Evaluation of Collagen Alterations in Early Precursor Lesions of High Grade Serous Ovarian Cancer by Second Harmonic Generation Microscopy and Mass Spectrometry

SIMPLE SUMMARY: The collagen architecture in the extracellular matrix (ECM) is highly remodeled in high grade serous ovarian cancer (HGSOC). Many of these tumors begin in the fallopian tubes (FT) before metastasizing to the ovaries and it is important to study ECM alterations in carcinogenesis. Here, we used Second Harmonic Generation (SHG) microscopy to classify changes in the collagen fiber morphology in normal FT, and precursor pure p53 signatures and serous tubal intraepithelial carcinoma (STICs) in tissues with no HGSOC. Using a machine learning approach based on image features, we were able to discriminate the tissue groups with good classification accuracy. We additionally performed mass spectrometry analysis of normal and HGSOC tissues to associate the differential expression of collagen isoforms with fiber morphology alterations. This work provides new insights into ECM remodeling in early stage HGSOC and suggests the combined use of SHG microscopy and mass spectrometry as a new diagnostic/prognostic approach. ABSTRACT: Background: The collagen architecture in high grade serous ovarian cancer (HGSOC) is highly remodeled compared to the normal ovary and the fallopian tubes (FT). We previously used Second Harmonic Generation (SHG) microscopy and machine learning to classify the changes in collagen fiber morphology occurring in serous tubal intraepithelial carcinoma (STIC) lesions that are concurrent with HGSOC. We now extend these studies to examine collagen remodeling in pure p53 signatures, STICs and normal regions in tissues that have no concurrent HGSOC. This is an important distinction as high-grade disease can result in distant collagen changes through a field effect mechanism. Methods: We trained a linear discriminant model based on SHG texture and image features as a classifier to discriminate the tissue groups. We additionally performed mass spectrometry analysis of normal and HGSOC tissues to associate the differential expression of collagen isoforms with collagen fiber morphology alterations. Results: We quantified the differences in the collagen architecture between normal tissue and the precursors with good classification accuracy. Through proteomic analysis, we identified the downregulation of single α-chains including those for Col I and III, where these results are consistent with our previous SHG-based supramolecular analyses. Conclusion: This work provides new insights into ECM remodeling in early ovarian cancer and suggests the combined use of SHG microscopy and mass spectrometry as a new diagnostic/prognostic approach.