Mass Spectrometric Imaging for the Molecular Analysis of FFPE Tumor Sections

Introduction: Mass spectrometric imaging is an upcoming technique for the investigation of tissues. It is a label-free technique that can directly measure molecular distributions in their histological context. For each pixel, a complete mass spectrometric profile is measured. These molecular phenotypes can be used for spatial segmentation or for classification of tissue samples. This offers the possibility to assess phenotypic heterogeneity in areas that appear histologically homogenous. Recent improvements in sample preparation and instrumentation have led to increased data quality. Unlike other multivariate molecular techniques, MALDI imaging maintains spatial integrity and allows a detailed comparison with histology. Here, we show a detailed analysis of 2 sections, a pancreatic tumor and a lung tumor, with a focus on spatial segmentation based on mass spectrometric molecular phenotypes. Methods: Formalin-fixed paraffin-embedded (FFPE) pancreatic tumor tissue was cut at 5 μm and subjected to deparaffinization and heat-induced epitope retrieval. Trypsin solution (0.1 μg/μl) was sprayed on the tissue slice using an automatic spray instrument. After 1.5 h digestion, alpha-cyano-4-hydroxycinnamic acid (CHCA) matrix was deposited using the same spray device and the tissue was analyzed for peptides with a matrix assisted laser desorption/ionization-time of flight (MALDI-TOF) mass spectrometer (rapifleX MALDI Tissuetyper, Bruker). Ion images and digital slides of the H&E stains were visualized using flexImaging software at 50 μm spatial resolution. Mass spectra were imported into SCiLS Lab software (SCiLS GmbH) for segmentation feature extraction and statistical analysis. Results: The objective of this study was to show detailed different histological structures by looking at the untargeted protein expression using a pancreatic cancer and a lung cancer section as examples. The entire pancreatic cancer section was subjected to hierarchical cluster analysis that allowed statistical grouping of similar spectra. Pixels belonging to a particular cluster were then assigned to a selected color and displayed as a spatial segmentation map. The clusters correlated with 8 different histological areas present in the tissue. In addition, for each cluster individual molecule species were found to be highly correlated to a distinct histopathological entity. Similar results were obtained for the lung section. Conclusions: The presented results show that it is feasible to obtain high-quality images and molecular data using mass spectrometry. Using this data, it is then possible to classify tissue sections according to their molecular profiles.