Classification of molecular subtypes of high-grade serous ovarian cancer by MALDI-Imaging.

e17544 Background: High-grade serous ovarian cancer (HGSOC) can be separated by gene expression profiling into four molecular subtypes with clear correlation of the clinical outcome. However, these signatures have not been implemented in practice to stratify patients for targeted therapy. This is mainly due a lack easy, cost-effective and reproducible methods, as well high heterogeneity HGSOC. Hence, we aimed examine potential unsupervised matrix assisted laser desorption/ionization imaging mass spectrometry (MALDI-IMS) patients, which might benefit from therapeutic strategies. Methods: Molecular subtyping paraffin-embedded tissue samples 279 HGSOC was performed NanoString analysis (ground truth labeling). Next, applied MALDI-IMS, novel technology identify distinct profiles on same sections paired machine learning algorithms proteomic signature. Finally, devised strategy annotate spectra stromal origin. Results: We elucidated MALDI-derived signature (135 peptides) able classify subtypes. Random forest classifiers achieved an area under curve (AUC) 0.983. Furthermore, demonstrated that exclusion stroma associated provides tangible improvements classification quality (AUC = 0.988). False discovery rates (FDR) were reduced 10.2% 8.0%. MALDI-based annotation near-perfect classifications 0.999, FDR < 1.0%). Conclusions: Here, present concept integrating MALDI-IMS according has great assign therapies.