Biochemical histology analysis of tissue samples by Desorption Electrospray Ionization (DESI) mass spectrometry imaging

Abstract

For over 100 years, the histopathological analysis of cytology, biopsy or resection specimens has been the final step in the process of diagnosing multiple diseases, including cancer. In recent years, standard clinical care is continuously becoming more complex, and as a result, diagnostic pathology workup is also more complex and extensive. Moreover, despite being considered a gold standard in making a diagnosis, histopathological investigations can be timeconsuming. Additionally, an examination of the stained slides is subject to intra-observer error. Therefore, it is evident that some additional techniques are required to complement making a diagnosis. Desorption electrospray ionisation mass spectrometric imaging (DESI-MSI) is an emerging mass spectrometry technique with great potential in tissue analysis, especially in histological settings. DESI-MSI enables visualising the spatial distribution of lipid species across tissue sections allowing a direct correlation of the metabolomic information with the morphological features. However, this technique has always relied on frozen sections, which are not required in routine histopathology settings very often. Moreover, some embedding media, e.g. OCT, a common choice in diagnostic laboratories, have been proven not to be very well suited for MSI. The main aim of this study was to make DESI-MSI more compatible with the standard pathology procedures. Therefore, the first step was to assess OCT's impact on the quality of DESI-MSI data. The acquired data suggested that this embedding medium could be used for histopathological and mass spectrometric analyses. There were no clear polymeric signals causing differences in the negative mode data, but some reduction in intensities might be attributable to polymer-induced ion suppression. In positive mode data, the interferences due to OCT were more overt but could be negated by removing the regular peaks of the various polymeric distributions. As formalin-fixed, paraffin-embedded (FFPE) samples are the gold standard in histopathology laboratories worldwide, the next step was to optimise the pre-DESI-MSI protocol to allow the analysis of specimens that have been processed that way. A new protocol has been adapted and successfully tested on FFPE mouse and human tissue samples for tissue classification. Additionally, DESI-MSI has been used to analyse fresh-frozen and FFPE colorectal samples. 88.5% accuracy for normal samples and 91.7% for tumours was achieved when a batch of 38 fresh-frozen samples was analysed. Tissue microarray (TMA) consisting of 54 cores was used further to test the application of DESI-MSI to FFPE samples. A 10μm thick sections were subjected to analysis in negative and positive modes, and accuracy of over 80% and 92% for tissue prediction was achieved, respectively. Equally good results were obtained for TMA sections which were 5μm thick. This last observation was crucial in the light of making DESIMSI as histology-friendly as possible, as 10μm tissue sections are not routinely prepared in histopathology laboratories. Lastly, a new statistical approach based on ion colocalisation features has been applied to DESI-MSI data acquired for cirrhotic liver diseases. It allowed to identify top correlations of ions, and their distribution within analysed tissue sections was visualised. It is possible that using this approach, some biochemical interactions that are distinguishing the three classes of cirrhotic liver diseases (metabolic, hepatitis and cholangiopathy) could be captured. The colocalisation patterns can potentially be used for data-driven hypothesis generation, suggesting possible local molecular mechanisms characterising the samples of interest.