MSACL 2018 US Abstract

Topic: Tissue Imaging & Analysis

Metabolic Phenotyping of Fresh Frozen and Formalin-Fixed, Paraffin-Embedded Colorectal Tissue Samples Using DESI-MSI

Anna Mroz (Presenter)
Imperial College London

Bio: I finished an MSc in Molecular Biology and Pathology of Viruses run by Imperial College London in 2011 and since then was employed by the College. Initially, I was involved in histopathology work during my time in the Centre for Pathology when I was working under the supervision of Professor of Liver and GI Pathology. Over four years ago, I started working with Professor Zoltan Takats’ group. First of all, I was mainly providing histology support to various projects run by his group but progressively became more and more involved in mass spectrometry itself. At the moment, I am a third year PhD student working with DESI-MSI and my projects are focused on assessing changes in lipids pathways in colorectal cancer as well as in various cirrhotic liver diseases. I am also working on histology-friendly DESI-MSI which is a great opportunity to combine my histology expertise with MS imaging.

Authorship: Anna Mroz1, Renata Soares1, James McKenzie1, James Alexander1, Liam Poynter1, Robert Goldin1, Zoltan Takats1
(1) Imperial College London, UK

Short Abstract

Colorectal cancer is one of the leading causes of cancer-related deaths and it is thought to be the third most commonly diagnosed type of cancer. In routine histopathology laboratories majority of samples are fixed in formalin and embedded in wax to allow further diagnosis. In this study we have assess the feasibility of DESI-MSI to distinguish between different tissue types in both fresh and paraffin-embedded colorectal samples. When normal versus tumour fresh samples were analysed, an accuracy of 87% and 91% was achieved, respectively. In case of the FFPE samples, metabolic information remaining in the tissue sections after standard histological processing was enough to discriminate between different tissue types in multiple samples (overall accuracy over 95%). This study has proven that DESI-MSI can be successfully employed not only for analysing fresh frozen but also FFPE samples.

Long Abstract

Introduction

Colorectal cancer (CRC) is one of the leading causes of cancer-related deaths worldwide. The molecular mechanisms are still not fully understood despite the fact that significant efforts have been undertaken to understand the pathogenesis of CRC. Its aetiology is multifactorial and has been linked to diet, genetic mutations, inflammatory processes, and more recently, the gut microbiota. It was established that more than 95% of colorectal cancers are sporadic and the individuals affected are not genetically predisposed to the disease. When a patient is diagnosed with cancer, biopsies of both solid tissue and the surrounding regions are taken to help doctors to create a “profile” of how the cancer looks and behaves. The outcome of this process acts as one of the key prognostic factors for majority of cancers. The biopsy report provides information on the tumour type, grade and any morphological abnormalities present. Manual histological evaluation of these post-operative tissue sections is the gold standard method that has been used over decades to enable clinicians to understand and characterise tumours. Whilst this approach is relatively simple and powerful, it has some significant drawbacks. Firstly, manual examinations are very time consuming and hence limit the speed and throughput of this procedure. Secondly, it is non-quantitative and relies on the subjective interpretation of both histopathological and clinical information which can be incomplete or misleading. It is therefore clear that alternative methods are required for accurate and timely diagnosis and characterisation of tumour grade and stage.

However, since formalin-fixed, paraffin-embedded (FFPE) clinical tissue samples are the standard for histopathological analysis, much effort has recently been spent on optimising and standardizing FFPE protocols. Paraffin blocks can easily be stored for many years and since these are clinical samples, corresponding patient data is also available. Implementing desorption electrospray ionisation mass spectrometry imaging (DESI-MSI) alongside histopathological analysis would help to expand the knowledge of the genetic and molecular content of analysed tissue samples, therefore it is clear that optimising the DESI-MSI protocol for FFPE tissue samples is well worth investigating.

Lipids, which are fundamental components of biological membranes, are functionally and structurally diverse class of metabolites. They play important roles in biological systems including composing membrane bilayers, signal transduction, storing energy, providing functional implementations of membrane proteins as well as their interactions. Therefore, it is crucial to understand the role played by lipids in both normal and diseased tissues as they may have prognostic value.

Mass spectrometry imaging (MSI) transformed the field of metabolomics, proteomics, and genomics over the past decade as it was demonstrated that it is a great scientific tool which is able to unambiguously detect multiple analytes during a single analysis from complex biological samples. MSI is a novel method which has proven to be an invaluable tool for localising peptides, metabolites, drugs, and proteins in biological tissues without any prior labelling. Mass spectrometry imaging, particularly DESI-MSI, has been proven to be a powerful technique which can create an image from individual mass spectra collected for tissue samples. This image is then used to show the distribution of particular compounds over the sample area and can significantly contribute to diagnosis and treatment of various diseases. This study aimed to use DESI-MSI in order to distinguish between different tissue classes present in both fresh frozen and formalin-fixed, paraffin-embedded colorectal tissue samples based on their metabolomics fingerprint.

Methods

In order to perform a DESI-MSI analysis, fresh frozen tissue samples obtained from patients undergoing surgery due to colorectal cancer diagnosis were cryosectioned at 10µm and mounted on glass slides which were then stored at -80°C prior to the measurements. Next, the sample glass slide was placed onto the DESI source and the area to be imaged was defined. This imaging analysis area was rastered line-by-line using the DESI sprayer with a mass spectrum collected at predefined x and y coordinates using the resolution of 100 µm.

33 normal, 10 adenoma and 20 colorectal tumour samples were analysed in negative ionization mode by DESI-MSI. Since this study relied on analysing the biochemical content of the tissue samples, mainly lipids, the m/z values of particular interest were those in the 600 – 1000 m/z range. Formalin-fixed, paraffin-embedded colorectal samples were also analysed by the same technique. Both single samples as well as tissue microarrays (TMA) were used. TMA contained many small representative tissue samples from dozens of different cases assembled on a single glass slide, and therefore allowed high throughput analysis of multiple specimens at the same time. The protocol for this part included duplicate washing of the tissue sections in Xylene for 8 minutes.

Following DESI measurements, tissue sections were stained with haematoxylin and eosin (H&E), digitalised and annotated by a histopathologist in order to allow the alignment of optical and MSI images. Therefore, efficient mapping of molecular ion patterns on the tissue regions of morphological interest was possible.

Results

Tissue samples from patients undergoing surgery were analysed, these included tumour tissue and corresponding normal tissue from the same patient. DESI-MSI profiling of the sample cohort was performed using high resolution DESI-MSI. This analysis was performed in an m/z range between 150 and 1500 Da in negative ion mode, which allowed collecting data for lipids. Raw DESI-MSI data were converted to imzML format and imported into MATLAB for pre-processing and analysis. The advantage of DESI-MSI is that the analysis is based on the biochemical content of the tissue samples, in this particular study lipids. Collected data has proven that in all analysed normal tissue samples the discrimination between mucosa and submucosa and mucosa and muscle was possible based on their lipid composition. Principal component analysis (PCA) of the individual regions of interest showed that mucosa and submucosa as well as mucosa and muscle clutters are distinct from each other. One component was enough to separate these two pairs of tissue classes in unsupervised PCA analysis. When all three tissue classes were taken into consideration, PCA showed a good separation of mucosa from submucosa and muscle. Supervised analysis using maximum margin criteria (MMC) was followed by leave-one-region-out cross validation with an accuracy of over 95% for mucosa tissue. When normal versus tumour samples were analysed, MMC revealed that discrimination of these two tissue types was possible with a leave-one-out cross validation accuracy of 87% for normal samples and 91% for tumour. The analysis of the acquired DESI data has also revealed that it is possible to use this technique to differentiate tissue classes within a single FFPE human colorectal tissue sample. Three tissue classes were taken into consideration and PCA showed a good separation of mucosa from muscle and tumour. Moreover, when the TMA data was analysed, it was confirmed that metabolic information remaining in the tissue cores after standard histological processing taking place prior embedding was enough to discriminate between different tissue types in multiple samples (overall accuracy over 95%). Some m/z peaks were assigned to particular tissue types and their identification was confirmed using an LC-MS approach.

Conclusions & Discussion

This study has proven that DESI-MSI shows the potential to differentiate between different tissue types in both fresh frozen and formalin-fixed, paraffin-embedded colorectal tissue samples.


References & Acknowledgements:


Financial Disclosure

DescriptionY/NSource
Grantsno
Salaryno
Board Memberno
Stockno
Expensesno

IP Royalty: no

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

no