Microscopia multifóton aplicada para a caracterização de câncer de próstata

Abstract

The interest on the use of non-linear microscopy techniques has increased recentely, mainly for the analysis and imaging of biological tissues. Among these techniques, second harmonic generation microscopy (SHG) and two-photon absorption photoluminescence (PL) have been the most important and most widely used when it comes to the analysis and prognosis of tumors. This increased interest is due to the fact that they are non-invasive technique, do not require the use of dyes or any other type of preparation in biopsies allowing the use in different situations such as in vivo samples, fixed tissues, in vitro 3D systems, among others. Prostate cancer has a slow evolution and the main prognostic markers currently used are the patient’s age, the prostate-specific antigen and the Gleason grading system that quantifies the histological grade. However, these markers cannot accurately predict the biological behavior of the tumor, as the tumor alters the environment around it. A relevant modification takes place in the stroma which is called reactive stroma. Thus, studying the stroma could help in histological grading and in determining more accurate prognoses. In this work, we used SHG and PL microscopy to study prostate tissue biopsies fixed on histological slides. SHG images allow us to analyze collagen fibers, which undergo changes in the presence of a tumor, both in the shape of the fibers and in their organization. And the PL images allow to obtain information about the cellular part of the tissue. We present results for 3 different analysis groups: the Gleason System and two groups that were proposed taking into account the structural alterations that occur in the acini and stroma of the prostate tissue in the presence of a tumor. We present also new methodologies for analysis of SHG and PL images developeing a software to automatically separate images into regions of fibers and cells: PyFibre. The metrics computed by the PyFibre were proposed with the aim of quantifying the changes in cellular and stroma structures observed by pathologists. The set of metrics was able to distinguish between non-neoplastic tissue and carcinoma areas by linear discriminant analysis (LDA) with an accuracy of 89 ± 3%, but between Gleason groups of only 46 ± 6%. The reactive stroma analysis improved the accuracy to 65 ± 5%, clearly demonstrating that stromal parameters should be considered as additional criteria for a more accurate diagnosis.