[Hemophagocytic Lymphohistiocytosis: Clinical Characteristics and Diagnostic Prediction Model].

Objective: To investigate the clinical characteristics of patients with hemophagocytic lymphohistiocytosis (HLH) and quantify the diagnostic value of various indexes in patients with elevated soluble interleukin-2 receptor (sCD25), so as to construct a diagnostic prediction model of HLH.

Methods: The clinical characteristics of 121 patients with elevated sCD25 (≥2 400 U/ml) in the Third Affiliated Hospital of Sun Yat-Sen University were analyzed retrospectively. The patients were divided into HLH group and non-HLH group according to the diagnostic criteria of HLH. The patients with HLH were divided into infection group, tumor group, macrophage activation syndrome (MAS) group and unknown etiology group according to their etiology. The basic data and treatment of the patients were collected for univariate and multivariate logistic analysis to establish a diagnostic prediction model of HLH.

Results: Among the 121 enrolled patients with elevated sCD25, 68 were diagnosed as HLH. The proportion of patients using vasopressors, the incidence rate of disseminated intravascular coagulation (DIC), and the HScore in the HLH group were higher than those in the non-HLH group ( < 0.05). Hepatomegaly, splenomegaly, and hemophagocytosis were more common in HLH patients( < 0.05). Compared with the patients in non-HLH group, patients in HLH group had lower levels of neutrophils, platelets, fibrinogen, IgG, and IgM, while the levels of triglycerides, ferritin (FER), sCD25, serum glutamic oxaloacetic transaminase (SGOT), alkaline phosphatase (ALP), total bilirubin (TBil), lactate dehydrogenase (LDH), and D-dimer were higher ( < 0.05). In subgroup analysis, the level of sCD25 in tumor group was higher than that in infection group. The level of sCD25/ferritin in tumor group was higher than that in infection group and MAS group. Compared with HLH patients in the tumor group, the procalcitonin (PCT) level, proportion of patients using vasopressors, positive rate of hemophagocytosis, and incidence rate of DIC were all higher in the infection group, and the differences were statistically significant ( < 0.05). The results of multivariate analysis showed that fever, splenomegaly, hemophagocytosis, cytopenias, IgM, M.sCD25 [multiple of sCD25 detection value relative to the diagnostic threshold (2400 U/ml)], fibrinogen, and triglycerides were independent predictive factors for HLH ( < 0.05).The diagnostic prediction model H constructed based on temperature, splenomegaly, hemophagocytosis, cytopenias, IgM, M.sCD25, fibrinogen, triglycerides showed good predictive accuracy. The optimal cutoff value of H was 39.45, the sensitivity of the model was 94.12%, the specificity was 83.02%.

Conclusion: sCD25, sCD25/FER, PCT, hemophagocytosis, hemodynamic instability and DIC could help to distinguish the underlying etiology of HLH. The prediction model H has high discrimination and calibration, which could be used as a relatively accurate clinical diagnostic tool for HLH.