噬血细胞综合征的临床特征及诊断预测模型的建立

Establishment of Diagnostic Prediction Models for Hemophagocytic Lymphohistiocytosis and Its Clinical Features

摘要: 目的：探讨sCD25等免疫炎症指标在噬血细胞综合征(HLH)发生发展过程中的变化及其临床意义，量化HLH患者各项实验室指标的诊断价值，从而构建HLH诊断预测模型。方法：回顾性分析HLH患者101例，非HLH患者203例可溶性炎症因子及其他免疫炎症等实验室指标。Logistic回归分析外周血细胞数量，sCD25等免疫炎症指标及凝血功能等指标探讨预测HLH疾病的影响因素。基于Logistic回归分析建立HLH的列线图预测模型，探究sCD25等免疫炎症指标的早期诊断预测价值。结果：与非HLH组相比，HLH组患者的白细胞计数、红细胞计数、血小板计数、血红蛋白水平、中性粒细胞计数、单核细胞计数和淋巴细胞计数显著降低；与非HLH组相比，HLH组炎症指标sCD25水平、丙氨酸转氨酶(ALT)、天冬氨酸转氨酶(AST)和乳酸脱氢酶(LDH)显著升高；与非HLH组相比，HLH组凝血功能指标活化部分凝血活酶时间(APTT)显著升高。多因素分析结果提示，sCD25、血小板、血红蛋白、活化部分凝血活酶时间(APTT)、为HLH的独立预测因素(P < 0.05)。并且基于二元Logistic构建的列线图预测模型显示出良好的预测效能，模型具有较好的判别能力(C指数为0.81)，校准曲线也显示列线图预测效能与实际拟合较好。决策曲线分析与临床影响曲线进一步证实，该模型在HLH的诊断评估中具有重要的临床应用价值。结论：基于sCD25、血小板、血红蛋白及活化部分凝血活酶时间(APTT)的实验室指标构建的列线图模型具有良好的区分度和校准度，在HLH诊断中展现出良好的预测价值。

Abstract: Objective: To investigate the changes and clinical significance of immune-inflammatory indicators such as sCD25 during the occurrence and progression of hemophagocytic syndrome (HLH), quantify the diagnostic value of various laboratory indicators in HLH patients, and thereby establish a diagnostic prediction model for HLH. Methods: A retrospective analysis was performed on soluble inflammatory cytokines and other immune-inflammatory laboratory indicators in 101 HLH patients and 203 non-HLH patients. Logistic regression analysis was used to explore the influence factors for predicting HLH, including peripheral blood cell counts, immune-inflammatory indicators such as sCD25, and coagulation function indicators. A Nomogram prediction model for HLH was established based on Logistic regression analysis to explore the early diagnostic and predictive value of immune-inflammatory indicators such as sCD25. Results: Compared with the non-HLH group, the HLH group showed significantly lower levels of white blood cell count, red blood cell count, platelet count, hemoglobin, neutrophil count, monocyte count, and lymphocyte count. In contrast, the HLH group exhibited significantly higher levels of the inflammatory indicators sCD25, alanine transaminase (ALT), aspartate transaminase (AST), and lactate dehydrogenase (LDH) compared with the non-HLH group. Additionally, the HLH group had significantly elevated coagulation function indicators, such as activated partial thromboplastin time (APTT), compared with the non-HLH group. Multivariate analysis revealed that sCD25, platelets, hemoglobin, and APTT were independent predictors of HLH (P < 0.05). The Nomogram prediction model was constructed based on binary Logistic regression and demonstrated good predictive performance. The model exhibits strong discriminative ability (C-index = 0.81), and the calibration curve indicates a close agreement between the nomogram’s predictions and actual outcomes. The decision curve analysis and clinical impact curve further confirm the significant clinical application value of this model in the diagnosis of HLH. Conclusion: The Nomogram model constructed based on laboratory indicators, including sCD25, platelets count, hemoglobin, and APTT, has good discrimination and calibration, and exhibits favorable predictive value for the diagnosis of HLH.