Serum spectral fingerprints reveal different forms of advanced multiple myeloma disease

Authors

PEČINKA Lukáš MORÁŇ Lukáš POROKH Volodymyr ADAMOVÁ Sabina GREGOROVÁ Jana POUR Luděk HAVEL Josef ŠEVČÍKOVÁ Sabina VAŇHARA Petr

Year of publication 2021
Type Conference abstract
Citation
Description INTRODUCTION: Multiple myeloma (MM) is a highly heterogeneous disease of malignant plasma cells (PC). Diagnosis of MM is based on bone marrow biopsies and on detection of abnormal immunoglobulin in serum and/or urine and clinical manifestation. MM can progress from bone marrow into devastating extramedullary multiple myeloma (EM). EM is diagnosed as two different forms: primary EM before any treatment while secondary EM usually forms in relapsed patients after treatment. However, the biological background of different forms of MM remains unclear. In our previous study we demonstrated that Matrix-Assisted Laser Desorption/Ionization Time-of Flight Mass Spectrometry (MALDI-TOF MS) of peripheral blood plasma coupled with artificial neural networks (ANNs) clearly discriminated MM patients from healthy donors.1 Here we show that spectral fingerprinting by MALDI-TOF MS can provide important insight into etiopathogenesis of extramedullary MM. OBJECTIVES: The goal of this study is to demonstrate if diverse forms of MM can be discriminated by MALDI-TOF MS of peripheral blood serum coupled with multivariate statistic methods and ANNs. METHODS: Mass spectra were acquired in positive linear mode in mass range 2-20 kDa that corresponds to peptides and small proteins. Mass spectra were pre-processed using R programming language and selected variables (m/z values) were introduced for statistical analysis. Various data mining approaches were examined. Data mining tools including unsupervised methods (e.g. principal component analysis – PCA, hierarchical clustering, and correlation analysis) as well as supervised methods (e.g. decision tree, random forest, and ANN) were used. All computations were done in R programming language. RESULTS: One hundred-fifty patients were included into a bicentric study. Samples were divided into the training data-set and validation data-set. Mass spectra were recorded with four technical replicates averaged to a representative spectrum. Several classification models based on supervised and unsupervised methods were used. While the unsupervised statistical methods have not been found effective enough for discrimination stages of MM forms, the ANNs and random forests can predict the classification and reveal potential biological and clinical links. CONCLUSION: Mass spectrometry coupled with data mining techniques were used for the classification of different forms of MM and revealed relevant clinical links. This work was supported by grants of the Ministry of Health of the Czech Republic, project nr. NV18-08-00299, NV18-03-00203, and NU21-03-00076, and by Masaryk University (project nr. MUNI/A/1390/2020). All rights reserved. LM is supported by Faculty of Medicine, Masaryk University (project nr. ROZV/28/LF/2020) and by Masaryk Memorial Cancer Institute (project nr. 00209805). 1. Deulofeu, M., Kolářová, L., Salvadó, V. et al. Rapid discrimination of multiple myeloma patients by artificial neural networks coupled with mass spectrometry of peripheral blood plasma. Sci Rep 9, 7975 (2019). https://doi.org/10.1038/s41598-019-44215-1.
Related projects:

You are running an old browser version. We recommend updating your browser to its latest version.