Comparison of biological materials to assess their suitability for Raman spectroscopic detection of chronic myeloid leukemia

Abstract

Chronic myeloid leukemia (CML) is a haematological malignancy that necessitates continuous monitoring for disease progression and resistance to tyrosine kinase inhibitors (TKIs). Current clinical assays are mechanism-specific, costly, and often inadequate for detecting resistance in the terminal phase, where underlying mechanisms remain poorly characterized. To address this limitation, we evaluated Raman spectroscopy (RS) as a mechanism-independent diagnostic approach capable of capturing comprehensive macromolecular profiles. Raman spectra were acquired from blood-derived cells, plasma, and serum samples of 32 CML patients and 12 healthy controls. Principal component analysis combined with linear discriminant analysis (PC-LDA) and multivariate curve resolution (MCR) revealed distinct biochemical differences between CML and control samples, including alterations in protein and fatty acid signatures. Classification accuracies achieved via PC-LDA were 70.73% (cells), 86.36% (plasma), and 88.64% (serum), with data fusion across sample types enhancing accuracy to 95.12–100%. Notably, MCR identified discriminative spectral components in plasma (component 6) and serum (component 5). These findings demonstrate the capacity of RS to differentiate CML from healthy states across multiple biological samples and identify serum as the most suitable material for future RS-based clinical diagnostics. This study provides a methodological foundation for future work toward developing mechanism-independent RS platforms for clinical monitoring in CML.