Objective: To investigate the Raman spectral characteristics of leukemia cells from 4 patients with acute promyelocytic leukemia (M3) and 3 patients with acute monoblastic leukemia (M5), establish a novel Raman label-free method to distinguish 2 kinds of acute myeloid leukemia cells so as to provide basis for clinical research.
Methods: Leukemia cells were collected from bone marrow of above-mentioned patients. Raman spectra were acquired by Horiba Xplora Raman spectrometer and Raman spectra of 30-50 cells from each patient were recorded. The diagnostic model was established according to principle component analysis (PCA), discriminant function analysis (DFA) and cluster analysis, and the spectra of leukemia cells from 7 patients were analyzed and classified. Characteristics of Raman spectra were analyzed combining with ultrastructure of leukemia cells.
Results: There were significant differences between Raman spectra of 2 kinds of leukemia cells. Compared with acute monoblastic leukemia cells, the spectra of acute promyelocytic leukemia cells showed stronger peaks in 622, 643, 757, 852, 1003, 1033, 1117, 1157, 1173, 1208, 1340, 1551, 1581 cm-1. The diagnostic models established by PCA-DFA and cluster analysis could successfully classify these Raman spectra of different samples with a high accuracy of 100% (233/233). The model was evaluated by "Leave-one-out" cross-validation and reached a high accuracy of 97% (226/233).
Conclusion: The level of macromolecules of M3 cells is higher than that of M5. The diagnostic models established by PCA-DFA can classify these Raman spectra of different cells with a high accuracy. Raman spectra shows consistent result with ultrastructure by TEM.