Strategies for managing multi-patient 3D mass spectrometry imaging data

D. R. N. Vos, I. Jansen, M. Lucas, M. R. L. Paine, O. J. de Boer, S. L. Meijer, C. D. Savci-Heijink, H. A. Marquering, D. M. de Bruin, R. M. A. Heeren, S. R. Ellis, B. Balluff

Research output: Contribution to journalArticleAcademicpeer-review

17 Citations (Scopus)


Mass spectrometry imaging (MSI) has emerged as a powerful tool in biomedical research to reveal the localization of a broad scale of compounds ranging from metabolites to proteins in diseased tissues, such as malignant tumors. MSI is most commonly used for the two-dimensional imaging of tissues from multiple patients or for the three-dimensional (3D) imaging of tissue from a single patient. These applications are potentially introducing a sampling bias on a sample or patient level, respectively. The aim of this study is therefore to investigate the consequences of sampling bias on sample representativeness and on the precision of biomarker discovery for histological grading of human bladder cancers by MSI. We therefore submitted formalin-fixed paraffin-embedded tissues from 14 bladder cancer patients with varying histological grades to 3D analysis by matrix-assisted laser desorption/ionization (MALDI) MSI. We found that, after removing 20% of the data based on novel outlier detection routines for 3D-MSI data based on the evaluation of digestion efficacy and z-directed regression, on average 33% of a sample has to be measured in order to obtain sufficient coverage of the existing biological variance within a tissue sample. Significance: In this study, 3D MALDI-MSI is applied for the first time on a cohort of bladder cancer patients using formalin-fixed paraffin-embedded (FFPE) tissue of bladder cancer resections. This work portrays the reproducibility that can be achieved when employing an optimized sample preparation and subsequent data evaluation approach. Our data shows the influence of sampling bias on the variability of the results, especially for a small patient cohort. Furthermore, the presented data analysis workflow can be used by others as a 3D FFPE data-analysis pipeline working on multi-patient 3D-MSI studies.
Original languageEnglish
Pages (from-to)184-191
JournalJournal of Proteomics
Early online date2018
Publication statusPublished - 20 Feb 2019

Cite this