Marker-independent imaging reveals a correlation of fibrotic and epigenetic alterations in endometriosis

Abstract

IntroductionEndometriosis describes the presence of endometrial glands outside of the uterus and can cause various symptoms such as chronic pelvic pain, hypermenorrhea and infertility. These complications pose an extreme burden on the patients, especially as up to date, the average time until diagnosis can consume several years and requires invasive laparoscopy. ObjectivesThe aim of this study is to molecularly characterize endometrium and endometriosis using marker-independent Raman microspectroscopy to identify potential biomarkers and validate its diagnostic potential. MethodsAfter histopathological characterization of tissue sections of human endometrium and endometriosis, Raman microspectroscopy was performed on the gland region. Multivariate analysis of the hyperspectral maps was used to localize major subcellular structures and further decipher their molecular composition. Samples from different anatomical regions and throughout all menstrual cycle phases were analyzed. ResultsRaman imaging enabled label-free visualization of tissue morphology and submolecular tissue characterization. Distinct differences between endometrium and endometriosis were found for collagen type I and nuclear signatures. Spectral deconvolution allowed identification of a Raman biomarker indicative of fibrotic changes in endometriosis samples. Additionally, a significant increase in epigenetic 5mC foci and an increased signal intensity relevant for methylations was detected in nuclei of endometriosis. Furthermore, a neural network-based classification of Raman data resulted in high accuracies in discriminating endometrial and peritoneal tissue from endometriosis. ConclusionThe non-destructive approach by hyperspectral Raman imaging enabled for molecular sensitive characterization of endometriotic lesions which could not only be an asset in complementing histopathological tissue evaluation but combined with data-driven classification models support in situ tissue diagnosis.