RESUMO
The high-throughput screening by flow injection coupled to high-resolution mass spectrometry (HTS-FIA-HRMS) is a powerful technique that enables the identification of several types of samples in a short period of time, either with qualitative or quantitative purposes. Sensory attributes of tobacco are affected by its chemical composition, and it is very important to quantify multi-analytes in a high-throughput methodology. HTS-FIA-HRMS coupled to multivariate analysis was used to create calibration models for 27 analytes, or group of compounds, of tobacco sensory interest. The models were validated by different approaches, including permutation test to avoid overfitting, evaluation of the equipment repeatability by control samples, reproducibility comparison of results from two different equipment and analysts, and with a blind test analysis. All tests demonstrated a good response to the proposed method. No statistical difference between the errors of both equipment was observed, with less than 7% error from the control samples, and a blind test error between 5.96% and 20.10%. The partial least squares (O-PLS) regression models were applied to 815 samples, and a principal component analysis (PCA) was performed from the predicted concentration values, aiming at the non-supervised classification based on tobacco type. We expect that this proposed methodology shows not only the applicability in tobacco samples, but also demonstrates a guideline to an efficient performance of multi-analytes target analysis using the flow injection mass spectrometry with reliable and robust validation steps.
Assuntos
Análise de Injeção de Fluxo/métodos , Espectrometria de Massas/métodos , Nicotiana/química , Reprodutibilidade dos Testes , Fatores de TempoRESUMO
Tobacco-specific nitrosamines (TSNAs), mainly the 4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone (NNK), are known carcinogens. Part of the NNK found in smoke is provided from matrix-bound NNK, and its determination is extremely relevant. However, the reference extraction procedure of matrix-bound NNK is time-consuming and labor-intensive and has a limited analytical capacity. Three different methodologies were proposed to predict matrix-bound NNK: simple linear regression (LR) with soluble NNK; multiple linear regression (MLR) considering soluble NNK and characteristic parameters of the samples; and orthogonal partial least-squares (O-PLS) regression using high-throughput screening by flow injection analysis coupled to high-resolution mass spectrometry (HTS-FIA-HRMS) data. Simple linear regression showed a high influence of matrix and leaf origin. Although an existing linearity trend has been observed ( R2 = 0.62) for the global model, higher correlation values were achieved for matrix and country segregation models. Multiple linear regression predicted matrix-bound NNK with more satisfactory efficiency than simple linear regression models. The coefficients of determination were 0.87 and 0.94 for flue-cured Virginia and air-cured Burley, respectively. However, this method has a limited application, since previous information about the sample is required. The proposed method based on HTS-FIA-HRMS and O-PLS has shown the most suitable performance in the prediction of matrix-bound NNK, with errors comparable to the reference method, and a higher throughput. In addition, this approach allows to determine other soluble nitrosamines, namely N'-nitrosoanatabine, N'-nitrosoanabasine, and N-nitrosonornicotine, with relative percentage errors between 5.25 and 11.98%. Therefore, the third approach is the best method for a large number of cured tobacco for accuracy in determination of TSNAs.