Semi-Targeted Ultra-High-Performance Chromatography Coupled to Mass Spectrometry Analysis of Phenolic Metabolites in Plasma of Elderly Adults

The identification and quantification of the bioactive phytochemicals that are absorbed after consumption of a food or food supplement are crucial for demonstrating and understanding the health benefits of these compounds and the foods containing them. In the present work, the UPLC-MS/MS method was developed, aimed only at the identification of the main phenolic compounds and their metabolites that increased in concentration in plasma after a 30-day nutritional intervention with two food products specially formulated for the elderly. It was assumed that, if one compound increased or appeared in plasma only after the intervention period, that compound had been absorbed and/or biotransformed from the foods used in the intervention.

UPLC-MS/MS is one of the preferred technologies for metabolomic studies in which many low molecular weight compounds are analyzed simultaneously in complex samples to assess the effect of some treatment or environmental alteration¹³. This can be done through targeted and non-targeted (or global) methods. Targeted analyses focus on a known, small number of metabolites of interest. They are the best option for quantifying compounds with available standards, providing the best specificity and sensitivity; however, these methods must be thoroughly optimized for the selected metabolites and are incapable of detecting unexpected compounds in the samples¹². In other studies, HPLC coupled with a UV-vis detector has been used to identify phenolic acids and metabolites in plasma, but this type of study has used analytical standards to identify and quantify the compound of interest¹⁵. In another study, the addition of an internal standard was proposed¹⁸; however, only ferulic and caffeic acids and their metabolites were analyzed in this work. On the other hand, the synthesis of internal standards of phenolic metabolites has been proposed as an alternative to the lack of commercial analytical standards¹⁹. Nevertheless, the synthesis of a large variety of phenolic compounds is difficult, time-consuming, and expensive. Non-targeted methods attempt to detect and identify as many molecular compounds as possible and are primarily hypothesis-generating¹². They can identify several hundreds of compounds in one sample by assigning formulas to detectable chromatographic signals with characteristic features, such as m/z or measured accurate mass, chromatographic retention time, isotopic fingerprint, etc., so the data generated by these analyses are very abundant and can be difficult to interpret²⁰. Since the objective of the present method was not to obtain a full metabolic profile of the plasma samples (which would be an untargeted approach) but to identify the major phenolic compounds and phenolic metabolites (previously unknown) in them, a semi-targeted approach was designed. For this, all the identified signals were extracted automatically with the instrument’s software and then compared with a self-created database that contained 645 phenolic compounds and their metabolites, which were obtained from free online databases. The database included reference MS/MS fragmentation patterns of the compounds that were used for the assignment of the compound name and formula, which allows for more accurate identification of compounds in the sample¹².

The most critical steps in the protocol were 1) sample pretreatment (extraction and concentration of phenolic compounds and metabolites from plasma samples); 2) the construction of a complete and specific database for the semi-targeted analysis of the samples and identification of all the possible compounds belonging to a specific compound class; and 3) the selection of the parameters used by the instrument’s qualitative software (mass match tolerance ppm, and score%) to accurately identify the compounds in the sample. In sample pretreatment, caution must be taken to avoid loss of the compounds of interest, recover them at a high final concentration, and be able to eliminate interfering compounds. In the creation of the specific database, it is critical to conduct a thorough literature search and then use free online databases to obtain specific chemical data that will be used for compound identification in the sample. The selection of the best values for mass match and score required previous validation of the analytical method using standards and known samples^21,22.

Once the protocol was implemented, the common problems and their recommended solutions were the following: 1) No compounds were identified in the samples. This was due to the low concentration of the compounds and could be solved by drying the sample and re-dissolving it in a smaller volume. 2) Signals appeared in the TIC, but compounds were not identified. This could be due to a failure in mass calibration, so the difference between experimental and theoretical masses was high. In this case, the instrument’s mass calibration should be performed according to the manufacturer and model of the equipment. A second reason could be an incomplete personal database, so it is critical to constantly check and maintain the database. Masses of compounds suspected to be of interest can be checked and compared with freely available online databases or new scientific literature. 3) Signals appeared in blank runs. This indicates contamination and can be solved by cleaning the autosampler needle and column. The standard cleaning protocol is to perform some runs using methanol, then isopropanol, and finally acetonitrile as the mobile phase. Then the column is re-equilibrated, running the mobile phase for at least 30 min.

The major difficulties in the development of this kind of method are 1) the low bioavailability, in general terms, of all phytochemicals, including phenolic compounds, which translates to very low plasma concentrations¹⁶; 2) the high diversity of phenolic compounds present in foods, which is increased by their metabolism and biotransformation that occur in human enterocytes and hepatocytes and the colonic microbiota²³; 3) the lack of standards (some standards exist but are very difficult to obtain) for many compounds, especially for the metabolites, and the existence of some unknown or uncharacterized compounds²³; and 4) variable responses among individuals, both in absorption and metabolism of phytochemicals¹⁴. To overcome the problem of low plasma concentrations of phenolic metabolites, they should be extracted and concentrated. This can be achieved by micro solid-phase extraction¹⁴ or, as in the present work, by the addition of solvents that precipitate proteins and dissolve phenolic compounds and metabolites, followed by solvent evaporation and resuspension in a smaller volume¹⁹. This technique is simple, economic, and adequate for a small number of samples. However, the recovery of compounds of interest could be easily improved by using larger plasma volumes, so the final concentration of all the compounds would be higher. The high diversity of phenolic metabolites and lack of standards is the reason why a semi targeted approach was used to identify the compounds increased by the intervention before attempting to quantify them. By using a specially created database instead of a completely untargeted analysis, it was possible to ignore the numerous primary metabolites in plasma, focusing only on the phenolic compounds and their metabolites. An important limitation of this database, specific for phenolic compounds and their human metabolites, was the lack of mass spectral information for many compounds and their metabolites, which reduces the accuracy of the compound identification.

Large variability among individuals is regularly observed in studies evaluating absorption and metabolism of phenolic compounds, both quantitively and qualitatively. In the present results, one individual presented 18 compounds in the plasma sample taken after the intervention, while the others showed only 9 or 10. Moreover, many phenolic compounds and metabolites were found only in the baseline samples (before the intervention), and they also varied among individuals. Overall, more than half of the identified compounds showed higher concentrations (AUC) before than after the intervention or were found only in the pre-intervention samples. Therefore, it was necessary to compare individual samples taken before and after the intervention to understand which compounds were actually increased by it. The only metabolite that consistently increased after the treatment was phenylacetic acid, a microbial catabolite of flavan-3-ols²⁴ that has been found in the plasma of individuals in acute studies of the consumption of phenolic-rich foods^14,25. Most of the studies that have identified and quantified phenolic metabolites in plasma were acute studies, in which the compound concentrations were monitored during the 24 h after consumption of the phenolic-rich meal, observing that, after 24 h, their concentrations were near the basal value^19,25. Therefore, it is understandable that the samples analyzed in the present work showed low numbers and concentrations of phenolic metabolites. Recently, Zhang et al.²⁵ evaluated the consumption of red raspberry, both in acute studies and after 4 weeks of supplementation. They observed that both acute and chronic interventions increased phenolic compounds and their metabolites, while after the 4-week supplementation, the basal plasma concentration of some metabolites increased (urolithins, cinnamic, phenylpropionic, and hippuric acids) and of others decreased (conjugated anthocyanin derivatives).

A final examination of the method developed in the present paper shows that, in most parts, it was well suited for the objective for which it was created. The sample pretreatment was simple and effective, and the UPLC separation and semi-targeted MS/MS identification of phenolic metabolites were achieved. This semi-targeted method can be useful as a first screening approach to identify phenolic compounds that can be absorbed from different food matrixes, as well as the phenolic metabolites present in human plasma. Moreover, the protocol is useful because it allows identification when no standards from phenolic compounds or metabolites are available. Finally, the method uses a small plasma quantity, which is critical in most in vivo studies where plasma samples are scarce. However, it is recommended that, for future studies, an acute study should be carried out before the chronic intervention to better identify the main phenolic compounds and metabolites absorbed from the specially formulated foods.