RAPID COMMUNICATIONS IN MASS SPECTROMETRY

Rapid Commun. Mass Spectrom. 2005; 19: 813–817

Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/rcm.1862

Liquid chromatography/tandem mass spectrometry for

pharmacokinetic studies of 20(R)-ginsenoside Rg3 in dog
Ke Li1, Xiaoyan Chen1, Jinghua Xu2, Xin Li3 and Dafang Zhong1,3*
1
Laboratory of Drug Metabolism and Pharmacokinetics, Shenyang Pharmaceutical University, Shenyang, China
2
School of Pharmacy, Shenyang Pharmaceutical University, Shenyang, China
3
Research Center for Drug Metabolism, Jilin University, Changchun, China
Received 15 November 2004; Revised 25 January 2005; Accepted 26 January 2005

A sensitive and specific liquid chromatography/tandem mass spectrometry (LC/MS/MS) method

was developed for the investigation of the pharmacokinetics of 20(R)-ginsenoside Rg3 in dog.
The plasma samples were pretreated by liquid-liquid extraction and analyzed using LC/MS/MS
with an electrospray ionization interface. Dioscin was used as the internal standard. The method
had a lower limit of quantitation of 0.5 ng/mL for Rg3 in 200 mL of plasma or 2 ng/mL in 100 mL of
plasma, which offered a satisfactory sensitivity for the determination of Rg3 in plasma. The intra-
and inter-day precisions were measured to be below 8% and accuracy between
1.5 and 1.4% for all
quality control samples. This quantitation method was successfully applied to pharmacokinetic
studies of Rg3 after both an oral and an intravenous administration to beagle dogs. No Rh2 and
protopanaxadiol were detected in plasma. Copyright # 2005 John Wiley & Sons, Ltd.

Panax ginseng is frequently used in Asian countries as a tradi- Rg3 in plasma, and the pharmacokinetics of 20(R)-ginseno-
tional medicine. The effective components of ginseng side Rg3 in dogs after oral and i.v. administrations were
are mainly dammarane triterpene O-glycosides, known as investigated. An attempt was also made to detect the possible
ginsenosides. metabolites 20(R)-ginsenoside Rh2 and 20(R)-protopanaxa-
Ginsenoside Rg3 (Fig. 1) has been shown to inhibit tumor diol in plasma.
metastasis in mice and in vitro tumor cell invasion.1,2 The
pharmacokinetics of Rg3 in humans were reported by Wang
EXPERIMENTAL
et al.3 Plasma concentrations of Rg3 were determined after an
oral dose of 3.2 mg/kg in healthy human volunteers, and Materials
some pharmacokinetic parameters were reported. In the 20(R)-Ginsenoside Rg3 (99.9% pure) and 20(R)-ginsenoside
high-performance liquid chromatography (HPLC) method Rh2 (98.8% pure) were supplied by Fusheng Medicine Co.,
used in this previous work,3 a large volume of plasma and a Ltd. (Liaoning, China). This company also supplied the gran-
complicated sample preparation method were applied in ules (5.4 mg/g) and the solution (0.3 mg/mL) of 20(R)-ginse-
order to achieve suitable sensitivity. Recently, a liquid noside Rg3 for oral and i.v. administration, respectively.
chromatography/mass spectrometry (LC/MS) method was Dioscin for use as internal standard (IS) was obtained from
applied to the study of metabolism and pharmacokinetics of Hengrui Pharmaceutical Co, Ltd. (Jiangsu, China). Acetoni-
Rg3 after intravenous (i.v.) administration of 5 mg/kg of Rg3 trile and methanol (HPLC grade) were purchased from
to rats.4 The determination of Rg3 and the identification of Fisher Chemicals (Fairlawn, NJ, USA), and water was doubly
in vitro major metabolites were performed using tandem distilled in the laboratory. All other chemicals were pur-
mass spectrometry (MS/MS) and high-resolution MS tech- chased from commercial sources and used as received.
niques. However, compared with typical plasma concentra-
tions after an oral dose of 50 mg/kg, the detection limit of Equipment
27 ng/mL was high so that Rg3 was not detected in the The HPLC was performed using an Agilent 1100 system (Palo
samples.4 Detailed pharmacokinetic descriptions of Rg3 in Alto, CA, USA) equipped with a G1313A autosampler, a
animals and humans are still required. It was reported that vacuum degasser unit, and a G1312A binary pump. The
the loss of glucose residues was the main metabolic pathway HPLC system was coupled to an API 4000 triple-quadrupole
of ginsenosides in vitro.5,6 When ginsenoside Rg3 was mass spectrometer (Applied Biosystems/MDS Sciex,
incubated with human fecal microflora it was metabolized Concord, ON, Canada) via a TurboIonSpray electrospray
to ginsenoside Rh2 and protopanaxadiol.7 ionization (ESI) interface for mass analysis and detection.
In the present work a sensitive and simple method using A 10-port switching valve (Rheodyne, Cotati, CA, USA)
LC/MS/MS was developed to determine 20(R)-ginsenoside was used to direct HPLC eluate to waste in the first 0.8 min
of the chromatographic run, and afterwards to the ionization
*Correspondence to: D. Zhong, Laboratory of Drug Metabolism source. Data were collected and analyzed by the Analyst
and Pharmacokinetics, Shenyang Pharmaceutical University,
103 Wenhua Road, Shenyang 110016, P. R. China. 1.3 data acquisition and processing software (Applied
E-mail: [email protected] Biosystems/MDS Sciex).

Copyright # 2005 John Wiley & Sons, Ltd.

814 K. Li et al.

Figure 1. Structures of 20(R)-ginsenoside Rg3, Rh2, 20(R)-protopanaxadiol and dioscin

(internal standard).

Chromatographic conditions Preparation of calibration standards and quality

Chromatographic separation was achieved on a Nucleosil control (QC) samples
ODS column (50  4.6 mm i.d., 5 mm; Dalian Johnsson Separa- Due to the low drug concentration in plasma after the oral
tion Science and Technology Corp., Liaoning, China) with a dose, two different sets of calibration curves were prepared
4.0  3.0 mm i.d. SecurityGuard C18 (5 mm) guard column for the determination of the analytes in plasma following
(Phenomenex, Torrance, CA, USA). The chromatography the oral and i.v. doses. For the orally or i.v. administered sam-
was performed at 208C. The mobile phase consisted of metha- ples, calibration standards and QC samples were prepared by
nol/acetonitrile/10 mM aqueous ammonium acetate spiking 100 mL of working solutions and 100 mL of internal
(47.5:47.5:5, v/v/v), without adjustment of pH, delivered at standard into 200 mL or 100 mL of drug-free plasma. Matrix-
a flow rate of 0.7 mL/min. matched calibration standards were obtained with concen-
trations of 0.5, 1, 2, 5, 10 and 20 ng/mL or 2, 4, 10, 20, 40,
Mass spectrometric conditions 100, 200 and 400 ng/mL of Rg3 in plasma. QC samples
The mass spectrometer was operated in the negative ion were obtained with concentrations of 1, 5 and 16 ng/mL or
mode. The tuning parameters were optimized for ginseno- 4, 40 and 320 ng/mL of Rg3 in plasma. The two high QC sam-
sides Rg3, Rh2 and the IS by infusing a solution containing ples (16 and 320 ng/mL) were also prepared using different
approximate 2 mg/mL of analytes at a flow rate of 30 mL/ weighed aliquots of the pure standard from those used for
min into the mobile phase (0.7 mL/min) using a T connection. the calibration samples and the other QC samples, and the
Following optimization of the settings, the instrument was values of the relative errors thus observed were 1.3 and
operated with an ion spray voltage of
3.5 kV, backpressures 1.4%, respectively.
for collision gas of 3 psi, curtain gas of 15 psi, nebulizer gas of For Rh2, calibration standards were prepared at concen-
30 psi, and heater gas of 30 psi; the heater gas temperature trations of 2, 5, 10 and 20 ng/mL or 10, 20, 40, 100, 200 and
was set at 5008C. All gases used were nitrogen. The fragmen- 400 ng/mL in plasma.
tation transitions for the multiple reaction monitoring (MRM)
were m/z 783.8 to 160.8 for Rg3, m/z 621.7 to 160.8 for Rh2, and Sample preparation
m/z 867.5 to 721.5 for the IS, with a dwell time of 200 ms per A volume of 100 mL of the IS (10 ng/mL), 100 mL of methanol
transition. and 500 mL of water were added to 200 mL of plasma from the
dogs dosed orally. This mixture was extracted with 3 mL of
Preparation of stock and working solutions ethyl acetate by shaking for 10 min. The organic and aqueous
The stock solution of Rg3 and Rh2 (400 mg/ml) was prepared phases were separated by centrifugation at 3000 g for 8 min.
in dimethyl sulfoxide and serially diluted to produce an 8 mg/ The upper organic phase was transferred to another tube and
mL stock solution in methanol, which was diluted to give evaporated to dryness at 408C under a gentle stream of air.
working solutions of 1, 2, 4, 10, 20, 32, 40, 100, 200, 320 and The residue was dissolved in 200 mL of the mobile phase,
400 ng/mL in methanol. The IS (dioscin) solutions of 10 and and vortex-mixed for 1 min. A 20-mL aliquot of the solution
100 ng/mL were similarly prepared in methanol. All stock was injected onto the LC/MS/MS system for analysis.
solutions and working solutions were stored at 48C. Plasma from the dogs dosed by i.v. injection was similarly

Copyright # 2005 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2005; 19: 813–817
 LC/MS/MS for pharmacokinetics of 20(R)-ginsenoside Rg3 815

pretreated, with the minor modifications that 100 mL of plas-

ma and 100 ng/mL of IS were used.

Application to pharmacokinetic study

The LC/MS/MS assay developed was used to investigate the
plasma concentration-time profile of ginsenoside Rg3 after an
oral dose of 2 mg/kg and an i.v. injection of 0.3 mg/kg of gin-
senoside Rg3. A study on 12 beagle dogs (both female and
male, body weight 10–12 kg) was performed. A single i.v.
dose of 20(R)-ginsenoside Rg3 was given to a group of six
dogs (3 female and 3 male), and another group of six dogs
that were fasted overnight but with access to water were
administered with a single oral dose. Blood (0.5 mL) was
withdrawn via the forelimb vein prior to dosing and at 0.25,
0.5, 1, 2, 3, 4, 6, 8, 12 and 24 h after the oral administration, and
at 0.033, 0.16, 0.5, 1, 2, 3, 5, 8 and 12 h after the i.v. injection.
Following centrifugation (3000 g, 10 min), plasma was sepa-
rated and stored at
208C until analysis.
Plasma-concentration data for individual dogs were
analyzed by non-compartmental analysis using the TopFit
2.0 software package (Thomae GmbH, Germany). Maximum
plasma concentration (Cmax) and the time-to-maximum
concentration (Tmax) were estimated by visual inspection
of semi-logarithmic plots of the concentration-time curves.
The area under the curve (AUC0-t) was calculated using
the linear-trapezoidal rule, with extrapolation to infinity
(AUC0-1) from the last detectable concentration using the Figure 2. Product ion mass spectra of [M–H]
ions of (A)
terminal elimination rate constant (ke) calculated by linear ginsenoside Rg3, (B) ginsenoside Rh2, and (C) dioscin (IS).
regression of the final log-linear part of the drug concentra-
tion-time curve. Apparent elimination half-life (t1/2) was
calculated as t1/2 ¼ 0.693/ke, total body clearance (CL) as respectively (the m/z 160.8 ion corresponds to deprotonated
dose/AUC0-1, and apparent volume of distribution (Vd) glucose). The intensities of these three fragmentation transi-
as CL/ke. tions of m/z 783.8 were optimized for MRM using flow
injections; the higher response for the transition yielding m/z
160.8 led to choice of this transition for the quantitative
RESULTS AND DISCUSSION
measurement of Rg3. Likewise, the product ion m/z 160.8 was
The low concentrations of Rg3 after an oral dose of 2 mg/kg chosen as the quantitation ion for Rh2. For the [M–H]
ion of
require a very sensitive assay to achieve the required quanti- dioscin (m/z 867.5) the major product ion m/z 721.5 (suggest-
tation limit. For this purpose, LC/MS/MS was considered to ing loss of a rhamnose) was monitored in the MRM analysis.
be a preferred technique, and the MS, chromatographic and These MS/MS fragmentations are shown in Fig. 2.
extraction conditions were optimized. During the optimization of chromatographic conditions,
columns packed with different types of C18 material
Chromatography and mass spectrometry (Nucleosil, Hypersil, Zorbax) were tried; ginsenosides Rg3
The responses of ginsenosides Rg3 and Rh2 to ESI were eval- and Rh2, and the IS, were extensively retained on these
uated by recording the full-scan mass spectra in both positive columns. To achieve symmetrical peak shapes and short
and negative ionization modes, introducing Rg3 and Rh2 chromatographic cycle times, a mobile phase consisting of
solutions via a syringe pump. The negative mode yielded a methanol/acetonitrile/10 mM aqueous ammonium acetate
signal more than 10-fold higher for the deprotonated mole- (47.5:47.5:5, v/v/v) was used, with the 50 mm Nucleosil ODS
cule of Rg3 (m/z 783.8) compared with the response for the column.
protonated molecule (m/z 785.8) in the positive mode. The
signal for Rh2 (m/z 621.7) in the negative mode was similar Sample preparation
to that (m/z 623.7) in the positive mode, and over 5-fold lower Different sample pretreatment procedures were evaluated,
than the signal for Rg3 (m/z 783.8). including solid-phase extraction (SPE), protein precipitation
In the full-scan MS/MS spectrum of the [M–H]
precursor and liquid-liquid extraction. SPE and protein precipitation
ion (m/z 783.8) of Rg3 the main product ions were observed at could not offer satisfactory sensitivities for the deter-
m/z 621.7, 459.5 and 160.8, consistent with neutral losses of mination of Rg3 and Rh2 in plasma. Finally, it was found
one glucose moiety, two glucoses, and one glucose plus the that these saponins in plasma could be extracted by ethyl
aglycone, respectively. Similarly, the main product ions of acetate with an over 80% extraction recovery. No pH adjust-
the [M–H]
precursor ion of Rh2 (m/z 621.7) were observed at ment was necessary due to the lack of acidic and basic groups
m/z 459.5 and 160.8, suggesting losses of glucose or aglycone, in the structures of these compounds.

Copyright # 2005 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2005; 19: 813–817
816 K. Li et al.

Table 1. Precision and accuracy results for Rg3 in dog

plasma (3 days, 6 replicates per day)
Nominal plasma Mean measured Relative Intra-day Inter-day
concentration concentration error RSD RSD
(ng/mL) (ng/mL) (%) (%) (%)

1.00 0.98
1.5 4.2 5.5

5.00 5.03 0.6 6.4 1.0
16.0 15.8
1.3 7.9 4.6
4.00 4.04 1.0 3.8 2.6
40.0 39.8
0.5 3.1 1.0
320 324 1.4 3.7 3.4

of plasma for Rg3. The lower limit of quantitation (LLOQ),

defined as the lowest concentration analyzed with accuracy
within 15% and a precision 15%, was 0.5 ng/mL in
200 mL of plasma or 2 ng/mL in 100 mL of plasma. For Rh2,
the ranges were 2–20 and 10–400 ng/mL in plasma.

Precision and accuracy

Precision and accuracy of the assay were determined by repli-
cate analyses (n ¼ 6) of QC samples at three concentrations,
by performing the complete analytical runs on the same
day and also on three consecutive days. The data from these
Figure 3. MRM chromatograms for (A) drug-free plasma; QC samples were examined by a one-way analysis of var-
(B) plasma (100 mL) spiked with 2 ng/mL of Rg3, 10 ng/mL of iance (ANOVA). The intra-day and inter-day precisions
Rh2 and 100 ng/mL of IS; and (C) a plasma sample (118 ng/ were less than 8% for each QC level of Rg3. The accuracy,
mL) 0.5 h after i.v. administration of 0.3 mg/kg of Rg3 to a dog. determined from QC samples, was within 2% for each QC
I: Rg3 (m/z 783.8 ! 160.8); II: IS (m/z 867.5 ! 721.5); III: level. The results are summarized in Table 1.
Rh2 (m/z 621.7 ! 160.8).
Extraction recovery and stability
The extraction recoveries of analytes were determined by
Assay specificity comparing the peak area of each analyte in plasma samples
The specificity of the method was demonstrated by compar- that had been spiked with the analytes prior to extraction
ing MRM chromatograms for Rg3, Rh2 and the IS for a drug- with those for samples to which the analytes had been added
free plasma sample, a spiked plasma sample, and a plasma
sample from a dog 0.5 h after i.v. administration. As shown
in Fig. 3, no significant peaks interfering with analytes were
observed in the drug-free rat plasma.
Matrix effects from co-eluting endogenous substances
provide another possible source of problems regarding assay
specificity, although matrix-matched calibration standards
were used. The ion suppression effect was evaluated by
comparing the peak areas of Rg3 (10 ng/mL), Rh2 (10 ng/
mL) and the IS (10 ng/mL) in six QC samples with those of
standard solutions that had been prepared in the same way as
the QC samples except that water was substituted for drug-
free plasma. For Rg3 and Rh2 the mean peak areas from the
six QC samples had relative errors of 5.2 and 3.8%,
respectively, when compared with that for these standard
solutions. For the IS, the relative error was
2.4%. These
observations indicate that no endogenous substances sig-
nificantly influenced the ionization of these analytes.

Linearity of calibration curves and lower

limit of quantitation
Linear calibration curves with correlation coefficients greater Figure 4. Mean plasma concentration-time profiles of
than 0.997 were obtained in the concentration ranges 0.5– ginsenoside Rg3 after (A) an oral dose of 2 mg/kg and (B)
20 ng/mL in 200 mL of plasma and 2–400 ng/mL in 100 mL an i.v. dose of 0.3 mg/kg (n ¼ 6).

Copyright # 2005 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2005; 19: 813–817
 LC/MS/MS for pharmacokinetics of 20(R)-ginsenoside Rg3 817

Table 2. The main pharmocokinetic parameters of After the oral dose of 2 mg/kg, the mean maximum plasma
ginsenoside Rg3 after oral and i.v. administrations to six concentration (standard deviation (SD)) of Rg3 was 7.30
dogs (mean  SD) (1.67) ng/mL occurring at 2.5 h post-dose. The mean
Administration mode
apparent plasma half-life (SD) was 5.99 (1.16) h. The mean
(SD) area under the plasma concentration versus time
Parameter Oral Intravenous curve was 45.9 (12.8) ng  h/mL. Following the i.v. dose of
0.3 mg/kg, the mean apparent plasma half-life (SD) was
Cmax (ng/mL) 7.30  1.67 1.71 (0.11) h, and the mean (SD) area under the plasma
Tmax (h) 2.50  0.84
t1/2 (h) 5.99  1.16 1.71  0.11 concentration versus time curve was 1458 (437) ng  h/mL.
ke (1/h) 0.12  0.02 0.42  0.05 The aglycone of Rg3, protopanaxadiol, was also checked by
AUC0-t (ng  h/mL) 45.9  12.8 1458  437 a fully developed LC/MS/MS method which cannot be
AUC0-1 (ng  h/mL) 50.0  13.1 1466  444 revealed due to commercial confidentiality. However, in this
MRT (h) 6.55  1.40 0.85  0.18 work, protopanaxadiol was not detected in plasma. Also, we
Vd (L/kg) 722  257 0.53  0.16
CL (ml/min/kg) 362  91.4 3.65  0.99 did not detect the glucuronides of Rg3, Rh2 and protopanax-
adiol. The results from an excretion study of Rg3 showed that
more than 70% of the dose was recovered in the form of
parent drug in bile after an i.v. dose. Therefore, it appears that
post-extraction. The results showed that the extraction recov- Rg3 is metabolized to a low degree in vivo.
eries of Rg3 were 87.1, 83.1 and 81.5% at concentrations of 1, 5
and 16 ng/mL, and 85.6, 75.2 and 78.3% at concentrations of 4,
CONCLUSIONS
40 and 320 ng/mL, respectively. For Rh2, the extraction
recoveries were 86.4 and 83.8% at concentrations of 5 and In conclusion, the present optimized method was validated
16 ng/mL, and 76.0 and 76.8% at concentrations of 40 and to guarantee a reliable determination of ginsenoside Rg3 in
320 ng/mL, respectively. The extraction recoveries of the IS dog plasma. It was then successfully applied to a pharmaco-
were 82.3 and 75.0% at concentrations of 10 and 100 ng/mL. kinetic study of Rg3 after both oral and intravenous adminis-
The stability of Rg3 and Rh2 in dog plasma was trations. No Rh2 or protopanaxadiol was detected in plasma.
investigated under a variety of storage and process condi-
tions. The analytes were found to be stable (RE < 2%) in dog
plasma after three freeze (
208C)/thaw (room temperature) REFERENCES
cycles. The analytes were also shown to be stable after 24 h of 1. Mochizuki M, Yoo YC, Matsuzawa K, Saiki I, Tonooka S,
storage in plasma (RE < 1%) and in reconstitution solutions Samukawa KI, Azuma I. Bio. Pharm. Bull. 1995; 18: 1197.
(RE < 2%) at room temperatures. 2. Shinkai K, Akedo H, Mukai M, Imamure F, Isoai A,
Kobayashi M, Kitagawa I, Jpn. J. Cancer Res. 1996; 87: 357.
3. Wang H, Zou H, Kong L, Zhang Y, Pang H, Su C, Liu G, Hui
Pharmacokinetic studies of Rg3 M, Fu L. J. Chromatogr. B 1999; 731: 403.
The plasma samples from the pharmacokinetic study were 4. Cai Z, Qian T, Wong RNS, Jiang ZH. Anal. Chim. Acta 2003;
492: 283.
analyzed by the present LC/MS/MS method. Rg3 was suc- 5. Wang Y, Liu TH, Wang W, Wang BX. Acta Pharm. Sinica 2000;
cessfully determined, but no Rh2 was detected in plasma 35: 284.
after the oral and i.v. administrations. Figure 4 shows mean 6. Ma JS, Zhou JL, Fei XF, Sun Y, Wang BX. Acta Pharm. Sinica
2001; 36: 603.
plasma concentration-time curves of Rg3 (n ¼ 6); the pharma- 7. Bae EA, Han MJ, Choo MK, Park SY, Kim DH. Biol. Pharm.
cokinetic parameters are presented in Table 2. Bull. 2002; 25: 58.

Copyright # 2005 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2005; 19: 813–817