Identification of Metabolic Biomarkers Associated with the Efficacy of Intravenous Immunoglobulin in the Treatment of Immune Thrombocytopenia Based on Machine Learning

Immune thrombocytopenia (ITP) is an acquired autoimmune bleeding disorder characterized by isolated thrombocytopenia, resulting from the breakdown of immune tolerance. Intravenous immunoglobulin (IVIG) is one of the first-line treatment strategies for ITP patients. Given that no studies have investigated the differences in response to IVIG treatment in ITP patients from a plasma metabolomics perspective, this prospective study aims to evaluate the efficacy of IVIG 5% in patients with persistent and chronic ITP from the National Longitudinal Cohort of Hematological Diseases. It also compares the plasma metabolic profiles of patients in the responsive and unresponsive IVIG treatment groups and uses machine learning algorithms to identify metabolic biomarkers that can effectively distinguish the efficacy of IVIG treatment. A total of 37 ITP patients (75.7% female, median age 36 years) were enrolled. Intent-to-treat analysis indicated that 32 out of 37 patients (86.5%) achieved a response, with a median time to response of 3 days and a median response duration of 9.5 days. From the plasma samples of 37 ITP patients, we quantified a total of 2,701 metabolites, including 1,280 metabolites in the positive mode and 1,421 metabolites in the negative mode. Metabolic patterns varied significantly between groups of responsive and unresponsive, with 35 metabolites upregulated and 173 metabolites downregulated in the IVIG-unresponsive group. 9-[(3S,4S,5R)-3,4-dihydroxy-5-(hydroxymethyl)oxolan-2-yl]-1H-purin-6-one was significantly upregulated in the IVIG-unresponsive group, while 3,4-Dihydroxymandelic acid, 2-Thiouracil, Gulose, Carnitine C6:1, and others were significantly downregulated. Enrichment factors and P-values revealed the top metabolically enriched pathways to be purine metabolism, arginine biosynthesis, the mTOR signaling pathway, ether lipid metabolism. In summary, this study revealed the metabolic pathways in plasma associated with differences in IVIG efficacy and identified potential metabolic biomarkers distinguishing IVIG response.

No relevant conflicts of interest to declare.