Desenvolvimento de modelos de calibração multivariada para determinação de compostos fenólicos e atividade antioxidante em extrato etanólico de própolis utilizando a técnica FT-NIR

Abstract

Propolis is produced by bees and has a complex chemical composition. Among the main constituents of propolis are phenolic compounds that contribute to the present propolis pharmacological properties, such as antioxidant activity (AA). Because of these beneficial properties, propolis extract (EEP) is commonly consumed as food supplement. Thus, in order to estimate the quality of the EEP, reference analyzes are used and, in most cases, they are time consuming, require specific reagents and trained technician. For the purpose of developing an alternative analysis to existing quantification methodologies, this work aims to the development of multivariate calibration models for regression by partial least squares (PLS) for quantification of total phenolic compounds and AA in ethanol extract of propolis (EEP) produced by selected Apis mellifera using the infrared spectroscopy technique in near region (NIR). First, the hydroalcoholic ethanol extracts: water (80:20 v / v) were analyzed by reference methods for the determination of the phenolics levels and total flavonoids and AA by sequestration methods ABTS radical, reduction of Fe3+ to Fe2+ (FRAP) and scavenging of free radical DPPH. After, the EEP were analyzed in Bruker FT-NIR spectrometer and models have been developed and validated in OPUS quant2 software through the correlation between the reference analysis and the generated spectra. The correlation coefficient parameter (R2) and reason RMSEP / RMSECV were used to evaluate the robustness and predictive ability of the models respectively. The model generated for total phenolic compounds showed the highest correlation coefficients and robustness, indicating an excellent predictive ability and robustness. The parameters evaluated for total flavonoid suggest that the model does not have a good ability to predict and robustness yet. However, the generated models for the determination of antioxidant activity suggest good predictive capability and a satisfactory robustness. Therefore, it can be understood that the flavonoids model is able to estimate the levels in the extracts and models of total phenolics, DPPH, ABTS and FRAP developed become an alternative to the presented reference analyzes, reducing the generation of residue and the analysis time.