Abstract / Summary
Abstract Background Patients with chronic active Epstein-Barr virus infection (CAEBV) face a remarkably high rate of early mortality. We aimed at developing a nomogram and a risk scoring system able to predict one-year acute mortality in patients with CAEBV. Methods A retrospective, two-center analysis was conducted on 180 CAEBV patients from January 2014 to December 2024, and identified independent risk factors for 1-year death of CAEBV. Using LASSO regression analysis, while multifactorial Cox regression was utilized to establish the nomogram model (Nomo) and the risk stratification score (Score). Calibration plots, receiver operating characteristic (ROC) curves, and decision curve analysis (DCA) were used to assess model’s clinical utility. Results Among 180 CAEBV cases, 49 patients (27.2%) died within 1 year and 131 patients (72.8%) survived. LASSO and COX regression together identified five independent risk factors (all P < 0.05): polyserositis (HR = 3.68, 95% CI: 1.89–7.18), cytopenia (HR = 3.29, 95% CI: 1.68–6.42), plasma EBV DNA (HR = 1.39, 95% CI: 1.16–1.66), LDH (HR = 1.01, 95% CI: 1.01–1.01), and TBIL (HR = 1.01, 95% CI: 1.01–1.02). Nomo and Score models for predicting 1-year acute mortality in CAEBV patients was subsequently developed. Both the Nomo and Score models demonstrated excellent discrimination (AUC > 0.8) and reliable calibration over a 1‑year period. The Score model exhibited predictive power comparable to Nomo, as indicated by internal cross-validation and improvement‑index analysis. Conclusion We developed and validated an evidence-based, weighted scoring system to stratify the mortality risk of CAEBV. These tools can assist clinicians in rapid patient triage and facilitate the implementation of tiered management, particularly during peak mortality periods.