MSACL 2019 EU Abstract

Cross-modality Single-pixel Correlation of Multimodal Imaging Mass Spectrometry of Prostate Cancer

Ambra Dreos (1), Patrick M. Wehrli (1), Wojciech Michno (1), and Jörg Hanrieder (1,2)
(1) Department of Psychiatry and Neurochemistry, the Sahlgrenska Academy at the University of Gothenburg, S-431 80 Mölndal, Sweden; (2) Department of Neurodegenerative Disease, UCL Queen Square Institute of Neurology, University College London, London WC1N 3BG, United Kingdom

Warning: Undefined variable $headshot in /var/www/html/view_abstract/view_abstract_in_program.php on line 704

Ambra Dreos (Presenter) Sahlgrenka Academy at the University of Gothenburg >> POSTER (PDF)

Presenter Bio: Ambra Dreos has a background within organic and physical organic chemistry, and recently started a PostDoc doing MALDI imaging at the University of Gothenburg, in the group of Ass. Prof. Jörg Hanrieder. She earned her PhD at Chalmers University of Technology (Sweden) in 2019, with a thesis exploring new organic materials for solar energy storage applications. She also has a MSc in Organic Chemistry and Biomolecular Chemistry from the University of Trieste (Italy), and a BSc in Chemistry from the same university. Having recently started doing research within analytical neurochemistry this conference is a great opportunity to get introduced to the clinical application of mass spectrometry.

Relevant Financial Disclosures (within past 24 months) No relevant financial relationship(s) to disclose. Conflict mitigation NOT REQUIRED.

Introduction
Using MALDI imaging mass spectrometry (IMS), our group have been investigating cancer associated distribution patterns of lipids and proteins in mice prostate cancer tissues. Prostate cancer is one of the most common cancer diagnosed, and it is one of the main causes of cancer-related death in men in the United States and Europe. While surgery, radiation and other therapies have increased survival rates, the 5 years survival rate is still about 30% for patients with the metastatic disease.
Here were present novel strategies for objective evaluation of data processing methods for multivariate analysis of MALDI IMS datasets of mice metastatic prostate cancer tissues. The strategy employs region-based active contours segmentation based on the PCA scores matrix to provide class information for Orthogonal Projections to Latent Structures discriminant analysis (OPLS-DA) modelling. Processing methods can then be compared by the quality metrics of the respective multivariate models. This workflow provides an unbiased strategy for sensitive annotation of anatomical regions of interest in prostate cancer tissues and quantitative comparison of processing procedures for multivariate analysis.

Methods
Fresh frozen tissue of rat prostate affected by highly metastatic Dunning R3327-MLL cancer cells was cut into 12 μm thick sections in a cryostat microtome and collected onto a glass slide. The 1,5-diaminonaphthalene matrix was applied using a dedicated HTX TM matrix sprayer. IMS lipid data acquisition in negative and positive ion mode was performed.

Results
Multivariate image analysis of the MALDI IMS date using principal component analysis provided structural features. This in turn allowed precise segmentation of anatomical features on the PCA data scale using a region-based active contour segmentation algorithm. This class information was supplied to OPLS-DA modelling, which provided quality metrics i.e. predictive ability of OPLS-DA models to quantitatively compare the applied processing methods. Logarithmic transformation reduced the heteroscedastic noise structure of the data resulting in improved multivariate models and the visualization of detailed anatomical features.

Discussion and Conclusions
We developed a multivariate strategy for robust regions of interest (ROI) feature detection, image segmentation, and classification that enabled quantitative comparison of ROI-associated biochemical localisation patterns in prostate cancer tissue. Quantitative comparison of the applied methods allow us to demonstrate how data processing methods impact feature detection and annotation.