Estudio de la presencia diferencial de exones en el ADN circulante (biopsia líquida) de pacientes con cáncer colorrectal como método de detección y estadificación

Abstract

Colorectal cancer (CRC) is one of the most lethal cancers in developed countries and is expected to increase due to population ageing and lifestyle changes, causing a major economic impact on healthcare systems. Improvements in disease diagnosis and more accurate prognoses of patient outcome are required. Traditionally, cancer analysis is performed based on tissue samples taken from the tumor, a highly costly and invasive procedure known as biopsy. In this context, and as an alternative to tissue biopsies, the concept of liquid biopsy emerged. A novel approach to tumor characterization that offers advantages such as non-invasive sampling and the capture of tumor heterogeneity. To date, most liquid biopsy studies have focused on sequencing clinically relevant cancer genes from panels. The current work provides a clinical evaluation of whole exome sequencing (WES) in circulating free DNA (cfDNA) present in plasma, with the aim of identifying differential clinical profiles in patients with localized and metastatic colorectal cancer (CRC). For this purpose, we applied the concept of "differential presence of exons" (DPE) and a new concept called "differential presence of genes" DPG. We determined differences in the levels of 1.760 exons and 1.730 genes from plasma cfDNA and used DPE and DPG analyses to group and classify patients with disseminated and localized disease. In turn, we identified distinctive clinical profiles between CRC patients and healthy controls. We identified a set of 510 exons whose differential presence in plasma allowed us to group and classify between the three cohorts. The resulting bioinformatics analysis allowed us to design a predictive, prognostic and diagnostic algorithm for DPE. Random forest classification (machine learning) was performed and an estimated out-of-bag (OOB) error rate of 28,85% was obtained and the predictive model had an accuracy of 70% with a confidence interval (CI) of 53,9–82,8. These results suggest that nucleic acids could play a role in the process of malignant transformation and in identifying a signature that could be associated with the occurrence of cancer. The DPE study is shown as a new technique for the study of plasma cfDNA by WES in CRC patients