Unveiling Cerebrospinal Fluid Protein Biomarkers in Pediatric Acute Lymphoblastic Leukemia Using a Proximity Extension Assay

Acute lymphoblastic leukemia (ALL) is a heterogeneous pediatric malignancy where relapse and central nervous system (CNS) involvement remain critical challenges. We combined high-throughput proteomics and machine learning (ML) to identify cerebrospinal fluid (CSF) biomarkers for treatment response and CNS status. Proteomic profiling of 276 proteins was performed on 82 CSF samples from 41 pediatric patients with ALL at diagnosis (VD) and in remission (VR) using Olink’s proximity extension assay. Differentially abundant proteins (q-value < 0.05, |log2FC| > 0.5) were identified via the Wilcoxon test, and Random Forest, LASSO, and SVM-RFE algorithms were integrated to select key predictive proteins for remission status and CNS involvement. Validation was performed using Pan-Cancer Atlas data sets. Remission was characterized by changes in ADGRG1, KYNU, CCL17, CD5, CD27, CXCL9, CXCL11, FASLG, GZMA, and TNFRSF9 between the two visits. Furthermore, distinct proteins associated with CNS involvement were shown, with CXCL9 and MMP7 being differentially abundant between CNS involvement groups at both visits. Integrating Olink proteomics with ML identified molecular signatures in ALL that have the potential to be considered as biomarkers for monitoring treatment response and guiding personalized therapeutic strategies, shifting the focus toward Precision One Health approaches.