Metabolomics Profiling Reveals Markers for Chemosensitivity and Clinical Outcomes in Pediatric AML Patients

Abstract Acute myeloid leukemia (AML) is a clinically challenging disease with high interpatient variability in response to chemotherapy. Despite continuing advances in treatment options, current 5-year survival rates for pediatric AML are suboptimal at ~60%. The heterogeneous nature of AML contributes significantly to the variability in treatment response and survival outcomes. Several known genetic lesions and cytogenetic features contribute to disease progression. However, our understanding of how molecular mechanisms contribute to variation in treatment outcomes is still limited. Previous metabolomics studies have successfully identified significant metabolic alterations in hematological malignancies, but very few metabolomics studies have been conducted for the pediatric AML patient population. In this study, we used global and targeted metabolomics to identify differential metabolite abundance associated with chemosensitivity and treatment outcomes in pediatric AML patients. Serum metabolomics profiles were generated with serum samples obtained at diagnosis from patients treated in the multicenter AML02 study (n=94, NCT00136084). Clinical outcomes tested for association included half-maximal inhibitory concentration (IC50) of cytarabine, minimal residual disease (MRD), relapse free survival (RFS), and overall survival (OS). Global metabolomics profiling was performed using liquid chromatography/mass spectrometry (LC/MS). Targeted metabolomics profiling was generated for a select group of organic acids and acylcarnitines. The organic acid panel included eight metabolites related to the tricarboxylic acid cycle and glycolysis. The acylcarnitine panel featured 20 varieties of acylcarnitines detectable in human serum. Statistical analyses were performed using MetaboAnalyst and various R packages. A total of 3205 features were detected in the global metabolome, with 124 known metabolites and 3081 unknown features. All metabolites were used for association analysis, while annotated metabolites were used in pathway analyses. Association analysis of clinical endpoints vs. metabolome identified 10 known metabolites significantly associated with IC50 values, 17 associated with MRD, 7 associated with RFS, and 7 associated with OS (p<0.05). Targeted metabolomics generated the absolute abundance profile of 8 organic acid metabolites and 20 acylcarnitine metabolites in patient samples. Spearman correlation analysis identified five acylcarnitines significantly correlated with IC50 values. Among the significant metabolites, the most interesting is pantothenic acid, showing higher serum abundance associated with poorer IC50, MRD, and RFS outcomes. Pantothenic acid is an essential component for Coenzyme A synthesis, leading into energy production through the tricarboxylic acid cycle. A previous study has shown a reduced capacity for pantothenic acid uptake in leukemia cells resistant to daunorubicin. Our results suggest a similar relationship for pantothenic acid uptake and cytarabine resistance. Pathway enrichment analysis identified 11 metabolic pathways showing significant association with IC50 values and 12 pathways associated with MRD (FDR<0.05). Some of the most significantly associated pathways included alanine, aspartate and glutamate metabolism, arginine and proline metabolism, and pantothenic acid based CoA biosynthesis. Overall, differences in chemosensitivity and clinical outcomes appear to be most closely related to amino acid synthesis and energy production. This study identifies several metabolites and metabolic pathways significantly associated with chemosensitivity and clinical endpoints in pediatric AML patients. These results help expand on previously conducted AML pilot studies, and metabolomics studies on other cancer types, to further clarify the metabolic differences associated with interpatient variability in chemotherapy response for AML patients. Continued metabolic profiling of AML patient populations can help establish targetable pathways that can be used to improve treatment efficiency for AML. In addition, in vitro functional modeling to validate results of the metabolomics study are currently underway. Disclosures No relevant conflicts of interest to declare.