Using Global Metabolomics to Identify Novel Biomarkers of Treatment-Associated Cognitive Impairment in Pediatric Acute Lymphoblastic Leukemia

Central nervous system-directed chemotherapy is a critical component of pediatric acute lymphoblastic leukemia (ALL) treatment, but is associated with long-term cognitive impairment. Because few strategies exist to identify children who are at risk for this adverse outcome, we employed global metabolomics to identify biomarkers of cognitive impairment using routinely collected, biologically relevant cerebrospinal fluid (CSF) samples from patients undergoing ALL chemotherapy. Methods: CSF samples were collected at 4 months post-induction on pediatric ALL patients enrolled in a multicenter prospective study of symptom toxicity. Metabolomics of CSF detected 314 metabolites by gas chromatography-mass spectrometry (MS) and liquid chromatography-MS/MS. Cognitive performance at 12 months post-induction was assessed with the validated Parent-Perceived Child Cognitive Function scale and categorized using published cut-points: mild (≥50), moderate (41–49), and severe (≤40). Ordinal logistic regression evaluated associations between normalized median-scaled metabolite values and cognitive function, adjusting for patient and treatment factors. A false discovery rate-corrected p-value (q) was calculated to account for multiple comparisons. Results: Among the 96 patients (diagnosed 2012–14), 43% exhibited moderate (n = 25) or severe (n = 16) cognitive impairment. Significant alterations were observed in several biomarkers, including methionine sulfoxide (Log2 Fold Change [FC] = 0.83, P = 5.2e−7, q = 0.001) and citramalate (FC = −0.29, P = 4.6e−5, q = 0.008). A pathway analysis of the top biomarkers (q < 0.3) revealed an enrichment of metabolites involved in methionine (P = 0.0015) and carnitine (P = 0.0052) metabolism. Finally, compared to clinical factors alone, CSF biomarkers significantly improved the ability to predict cognitively impaired patients (area under curve [AUC] = 0.68 vs. AUC = 0.91, P < 0.001). Conclusion: We identified several novel CSF biomarkers of treatment-related cognitive impairment among pediatric ALL patients. These findings may translate to clinical improvements in the management of cognitive outcomes by introducing the opportunity to deliver targeted interventions to high-risk patients prior to detectable or irreversible cognitive impairment.