Cerebrospinal fluid proteome during chemotherapy for childhood leukemia: Identifying pathways associated with treatment and system toxicity.

Abstract

10030 Background: Acute Lymphoblastic Leukemia (ALL) is the most common childhood cancer with a 5-year survival rate > 90%. Survivors treated on contemporary chemotherapy-only protocols are at heightened risk for musculoskeletal, endocrine, cardiac, and neurological/neurocognitive late-effects. Chemotherapy treatment doses and administration routes (i.e., intrathecal) are associated with the late-effects and alterations in targeted cerebrospinal fluid (CSF) proteins. We aimed to identify CSF proteome pathways linked to patient variables, treatment exposures, and early adverse events (AE). Methods: At diagnosis and the end of induction, CSF samples were collected from 178 ALL patients (71 females, mean age [range] 7.6 [0.5-18.8] years at diagnosis) treated on a chemotherapy-only protocol. Expression of 3188 proteins was measured via tandem-mass-tag mass spectrometry and clustered via Weighted Gene Co-expression Network Analysis (WGCNA). Severe/life threatening (CTCAE grade 3/4) AEs across multiple organ systems were compiled per patient. Trait-Cluster association was assessed by generalized linear models, linking cluster-specific eigenvalues to final treatment risk stratum (Standard/High vs Low) and AE occurrence, with false discovery rate corrected significance threshold of p < 0.2. Protein-Protein Interaction (PPI) network and enrichment analysis were performed by STRING within target clusters. Results: Patients experienced AEs in 24 organ systems (86% post-induction) and > 5% of patients experienced toxicities in 6 systems during and 11 after induction. A total of 1770 proteins were measured at both time points and WGCNA revealed 8 clusters at diagnosis (T1; 1046 proteins) and 13 after induction (T2; 1184 proteins). Four T1 clusters were associated with post-induction Hepatobiliary/Pancreas AEs (p < 0.022). Five T2 clusters were associated with treatment risk (p < 0.055) and eight were associated with AEs: Hepatobiliary/Pancreas (p < 0.184), Musculoskeletal/Soft Tissue (p < 0.151), or Neurology (p < 0.196). In four T2 clusters associated with both risk and AEs, PPI analysis revealed 251 pathways involved in nervous system development, skeletal/cardiac development, and immune regulation. Conclusions: The associations of Musculoskeletal/Soft Tissue, Neurology, and Hepatobiliary/Pancreas AEs with risk-associated clusters suggest pathological protein dynamics exacerbated by treatment intensity. These changes are reflected in different T1 and T2 cluster composition and the emergence of new clusters after induction. Associations between protein clusters, treatment risk, and AEs biologically connect treatment exposures to toxicities, providing mechanistic targets to reduce late-effects. Future work includes isolating enriched pathways and hub proteins to gain insight into specific protein dynamics contributing to AEs.