Time-Dependent Prognostic Value of the sCD25-to-Ferritin Ratio in Chronic Active Epstein–Barr Virus-Associated Hemophagocytic Lymphohistiocytosis

Abstract

INTRODUCTION: Chronic active Epstein-Barr virus infection (CAEBV) is an EBV-positive T/NK-cell lymphoproliferative disease characterized by persistent EBV infection in T or NK cells. CAEBV-HLH refers to hemophagocytic lymphohistiocytosis secondary to CAEBV. As important markers of HLH activity, soluble CD25 (sCD25) and serum ferritin reflect lymphocyte activation and macrophage-mediated inflammation, respectively. The sCD25-to-ferritin ratio may therefore integrate information between the underlying EBV-driven T/NK-cell lymphoproliferative process and HLH-related hyperinflammation. Considering the heterogeneous clinical course varying over time, we evaluated the time-dependent prognostic value of baseline sCD25/ferritin in CAEBV-HLH. METHODS: This multicenter retrospective study included 108 patients with CAEBV-HLH diagnosed between January 2015 and February 2026. Baseline sCD25/ferritin was calculated at diagnosis. All-cause mortality within the first year was assessed across cumulative monthly follow-up. Time-restricted Cox regression and time-dependent receiver operating characteristic (ROC) analysis were performed to evaluate prognostic association and discriminative performance over time. The optimal cutoff was determined using the Youden Index, and the AUC was validated by bootstrap resampling (1000 iterations) to correct for optimism bias. Survival was compared by Kaplan-Meier analysis. The discriminative performance of the sCD25/Fer ratio was further compared against the existing HScore using DeLong's test for paired ROC curves. Analyses were repeated in patients who did not undergo hematopoietic stem cell transplantation (HSCT). RESULTS: During follow-up censored at 12 months, 43 (39.81%) deaths occurred. Baseline sCD25/ferritin was significantly higher in non-survivors than in survivors (P = 0.002). In the overall cohort, the sCD25/ferritin was significantly associated with mortality from month 3 onward (HR, 1.23; 95% CI, 1.02-1.47), with stable estimates through 12 months. The 4-month window was selected as an early clinically informative time point, with an AUC of 0.705 in conventional ROC analysis. A cutoff of 27.98 stratified the overall cohort into distinct survival groups (P < 0.001). In the non-HSCT subgroup, predictive performance became significant from month 4 onward, and the 4-month AUC increased to 0.766, with a cutoff of 22.80. Notably, sCD25/Ferritin ratio significantly outperformed the conventional HScore for 4-month mortality prediction (AUC 0.730 vs. 0.584; P = 0.022 by DeLong's test), with consistent superiority in the non-HSCT subgroup (AUC 0.768 vs. 0.597; P = 0.029). CONCLUSIONS: The baseline sCD25/ferritin ratio is a disease-relevant biomarker in CAEBV-HLH. Its prognostic value became statistically significant from month 3 onward and was more pronounced in patients without HSCT, indicating that this ratio may better reflect the natural disease course when outcomes are less influenced by transplantation.