a p-value 0.009. The calibration curve indicated that the model prediction probability is in well line with the actual observation probability, and decision clinical analysis showed a good net return.
Conclusion: This constructed nomogram could well predict the possibility of SPN recurrence.
Keywords: Prediction; Recurrence; Solid pseudopapillary neoplasm of the pancreas.
© 2025. The Author(s).