Caracterização química por cromatografia líquida e análise quimiométrica de espécies vegetais de Bauhinia com aplicação em controle de qualidade de amostras comerciais de "pata-de-vaca"

Abstract

Chemical characterization by liquid chromatography and chemometric analysis of Bauhinia species (popularly known as pata-de-vaca ) with applications in quality control of commercial samples. The infusion of leaves from Bauhinia species, is commonly used in the treatment to combat various diseases, among them diabetes. Studies show that B. forficata species possesses the most hypoglycemic activity among the species of the genus. However, due to the extreme morphological similarity between its leaves in comparison to other species , occasionally the popular use in the decoction of leaves is improper. Thus, it is necessary to develop methods of quality control of herbal drugs, in order to certify the authenticity of them. The chemical profiles (fingerprints) obtained by chromatographic analysis of plant extracts, brings a large amount of information, providing multivariable systems that can be used as a good parameter for quality control. The present study reports the separation optimization of the constituents obtained from aqueous extract on Bauhinia species by gradient elution on reverse phase. The chromatograms were first aligned using the MatLab® software and the COW algorithm (Correlation Optimized Warping) and after, the chromatograms were subjected on exploratory chemometric analysis Principal Component Analysis (PCA) and Hierarchical Cluster Analysis (HCA) and used to create models for supervised chemometric classification (KNN, SIMCA and PLS-DA) performed using the Pirouette® software. 170 chromatograms of extracts obtained from leaves of botanically identified trees from three different species - B. forficata, B. longifolia and B. variegata - were used on analysis by PCA and HCA, which allowed the visualization of the distribution of 15 commercial samples of "pata-de-vaca". Those samples were also analyzed using the supervised classification models, and among them, 3 were classified as B. forficata, 6 were classified as B. longifolia and 5 of them were classified as B. variegata. For one of the samples it was not possible to classify its species due to large difference among the determinations obtained from the three distinct analytic tools