Journal of Chromatography B, 877 (2009) 1698–1704

Contents lists available at ScienceDirect

Journal of Chromatography B
journal homepage: www.elsevier.com/locate/chromb

A rapid and robust liquid chromatography/tandem mass spectrometry method

for simultaneous analysis of anti-tuberculosis drugs—Ethambutol and
pyrazinamide in human plasma
Zhilong Gong ∗ , Yousef Basir, David Chu, Melanie McCort-Tipton
Covance Bioanalytical Laboratory Service, LLC, 8211 Scicor Drive, Suite B, Indianapolis, IN 46214, United States

a r t i c l e i n f o a b s t r a c t

Article history: Ethambutol and pyrazinamide are two first-line anti-tuberculosis drugs. Though they are normally
Received 3 December 2008 combined for the treatment, their highly different polarity complicates simultaneous liquid chromatog-
Accepted 14 April 2009 raphy/tandem mass spectrometry (LC/MS/MS) analysis of these two drugs in human plasma with decent
Available online 18 April 2009
peak shape and retention. Here we report a rapid and robust LC/MS/MS method for the simultaneous
determination of these two drugs in human plasma. Human plasma samples, together with the iso-
Keywords:
topically labeled internal standards were extracted using protein precipitation, and then separated on a
Ethambutol
Chromolith SpeedROD RP-18e column and detected with mass spectrometry. The mobile phase is 0.1%
Pyrazinamide
Human plasma
trifluoroacetic acid in water and 0.1% trifluoroacetic acid in methanol. Addition of trifluoroacetic acid in
Liquid chromatography the mobile phases was found to be able to improve peak shape as well as to increase the retention of
Mass spectrometry ethambutol, thus being able to analyze these two drugs at the same time with both drugs having decent
peak shape and enough retention on a C18 column. An atmospheric pressure chemical ionization inter-
face was chosen to reduce ion suppression from sample matrix components and provide high sensitivity.
The standard curve range was 10.0–5000 ng/mL for ethambutol and 50.0–25,000 ng/mL for pyrazinamide
using a plasma sample volume of 50.0 ␮L. This method has a very short run time of 3.8 min. The method
has been fully validated, and <15% relative standard deviation was obtained for both analytes.
© 2009 Elsevier B.V. All rights reserved.

1. Introduction drugs, it is desirable to develop an analytical method to allow both

drugs to be quantified simultaneously in human plasma.
Tuberculosis remains one of the most common infectious dis- The purpose of this project was to develop a liquid chromatog-
eases, which caused death of millions of people each year [1,2]. raphy/tandem mass spectrometry method for the simultaneous
Tuberculosis is normally treated with combination therapy due quantitation of ethambutol and pyrazinamide in human plasma.
to rapid development of resistance when single drug is used The method was expected to have a short run time, thus enabling
[2]. Ethambutol and pyrazinamide, together with isoniazid and efficient analysis of large number of plasma samples obtained for
rifampicin, are first-line drugs for tuberculosis treatment. These pharmacokinetics studies, or for therapeutic drug monitoring with
four drugs are normally combined in the first few months treatment combinational drug therapy containing ethambutol and pyrazi-
to avoid development of resistance. Then the number is usually namide.
reduced to two drugs for the remainder of the treatment based on
drug sensitivity testing that is usually available by this time in the 2. Experimental
course. Ethambutol and pyrazinamide combination is also one of
the recommended treatments for latent tuberculosis infection [3]. 2.1. Materials and reagents
A number of quantitative methods for the determination of anti-
tuberculosis drugs in biological matrices have been reported [4–8]. Ethambutol dihydrochloride and pyrazinamide were purchased
However, many of the reported methods have limitations such as from Sigma–Aldrich (St. Louis, MO). The internal standards,
low sensitivity or long chromatographic run times. For the com- ethambutol-d4 dihydrochloride and pyrazinamide-d3 , were from
bination of ethambutol and pyrazinamide with or without other Covance in-house synthesis (Covance, Madison, WI). Acetonitrile
(HPLC grade), formic acid (96%) and trifluoroacetic acid (1-mL
ampule) were all purchased from Sigma–Aldrich (St. Louis, MO).
∗ Corresponding author. Methanol (HPLC grade) was purchased from Fisher Scientific (Fair
E-mail address: [email protected] (Z. Gong). Lawn, NJ). Water was purified by a Barnstead system.

1570-0232/$ – see front matter © 2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.jchromb.2009.04.023
 Z. Gong et al. / J. Chromatogr. B 877 (2009) 1698–1704 1699

2.2. Instrumentation

The high-performance liquid chromatography (HPLC) system

consists of a Shimadzu SIL-HTc autosampler, LC-10ADVP solvent
delivery system and a column heater (Columbia, MD). The mass
spectrometer was an Applied Biosystems MDS Sciex (Toronto,
Canada) API-4000 triple quadrupole mass spectrometer. An atmo-
spheric pressure chemical ionization (APCI) source was used. Data
was collected and processed using Sciex Analyst 1.1 data collection
and integration software.

2.3. Chromatographic conditions

The HPLC column was a Chromolith SpeedROD RP-18e

(50 mm × 4.6 mm) from Merck (Darmstadt, Germany). Column
temperature was held at 25 ◦ C. The mobile phase A was 0.1% triflu-
oroacetic acid in water, and mobile phase B was 0.1% trifluoroacetic
acid in methanol. The gradient was as follows: 2% mobile phase B
was first held for 0.5 min, then mobile phase B was raised up to 15%
in 1.7 min, at 2.21 min, mobile phase B was adjusted to 85% and held
at this level until 2.9 min, at 2.91 min, mobile phase B was switched
back to 2% until 3.8 min. The flow rate was 1.00 mL/min. A typical
injection volume was 5 ␮L.

2.4. MS/MS detection

Precursor ions for analytes and internal standards were deter-

mined from mass spectra obtained by teeing-in the neat solution of
each individual compound into an LC stream and then into the API-
4000 mass spectrometer. Using APCI source, the mass spectrometer
was operated in the positive ionization mode. With Q1 full scan,
protonated molecular ions of all compounds were observed. Each of
the precursor ions were subjected to collision-induced dissociation
to determine the resulting product ions. The determined multiple Fig. 1. Product ion mass spectra of ethambutol (A), and pyrazinamide (B).
reaction monitoring (MRM) transitions were 205 → 116 for etham-
butol, 124 → 81 for pyrazinamide, 209 → 120 for ethambutol-d4 ,
and 127 → 84 for pyrazinamide-d3 . Fig. 1A and B shows the frag- intermediate standard solutions of 125/625, 100/500, 62.5/313,
mentation of ethambutol and pyrazinamide, respectively. In these 20.0/100, 5.00/25.0, and 2.00/10.0 ␮g/mL into blank human
two figures, the parent masses and their fragmentation patterns are plasma to provide concentrations (ethambutol/pyrazinamide) of
clearly shown. 5000/25,000, 4000/20,000, 2500/12,500, 800/4000, 200/1000,
Interface independent parameters and heated nebulizer param- 70.0/350, 20.0/100 and 10.0/50.0 ng/mL. The quality control plasma
eters were optimized during the infusion of a solution of all samples were prepared at 25,000/125,000, 3750/18,800, 400/2000,
compounds with HPLC mobile phase. These were (arbitrary units if 30.0/150 and 10.0/50.0 ng/mL (ethambutol/pyrazinamide) by spik-
not specified) CAD: 8, Gas 1: 30, CUR: 30, CE: 23 V for ethambutol ing human plasma with the second sets of standard stock solutions,
and ethambutol-d4 and 26 V for pyrazinamide and pyrazinamide- which were for dilution QC preparation, and the intermediate
d3 , current: 3 mA, ionization source temperature: 500 ◦ C, and dwell standard solutions of 250/1250, 40.0/200 and 2.50/12.5 ␮g/mL.
time: 150 ms. Unit resolution was used for both Q1 and Q3. For the spiking, typically, the intermediate standard solutions of
20.0–200 ␮L were spiked into 5.00 mL of human plasma. Inter-
2.5. Preparation of standards and quality control samples nal standard stock solutions of 100 ␮g/mL of ethambutol-d4 and
pyrazinamide-d3 each were prepared in methanol. Working inter-
Stock standard solutions for ethambutol and pyrazinamide were nal standard solution was prepared in 1.25% formic acid in
prepared in duplicate and compared using the appropriate weight acetonitrile containing 125 ng/mL of each of ethambutol-d4 and
for the lot purity, moisture, and salt correction. The calibration pyrazinamide-d3 .
standards and the quality control (QC) samples were prepared
from separate stock standard solutions. Stock standard solutions of 2.6. Sample preparation
ethambutol of 2.50 mg/mL and pyrazinamide of 12.5 mg/mL were
prepared in 50% methanol in water. Intermediate working solu- The sample pre-treatment is based on protein precipitation.
tions (ethambutol/pyrazinamide) of 125/625, 100/500, 62.5/313, 50.0 ␮L of blank matrix, control-0 (blank with internal standard),
20.0/100, 5.00/25.0 and 2.00/10.0 were prepared from one set QC samples and calibration standard samples were acidified with
of stock standard solutions and were used for calibration stan- 50.0 ␮L of 5% formic acid aqueous solution, after vortex-mixing,
dard preparation. Intermediate standard solutions of 250/1250, 400 ␮L of 1.25% formic acid in acetonitrile were added to all blank
40.0/200 and 2.50/12.5, prepared from another set of stock stan- matrix samples, 400 ␮L of intermediate internal standard solution
dard solutions, were for quality control samples preparation. They was added to all other samples. This mixture was then vortex-mixed
were all prepared in 50% methanol in water by dilution from vigorously for ∼2 min, and then centrifuged at 3000 rpm at room
the stock standard solutions. The working calibration standard temperature for approximately 5 min. 100 ␮L of the supernatant
human plasma samples were prepared by spiking corresponding was transferred and dried down under nitrogen stream at 45 ◦ C.
1700 Z. Gong et al. / J. Chromatogr. B 877 (2009) 1698–1704

a freshly prepared calibration curve and six replicates of QC-LLOQ,

low QC, mid QC and high QC. Dilution QC of six replicates were
also included in one batch to evaluate impact of the dilution pro-
cess to precision and accuracy to ensure that dilution of study
samples will not affect accuracy and precision. Dilution QC con-
centration was at five times the upper limit of the standard curve.
They were diluted 10× into calibration curve with blank matrix.
By Covance SOP, these three consecutive batches should pass all
criteria. Failure of any batch will fail the whole validation. Intra-
assay precision was calculated by obtaining the relative standard
deviation (RSD) of the six replicates of each QC level, and intra-
assay accuracy was calculated by averaging the accuracies of six
replicates of each QC level against the fresh curve. For inter-assay,
precision was calculated by obtaining the RSD of all 18 replicates
at each QC level from all three batches, accuracy was obtained by
averaging the accuracies of all 18 replicates at each QC level from
Fig. 2. Chemical structure of ethambutol (A) and pyrazinamide (B). all three batches. Acceptable accuracy and precision were ≤15.0%
accuracy and ≤15.0% RSD at every concentration level except for
The residue was then reconstituted with 400 ␮L of 0.1% formic acid the LLOQ where ≤20.0% accuracy and ≤20.0% RSD were accept-
in methanol:water (5:95, v:v). able.
For QCs used for validation, Covance SOP does not specify
2.7. Validation procedure whether they should be freshly prepared or can also be stored. In
practice, both freshly prepared QCs and stored QCs can be used.
The whole validation is based on Covance Bioanalytical Method For stored QCs, they were typically used within two weeks of stor-
Validation SOP. age. Due to usage of stored QCs, the subsequent long-term matrix
stability should cover the storage of this period of time. In some
2.7.1. Calibration curve and linearity cases when client’s SOPs were used, the SOP specified QCs were
The eight-point calibration curve was constructed by plot- then used.
ting peak area ratio of ethambutol and pyrazinamide to their
corresponding internal standard versus ethambutol and pyrazi- 2.7.3. Selectivity
namide concentrations. A linear regression with weighing factor Blank samples from six different lots of blank matrix were pro-
of 1/x2 was applied. The concentrations of the standards were then cessed without internal standard to evaluate presence of interfering
back-calculated. Standards with back-calculated concentrations of peaks. Interfering peaks at the retention time of interest should not
85.0–115.0% of their theoretical values were considered acceptable. exceed 20.0% of LLOQ. At maximum, one in six lots is allowed to
have more than 20.0% of LLOQ interference.
2.7.2. Intra-assay and inter-assay precision and accuracy These blank matrix lots were also separately spiked at mid QC
To evaluate intra-assay and inter-assay precision and accu- levels for both ethambutol and pyrazinamide in the same way as
racy, three consecutive batches were done. Each batch contained regular mid QC preparation, and single replicate of each QC lot were

Fig. 3. Representative chromatogram of ethambutol (70.0 ng/mL) (left) and its internal standard (right) in human plasma.
 Z. Gong et al. / J. Chromatogr. B 877 (2009) 1698–1704 1701

Fig. 4. Representative chromatogram of pyrazinamide (350 ng/mL) (left) and its internal standard (right) in human plasma.

extracted as regular QCs. Regular QC acceptance criteria also apply final concentrations of both compounds and their internal stan-
to the selectivity QCs. dards in the final extract for mid QC. These samples refer to post
extraction spike samples. Water was aliquotted in triplicates and
extracted and reconstituted in the same way as the above post
2.7.4. Recovery
extraction spike samples. These samples refer to pure solution sam-
To evaluate the loss of drugs and/or internal standards during
ples. The area counts for all the compounds from the post extraction
sample preparation, additional blank matrix samples were also
spike samples were then compared with the area counts from the
extracted without addition of internal standard. At reconstitution
pure solution samples to evaluate the matrix effect. Matrix effect
step during extraction, instead of adding reconstitution solutions,
is calculated as follows: mean peak area from the post extraction
three solutions containing all drugs and internal standards at con-
spike samples subtracts the mean peak area from the pure solution
centrations that mimic the concentrations of low QC, mid QC and
samples, then divided by mean peak area from the pure solution
high QC in their final extracts were added. Three replicates for
samples and multiplied by 100. Positive value indicates percent
each QC level were prepared. The drugs and internal standards area
enhancement, and negative value indicates percent suppression.
counts from these samples were compared with corresponding pre-
cision/accuracy QCs of six replicates to evaluate any loss of either
drugs or internal standards. No acceptance criteria were applied to
2.7.6. Stability
this parameter, but it is preferable to observe consistent recovery
Stability of both drugs in different matrices and under different
for all three QC levels.
conditions was evaluated. The detailed tests are described below.
Room temperature matrix stability was assessed by leaving
2.7.5. Matrix effect low and high QC samples at room temperature for 41 h before
To evaluate matrix effect, blank matrix samples from six dif- extraction. Freeze/thaw matrix stability was assessed by repeatedly
ferent lots were processed without internal standard. At the freezing/thawing low and high QC samples for three cycles before
reconstitution step during extraction, instead of using reconstitu- extraction. The initial freezing was at least 24 h before first thawing.
tion solution, a neat solution was added. The neat solution was The second and third freezings were at least 12 h. Each thawing step
prepared by spiking both compounds and their internal standards was around 1 h at room temperature. For both room temperature
in the reconstitution solution at concentrations that mimic the and freeze/thaw matrix stability evaluation, six replicates of each

Table 1
Precisions and accuracies of calibration standards of ethambutol in human plasma from three validation batches.

Analysis group Theoretical concentration (ng/mL)

10.0 20.0 70.0 200 800 2500 4000 5000

Measured concentration (ng/mL)

001 10.6 17.3 74.8 201 781 2590 4070 4880

002 9.69 20.7 76.1 FC 809 2510 3900 4590
003 9.85 20.1 73.2 222 832 2510 3740 4370

n 3 3 3 2 3 3 3 3
Mean 10.0 19.4 74.7 212 807 2540 3900 4610
RSD (%) 4.9 9.3 1.9 NA 3.2 1.8 4.2 5.6
Accuracy 100.0 97.0 106.7 106.0 100.9 101.6 97.5 92.2

FC: failed acceptance criteria.

1702 Z. Gong et al. / J. Chromatogr. B 877 (2009) 1698–1704

Table 2
Precisions and accuracies of calibration standards of pyrazinamide in human plasma from three validation batches.

Analysis group Theoretical concentration (ng/mL)

50.0 100 350 1000 4000 12500 20000 25000

Measured concentrations (ng/mL)

001 50.3 99.5 340 1040 3710 13100 21000 24300

002 49.2 103.0 350 FC 3990 12700 20100 24100
003 52.7 87.6 350 1130 4350 12900 18900 21700

n 3 3 3 2 3 3 3 3
Mean 50.7 96.7 347 1090 4020 12900 20000 23400
RSD (%) 3.5 8.3 1.7 NA 8.0 1.6 5.3 6.2
Accuracy 101.4 96.7 99.1 109.0 100.5 103.2 100.0 93.6

FC: failed acceptance criteria.

QC level were tested, and these two tests were included in a preci- can be seen that ethambutol is extremely polar, this should explain
sion and accuracy batch. To evaluate processed-sample re-injection why it did not have much retention on a C18 column and caused
reproducibility, a precision and accuracy batch, after initial injec- the peak tailing. In the end, still with a C18 column, trifluoroacetic
tion that passed all criteria, the batch was then stored at 2–8 ◦ C for acid was added in the mobile phase, a decent peak shape as well
73 h and re-injected. Long-term frozen matrix stability was evalu- as stronger retention was then obtained for ethambutol. This peak
ated by storing low mid and high QC samples at both −10 to −30 ◦ C shape improvement and increase of retention could be interpreted
and −60 to −80 ◦ C storage conditions for 56 days before extraction. in two ways: first, addition of trifluoroacetic acid might suppress
Six replicates of each QC level, a freshly prepared calibration curve the ionization of the hydroxyl groups on the silica column, thus
as well as two replicates of freshly prepared low, mid and high QC reduced the polar interaction between ethambutol and the nega-
samples that served as batch acceptance QCs were used for long- tively charged hydroxyl groups on the silica, resulting in a better
term frozen matrix stability evaluation. Stability of stock standard peak shape, and second, trifluoroacetic acid in the mobile phase
solutions and intermediate standard solutions was also evaluated might serve as an ion-pairing agent, which paired with etham-
by comparing the stored stock and intermediate standard solutions butol and increased its retention on C18 column. With addition
to freshly prepared stock and intermediates solutions. of trifluoroacetic acid in the mobile phases, both ethambutol and
pyrazinamide can be analyzed on a C18 column simultaneously.
The typical chromatograms for ethambutol and pyrazinamide are
3. Results and discussions
shown in Figs. 3 and 4.
3.1. Liquid chromatography separation
Table 4
During method development, it was observed that ethambutol Precisions and accuracies of quality control samples for pyrazinamide in human
and pyrazinamide showed distinctively different chromatographic plasma.
behavior. For example, on a C18 column, when using widely used
Analysis Statistics Theoretical concentration (ng/mL)
mobile phase components like formic acid, water, methanol, ace- group
tonitrile etc., ethambutol showed very weak retention on this type 50.0 150 2000 18800
of column, causing it being eluted into the injection salt valley, caus-
n 6 6 6 6
ing potential ion suppression. In addition, it showed serious peak Intra-assay mean 44.1 129 1950 18800
tailing. Pyrazinamide showed decent peak shapes as well as decent 001
RSD (%) 8.8 5.6 6.3 3.4
retention. From the chemical structure as shown in Fig. 2A and B, it Accuracy (%) 88.24 86.0 97.5 100.0

n 6 6 6 6
Intra-assay mean 53.8 140 1860 18200
Table 3 002
RSD (%) 5.8 5.5 5.8 8.5
Precisions and accuracies of quality control samples for ethambutol in human Accuracy (%) 107.6 93.3 93.0 96.8
plasma.
n 6 6 6 6
Analysis Statistics Theoretical concentration (ng/mL) Intra-assay mean 54.2 144 2280 16100
group 003
RSD (%) 10.2 5.7 1.4 4.2
10.0 30.0 400 3750 Accuracy (%) 108.4 96.0 114.0 85.6

n 6 6 6 6 n 18 18 18 18
Intra-assay mean 9.04 27.3 427 4050 Inter-assay mean 50.7 138 2030 17700
001 Overall
RSD (%) 14.7 8.8 5.2 4.0 RSD (%) 12.4 7.0 10.1 8.6
Accuracy (%) 90.4 91.0 106.8 108.0 Accuracy (%) 101.4 92.0 101.5 94.1

n 6 6 6 6
Intra-assay mean 11.4 30.6 416 3950
002 Table 5
RSD (%) 8.7 8.6 5.2 7.8
Accuracy (%) 114.0 102.0 104.0 105.3 Precisions and accuracies of dilution QCs for ethambutol and pyrazinamide in human
plasma.
n 6 6 6 6
Intra-assay mean 10.4 31.0 432 3440 Analysis Statistics Theoretical concentration (ng/mL)
003 group
RSD (%) 11.6 3.7 2.8 5.0
Accuracy (%) 104.0 103.3 108.0 91.7 Ethambutol, 25000 Pyrazinamide, 125000

n 18 18 18 18 n 6 6
Inter-assay mean 10.3 29.6 425 3810 Intra-assay mean 24600 135000
Overall 001
RSD (%) 14.5 9.0 4.5 9.1 RSD (%) 2.9 2.7
Accuracy (%) 103.0 98.7 106.3 101.6 Accuracy (%) 98.4 108.0
 Z. Gong et al. / J. Chromatogr. B 877 (2009) 1698–1704 1703

Table 6
Recoveries at low, mid and high QC levels for ethambutol, ethamtubol-d4 , pyrazinamide and pyrazinamide-d3 in human plasma.

QC levels Statistics Ethambutol Ethamtubol-d4 Pyrazinamide Pyrazinamide-d3

a
RSD (%) 5.4/5.9 3.0/1.2 10.5/2.0 8.7/1.7
Low
Recovery (%) 119.9 106.5 44.8 47.2

RSD (%)a 4.0/2.9 2.7/1.6 3.8/0.2 3.0/0.1

Mid
Recovery (%) 102.2 96.0 39.9 41.0

RSD (%)a 2.8/2.1 3.9/2.5 3.5/0.2 3.9/0.6

High
Recovery (%) 99.5 95.7 50.2 45.9

Overall Recovery (%) 107.2 99.4 45.0 44.7

a
RSD: RSD of six replicates of extracted QC samples/RSD of three replicates of recovery samples.

Table 7
Matrix effect for ethambutol, ethamtubol-d4 , pyrazinamide and pyrazinamide-d3 in human plasma.

Sample type Post extraction spike Pure solution Post extraction spike Pure solution

Compound Ethambutol Ethambutol-d4

RSD (%)a 2.9 2.8 2.4 1.7
Matrix effect (%) 15.4 13.1

Compound Pyrazinamide Pyrazinamide-d3

RSD (%)a 5.2 0.9 5.0 0.8
Matrix effect (%) −6.4 −6.2
a
RSD for post extraction spike is from six different matrix lots with one replicate of each, and RSD for pure solution is from three replicates.

Trifluoroacetic acid is well known for its ion suppression. To around 200–1000 ng/mL for ethambutol and 1000–100,000 ng/mL
evaluate its effect on ethambutol and pyrazinamide, the chosen for pyrazinamide, which were obtained from a cohort study by
trifluoroacetic acid mobile phases were compared with formic McIlleron et al. [9].
acid mobile phases, which have exactly the same solvent and
same percentage of acid in the mobile phases. Under same liq-
uid chromatography and mass spectrometry conditions, compared 3.3. Precision and accuracy
to formic acid, trifluoroacetic acid showed ∼7% enhancement for
ethambutol and ∼27% enhancement for pyrazinamide. Tables 3 and 4 show the inter- and intra-assay precision and
Under the optimized HPLC and MS conditions, ethambutol and accuracy. The method was found to be highly accurate and precise.
pyrazinamide were baseline separated with retention times of 1.26 For ethambutol, accuracy of 90.4–114.0% and precision of 2.8–14.7%
and 2.26 min, respectively. RSD for intra-assay, and accuracy of 98.7–106.3% and precision of
4.5–14.5% RSD for inter-assay were obtained for all QC levels includ-
ing LLOQ. For pyrazinamide, accuracy of 86.0–114.0% and precision
3.2. Calibration curve and linearity of 1.4–10.2% RSD for intra-assay, and accuracy of 92.0–101.5% and
precision of 7.0–12.4% RSD for inter-assay were obtained for all QC
For three consecutive batches, the calibration curves showed levels including LLOQ.
an overall accuracy of 92.2–106.7% with RSD of less than 10%. Table 5 shows the precision and accuracy of dilution QCs for
The detailed results are shown in Tables 1 and 2. The applied both ethambutol and pyrazinamide. Accuracy of 98.4% with an RSD
ranges for ethambutol and pyrazinamide, together with the dilu- of 2.9% for ethambutol and accuracy of 108.0% with an RSD of 2.7%
tion QC levels, fully cover the clinical concentration levels of for pyrazinamide were obtained.

Table 8
Stability of ethambutol and pyrazinamide in human plasma under different conditions.

Statistics Ethambutol Pyrazinamide

Low QC Mid QC High QC Low QC Mid QC High QC

Three freeze/thaw matrix stability

n 6 6 6 6
RSD (%) 4.1 NA 3.1 2.9 NA 2.4
Accuracy (%) 103.0 91.2 98.0 87.8

41 h room temperature matrix stability

n 6 6 6 6
RSD (%) 5.8 NA 1.8 3.9 NA 0.9
Accuracy (%) 106.0 92.8 95.3 87.8

56 days −10 to −30 ◦ C matrix stability

n 6 6 6 6 6 6
RSD (%) 3.8 2.1 0.9 4.1 2.2 1.8
Accuracy (%) 108.3 110.0 103.7 89.3 96.0 89.4

56 days −60 to −80 ◦ C matrix stability

n 6 6 6 6 6 6
RSD (%) 2.4 2.3 1.1 3.1 0.8 1.9
Accuracy (%) 108.3 109.3 102.1 90.0 95.5 88.8
1704 Z. Gong et al. / J. Chromatogr. B 877 (2009) 1698–1704

3.4. Selectivity injection reproducibility have been established. In addition, 58 days

stability for stock standard solutions and 54 days for intermediate
In all the tested six lots of blank matrix samples, the retention standard solutions stored at 2–8 ◦ C were established. All of these
regions of ethambutol, pyrazinamide and their internal standards demonstrate the robustness and ruggedness of the method.
were all free of significant interference peaks. For selectivity QCs,
accuracy of 96.5–114% with an RSD of 5.8% was obtained for etham- 4. Conclusions
butol while 102.8–118% accuracy and 5.0% RSD were seen for
pyrazinamide. Addition of trifluoroacetic acid in mobile phase greatly improved
the peak shape as well as increased the chromatographic retention
3.5. Recovery of ethambutol, enabling simultaneous analysis with pyrazinamide
together on a C18 column. Thus, a rapid and robust LC/MS/MS
As shown in Table 6, overall recovery of 107.2% for ethambu- method has been developed for these two anti-tuberculosis drugs
tol, 99.4% for ethambutol-d4 , 45.0% for pyrazinamide and 44.7% for in human plasma. This method has been fully validated and can be
pyrazinamide-d3 were obtained. Both compounds show consistent directly applied to patient sample analysis.
recovery results for all three QC levels. Pyrazinamide showed much
lower recovery compared to ∼100% for ethambutol. Since it is con- Acknowledgement
sistent at all three QC levels for both drug and internal standard, no
further investigation was done for the cause. The author would like to thank Covance Organic Synthesis lab
for providing the internal standards for both ethambutol and pyraz-
3.6. Matrix effect inamide.

As shown in Table 7, matrix effect of 15.4% for ethambutol, References

13.1% for ethambutol-d4 , −6.4% for pyrazinamide and −6.2% for
pyrazinamide-d3 were observed. All six matrix lots showed very [1] A. Kochi, Tubecle 72 (1991) 1.
[2] B. Blomberg, S. Spinaci, B. Fourie, R. Laing, Bull. World Health Organ. 79 (1) (2001)
similar matrix effect for both analytes and their corresponding
1.
internal standards. [3] A.B. Younossian, T. Rochat, J.-P. Ketterer, J. Wacker, J.-P. Janssens, Eur. Respir. J. 26
(3) (2005) 462.
[4] X. Chen, B. Song, H. Jiang, K. Yu, D. Zhong, Rapid Commun. Mass Spectrom. 19
3.7. Stability (2005) 2591.
[5] M.Y. Khuhawar, F.M.A. Rind, J. Chromatogr. B 766 (2002) 357.
Stability of ethambutol and pyrazinamide in human plasma [6] S. Guermouche, M.H. Guermouche, J. Chromatogr. Sci. 40 (2004) 250.
[7] P.J. Smith, J. van Dyk, A. Fredericks, Int. J. Tuberc. Lung Dis. 3 (11) (1999) 5325.
under different conditions were evaluated. The detailed results are [8] G. Bhutani, S. Singh, S. Vir, K.K. Bhutani, R. Kumar, A.K. Chakraborti, K.C. Jindal, J.
shown in Table 8. As seen from the table, three freeze/thaw cycles, Pharm. Biomed. Anal. 43 (2007) 1213.
41 h room temperature storage, 56 days storage at −10 to −30 ◦ C, [9] H. McIlleron, P. Wash, A. Burger, J. Norman, P.I. Folb, P. Smith, Antimicrob. Agents
Chemother. 50 (4) (2006) 1170.
56 days storage at −60 to −80 ◦ C, and 73 h processed-sample re-