Development and validation of a multi-parameter nomogram for venous thromboembolism in gastric cancer patients: a retrospective analysis
Hang Zhou,Haike Lei,Huai Zhao,Kaifeng Huang,Yundong Wang,Ruixia Hong,Jishun Huo,Li Luo,Fang Li
DOI: https://doi.org/10.7717/peerj.17527
IF: 3.061
2024-06-25
PeerJ
Abstract:Objective: Gastric cancer (GC), one of the highest venous thromboembolism (VTE) incidence rates in cancer, contributes to considerable morbidity, mortality, and, prominently, extra cost. However, up to now, there is not a high-quality VTE model to steadily predict the risk for VTE in China. Consequently, setting up a prediction model to predict the VTE risk is imperative. Methods: Data from 3,092 patients from December 15, 2017, to December 31, 2022, were retrospectively analyzed. Multiple logistic regression analysis was performed to assess risk factors for GC, and a nomogram was constructed based on screened risk factors. A receiver operating curve (ROC) and calibration plot was created to evaluate the accuracy of the nomogram. Results: The risk factors of suffering from VTE were older age (OR = 1.02, 95% CI [1.00-1.04]), Karnofsky Performance Status (KPS) ≥ 70 (OR = 0.45, 95% CI [0.25-0.83]), Blood transfusion (OR = 2.37, 95% CI [1.47-3.84]), advanced clinical stage (OR = 3.98, 95% CI [1.59-9.99]), central venous catheterization (CVC) (OR = 4.27, 95% CI [2.03-8.99]), operation (OR = 2.72, 95% CI [1.55-4.77]), fibrinogen degradation product (FDP) >5 µg/mL (OR = 1.92, 95% CI [1.13-3.25]), and D-dimer > 0.5 mg/L (OR = 2.50, 95% CI [1.19-5.28]). The area under the ROC curve (AUC) was 0.82 in the training set and 0.85 in the validation set. Conclusion: Our prediction model can accurately predict the risk of the appearance of VTE in gastric cancer patients and can be used as a robust and efficient tool for evaluating the possibility of VTE.