Browsing by Author "Bravo, Maria Loreto"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Machine learning analysis of a Chilean breast cancer registry
    (2021) Acevedo, Francisco; Causa, Leonardo; Bravo, Sebastián; García, Pablo; Cuevas, Ricardo; Bravo, Maria Loreto; Avellaira, Carla; Muñiz, Sabrina; Petric, Militza; Martinez, Raúl; Guerra, Constanza; Navarro, Marisel; Taramasco, Carla; Sanchez, Cesar
    In recent years, artificial intelligence (AI) and machine learning (a form of AI) have offered valuable tools for medicine by applying and training algorithms in order to make predictions. Herein, we applied a machine learning algorithm to analyze data from a >20 year breast cancer (BC) registry elaborated in two Chilean health institutions (a public hospital and a private center) that includes a total of 4838 patients and their basic clinicalpathological characteristics. Preliminary results suggest that this cohort of patients can be subdivided into five clusters according to key variables that also correlate with overall survival and disease-free survival rates. To our knowledge this is the first Latin American report of its kind. Our laboratory is currently expanding these analyses.
  • No Thumbnail Available
    Item
    Validation of an NGS Panel Designed for Detection of Actionable Mutations in Tumors Common in Latin America
    (2021) Salvo, Mauricio; Gonzalez-Feliu, Evelin; Toro, Jessica; Gallegos, Ivan; Maureira, Ignacio; Miranda-Gonzalez, Nicolas; Barajas, Olga; Bustamante, Eva; Ahumada, Monica; Colombo, Alicia; Armisen, Ricardo; Villaman, Camilo; Ibanez, Carolina; Bravo, Maria Loreto; Sanhueza, Veronica; Spencer, M. Loreto; de Toro, Gonzalo; Morales, Erik; Bizama, Carolina; Garcia, Patricia; Carrasco, Ana Maria; Gutierrez, Lorena; Bermejo, Justo Lorenzo; Verdugo, Ricardo A.; Marcelain, Katherine
    Next-generation sequencing (NGS) is progressively being used in clinical practice. However, several barriers preclude using this technology for precision oncology in most Latin American countries. To overcome some of these barriers, we have designed a 25-gene panel that contains predictive biomarkers for most current and near-future available therapies in Chile and Latin America. Library preparation was optimized to account for low DNA integrity observed in formalin-fixed paraffin-embedded tissue. The workflow includes an automated bioinformatic pipeline that accounts for the underrepresentation of Latin Americans in genome databases. The panel detected small insertions, deletions, and single nucleotide variants down to allelic frequencies of 0.05 with high sensitivity, specificity, and reproducibility. The workflow was validated in 272 clinical samples from several solid tumor types, including gallbladder (GBC). More than 50 biomarkers were detected in these samples, mainly in BRCA1/2, KRAS, and PIK3CA genes. In GBC, biomarkers for PARP, EGFR, PIK3CA, mTOR, and Hedgehog signaling inhibitors were found. Thus, this small NGS panel is an accurate and sensitive method that may constitute a more cost-efficient alternative to multiple non-NGS assays and costly, large NGS panels. This kind of streamlined assay with automated bioinformatics analysis may facilitate the implementation of precision medicine in Latin America.