Development of Laboratory Parameters-Based Formulas in Predicting Short Outcomes for Adult Hemophagocytic Lymphohistiocytosis Patients with Different Underlying Diseases.

Abstract

Hemophagocytic lymphohistiocytosis (HLH) is a severe disease with high mortality. The purpose of this investigation was to build models to predict 30-day death in total and subgroup HLH patients based on available and cheap laboratory parameters. The research contained 431 adults HLH patients from January 2015 to September 2021 in the hospital. Logistic regression and receiver operating characteristic (ROC) were utilized to build models. Results suggested that age, ferritin, lymphocyte (LY), international normalized ratio (INR), thrombin time (TT), globulin, uric acid (UA), chloride, activated partial thromboplastin time (APTT), aspartate aminotransferase (AST), triglycerides (TG), total bilirubin (TB), and indirect bilirubin (IB) were independent factors in HLH and subgroups. Then, models adapted to patients with different underlying diseases were established based on these factors. Area under curve (AUC) of these models was excellent: HLH patients: 0.838 (p < 0.001); infection-associated HLH (I-HLH) patients: 0.913 (p < 0.001); malignancy-associated HLH (M-HLH): 0.921 (p < 0.001) and 0.809 (p < 0.001) for two or more different etiologies-associated HLH (Mix-HLH patients). In addition, UA, TT, and chloride were firstly confirmed as independent factors in adult HLH. Four models depending on biomarkers that available and affordable in clinical practice were built. With these models, high-risk patients with different underlying diseases could be easily identified.