Chemometric tools for quality control of Cannabis oils from the community

CABEZUDO, IGNACIO; BORTOLATO, SANTIAGO; PISANO, PABLO; HOURCADE, MÓNICA

Rosario

Congreso; 7ma Reunión Internacional de Ciencias Farmacéuticas (RICiFa 2023); 2023

FBIOYF-UNR

As interest and regulationsurrounding the medicinal use of Cannabis continue to grow, there is an urgentneed to improve quality control methods for pharmaceutical products derivedfrom Cannabis. This is particularly important since in our country the greatestCannabis medicines are obtained in the form of homemade oils from thecommunity. Since 2017, the School of Biochemical and Pharmaceutical Sciences (FBIOYF-UNR)has been offering a service for analysing Cannabis oils1. Thisservice employs gas chromatography coupled with mass spectrometry (GC-MS) forthe determination of cannabinoids. About 3000 samples have been analysed so far,revealing specific cannabinoids, and more recently terpenes. However, thesedata may contain critical information about product quality, such ascontaminants, additives, plant origin, extraction methods, and more, which we havejust started to explore.In this study, we propose usingchemometric tools to process data from Cannabis oils obtained through GC-MS,with the aim of rigorously classifying samples based on all availableinformation. Our method involves the use of Matlab software and the MCR-ALSalgorithm to decompose the data into matrices containing information about thechromatographic and spectral profiles of the most relevant components in thesamples2. The non-negativity constraint ensured that both thespectral profile and the concentration did not contain negative values. Also, unimodalitywas applied for component concentrations not to exhibit multimodal behaviour. Additionally,we prepared a score matrix that reflects the quantity of each component in eachsample. These mathematical arrangements simplify processing withoutcompromising data quality. Subsequently, we applied principal componentanalysis routine (PCA) to the score matrices of all samples to achieveclassification based on common profiles. By incorporating additionalinformation about each sample, such as origin, extraction method, andgeographical location, our method has enabled the classification of samplesinto groups that reflect differences. Using PCA, four classes of country oforigin of the samples could be classified plotting principal components one andtwo. This provides valuable insights beyond specific cannabinoids and offersthe possibility of detecting adulteration, enhancing quality control andproduct traceability. This approach holds the promise of further optimizing thequality control of Cannabis-derived medications, contributing to the safety andefficacy of these products. 1 https://www.fbioyf.unr.edu.ar/?page_id=1022. 2 Jaumot J., Gargallo R., de Juan A., Tauler R.,Chemometrics and Intelligent Laboratory Systems, 76(1) (2005) 101-110.