Metabolomic profiling of cerebrospinal fluid reveals an early diagnostic model for central nervous system involvement in acute lymphoblastic leukaemia

Zhiqiang Song,Gusheng Tang,Chunlin Zhuang,Yang Wang,Mian Wang,Diya Lv,Guihua Lu,Jie Meng,Min Xia,Zhenyu Zhu,Yifeng Chai,Jianmin Yang,Yue Liu
DOI: https://doi.org/10.1111/bjh.18307
2022-06-18
British Journal of Haematology
Abstract:Summary The pathogenesis of central nervous system involvement (CNSI) in patients with acute lymphoblastic leukaemia (ALL) remains unclear and a robust biomarker of early diagnosis is missing. An untargeted cerebrospinal fluid (CSF) metabolomics analysis was performed to identify independent risk biomarkers that could diagnose CNSI at the early stage. Thirty‐three significantly altered metabolites between ALL patients with and without CNSI were identified, and a CNSI evaluation score (CES) was constructed to predict the risk of CNSI based on three independent risk factors (8‐hydroxyguanosine, l‐phenylalanine and hypoxanthine). This predictive model could diagnose CNSI with positive prediction values of 95.9% and 85.6% in the training and validation sets respectively. Moreover, CES score increased with the elevated level of central nervous system (CNSI) involvement. In addition, we validated this model by tracking the changes in CES at different stages of CNSI, including before CNSI and during CNSI, and in remission after CNSI. The CES showed good ability to predict the progress of CNSI. Finally, we constructed a nomogram to predict the risk of CNSI in clinical practice, which performed well compared with observed probability. This unique CSF metabolomics study may help us understand the pathogenesis of CNSI, diagnose CNSI at the early stage, and sequentially achieve personalized precision treatment.
hematology
What problem does this paper attempt to address?