Serum metabonomics of acute leukemia using nuclear magnetic resonance spectroscopy

Abstract

Acute leukemia is a critical neoplasm of white blood cells. In order to differentiate between the metabolic alterations associated with two subtypes of acute leukemia, acute lymphoblastic leukemia (ALL) and acute myeloid leukemia (AML), we investigated the serum of ALL and AML patients and compared with two controls (healthy and aplastic anemia) using (1)H NMR (nuclear magnetic resonance) spectroscopy. Thirty-seven putative metabolites were identified using Carr-Purcell-Meiboom-Gill (CPMG) sequence. The use of PLS-DA and OPLS-DA models gave results with 84.38% and 90.63% classification rate, respectively. The metabolites responsible for classification are mainly lipids, lactate and glucose. Compared with controls, ALL and AML patients showed serum metabonomic differences involving aberrant metabolism pathways including glycolysis, TCA cycle, lipoprotein changes, choline and fatty acid metabolisms.

Figure 1. Stacked view of ¹H NMR spectra of blood serum from ALL (green), AML (blue), APA (red) and healthy control (yellow).

(A) standard 1D, (B) T₂-edited (CPMG), (C) diffusion edited. The low field region (δ 6–9) is vertically projected 10 times relative to the rest of the spectrum.

Figure 2. Assignments of the ¹H-NMR signals of a representative 500 MHz 1D-CPMG ¹H-NMR average spectrum of a healthy serum sample measured at 310 K.

A, full spectrum (δ 5.50–0.5 ppm) and magnification of aromatic region (δ 9.00–6.50 ppm). Peak assignments: 0, unidentified; 1, cholesterol; 2, lipids (–CH₃) (mainly LDL/VLDL); 3, leucine; 4, valine; 5, 3-hydroxybutyrate; 6, lipids (CH₂)n (mainlyLDL/VLDL); 7, lactate; 8, alanine; 9, adipicacid; 10, arginine; 11, lysine; 12, acetate; 13, lipids (CH₂–C=C); 14, acetyl signals from glycoproteins; 15, glutamine; 16, lipids (CH₂–CO); 17, citrate; 18, lipids (CH=CH–CH₂–CH=CH–); 19, Albumin lysyl; 20, creatine; 21, choline; 22, Trimethylamine N-oxide; 23, Proline; 24, glucose; 25, glycerol; 26, Myo-inositol; 27, creatinine; 28, threonine; 29, β-glucose; 30, glycerol of lipids; 31, α-glucose; 32, lipids (–CH=CH–); 33, tyrosine; 34, phenylalanine; 35, histidine; 36, 1-methylhistidine; 37, formate.

Figure 3

Scores scatter plots (A) PCA, (B) PLS-DA and (C) OPLS-DA of ¹H CPMG NMR spectra of serum from ALL (green), AML (blue), APA (red) and healthy control (yellow).

Figure 4. Permutation plots for the OPLS-DA model showing R2 (green) and Q2 (blue) values.

Figure 5. OPLS-DA loadings plot colored as a function of VIP.

Assignment of main signals is indicated (unassigned signals with high VIP are marked with an asterisk).

Figure 6. Average changes relative to healthy control of main metabolites contributing to the discrimination between serum of cancer patients and of healthy subjects.