Bayesian estimation of RGP90 ELISA parameters for diagnosis of equine infectious anemia

• Autor(a) principal: Rejane Silva Diniz
• Data de Publicação: 2020
• Outros Autores: Jenner Karlisson Pimenta dos Reis, João Paulo Amaral Haddad
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Institucional da UFMG
• Texto Completo: http://hdl.handle.net/1843/51700; https://orcid.org/0000-0003-0395-358X; https://orcid.org/0000-0003-2823-6288
• topic: EIA; Sensitivity; Specificity; Bayesian model; ROC curve; Virologia veterinária; Veterinária - Diagnóstico; Cavalo; Anemia infecciosa equina; Testes sorológicos; Curva ROC; Ensaio de imunoadsorção enzimática

Objective: Equine Infectious Anemia (EIA) is caused by a retrovirus. The infected animal is the main source of the virus, and a laboratory diagnostic test is essential for the identification of infected horses when EIA cannot be definitively diagnosed clinically. EIA can be diagnosed based on serology, and serology methods often have limitations due to the uncertainty of sensitivity and specificity estimates. Our aim was to investigate the accuracy of these serological tests with a Bayesian model, as a gold standard for the identification of EIAV does not exist. Methods: Validation studies for serological tests for EIA diagnosis are necessary. Using ROC curve analysis, we examined three possible cut-off values, 0.220, 0.228 and 0.232, for the rgp90 ELISA. In this study, we performed a Bayesian analysis of diagnostic data from an enzyme-linked immunosorbent assay (ELISA) of recombinant envelope glycoprotein gp90 and the classical agar gel immunodiffusion (AGID) test. For each scenario cut-off, we estimated the sensitivity and specificity of each test separately and of the two tests in combination. Results: The upper limits of the posterior equally tailed 95% credible intervals for the Sensitivities (Se) and Specificities (Sp) of these two tests were as follows: AGID test alone, Se 85% and Sp 99%; ELISA alone, Se 99% and Sp 97%; and for the tests in combination, AGID test, Se 99% and Sp 100%; and ELISA, Se 99% and Sp 97%. Conclusion: In this study, the Bayesian method was found to be a valuable tool for estimating the sensitivities and specificities of ELISA and AGID tests. In addition, the combination of those two tests was found to have better diagnostic accuracy than either test alone.