development of a low volume plasma sample precipitation procedure for liquid chromatography/tandem...
TRANSCRIPT
RAPID COMMUNICATIONS IN MASS SPECTROMETRY
Rapid Commun. Mass Spectrom. 2005; 19: 2131–2136
Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/rcm.2040
Development of a low volume plasma sample
precipitation procedure for liquid chromatography/
tandem mass spectrometry assays used for drug discovery
applications
Xiaoying Xu*, Qiao Zhou and Walter A. KorfmacherDepartment of Drug Metabolism and Pharmacokinetics, Schering-Plough Research Institute, Kenilworth, NJ 07033, USA
Received 25 February 2005; Revised 2 June 2005; Accepted 2 June 2005
The demand for high sensitivity bioanalytical methods has dramatically increased in the drug dis-
covery stage; in addition, there has been a growing trend of reducing the sample volume that is
required for these assays. A sensitive high-performance liquid chromatography/tandem mass spec-
trometry (HPLC/MS/MS) procedure has been developed and tested to meet these needs. The assay
requires only a low plasma sample volume (10mL) and employs a protein precipitation procedure
using a 1:6 plasma/acetonitrile ratio. The supernatant is injected directly into the LC/MS/MS system
using the selected reaction monitoring (SRM) procedure for detection. A generic HPLC gradient
based on a methanol/water mobile phase with a flow rate set to 0.8mL/min was used. The test meth-
od showed very good linearity between 0.1–1000ng/mL (R2¼ 0.9737), precision (%RSD¼ 6–9),
accuracy (%RE¼�2) and reproducibility (%RSD¼ 11). A drug discovery IV/PO study was assayed
using both the new low volume method and our standard volume (50mL) method. The correlation
of the two sets of data from the two methods was excellent (R2¼ 0.9287). This new assay procedure
has been successfully used in our laboratory for over 100 different rat or mouse discovery PK stu-
dies. Copyright # 2005 John Wiley & Sons, Ltd.
Many pharmaceutical companies are investing large
amounts of money in their research and development pro-
grams. The main driving force is the need to discover new
drugs that can cure health problems and be marketable to
allow the companies to eventually recover their investments
and make a profit. Drug metabolism and pharmacokinetics
(DMPK) research departments play a critical role in the
lead optimization phase of drug discovery by helping to
improve the potential candidate compound before it is
selected for development.1,2 Various absorption, distribu-
tion, metabolism and excretion (ADME) and pharmacoki-
netic (PK) screens are now key steps in early drug
discovery.1–5 Developing fast and cost-effective analytical
methods for ADME and PK studies has become a crucial
element for major pharmaceutical companies. Using high-
performance liquid chromatography combined with tandem
mass spectrometry (HPLC/MS/MS) has proven to be the
bioanalytical technique of choice for most of these ADME
and PK screens that are now used in the lead optimization
phase of drug discovery.1,4,6,7
During the drug discovery phase, the DMPK studies are
often performed in small rodents (mouse and rat). With
multiple requests from different assays (e.g., biological
testing, PK study, metabolism, toxicity, etc.), the available
plasma sample volumes for the bioanalytical analysis are
often very small.8 In recent years, the typical amount of
plasma that bioanalytical scientists have used for their assays
has dropped from 200 to 50 mL.9–13 Recent publications have
suggested that there is a need to decrease the sample volume
yet further.14–16 On the other hand, as new chemical entities
show increased potency, lower dosage of a drug in the
animals requires methods that have even higher sensitivity.
In the last decade, the limit of quantitation (LOQ) required for
a discovery PK bioanalytical assay was in the 10 ng/mL or
higher range.6,8 More recently, a LOQ of 1–10 ng/mL (based
on 50–200mL of plasma sample) has been needed for typical
drug discovery PK applications.7,17 For some highly potent
compounds, an LOQ of 0.1 ng/mL may be required even in a
drug discovery PK setting.
Another important aspect of drug discovery is that
discovery bioanalytical methods must be developed in a
short timeframe, with good levels of precision and a high
certainty of success. Protein precipitation is the simplest
approach for removing the majority of the protein
matrix.6,7,18 Protein precipitation has the advantage that it is
a generic procedure that works for most compounds; it has
the disadvantage that it results in a sample that is not as clean
as those produced by other extraction techniques. Fortu-
nately, because of the high selectivity of MS/MS with the
selected reaction monitoring (SRM) setting, protein precipi-
tation is normally an acceptable sample preparation proce-
dure for a discovery PK assay.6,7 By monitoring not only the
Copyright # 2005 John Wiley & Sons, Ltd.
*Correspondence to: X. Xu, Schering-Plough Research Institute,2015 Galloping Hill Road, K-15-2-2945, Kenilworth, NJ 07033,USA.E-mail: [email protected]
specific precursor mass of the drug ion but also a character-
istic product ion, interference from other matrix components
within the sample can be eliminated in most cases, although
matrix effects can still be an issue.6,7,17 Once an analytical
method has been developed, it is desirable that the method
performance remains reasonably consistent over time. The
results generated based on the established method should be
relatively free from systematic error and the relative error
should be kept to acceptable limits.7
Currently, for most higher-throughput bioanalytical
assays used in a drug discovery setting, 96-well plates are
used for the sample-handling procedures.7 These 96-well
plates have the advantage that they are relatively easy to
work with whether one is performing manual or robotic
sample-handling procedures. While some efforts have been
made to switch to 384-well plates for bioanalytical applica-
tions, these plates can be difficult to work with when the
operation is not fully automated. Therefore, our goal was to
demonstrate that one could use 96-well plates to perform
discovery PK assays with a 10-mL plasma sample in a routine
manner.
In this report, a new HPLC/MS/MS bioanalytical proce-
dure is described that is based on using very low volumes of
plasma samples (10mL). A partial validation of the procedure
was performed to ensure that the new method was robust
and reproducible. As a validation exercise, a drug discovery
IV/PO PK study was assayed using both the previous
standard method (based on a 50-mL sample volume) and the
new low volume procedure (based on a 10-mL sample
volume) to confirm the reliability of the new method.
EXPERIMENTAL
ReagentsFor sample preparation, acetonitrile (Otima; Fisher Scientific,
Pittsburgh, PA, USA) was used. For the high-performance
liquid chromatography (HPLC) mobile phase, methanol
(Optima; Fisher Scientific), ammonium acetate and acetic
acid (glacial, 99.99þ%; Aldrich Chemical Co., Inc., St. Louis,
MO, USA) were used. Deionized water was purified using a
compact ultrapure water system (EASYpure UV; Fisher
Scientific).
InstrumentationThe HPLC system consisted of Shimadzu (Columbia, MD,
USA) LC-10ADvp pumps with a Leap (Carrboro, NC, USA)
HTS PAL autosampler. A Phenomenex (Torrance, CA, USA)
Synergi Max reversed-phase HPLC column (C18,
30� 2.0 mm i.d., 5 mm) was used as the analytical column.
The mobile phase consisted of A (MeOH/H2O 20:80, 0.01 M
ammonium acetate, pH 6.0) and B (0.01 M ammonium acetate
in MeOHþ 0.6 mL/L 10% acetic acid, pH 6.0). A 1.5-min gra-
dient from 5% B to 95% B was employed at a flow rate of
0.8 mL/min. Under these conditions, the test compound
and the internal standard (IS) eluted at 0.97 and 1.02 min,
respectively. The HPLC eluant passed through a divert valve
and was then introduced directly into the source of the mass
spectrometer.
The HPLC/MS/MS measurements were performed using
a ThermoFinnigan (San Jose, CA, USA) Quantum triple-
quadrupole mass spectrometer. The mass spectrometer was
operated in the positive electrospray ionization (ESI) mode
with a spray voltage of 4000 V, capillary temperature of
3508C, sheath gas 80 psi, and auxiliary gas 20 psi. The MS/MS
measurements were performed using the SRM mode. The
SRM transitions were selected by subjecting the protonated
molecules to collision-induced dissociation (CID) in the
collision cell where the argon gas was maintained at a
constant pressure of 1.3 mTorr. The product ions were
monitored by scanning the third quadrupole of the MS/MS
system.6,7,17
Calibration standards, IS and quality controlsamples (QCs)In the standard volume 96-well plate procedure, the stock
solution of the test compound was prepared as a 1 mg/mL
solution in methanol and diluted with methanol to make
working solutions at 0.01, 0.1, 1, 10 and 100 ng/mL. The IS
was prepared in acetonitrile at a final concentration of
0.01 ng/mL. The low, medium and high QCs (1, 50,
1000 ng/mL) were prepared by spiking known quantities of
the working solutions (10–50 mL) into control rat plasma to
make a final volume of 1 mL and stored in a freezer at
�208C. The calibrators were prepared on the day of each
run. Calibration standards (0, 0.1, 0.25, 0.5, 1, 2.5, 5, 10, 25,
50, 100, 250, 500 and 1000 ng/mL) were prepared by spiking
known quantities of the working solutions (10–50mL) into
control rat plasma to make a final volume of 1 mL.
In the low volume procedure, the stock solution of the test
compound was prepared as a 1 mg/mL solution in methanol
and diluted with methanol to make the standard working
solutions at 0.002, 0.005, 0.01, 0.02, 0.05, 0.1, 0.2, 0.5, 1, 2, 5, 10
and 20 ng/mL. The IS was prepared in acetonitrile with a final
concentration of 0.01 ng/mL. The low, medium and high QCs
(1, 50, 1000 ng/mL) were prepared by spiking 5mL of the
appropriate standard working solution into 95mL of control
rat plasma to make a final volume of 100 mL and stored in a
freezer at�208C. The calibrators were prepared on the day of
each run. Calibration standards (0, 0.1, 0.25, 0.5, 1, 2.5, 5, 10,
25, 50, 100, 250, 500 and 1000 ng/mL) were prepared by
spiking 5mL of the appropriate working solution into 95mL of
control rat plasma to make a final volume of 100 mL.
Sample preparationSample preparation in the standard volume 96-well plate
procedure has been described previously.7,13 Briefly, aliquots
of the plasma samples (50mL) were placed into a Strata 96-
well plate (AHO-7193; Phenomenex, CA, USA) and 150 mL
of the IS solution (0.01 ng/mL of IS in acetonitrile) were added
to each well (the protein precipitation ratio was 1:3). After
vortexing for 30 s, the plate was centrifuged (Eppendorf
5810; Westbury, NY, USA) for 10 min at 4000 rpm. Finally,
the supernatant was transferred to a second standard 96-
well plate (AHO-7192; Phenomenex) using the Tomtec
(Hamden, CT, USA) Quadra 96 instrument. For the assay,
an aliquot of the supernatant (5 mL) was injected into the
HPLC/MS/MS system for the assay.
Sample preparation using the low sample volume proce-
dure required only 10 mL of the plasma samples. The plasma
samples (e.g., standards, QCs, study samples) were placed
Copyright # 2005 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2005; 19: 2131–2136
2132 X. Xu, Q. Zhou and W. A. Korfmacher
into a v-bottom 96-well plate (AB-1058; ABgene House, UK)
and then IS solution (60mL; 0.01 ng/mL of IS in acetonitrile)
was added to each well (the protein precipitation ratio was
1:6). After vortexing for 30 s, the plate was centrifuged
(Eppendorf 5810) for 10 min at 4000 rpm. Finally, the super-
natant was transferred to a second v-bottom 96-well plate
(AB-1058) using the Tomtec Quadra 96 instrument. An
aliquot of the supernatant (5 mL) was injected into the
HPLC/MS/MS system for the assay.
Method testingThe quantitation range was from 0.1–1000 ng/mL. Two cali-
bration curves over this range were prepared on three sepa-
rate days. Each analytical accuracy and precision run
included calibration standards in duplicate within the 0.1–
1000 ng/mL range. The low, medium and high QCs (1, 50
and 1000 ng/mL) were assayed in replicates of six on three
separate days.
RESULTS AND DISCUSSION
The limit of quantitation (LOQ)The standard sample preparation method routinely used in
the laboratory is protein precipitation with a sample-to-sol-
vent ratio of 1:3. The test compound, blank control plasma,
0.1 ng/mL and 0.5 ng/mL standard plasma samples were
prepared via the standard volume (50mL plasma) or the
new low volume (10 mL plasma) sample preparation proce-
dure. After injecting these samples into the HPLC/MS/MS
system, the resulting mass chromatograms are shown in
Figs. 1–3. The expected retention time for the test compound
was 0.97 min.
In control plasma, the standard volume (50mL plasma)
procedure (protein precipitation 1:3 ratio) gave a background
peak intensity of 4.00 E2 counts per second (cps) (Fig. 1).
Using the low volume (10mL plasma) sample preparation
procedure (protein precipitation 1:6 ratio) showed a ten-fold
lower background intensity of 3.18 E1 cps (Fig. 1), which
clearly indicated that the low volume procedure resulted in a
significantly reduced level of background interference for
this test compound.
Analysis of the prepared 0.1 and 0.5 ng/mL standard
plasma samples by HPLC/MS/MS following the two sample
preparation procedures resulted in an impressive difference
between the two procedures. As shown in Fig. 2(A), there was
significant background interference and the compound was
not baseline separated from the chemical noise at the 0.1 ng/
mL concentration. As shown in Fig. 3(A), at 0.5 ng/mL, the
signal-to-noise (S/N) ratio achieved was 10 and it was set as
the LOQ for the standard volume (50mL plasma) sample
preparation procedure. When the same samples (0.1 and
0.5 ng/mL) were analyzed by the new low volume (10mL
plasma) sample preparation procedure, the results were
improved compared to the standard procedure. As shown in
Fig. 2(B), at the 0.1 ng/mL concentration of the test
compound, a S/N ratio of 10 was obtained, which resulted
in a five-fold lower LOQ for this assay. Even though the
absolute peak intensity of the test compound was smaller in
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Time (min)
0
10
20
30
40
50
60
70
80
90
100
Rel
ativ
eA
bund
ance
RT: 0.97
RT: 0.97
NL: 4.00E2 (1: 3)
NL: 3.18E1 (1: 6)
Figure 1. Normalized mass chromatograms of blank control
plasma obtained using protein precipitation ratios of 1:3 and
1:6.
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Time (min)
0
20
40
60
80
100
Rel
ativ
eA
bund
ance
RT: 0.99
NL: 4.85E2
RT: 0.97
NL: 3.84E2
0
20
40
60
80
100
A (1:3)
B (1:6)
Figure 2. Normalized mass chromatograms of test com-
pound (0.1 ng/mL) in plasma obtained using protein pre-
cipitation ratios of 1:3 (A) and 1:6 (B).
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
Time (min)
0
20
40
60
80
1000
20
40
60
80
100R
elat
ive
Abu
ndan
ceRT: 0.98
NL: 1.62 E3
RT: 0.97
NL: 8.81 E2
A (1:3)
B (1:6)
Figure 3. Normalized mass chromatograms of test com-
pound (0.5 ng/mL) in plasma obtained using protein pre-
cipitation ratios of 1:3 (A) and 1:6 (B).
Low volume plasma sample precipitation procedure for LC/MS/MS 2133
Copyright # 2005 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2005; 19: 2131–2136
the new low volume sample preparation procedure than in
the standard volume sample preparation procedure, the new
procedure resulted in a much lower noise level, thereby
increasing the S/N ratio. For comparison, Fig. 3(B) shows the
results obtained with the new low volume (10mL plasma)
sample preparation procedure for the 0.5 ng/mL sample.
Therefore, even though the absolute analyte peak intensity
dropped by a factor of two (as would be expected due to the
higher dilution), the new low volume sample procedure
provided a much higher S/N ratio due to the decrease in the
matrix signal.
Linearity of the standard curvesThe low plasma volume calibration curve showed a very
good linear range up to 1000 ng/mL (Fig. 4). The curves
were fit by the linear regression method using a 1/x2 weight-
ing. The correlation coefficient (R2) was 0.9737, which met our
discovery analytical criteria.7
ReproducibilityThe reproducibility of the low volume assay was performed
using either stock solution or a plasma standard with a con-
tinuous series of 96 and 192 injections, respectively. Figures 5
and 6 show the distribution of the injections vs. peak area
ratio of test compound and IS. The standard deviations
were 7 and 11%, respectively, which meets the criteria for
reproducibility of a bioanalytical assay.
Precision and accuracyThe precision of the method was defined as the percent rela-
tive standard deviation (%RSD) calculated from replicate
measurements. The accuracy of the assay was defined as
the percent relative error (%RE) of the mean of the replicate
measurements from the theoretical values. Precision and
accuracy were determined by analyzing QCs prepared at
three concentrations (1, 50, 1000 ng/mL) with six replicates
on three separate days. Tables 1 and 2 summarize the intra-
day and inter-day precision and accuracy for the new low
volume assay. The intra-day precision and accuracy were
between 5 to 6% and �7 to 6% (RE), respectively. The inter-
day precision and accuracy were between 8 to 9% and �4 to
3% (RE), respectively. All intra- and inter-day precision and
accuracy values were acceptable and spanned the entire con-
centration range, which indicated the assay was very robust,
reproducible and could be used to accurately quantify the
biological samples in a drug discovery environment.
ApplicationThe new low volume method was tested and compared with
the previous standard volume method in one drug discovery
PK study. The plasma samples were collected at 0.117, 0.25,
0.5, 1, 2, 4, 6, 8, and 24 h post-IV dose and at 0.25, 0.5, 1, 2, 4,
6, 8, 24 h post-PO dose. There were three animals in the IV and
PO groups, respectively. A total of 51 plasma samples (27 IV
and 24 PO samples) were analyzed under both the standard
and the low volume methods. Figures 7 and 8 show represen-
tative plasma concentration time profiles after IV and PO dos-
ing in animals using the standard and the low volume
methods. In both IV and PO dosages, the same PK profiles
were observed regardless of the sample preparation proce-
dure. Figure 9 shows that a very good correlation of the
y = 0.00156x - 0.0006R2 = 0.9737
0.0
0.4
0.8
1.2
1.6
2.0
0 200 400 600 800 1000 1200
Concentration (ng/mL)
Pea
kA
rea
Rat
io
Figure 4. Calibration curves of test compound (0.1–
1000 ng/mL) using the low volume method.
0
1
2
3
4
5
0 10 20 30 40 50 60 70 80 90 100
Injection
Rat
io
Figure 5. Reproducibility of the stock solution of a test compound using the low volume method.
2134 X. Xu, Q. Zhou and W. A. Korfmacher
Copyright # 2005 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2005; 19: 2131–2136
Table 2. Inter-day precision and accuracy of test compound
in rat plasma in the low volume assay (n¼ 6 per day, 3 days)
Precision (%RSD) Accuracy (%RE)
1 ng/mL 9 �450 ng/mL 8 �31000 ng/mL 9 3Mean 9 �2
Table 1. Intra-day precision and accuracy of the test
compound in rat plasma in the low volume assay (n¼ 6)
Precision (%RSD) Accuracy (%RE)
1 ng/mL 5 �750 ng/mL 6 61000 ng/mL 6 �5Mean 6 �2
0
1
2
3
4
5
0 20 40 60 80 100 120 140 160 180 200
Injection
Rat
io
Figure 6. Reproducibility of plasma standard of a test compound using the low volume method.
1
10
100
1000
0 5 10 15 20 25 30Time (h)
Plasma concentration (ng/mL)
Standard method
Low volume method
Figure 7. Representative plasma concentration time profile
after IV dose of a test compound in animal #2 (standard
method results vs. low volume method results).
1
10
100
1000
0 5 10 15 20 25 30Time (h)
Plasma concentration (ng/mL)
Standard method
Low volume method
Figure 8. Representative plasma concentration time profile
after PO dose of a test compound in animal #4 (standard
method results vs. low volume method results).
y = 1.0762x - 5.2276R2 = 0.9287
0
100
200
300
400
500
0 100 200 300 400 500Regular Plate
v-B
ott
omP
late
Figure 9. Correlation of the plasma concentrations (n¼ 51)
after IV and PO doses using the standard method (regular
plate) vs. the low volume method (v-bottom plate).
Low volume plasma sample precipitation procedure for LC/MS/MS 2135
Copyright # 2005 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2005; 19: 2131–2136
plasma concentrations (n¼ 51) was found for the two
methods; the correlation coefficient (R2) was 0.9287.
This new assay procedure has been successfully used in
our laboratory for over 100 different rat or mouse discovery
PK studies. The main advantage of this new low volume
assay is that smaller sample volumes can be utilized. We have
found this to be very helpful for discovery PK studies based
on the mouse in which study sample volumes are often small.
A second advantage of this new low sample volume
procedure is that only 10% of the control plasma is required
to make the calibration standards. The previous procedure
required 14 mL of control plasma for the calibration
standards while this new procedure requires only 1.4 mL of
control plasma—this translates into a significant saving when
a laboratory assays hundreds of discovery PK studies each
year.
CONCLUSIONS
The demand for high sensitivity bioanalytical methods,
coupled with a current trend of reducing the sample volume,
has dramatically increased in the drug discovery stage. As
drugs become more potent, the level of sensitivity of a bioa-
nalytical method has to be continuously increased and a low-
er LOQ is then required. On the other hand, the available
sample volume has often decreased so that a method based
on 50 mL of plasma is no longer preferred. Therefore, a low
sample volume (10 mL) HPLC/MS/MS method was devel-
oped to meet these needs. The new method was tested and
partially validated to ensure that the method’s performance
remained reasonably consistent over time and had acceptable
reproducibility. The good reliability and robustness of the
procedure was demonstrated using a drug discovery PK
study. Not only did this method provide excellent sensitivity
and linearity for quantitation of biological samples, but also it
reduced the control plasma costs to conduct the study due to
the lower volume of control plasma required for the calibra-
tion curve preparation.
REFERENCES
1. Korfmacher WA. Curr. Opin. Drug Discov. Develop. 2003; 6:481.
2. Eddershaw PJ, Beresford AP, Bayliss MK. Drug Discov.Today 2000; 5: 409.
3. Jenkins KM, Angeles R, Quintos MT, Xu R, Kassel DB,Rourick RA. J. Pharm. Biomed. Anal. 2004; 34: 989.
4. Kassel DB. Curr. Opin. Chem. Biol. 2004; 8: 339.5. Thompson TN. Curr. Drug Metab. 2000; 1: 215.6. Korfmacher WA, Cox KA, Bryant MS, Veals J, Ng K, Lin CC.
Drug Discov. Today 1997; 2: 532.7. Korfmacher WA. Bioanalytical analysis in a drug discovery
environment. In Using Mass Spectrometry for Drug Metabo-lism Studies, Korfmacher WA (ed). CRC Press: Boca Raton,2005; 1–34.
8. Mallet CR, Lu Z, Fisk R, Mazzeo JR, Neue UD. RapidCommun. Mass Spectrom. 2003; 17: 163.
9. Allen GD, Brookes ST, Barrow A, Dunn JA, Grosse CM.J. Chromatogr. B Biomed. Sci. Appl. 1999; 732: 383.
10. Parise RA, Sparrow BR, Merrill JW, Grossi IM, Covey JM,Peggins JO, Egorin MJ. J. Chromatogr. B Anal. Technol.Biomed. Life Sci. 2004; 810: 35.
11. Liang HR, Foltz RL, Meng M, Bennett P. J. Chromatogr. BAnal. Technol. Biomed. Life Sci. 2004; 806: 191.
12. Beaudette P, Bateman KP. J. Chromatogr. B Anal. Technol.Biomed. Life Sci. 2004; 809: 153.
13. Korfmacher WA, Cox KA, Ng KJ, Veals J, Hsieh Y,Wainhaus S, Broske L, Prelusky D, Nomeir A, White RE.Rapid Commun. Mass Spectrom. 2001; 15: 335.
14. Humphries D, Ruterbories K, Chan C, Narayanan R.J. Chromatogr. B Anal. Technol. Biomed. Life Sci. 2004; 810: 229.
15. Heinig K, Bucheli F. J. Pharm. Biomed. Anal. 2004; 34: 1005.16. Guo P, Wang X, Zhou F, Gallo JM. J. Chromatogr. B Anal.
Technol. Biomed. Life Sci. 2004; 809: 273.17. Xu X, Veals J, Korfmacher WA. Rapid Commun. Mass Spec-
trom. 2003; 17: 832.18. Rossi DT, Sinz M. Mass Spectrometry in Drug Discovery.
Marcel Dekker, Inc.: New York, 2002.
2136 X. Xu, Q. Zhou and W. A. Korfmacher
Copyright # 2005 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2005; 19: 2131–2136