Chronic myeloid leukemia patients sensitive and resistant to imatinib treatment show different metabolic responses

Abstract

The BCR-ABL tyrosine kinase inhibitor imatinib is highly effective for chronic myeloid leukemia (CML). However, some patients gradually develop resistance to imatinib, resulting in therapeutic failure. Metabonomic and genomic profiling of patients' responses to drug interventions can provide novel information about the in vivo metabolism of low-molecular-weight compounds and extend our insight into the mechanism of drug resistance. Based on a multi-platform of high-throughput metabonomics, SNP array analysis, karyotype and mutation, the metabolic phenotypes and genomic polymorphisms of CML patients and their diverse responses to imatinib were characterized. The untreated CML patients (UCML) showed different metabolic patterns from those of healthy controls, and the discriminatory metabolites suggested the perturbed metabolism of the urea cycle, tricarboxylic acid cycle, lipid metabolism, and amino acid turnover in UCML. After imatinib treatment, patients sensitive to imatinib (SCML) and patients resistant to imatinib (RCML) had similar metabolic phenotypes to those of healthy controls and UCML, respectively. SCML showed a significant metabolic response to imatinib, with marked restoration of the perturbed metabolism. Most of the metabolites characterizing CML were adjusted to normal levels, including the intermediates of the urea cycle and tricarboxylic acid cycle (TCA). In contrast, neither cytogenetic nor metabonomic analysis indicated any positive response to imatinib in RCML. We report for the first time the associated genetic and metabonomic responses of CML patients to imatinib and show that the perturbed in vivo metabolism of UCML is independent of imatinib treatment in resistant patients. Thus, metabonomics can potentially characterize patients' sensitivity or resistance to drug intervention.

Figure 1. A typical GC/TOFMS chromatogram of blood plasma.

Several hundreds of peaks can be detected at only one analysis and some of the peaks were identified. Part of the chromatogram was zoomed in for clear inspection of the peaks, 15–19, retention time from 335 to 355 s. 1, Lactate; 2, butyamine; 3, 3-hydroxybutyrate; 4, valine; 5, urea; 6, phosphate; 7, serine; 8, threonine; 9, pyroglutamate; 10, hydroxyproline; 11, creatinine; 12, ornithine; 13, phenylalanine; 14, methyl myristate (QC reference standard); 15, glutamine; 16, glycerol-3-phosphate; 17, azelate; 18, citrate; 19, [¹³C₂]-myristic acid (stable isotope internal standard); 20, glucose; 21, palmitic acid; 22, uric acid; 23, linoleic acid; 24, oleic acid; 25, stearic acid; 26, alpha-tocopherol; 27, cholesterol; U1-5, unidentified peaks of retention indices at 1163, 1876, 1986, 2015, 2415, 2448, respectively.

Figure 2. The scores plot of PLS-DA model (3 principal components) of the four groups.

UCML, SCML, RCML and the healthy control (HC). This figure shows the difference between UCML and the healthy control and SCML and UCML, respectively. The overlapping of UCML and RCML suggested similar metabolic phenotype and indicated the therapeutic ineffectiveness of imatinib on RCML. The overlapping of SCML and HC suggested similar metabolic phenotype and the therapeutic effectiveness of imatinib on SCML.