Identification of active compounds in vegetal extracts based on correlation between activity and HPLC–MS data

We propose a method identifying candidates for active compounds in vegetal extracts. From a collection of samples, the method requires, for each sample, a HPLC–MS analysis and a measurement of the activity. By applying a correlation analysis between the activity and the chromatographic area for each interval of elution time and m/z ratio, the peaks corresponding to candidates for active compounds can be identified. Additionally, when peaks are identified, a model can be estimated to predict the activity in new samples. Both methods are evaluated in one experiment involving the phenolic extract (PE) from 22 samples of extra virgin olive oil (EVOO) where the activity is a cytotoxicity index against JIMT-1 breast cancer cells. In this experiment, the samples were separated into two disjunct partitions: one was used for training (identification of candidates and estimation of prediction model), while the other was used for validation (by comparing the predicted and the measured activities). Three compounds were identified as candidates to be responsible for the cytotoxicity of the EVOO-PE against JIMT-1 cells. The prediction model provided an accurate estimation of the activity.