Nomogram for Predicting Distant Metastasis of Pancreatic Ductal Adenocarcinoma: A SEER-Based Population Study

• Published in: Current oncology (Toronto) Vol. 29; no. 11; pp. 8146 - 8159
• Main Authors: Li, Weibo, Wang, Wei, Yao, Lichao, Tang, Zhigang, Zhai, Lulu
• Format: Journal Article
• Language: English
• Published: Switzerland MDPI AG 28.10.2022; MDPI
• Subjects: Adenocarcinoma; Analysis; Carcinoma, Pancreatic Ductal; distant metastasis; Epidemiology; Humans; Medical research; Medicine, Experimental; Metastasis; Neoplasm Staging; Nomograms; Pancreatic cancer; pancreatic ductal adenocarcinoma; Pancreatic Neoplasms; Pancreatic Neoplasms - pathology; Prognosis; Retrospective Studies; risk; Risk factors; Taiwan; pancreatic cancer; risk; prognosis; distant metastasis; pancreatic ductal adenocarcinoma
• ISSN: 1718-7729; 1198-0052; 1718-7729
• DOI: 10.3390/curroncol29110643

(1) Background: The aim of this study was to identify risk factors for distant metastasis of pancreatic ductal adenocarcinoma (PDAC) and develop a valid predictive model to guide clinical practice; (2) Methods: We screened 14328 PDAC patients from the Surveillance, Epidemiology, and End Results (SEER) database between 2010 and 2015. Lasso regression analysis combined with logistic regression analysis were used to determine the independent risk factors for PDAC with distant metastasis. A nomogram predicting the risk of distant metastasis in PDAC was constructed. A receiver operating characteristic (ROC) curve and consistency-index (C-index) were used to determine the accuracy and discriminate ability of the nomogram. A calibration curve was used to assess the agreement between the predicted probability of the model and the actual probability. Additionally, decision curve analysis (DCA) and clinical influence curve were employed to assess the clinical utility of the nomogram; (3) Results: Multivariate logistic regression analysis revealed that risk factors for distant metastasis of PDAC included age, primary site, histological grade, and lymph node status. A nomogram was successfully constructed, with an area under the curve (AUC) of 0.871 for ROC and a C-index of 0.871 (95% CI: 0.860-0.882). The calibration curve showed that the predicted probability of the model was in high agreement with the actual predicted probability. The DCA and clinical influence curve showed that the model had great potential clinical utility; (4) Conclusions: The risk model established in this study has a good predictive performance and a promising potential application, which can provide personalized clinical decisions for future clinical work.