Aplicação da espectroscopia no infravermelho médio para determinação de CBD e THC em produtos de Cannabis sativa

Abstract

The increasing demand and regulation of the Cannabis sativa industry require effective methods for determining cannabinoids in derivatives such as oral solutions and extracts. In this study, mid-infrared (MIR) spectroscopy combined with chemometrics was used to quantify cannabidiol (CBD) and tetrahydrocannabinol (THC) in oral solutions and extracts. The results were compared with those obtained by highperformance liquid chromatography coupled with a diode-array detector (HPLC-DAD). Principal component analysis (PCA) revealed a clear separation between extracts and oral solutions. Subsequently, regression models were developed using the partial least squares (PLS) chemometric regression tool. The results indicate high accuracy and reliability, with determination coefficients (R²) greater than 0.93. For the models developed for CBD quantification, RPD values ranged from 7.9 to 27.0, with RMSEC, RMSECV, and RMSEP values ranged from 0.35 to 0.46 for the model obtained only with oral solutions samples, from 2.968 to 4.832% for the model obtained only with extracts samples, and from 0.785 to 0.948% for the model obtained with all samples. Similarly, the models for THC quantification showed excellent results with RPD values ranged from 4.01 to 39. For the model obtained only with extracts samples, the error ranged from 2.838 to 1.834%, while for the model obtained only with oral solutions samples, the errors ranged from 0.114 to 0.145%. For the model containing all samples, the errors ranged from 0.555 to 0.670%. Additionally, a Partial Least Squares Discriminant Analysis (PLS-DA) model was implemented to classify oral solutions according to the regulatory specifications of the National Health Surveillance Agency (ANVISA) for THC (<0.2%). The results revealed a high accuracy rate in classification, with sensitivity and specificity values of 100%.