The Efficacy of the Systemic Immune-Inflammation Index and Prognosis Nutritional Index for the Diagnosis of Venous Thromboembolism in Gastrointestinal Cancers.

Purpose: This study aimed to analyze the association between venous thromboembolism (VTE) and inflammatory markers like systemic immune-inflammation index (SII) and prognosis nutritional index (PNI), and to evaluate their efficacy for the diagnosis of VTE in patients with gastrointestinal malignancies. Patients and Methods: A total of 1326 patients with the initial diagnosis of gastrointestinal cancer in the First Affiliated Hospital of Anhui Medical University (AHMU) were enrolled in the training cohort. Univariate and multivariate analysis was used to pinpoint independent predictors of VTE, which were eventually visualized as the nomogram models. The Akaike Information Criterion (AIC) was used to screen the best model. The receiver operating characteristic curve (ROC) and the clinical decision curve analysis (DCA) were utilized to evaluate the models' predictive performance in the training queue and another external sample of 250 patients at the Second Affiliated Hospital of AHMU. Results: A total of 476 patients were complicated with VTE in the training cohort. Multifactorial analysis of clinical characteristics and inflammatory markers showed that PNI, SII, age, tumor location, and therapy were independent risk factors of VTE, visualized as model A. Another model B was constructed by adding coagulation markers to the previous analysis. Model B was the best prediction model with the minimum AIC value, followed by model A with an AUC of 0.806 (95% CI 0.782~0.830) which was similar to model B's 0.832 (95% CI 0.810~0.855) but significantly higher than the currently widely used Khorana score's 0.592 (95% CI 0.562~0.621) and the CATS score's 0.682 (95% CI 0.653~0.712). The external verification yielded similar findings, with the AUC being 0.792 (95% CI 0.734~0.851), 0.834 (95% CI 0.778~0.890), 0.655 (95% CI 0.582~0.729), and 0.774 (95% CI 0.699~0.849) respectively. The DCA curves demonstrated that new models had excellent usefulness in screening patients with a high VTE risk. Conclusion: The SII and PNI were simple and viable inflammatory markers associated with VTE, and the nomogram based on them and clinical features had a meaningful clinical utility for VTE in patients with gastrointestinal malignancies.