Distinct Bone Marrow Metabolic Profiles at Relapse Compared with Initial Diagnosis in Pediatric Acute Lymphoblastic Leukemia—An LC-MS Metabolomics Study

Abstract

Background/Objectives: Relapse remains the leading cause of mortality in pediatric acute lymphoblastic leukemia (ALL), underscoring the need for novel biomarkers and therapies. The objective of this study is to identify distinct metabolomes associated with relapsed ALL, using bone marrow aspirated at the relapse episode itself, a specimen that is rarely banked and whose metabolome has not previously been reported. Method: This study utilized untargeted liquid chromatography–mass spectrometry (LC-MS) metabolomics on bone marrow aspirates from 21 children with B-lineage ALL. Aliquots from the 12 non-relapsed patients were collected at initial diagnosis and those from the 9 relapsed patients at relapse, so group and disease phase are collinear by design. Results: The resulting metabolic profile clearly distinguished bone marrow sampled at relapse from bone marrow sampled at initial diagnosis. Partial least-squares discriminant analysis separated the two groups, completely in positive ion mode and with 18 of 21 samples correctly classified in negative ion mode (1000-permutation test, p = 0.001 in both modes), and 31 of 92 metabolites in positive ion mode and 3 of 44 in negative ion mode differed after Benjamini–Hochberg correction (q < 0.05), of which 26 and 3 respectively also exceeded the fold-change threshold of |log2FC| > 1. The relapse-associated differences involved carnitine, amino acid and lipid metabolites. O-acetylcarnitine was elevated, which is compatible with increased fatty acid oxidation; dipeptides were higher but not significantly (q = 0.075); L-allo-threonine was elevated, which is compatible with increased flux through the serine–glycine one-carbon pathway; adenosine monophosphate (AMP) was decreased, although the abundance of AMP alone does not provide the AMP: adenosine triphosphate (ATP) ratio and cannot establish the activity of AMP-activated protein kinase (AMPK) or of mammalian target of rapamycin complex 1 (mTORC1); and individual lipid species differed in both directions. These are hypotheses generated by the observed differences in metabolite abundance and are not mechanisms demonstrated by this study. Conclusions: This exploratory comparison describes the metabolite composition of bone marrow at relapse and identifies fatty acid oxidation, the serine–glycine one-carbon pathway and ether-lipid metabolism as priorities for future investigation rather than as established therapeutic dependencies. Because the relapsed specimens were obtained at relapse, the signature describes relapsed disease and is not evidence that relapse can be predicted at diagnosis.