MSACL 2017 EU Abstract

Diagnosis of Multiple Myeloma by Mass Spectrometry of Peripheral Blood Plasma and Artificial Intelligence

Meritxell Deulofeu (Presenter)
Masaryk University

Bio: I am a PhD student in Molecular Biology, Biomedicine and Health. I am graduated in Biomedical Science by the University of Barcelona and I have a Master in Molecular Biology and Biomedicine by the University of Girona.

Authorship: Meritxell Deulofeu (1, 2), Lenka Kolářová (2), Victoria Salvadó (3), Eladia M. Peña-Méndez (4), Pere Boadas-Vaello (5), Luděk Pour (6), Sabina Ševčíková (7), Martina Almasi (8), Aleš Hampl (1, 9), Petr Vaňhara (1, 9), Josef Havel (2)
(1)Department of Histology and Embryology, Faculty of Medicine, Masaryk University, Brno, Czech Republic (2)Department of Chemistry, Faculty of Science, Masaryk University, Brno, Czech Republic (3)Department of Chemistry, Faculty of Science, University of Girona, Girona, Spain (4)Department of Chemistry, Analytical Chemistry Unit, Faculty of Chemistry, University of La Laguna, San Cristóbal de La Laguna, Tenerife, Spain (5)Department of Medical Sciences, University of Girona, Girona, Spain (6)Department of Internal Medicine - Hematooncology, University Hospital Brno, Brno, Czech Republic (7)Babak Myeloma Group, Department of pathological physiology, Faculty of Medicine, Masaryk University, Brno, Czech Republic (8)Department of Clinical Hematology, University Hospital Brno, Brno, Czech Republic (9)International Clinical Research Center, St. Anne’s University Hospital, Brno, Czech Republic

Short Abstract

A fast and simple method for the diagnosis of multiple myeloma by the analysis of peripheral blood plasma mass spectra has been developed. It is based on recording the Matrix Assisted Laser Desorption Ionisation Time Of Flight (MALDI TOF) mass spectra of low mass metabolites/compounds (below 2000 Daltons) and the evaluation of these data using Artificial Neural Networks (ANNs). The method, which does not require the identification of biomarkers, has been verified using clinical database of myeloma positive and negative patients.

Long Abstract


Multiple Myeloma (MM), characterized by clonal expansion of plasma cells within the bone marrow resulting in lytic bone lesions and accumulation of monoclonal proteins, is the most severe and frequent disease of the monoclonal gammopathies. Clinical manifestations of MM are frequently presented only at advanced stage making the early diagnosis a difficult task. Current diagnosis and monitoring of MM is based on analysis of bone marrow aspirates that is, in principle, invasive (1). Thus, low-invasive, new diagnostic tests for early detection and monitoring of MM are needed. Mass Spectrometry (MS) techniques are widely used in both the proteome and the metabolome, making them extremely useful also for the identification of new biomarkers. The main issues of these techniques are the complexity of the data and necessity of biomarkers identification while there is potentially a bias of random correlations. In deciphering the complexity of the data, mathematical methods based on artificial intelligence such as Artificial Neural Networks (ANNs) have previously been shown highly effective, cf. review (2). In this study, (i) we analysed statistically MALDI TOF MS mass spectrometric data concerning analysis of peripheral blood plasma - from low mass range (below 2000 Dalton) and we have proved that they contain information able to detect MM. (ii) We have examined the possibility of data explotation using Artificial Neural Networks for MM diagnosis without any identification of MM biomarkers.


Peripheral blood plasma samples were obtained from 20 positively diagnosed patients and 20 healthy donors upon informed consent. The sample preparation method consisted of a ten times dilution of plasma samples with double distilled water, followed by the addition of sinapinic acid (SA) in 1:1 volume ratio. One microliter of this mixture was then spotted on a stainless steel target plate. Mass spectra for three replicates were acquired in linear positive ion mode using Matrix Assisted Laser Desorption/Ionisation Time Of Flight (MALDI TOF) mass spectrometer (Axima CFR with TOF analyser from Kratos Analytical Ltd) and evaluated with Launchpad software. The data were statistically evaluated applying eigenvalues analysis, principal component and cluster analysis, for example. Finally, ANN-based data evaluation was used to perform computer-aided diagnosis.


Using statistical analysis (as mentioned above) we have identified significant differences between mass spectra obtained from MM patients and healthy donors. Then, a dataset containing the most relevant peaks, selected by independent variable selection approaches, was created. Using ANNs analysis we were able to classify correctly all the samples in the process of “training”. Verification of 20 unknown samples (patients) as done using cross validation approach was yielding 100% success. Even when we have arbitrarily used half of the data for the verification, classification was also completely correct.

Conclusions & Discussion

New fast, simple and low invasive method for discrimination of MM from healthy donors by artificial intelligence-driven analysis of mass spectra concerning peripheral blood plasma has been developed. The approach can be applied for diagnosis, even at earlier stage of the illness. The analysis of low mass metabolites/compounds using MALDI TOF MS in combination with ANNs enables reliable diagnosis without the need to identify the markers (3).

References & Acknowledgements:


(1) Raab MS, Podar K, Breitkreutz I, Richardson PG, Anderson KC. (2009). Multiple myeloma. Lancet 374: 324-339.

(2) Amato F, Lopez A, Pena-Mendez EM, Vanhara P, Hampl A, Havel J (2013) Artificial neural networks in medical diagnosis. Journal of Applied Biomedicine 11: 47–58.

(3) Deulofeu M, Kolářová L, Salvadó V, Peña-Méndez, EM, Boadas-Vaello P, Pour L, Ševčíková S, Almasi M, Hampl A, Vaňhara P, Havel J. Diagnosis of multiple myeloma by mass spectrometry of peripheral blood plasma and artificial intelligence, CTT Technology transfer office, Masaryk University, Brno. Patent pending.


This work was supported by Masaryk University (MUNI/A/1369/2016), Czech health research council (17-29343A). M. Deulofeu would like to thank to Erasmus+ programme, which enables her research stay at Masaryk University.

Financial Disclosure

Board Memberno

IP Royalty: no

Planning to mention or discuss specific products or technology of the company(ies) listed above: