Predição de micotoxinas via espectroscopia no infravermelho próximo (NIRs) para gerenciamento micotoxicológico em milho

• Ano de defesa: 2020
• Autor(a) principal: Vidal, Juliano Kobs
• Orientador(a): Mallmann, Carlos Augusto
• Banca de defesa: Pötter, Luciana, Meinerz, Gilmar Roberto
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Santa Maria; Centro de Ciências Rurais
• Programa de Pós-Graduação: Programa de Pós-Graduação em Medicina Veterinária
• Departamento: Medicina Veterinária
• País: Brasil
• Palavras-chave em Português: Aflatoxinas; Zearalenona; Deoxinivalenol; Quimiometria
• Palavras-chave em Inglês: Z-score; Zea mays; Aflatoxins; Zearalenone; Deoxynivalenol; Chemometrics
• Área do conhecimento CNPq: CNPQ::CIENCIAS AGRARIAS::MEDICINA VETERINARIA
• Link de acesso: http://repositorio.ufsm.br/handle/1/21267

The present study was aimed at evaluating the performance of Near Infrared Spectroscopy (NIRs) in the prediction of mycotoxins in silo-stored lots of maize. We analyzed 240 samples from 4 silos, which were collected with the aid of a pneumatic probe using 2 sampling processes: A and B. In process A, three collective samples were taken (upper, middle and lower third of the silo depth). In process B, only one sample composed of grains from the whole depth of the silo was obtained. Five points were collected from each silo: surface center and center of each surface quadrant (north, south, east and west). Analyses of Aflatoxin B1 (AFB1), Zearalenone (ZON) and Deoxynivalenol (DON) were performed by high performance liquid chromatography coupled to mass spectrometry (LC-MS/MS) using an Infinity 1200 Series HPLC (Agilent, Palo Alto, USA), coupled to a 5500 QTRAP mass spectrometer (Applied Biosystems, Foster City, CA, USA). Spectra were obtained via a NIRs equipment, model XDS (Foss, Hilleroed, Copenhagen, DK). The spectrum of each sample was sent to Pegasus Science Olimpo platform to obtain the results of mycotoxicological prediction. The value of the samples analyzed for DON was lower than the NIRs quantification limit, which is 350 μg.kg-1. Acceptable contamination ranges were determined for each mycotoxin, and the result via NIRs was considered correct when it was within those ranges compared to the LC-MS/MS result. The accepted variability, upwards or downwards (±), was ±10 μg.kg-1 for AFB1 and ±100 μg.kg- 1 for ZON. In addition, the sampling processes for each sample collection point in the silo were compared: the mean of the prediction of three samples via NIRs (plan A) was compared with the result of one analysis via LC-MS/MS (plan B). The analysis of one sample via LC-MS/MS versus the prediction of one sample via NIRs showed 91, 95 and 100% accuracy for AFB1, ZON and DON, respectively. When comparing the mean of the prediction of three samples via NIRs with the analysis of one sample via LC-MS/MS, there was 100% accuracy for AFB1, ZON and DON. The Z-Score of the results via NIRs was calculated for the evaluation, taking the LC-MS/MS results as standard. Data were classified as satisfactory, questionable and unsatisfactory, being satisfactory in 81%, 90% and 100% of the samples for AFB1, ZON and DON, respectively. The average concentration of each silo for the analyses through LC-MS/MS and prediction via NIRs were, respectively: silo 1= AFB1: 0.6 and 2.2 μg.kg-1; and ZON: 13 and 26 μg.kg-1; silo 2= AFB1: 0.5 and 2.7 μg.kg-1 and ZON: 18 and 18 μg.kg-1; silo 3= AFB1: 5.3 and 6.1 μg.kg-1 and ZON: 38 and 57 μg.kg-1; and silo 4= AFB1: 2.1 and 4 μg.kg-1 and ZON: 46 and 39 μg.kg-1. It may be concluded that the NIRs methodology can be used as a practical, accurate, fast and non-destructive mycotoxicological monitoring tool for lots of maize stored in silos.