Mass spectrometry imaging (MSI) on tissue microarrays (TMAs) enables the detection of large number of biomolecules from several samples at the same time, making it an attractive tool for biomarker discovery.
In this study, we investigated ovarian cancer subtyping of over 370 samples using MALDI-MSI data from specifically prepared TMA sections. Ovarian carcinomas are largely of high-grade serous histology, which is associated with poor prognosis. Surgery and chemotherapy are the mainstay of treatment, and accurate molecular characterization is necessary to lead the way to targeted therapeutic options. To this end, various methods for gene expression-based subtyping of high-grade serous ovarian carcinoma have been proposed, but they overlap and provide poor robustness. We exploit the glycan heterogeneity observed between similar but distinct cell-types in the classification of ovarian carcinomas.
MALDI-MSI measurements were performed on TMAs and then filtered based on glycan spectral intensity, percentage of tumor cells and histotype classification, after which the resulting data was further preprocessed. Two univariate feature selection methods are applied to reduce the dimensionality of the MALDI-MSI data. The selected features are then used in combination with three classifiers. The best classification scores are obtained with a decision tree classifier, which classified about 84% of samples correctly. Almost all the predictive power originated from a few glycan peaks. The sensitivity of our classification approach, which can be generically used to identify biomarkers was further investigated and revealed to be of higher sensitivity then current approaches. The simultaneous characterization of morphological and glycomic features in the same tissue section adds unique information for histopathological diagnostics, which relies at present on conventional hematoxylin and eosin staining, histochemical, IHC and lengthy molecular methods. The increased understanding of the molecular abnormalities associated with each subtype will lead to further exploration and introduction of more subtype-specific treatment of ovarian carcinoma.