ETA2024 Oral Presentations Oral Session 6: Translational thyroid cancer research (7 abstracts)
Performance of the mir-thype new version algorithm in classification of thyroid indeterminate nodules: latin american multicenter validation study
Santos Marcos Tadeu 1,2 , Oliveira Andrei Felix 1 , Vilela Diego Nogueira 1 , Rodrigues Bruna Moretto 1 , Oliveira Miriane 1 , Silva Eduardo 2 , Oliveira Cristiano 3 , Dufloth Rozany Mucha 4 , Soares Fernando Augusto 4 , Schneider Leandra 5 , Netto Mário Rodrigues Montemór 5 , Kozonoe Milena Massumi 6 , Ioshii Sérgio Ossamu 6 , Gama Ricardo Ribeiro 2 , Viana Acklei 7 , Ortellado Daniel Knabben 7 , Santos Gustavo Philippi de Los 7 , Aust Jalmir Rogério 7 , Melo Giulianno Molina 8 , Silva Rubens Adão 9 , Figueiredo David Livingstone Alves 10 , Soto Miluska Huachín 11 , Saieg Mauro Tadeu Ajaj 12 , Callegari Fabiano Mesquita 13 , Maciel Léa Maria Zanini 14 , Mazeto Gláucia Maria F. da Silva 3 , Brenta Gabriela 15 & Kimura Edna Teruko 16
1 ONKOS Molecular Diagnostics; 2 Barretos Cancer Hospital; 3 Botucatu Medical School, São Paulo State University (Unesp); 4 Rede D’Or Hospitals Network; 5 State University of Ponta Grossa (UEPG); 6 Erasto Gaertner Hospital; 7 Integrated Center for Head and Neck Surgery (NICAP); 8 The Portuguese Beneficence of São Paulo (BP); 9 Surgical Cancerology, Complexo ISPON; 10 Midwestern State University (UNICENTRO); 11 Peruvian University Cayetano Heredia; 12 Santa Casa de São Paulo School of Medical Sciences; 13 São Paulo Federal University (Unifesp/EPM); 14 Ribeirão Preto Medical School, University of São Paulo (USP/RP); 15 Dr. Cesar Milstein Hospital; 16 Institute of Biomedical Sciences, University of São Paulo (USP)
Background: Thyroid nodules are present in nearly 60% of the population, representing the most frequent endocrine disease. Fine-needle aspiration (FNA) cytopathology classifies 20%–30% of nodules as indeterminate (Bethesda 3 and 4), a scenario in which molecular tests are recommended aiming to avoid unnecessary diagnostic surgeries.
Objective: The aim was to evaluate the diagnostic performance of the new v2 algorithm of mir-THYpe full molecular classifier test for preoperative diagnosis of cytologically indeterminate thyroid nodules, optimized by the use of new machine learning techniques, larger sample cohort size, multicentricity and association of DNA-mutation analysis.
Methods: A multicenter validation study was conducted on a set of 2.372 thyroid nodules with Bethesda 3 and 4 cytology from 15 academic, community and private centers in Brazil, Argentina and Peru. Eligibility criteria were met in 510 nodules. The FNA smear slides were used to obtain and analyze microRNA expression and DNA mutations (BRAF V600E and pTERT C228/250T) by qPCR. Molecular data from 306 (150 benign/156 cancer+NIFTP) nodules were used to retrain and optimize the mir-THYpe v2 algorithm, using random forest, SVM and neural networks machine learning techniques. For final validation, molecular data from 204 (151 benign/53 cancer+NIFTP) thyroid nodules were used to measure diagnostic performance. Anatomopathological data were used as gold-standard for blinded comparison.
Results: In the validation set, 61.8% of the samples were assigned as Bethesda 3 (126) and 38.2% as Bethesda 4 (78). The v2 algorithm had a specificity of 94% (95% Cl, 84-99), a sensitivity of 89% (95% Cl, 83-94) and an accuracy of 91% (95% Cl,86-94). At 26% cancer prevalence, the negative predictive value was 98% (95% Cl, 94-99) and the positive predictive value was 76% (95% Cl, 66-83) with a benign call rate of 68% (138/204). The v2 algorithm was able not only to classify but also to identify as medullary thyroid carcinoma (MTC) in all the MTC samples (5/5).
Conclusion: The optimized classifier demonstrated a high diagnostic performance for identifying benign nodules, which may potentially obviate diagnostic surgery in 68% of patients with indeterminate nodules, and up to 89% of all benign nodules cytologically indeterminate. The BRAF and pTERT status analysis, added to the ability to rule-in cancer samples may help to guide prognostic decisions, including surgery extension and individualized treatments.
Volume 101
Article tools
My recent searches
My recently viewed abstracts
Authors
Endocrine Abstracts
ISSN 1470-3947 (print) | ISSN 1479-6848 (online)
© Bioscientifica 2024 |
Privacy policy |
Cookie settings
BiosciAbstracts
Bioscientifica Abstracts is the gateway to a series of products that provide a permanent, citable record of abstracts for biomedical and life science conferences.
Find out more