Abstract
BACKGROUND: Currently, there is no validated model for predicting the occurrence and prognosis of lung metastases (LM) in patients with pancreatic ductal adenocarcinoma (PDAC). We aimed to construct a nomogram for risk prediction and a prognostic model to guide clinical practice. METHODS: In total, 10,813 patients were enrolled from the Surveillance, Epidemiology, and End Results (SEER) database between 2010 and 2015 and divided into training and internal validation cohorts at a ratio of 7:3. Following nomogram construction, data of patients diagnosed with PDAC were retrospectively collected for external validation. RESULTS: Using multivariate logistic regression analysis, larger tumour size, primary tumour site in the body or tail of the pancreas, bone metastasis, and liver metastasis were associated with LM. Furthermore, through multivariate Cox analysis, we found that LM was associated with a poor prognosis in patients with PDAC. Patients who underwent surgery or chemotherapy had better prognoses. CONCLUSION: The two nomograms showed excellent performance in the training and internal validation cohorts and a favourable performance in the external validation. The prognostic nomogram divided the patients into high- and low-risk groups based on mortality. The LM risk and prognostic prediction model in PDAC showed high accuracy and reliable clinical application.