After identification of the phenolic components in a food, quantification can be achieved by selecting appropriate calibration standards (pure authentic compounds) that are identical or closely related to the compounds of interest [46, 48, 50, 57, 58, 60, 61, 66, 72, 92, 93, 96, 97, 122, 281-283]. Quantification is important for accurate estimation of dietary flavonoid/phenolic compond intake and to evaluate the relationship between flavonoid ingestion and human health. Numerous reports have characteized phenolic compounds in
LC-MS Profiling and Quantification of Food Phenolic Components ... 61 many foods. These data has been comprehensively surveyed, evaluated with respect to quality, and published in the USDA database [16]. This database is available on-line for no charge.
4.1. Quantitative Extraction
The extraction efficiency was evaluated using a series of 3 extractions. The first extraction was made as described in the Sample Preparation section; between 100 mg and 250 mg of sample were sonicated with 5.00 mL of solvent, centrifuged and then 4.00 mL of the solution (Extract 1) was drawn off. The volume of solvent removed was accurately determined by weight. Then 4.00 mL of fresh solvent was added to the sample and remaining solvent from the first extraction. After sonication, 4.00 ml was again drawn off (Extract 2). A final volume of 4.00 mL of fresh solvent was added to the samle, sonicated, and 4.00 mL removed for a third time (Extract 3). All 3 extracts were analyzed by HPLC, and peak areas determined for each peak in the chromatogram. The mass in the first extract (corrected for volume) was compared to the sum of all three. If 95%, or more, of the mass was seen in the first extract this indicated that a single extraction was suitable for a quantitative determination.
4.2. Quantitative Absorption Measurements
Quantification of phenolic compounds is often difficult because of the lack of appropriate glycosylated and acylated standards. A general solution to this problem is to use the aglycone for calibration provided glycolsylation and acylation has a negligable effect on the UV λmax and absorption coefficient. In general, aglycones, or common glycosides (glucoside or rutinoside), are more available than other flavonoid forms. For example, if flavonols are present, quercetin 3-O-glucoside, which is commercially available, can be used for quantification of all quercetin 3-O-glycosides (i.e. those with different saccharides at the same position) and their malonylated and acetylated forms. All of these compounds have the same UV band I λmax and absorption coefficients. The calculated concentrations only have to be corrected for the molecular weight [46,48,50,57,58,60, 61,66,72,92,93,96,97,122,281-283].
For example, quantitative results for the flavonoid content of celery, Chinese celery, and celery seeds are presented in Table 7 [122]. In this case, apiin (apigenin 7-O-apiosylglucoside) was used as a standard for quantifying the malonates of apiin and luteolin 7-O-glucoside was used as a standard for the 7-O-apiosylglucosides and malonates of luteolin and chrysoeriol. In both cases, the concentrations had to be corrected by the molecular weight ratios of the analyte and standard compounds.
Our approach to calibration is to prepare a stock solution of the standard and to prepare at least 3 different dilutions (calibration standards) that provide a range of signals (absorbance or ion counts) that encompasses the signals for the food extracts. Each of the calibration standards should be injected three times. The relative standard deviations ( RSDs) for the peak areas should be less than 3%, and the average intensities should have a linear relationship. The RSDs ion counts (MS detection in SIM mode) may exceed 3% but should be less than 5%. Each sample extract is prepared in triplicate and each preparation is analyzed
Long-Ze Lin and James M. Harnly 62
in duplicate or triplicate. Concentrations, expressed as mg/100 g fresh food or mg/g dried food, are then calculated.
In general, UV detection is used for quantification because of its stability and robustness.
When two compounds are not well separated, however, MS detection in the SIM mode is the only means to accurately measure the peak area. For quantification using SIM, the standard should be structurally identical to the compound being analyzed. To assure a stable MSD fragmentation voltage, all standard and sample extracts should be run at least twice using as short a sample injection sequence as possible.
Two of the peaks in Figure 13 are composed of unresolved, overlapping peaks (14A and 14B and 15A and 15B) that could not be quantified directly by UV absorption. For these peaks, MS detection in the SIM mode was used to determine the percent contribution of each compound to the peak. The total concentration of the peak, based on UV absorption, was then multiplied by the appropriate fraction to provide a concentration for each compound.
4.3. A General Approach
We have developed an approach to quantification using a limited number of common standards to quantify most of the phenolic compounds in foods (flavones, flavonols, flavanones, isoflavones, anthocyanins, hydroxycinnamic acids, and hydroxybenzoic acids) [150]. The basis for this approach are: 1) a great number of non-UV sensitive substitutions (such as O-methyl, O-glycoside, O-malonylglycoside or O-acetylglycoside) occur at the C3 position of the flavonols, and the C7 position of the flavonones, isoflavones, flavanones and 2) the sites of substitution follow very predictable patterns for the flavonoid subgroups. For example, as mentioned previously the C3 position of quercetin may have many different substitutions, but all of the resulting compounds have the same UV band I λmax and absorption coefficients as that of quercetin 3-O-glucoside. The polymeric flavan-3-ols made through a single bond linkage between the units, have the same or very similar λmax and absorption coefficients as their monomers. The conjugates of hydroxycinnamic or hydroxybenzoic acids formed with non-UV sensitive aliphatic acids (quinic acid, tartaric acid, malic acid, and glucaric acid) also have the same or very similar λmax and absorption coefficients as those of their parent phenolic acids.
The proceeding discussion illustrates how quantification can be achieved using a related, but different, compound. It is only necessary to correct the results by the ratio of their molecular weights (i.e. their absorption coefficients only differ by their molecular weight). In some cases, however, the substitutions may cause minor shifts in the λmax of the absorption band or a change in the absorption coefficient. Then it is necessary to establish correction factors for specific compounds or groups of compounds. Correction factors that relate the absorption coefficients of quercetin 3-O-glucoside to other quercetin glycosides, either singly (e.g. 7-, or 4'-glucoside) or multiply (e.g. 3,7-, or 3,4'-, 7,4'-diglucoside) substituted, to substituted flavonols, to substituted flavones (e.g. glycosides of kaempferol and luteolin), and to aglycones, will allow quantification with a single standard. In principle, only a few standard compounds will be necessary to quantify most of the food flavonoids. The critical aspect of this approach is selecting those few standards and establishing the correction factors.
LC-MS Profiling and Quantification of Food Phenolic Components ... 63 In some cases, substitutions may produce significant differences in the UV band I λmax
and absorption coefficients. For example, the acylated flavonoids formed with hydroxycinnamoyls and hydroxybenzoyls have absorption spectra that are significantly different from their parent compounds. Some biflavones have large wavelength shifts in the UV band I λmax compared to their monomers. Fortunately, these compounds are not commonly found among food flavonoids. Still, it is possible to establish correction factors that permit quantification within the desired error ranges. Alternatively, it is possible to convert the substituted compounds into their parent glycosides that can be quantified.
4.4. Estimation of the Range of Error
It is well known that commercially available standards (usually as powders), especially the polar flavonoid glycosides, polyhydroxyflavonoid aglycones, and the chlorides of the anthocyanins, contain crystalline water or other solvents. For example, a Sigma commercial standard for quercetin dihydrate with a minimum purity of 98%, contained two crystalline waters per molecule. A Sigma standard for rutin trihydrate with a minimum purity of 95%, contained 3 crystalline waters per molecule. Thus, the crystalline water ranged from 7% to 11%. Unfortunately, most of other suppliers do not specify the degree of hydration. Further complicating the accuracy of an analysis is the fact that the polar phenolics readily absorb moisture from air. Thus, standard purity is usually lower than expressed on most labels and will be further dependent on storage and handling conditions.
Indigenous and absorbed water in standards cannot be easily determined. Consequently, the accuracy of standards and the resultant quantification is problematic. In general, little attention has been paid to the moisture content of the standards. In our opinion, the accuracy of the data for most reported flavonoids is biased low by 85-90 %.