Quantification by LC-MS/MS of astragaloside IV and isoflavones in Astragali radix can be more accurate by using standard addition.

INTRODUCTION
Astragali radix (AR), the root of Astragalus, is an important medical herb widely used in traditional Chinese medicine. Bioactive components include isoflavones and a unique class of triterpenoid saponins (named astragalosides).


OBJECTIVES
Accurate measurement of bioactive components, especially astragaloside IV, is necessary for confirming AR authenticity, quality control and future medical research.


METHODOLOGY
Liquid chromatography coupled with tandem mass spectrometry (LC-MS/MS) is a suitable technique but suffers from ion suppression effects due to sample matrix. This can be corrected by using isotopic labelled internal standards, but these are not available for many phytochemicals. We explored the use of standard addition to circumvent this issue.


RESULTS
LC-MS/MS and liquid chromatography coupled with ultraviolet (LC-UV) detection provided linear calibration curves (R2 > 0.99). LC-MS/MS provided superior selectivity and detection limits below 10 ng/mL, which was 2-3 magnitudes lower than LC-UV detection. Precision and accuracy were overall improved by using LC-MS/MS with diluted sample extracts, resulting in an inter series coefficient of variation (CV) of 12% or less and mean recovery estimates in the 85-115% range. LC-MS/MS quantification by standard addition resulted in significantly higher concentrations of astragaloside IV measured in the samples. Concentrations calculated by standard addition were unaffected by large variation in signal response caused by matrix effects, independent of variation in slope of the standard addition curves.


CONCLUSION
Sample dilution was helpful but not sufficient for reducing effects of ion suppression. We have shown that LC-MS/MS quantification by standard addition can be a powerful approach for accurate measurement of phytochemicals in the absence of isotopic labelled internal standards.


| INTRODUCTION
Astragalus mongholicus Bunge (English: Mongolian milkvetch, Chinese: huáng qí) and Astragalus propinquus Schischkin are important medicinal plants in traditional Chinese medicine (TCM). The root of these plants is known as Astragali radix (AR) and is often clinically applied as a medical herb in TCM and Western phytomedicine. 1 AR preparations are classified according to TCM, as tonifiers of vital energy (Qi). AR is the second strongest TCM medicine in this category, after the much more expensive and rare Ginseng. 1,2 Meta-studies and reviews of randomised controlled trials (RCTs) have claimed positive effects of AR in breast cancer, 3 acute myocardial infraction, 4 diabetic nephropathy 5 and many more diseases.
AR is known to be a rich source of astragalosides (AG) which are triterpenoid saponins unique to the Astragalus species. There are at least seven AG components present in different plant tissues, 6 with astragaloside IV (AG-IV) being the most well-known because of its high bioactivity. This component is a 9,19-cycloartane type major active triterpene glycoside, which has been reported to increase T and B lymphocyte proliferation. 7 Other effects of AG-IV include cardioprotective, neuroprotective, immune stimulating and anti-inflammatory properties. 8 Upon ingestion, AG-IV is metabolised to cycloastragenol (CAG), which is its corresponding aglycon (glycoside-free) sapogenin. 9 CAG extend T-cell proliferation by increasing the telomerase activity, a vital process to delay cellular ageing. CAG has been marketed for human applications by T.A. Sciences (New York, NY, USA) as a dietary supplement for anti-ageing under the brand name TA-65 ® . 10 Herbal medicinal products, including TCM herbs, are regulated in the European Union (EU) by the EU medicine directive 2001/83/EC amended by the 2004/24/EC directive. TCM products need approval by the European Medicines Agency (EMA). Both for quality control and to correlate pharmaceutical effects with consumption of TCM herbs, it is essential to know their chemical composition, accurately measure concentrations of their active components and study the pharmacokinetics in humans. Most pharmacopeias state that AG-IV should be above 0.04%. The European pharmacopeia suggest using an ammonia solution during sample preparation for increasing the level of AG-IV in the extract. This occurs by hydrolysis of several other AGs into AG-IV. 11 However, by adding ammonia we might not measure the true concentration of AG-IV naturally present in samples intended for intake. Consequently, it is questionable to perform pharmacokinetic studies based upon ammonia-based results. 12 Other components in AR include isoflavones such as calycosin, ononin and formononetin ( Figure 1). Their effects include boosting of the immune system, as well as having impact on glucose homeostasis and lipid metabolism. 13 Isoflavone concentrations in AR can be assessed by liquid chromatography coupled with ultraviolet (LC-UV) detection. However, AG-IV and CAG do not contain chromophores, and UV detection is therefore not possible. Instead, their quantification can be done by LC coupled with evaporative light scattering detector (ELSD) or with mass spectrometry (MS) detection. For quality control of AR, it was recently shown that LC-MS provides a more sensitive and convenient quantification of AG-IV than with LC-ELSD. 14 MS and especially tandem mass spectrometry (MS/MS) provide superior analytical selectivity when compared with ELSD, which improves the accuracy of the method. 15,16 Although LC-MS/MS performed with atmospheric pressure chemical ionisation (APCI) can quantify AGs in the negative mode 17 it is more common to increase sensitivity by applying electrospray ionisation (ESI) in the positive F I G U R E 1 Molecular structures of (A) astragaloside IV, (B) cycloastragenol, (C) formononetin, (D) ononin, and (E) calycosin 7-O-β-D-glucoside mode. 8,14 However, ESI is known to suffer from ion suppression due to various effects of the sample matrix. 18 This is typically seen as a variation in the signal response caused by sample composition, making accurate quantification more difficult. Isotope labelled internal standards are applied to correct for ion suppression in the LC-MS/MS analysis, but no such internal standards exist for AG-IV and CAG.
Most studies on the analysis of AR samples have used UV detectors, 19,20 ELSDs 12,21 and some have used MS detectors. 6,14 Reports indicate extensive use of MS/MS detectors for AR samples and even for bioavailability of compounds during animal studies. 16,17,22,23 None of these have utilised standard addition to compensate for ion suppression effects in the LC-MS/MS analysis of AR, although standard addition was successfully reported for quantification of AGs by LC-ELSD. 24 In the absence of an isotope labelled internal standard, we explored the use of standard addition to improve the accuracy of quantitative assessment of bioactive components extracted from EU-approved AR. By compensating for various matrix effects of the samples, standard addition is expected to provide concentration measurements that are closer to the true levels in samples. Purified water was obtained from Ultra purified water purification system (Purelab Flex from ELGA LabWater, High Wycombe, United Kingdom). All the samples prepared were centrifuged at 4000 rpm for 10 min using Eppendorf Centrifuge 5702 (Eppendorf, Hamburg, Germany) before injection into LC-UV detector and LC-MS/MS.

| Sample extracts
The analysed AR samples were dried herb, granulate or hydrophilic concentrate of Astragalus mongholicus Bunge, all purchased from Natuurapotheek (Pijnacker, the Netherlands). Sample extracts (Table 1) were prepared from solid samples (5 g) either by boiling in water for 60 min or by ultrasonication for 60 min in 70% methanol at 40 C. Sample extracts were centrifuged at 4000 rpm for 10 mins twice to remove impurities (Eppendorf Centrifuge 5702), followed by drying in a rotary evaporator (IKA HB 10, VWR International), and reconstitution in methanol to a final volume of 20 mL. The hydrophilic concentrate liquid was analysed directly without extraction.

| Standard solutions
Pure analytical standard chemicals were weighed and dissolved in methanol to a concentration of 1 mg/mL. Addition of 4 to 5% of acetone in methanol, as well as heating to 40 C, was used to dissolve   Components were separated on a 100 mm long and 2.1 mm ID reversed-phase BEH C18 column with 1.7 μm particle size and 130 Å pore size (Waters). Column temperature was 50 C. The mobile phase delivered at 0.5 mL/min was a mixture of (A) 0.2% formic acid and

| Method validation
The methods were validated by following the guidelines provided by Eurachem Guide 25 (Table 4) and all five analytes by LC-MS/MS were assessed by analysis of samples spiked to concentrations in the 2.5-10 μg/mL range ( Table 5). The LC-MS/MS method was found to be more precise and accurate, with mean recovery for compounds found to be 102 ± 6% for AG-IV, 97 ± 9% for cycloastragenol, 101 ± 8% for formononetin, 97 ± 8% for ononin, and 95 ± 7% for calycosin 7-O-β-D-glucoside. However, to achieve such high recoveries, it was necessary to quantify by applying the slope obtained from the standard addition calibration curves described later.

| Standard addition
For quantification by standard addition, 100 μL of the sample extract was added to four autosampler vials together with 100 μL of standard mixture containing 0, 5, 10 or 20 μg/mL of each analyte, followed by dilution to 1 mL with methanol. Figure 2    The AR samples analysed were selected from various formulations and raw herbs from one specific vendor. The samples were extracted by ultrasonication or boiling in water, instead of traditional Soxhlet extraction, which requires large quantity of samples and solvents. 26 We hypothesised that the extraction technique would affect the measured concentration levels. 27 Indeed, this was observed for the dried root herb where more AG-IV was extracted into boiling water (sample B1) than into 70% methanol by sonication (sample B).
Hot water seemed to be the better solvent for this rather polar substance.
Accurate identification and quantification of AR samples is needed, because of the presence of complex and multiple bioactive components and their therapeutic mechanisms. These can be helpful for optimising the dosage regimens, as well as pharmacological and pharmacokinetic profiling of the components. 28 As shown in the present work, the superior analyte detectability of MS/MS over UV detection, illustrates the advantages of using LC-MS/MS, not only for AR quality control, but potentially also in future pharmacokinetic studies that require low LODs.