Estudio del valor pronóstico del microambiente inmune tumoral en el carcinoma seroso de alto grado de ovario mediante análisis digital de imagen

  1. Machuca Aguado, Jesús
Supervised by:
  1. Miguel-Ángel Idoate Director
  2. Juan José Ríos Martín Director

Defence university: Universidad de Sevilla

Fecha de defensa: 04 March 2024

Type: Thesis

Sustainable development goals

Abstract

Introduction: High-grade serous ovarian carcinoma is a neoplasm with a bleak prognosis. Despite advancements in therapies, altering the disease trajectory has proven elusive. In this context, immunotherapy emerges as a prospective alternative. Investigation into the tumor immune microenvironment, particularly the characterization and quantification of tumor-infiltrating lymphocytes (TILs), assumes critical significance. Two types of TILs are discerned based on their location within the neoplasm—stromal and intraepithelial. The latter is defined as TILs strongly associated with tumor cords. To precisely quantify TILs, novel methodologies such as digital image analysis are imperative, enabling the derivation of algorithms useful in estimating the prognostic role of TILs, with the potential to identify patients whose tumor microenvironment may be more conducive to immunotherapy. Objectives: Given the aforementioned context, our aim was to quantify the tumor immune microenvironment in a substantial cohort of patients diagnosed with high-grade serous ovarian carcinoma and establish its correlation with the BRCA gene mutation and pertinent oncological parameters. These parameters include the impact of neoadjuvancy, degree of tumor regression, overall survival, and platinum-free interval. To achieve this, original algorithms were devised to discern the prognostic value of TILs based on their location and quantity, with particular emphasis on CD8+ effector T lymphocytes. Materials and Methods: We conducted an analytical observational study involving 76 patients diagnosed with high-grade serous ovarian carcinoma between 2013 and 2022, with updated clinical follow-up. Techniques for quantifying intraepithelial and stromal TILs, encompassing tissue segmentation and cell classification through algorithms based on automatic image application (machine learning), were employed. This comprehensive evaluation was complemented by an immunohistochemical study focusing on CD8+ T lymphocytes. Subsequently, a meticulous statistical analysis was undertaken to correlate digital analysis results with clinically and molecularly relevant parameters. Results: An intriguing morphological phenomenon was observed—colonization of tumor cords by intraepithelial lymphocytes (ieTILs), representing the apex of the tumor-immune response interaction, correlated with the response to neoadjuvancy. In the multivariate analysis, the quantity of ieTILs emerged as an independent prognostic factor for overall survival and platinum-free interval in patients not subjected to neoadjuvancy, particularly significant in the context of complete surgical resection. Stromal TILs (sTILs) did not manifest such prognostic value. Furthermore, the quantity of CD8+ ieTILs demonstrated a prognostic correlation for both overall survival and platinum-free interval. Discussion: The methodology of digital analysis based on machine learning has proven its utility in the precise quantification of TILs. A substantial correlation was observed between neoadjuvancy and the colonization of tumor cords by ieTILs. Tumors rich in ieTILs exhibited superior clinical behavior compared to neoplasms with a sparse number of them or sTILs. Consequently, this parameter emerges as an independent prognostic factor with at least comparable value to the degree of surgical resection. Conclusions: The developed digital analysis methodology serves as an incredibly useful tool for the precise quantification of TILs. Thus, it has been established that the quantity of ieTILs in high-grade serous ovarian carcinoma is an independent prognostic parameter with a clinical impact similar to that of complete surgical resection, a phenomenon not observed in sTILs. This finding could have significant implications in the selection of candidates for immunotherapy.