Journal of Chromatography A, 1426 (2015) 226232

Contents lists available at ScienceDirect

Journal of Chromatography A
journal homepage: www.elsevier.com/locate/chroma

An on-line stacking capillary electrophoresis method for the analysis

of 9 -tetrahydrocannabinol and its metabolites
Hui-Ling Cheng a , Yi-Hsuan Tsai b , Wan-Ling Hsu b , Yi-Hui Lin b,
a
Pharmaceutical Optimization Technology Division, Biomedical Technology and Device Research Laboratories, Industrial Technology Research Institute,
Hsinchu 300, Taiwan
b
School of Pharmacy, College of Pharmacy, China Medical University, Taichung 404, Taiwan

a r t i c l e

i n f o

Article history:
Received 7 August 2015
Received in revised form 31 October 2015
Accepted 9 November 2015
Available online 14 November 2015
Keywords:
Large-volume sample stacking
LVSS
Anion selective exhaustive injection
ASEI
Sweeping
Urine
Abused drug
On-line preconcentration

a b s t r a c t
The objective of this study was to establish a practical and reliable analytical method for monitoring
trace amounts of 9 -tetrahydrocannabinol (THC) and its metabolites in biological samples. A novel
on-line preconcentration capillary electrophoresis method combining large volume sample injection,
anion selective exhaustive injection and sweeping was developed to enhance analytical sensitivity.
A background buffer composed with 30 mM phosphate buffer (pH 2.5) containing 40% methanol and
100 mM SDS was used to suppress the electroosmotic ow of the uncoated fused silica capillary
(40 cm 50 m i.d.). High conductivity buffer (200 mM phosphate, pH 2.5) was injected for analyte accumulation. The samples, prepared in phosphate buffer or Tris buffer, were introduced by hydrodynamic
injection and electrokinetic injection. After sweeping, the separation was performed in micellar electrokinetic chromatography (MEKC) mode at 15 kV. During the method validation, the coefcient of
determination of the regression curve was measured at greater than 0.993, and the relative standard
deviation and relative error were lower than 11.06% and 9.24%, respectively. Under optimized conditions, an improvement of up to 2000-fold higher sensitivity was achieved. This method was applied to
the analysis of urine samples, indicating that it could be satisfactorily utilized in the toxicological and
clinical monitoring of cannabis.
2015 Elsevier B.V. All rights reserved.

1. Introduction
Cannabis is one of the most commonly abused drugs in the
world with its increased use due to a low perceived risk [1]. In
Taiwan, cannabis is classied as a second-class drug of abuse and
its use is strictly banned. The nationwide statistical data from
the National Health Administration in 2014 showed that cannabis
ranked 8th in the most commonly abused drugs in Taiwan. Compared to 2013, cannabis abuse increased by 4.5%, particularly with
young people. In some countries, cannabis was used as medicine to
relieve pain, promote appetite in AIDS patients, reduce nausea and
vomiting in patients undergoing chemotherapy, and treat neurodegenerative diseases such as Parkinsons and Huntingtons diseases
[2,3]. However, long-term use of cannabis may cause problems
in the respiratory tract, neurological and cardiovascular systems,
and lead to physical dependence. Withdrawal symptoms include

Corresponding author at: School of Pharmacy, College of Pharmacy, China Medical University, No. 91, Hsueh-Shih Rd., Taichung 40402, Taiwan.
E-mail address: yihulin@mail.cmu.edu.tw (Y.-H. Lin).
http://dx.doi.org/10.1016/j.chroma.2015.11.032
0021-9673/ 2015 Elsevier B.V. All rights reserved.

irritability, restlessness, anxiety, sleep disorders and appetite

disorders [4].
Delta-9-tetrahydrocannabinol (THC) is the main psychoactive
component in cannabis. In humans, THC is metabolized primarily in the liver by the cytochrome P450 isozyme system [5]. More
than 100 metabolites of THC have been identied [6]. Among
them, 11-hydroxy-9 -tetrahydrocannabinol (THC-OH) is the most
important metabolite of THC and is equipotent to THC in its
psychoactivity [7]. THC-OH is further metabolized to 11-nor-9carboxy-9 -tetrahydrocannabinol (THC-COOH), which does not
possess a psychotropic effect but dose exhibit anti-inammatory
and analgesic properties [8]. The major metabolic pathways of THC
are shown in Fig. 1. More than 55% of the THC is excreted in feces,
and 2035% is eliminated in the urine. The most abundant metabolites in urine are THC-COOH and its glucuronide, while only trace
amounts of THC-OH are present [7]. The detection of THC-COOH in
urine can be used to prove cannabis use.
The analysis of THC and its metabolites was routinely conducted through the use of immunoassays and gas chromatography
mass spectrometry (GCMS). Immunoassays are subject to interferences and may generate false positive screening results [9]. Gas

H.-L. Cheng et al. / J. Chromatogr. A 1426 (2015) 226232

227

Fig. 1. Major metabolic routes of 9 -tetrahydrocannabinol (THC) in humans.

chromatography exhibit good sensitivity and specicity, but timeconsuming pre-column derivatization is unavoidable [10,11]. The
rst application of capillary electrophoresis (CE) in the analysis
of cannabinoids in hair was reported using non-aqueous CE with
electrochemical detection (ECD) [12]. With the high selectivity
of ECD, the lower limit of detection of THC can be decreased to
37 ng/mL. A CE-MS method was reported for the direct analysis
of THC-COOH and its glucuronide in urine. The total analysis time
including sample-pretreatment was approximately 30 min [13].
In spite of many advantages, such as low consumption of
reagents, simplicity and high-efciency, CE has lower sensitivity
due to the short optical length and small sample size. Many
on-line preconcentration methods have been developed in order
to increase sensitivity. Several to thousand-folds improvement
in detection sensitivity was achieved without any alternation to
the current instrumentation. Sweeping mode was introduced in
1998 with a focusing effect that was due to the pseudostationary
phase penetrating the sample zone and picking and accumulating analytes [14]. The combination of sweeping with other
preconcentration methods provides an additive effect on sensitivity. Various methods including dynamic pH junctionsweeping
[15,16], selective-exhaustive injectionsweeping (SEI-sweep)
[17,18], transient trapping (tr-trapping) [19] and analyte focusing
by micelle collapse (AFMC) [20] were successfully established,
which expanded the application of CE to many elds. Among
these methods, SEI-sweep showed the greatest improvements to
sensitivity [21].
Owing to extensive metabolism of cannabinoids after consumption, trace amounts of the unchanged drug and various
metabolites were found in the body. We attempted to apply
the anion selective-exhaustive injectionsweeping micellar electrokinetic chromatography (ASEI-sweeping-MEKC) model for the
analysis of THC and its major metabolites. Although all analytes
were baseline separated, the detection sensitivity of THC and THCOH was unsatisfactory. In solution, analytes with varying degrees

of dissociation may affect the individual sampling efciency

during electrokinetic injection. To overcome this limitation, we
developed a method named large volume sample injection-anion
selective exhaustive injectionsweeping (LVSI-ASEI-sweeping) for
the on-line preconcentration of THC, THC-OH and THC-COOH.
The factors that inuence sensitivity and selectivity were carefully evaluated. The developed method may be complementary
to immunoassays or GC-MS for fast screening of cannabis use in
forensic analysis and may be applicable in toxicological and clinical
monitoring.
2. Materials and methods
2.1. Materials and chemicals
THC, its metabolites, THC-OH and THC-COOH, and indomethacin
as an internal standard (IS) were purchased from SigmaAldrich (St. Louis, MO, USA). Sodium dihydrogen phosphate
monohydrate (NaH2 PO4 H2 O), sodium dodecyl sulfate (SDS),
methanol (MeOH), n-hexane, ethyl acetate (EA), sodium
hydroxide (NaOH), acetonitrile (ACN), tetrahydrofuran (THF),
tris(hydroxymethyl)aminomethane (Tris) were all analytical grade
and purchased from Merck (Darmstadt, Germany). Milli-Q water
(Millipore, Bedford, MA, USA) was used for preparing buffer and
related aqueous solution.
2.2. Preparation of solutions
Stock solutions of analytes were obtained in MeOH at the
concentration of 1 mg/mL. They were suitably diluted with the
prepared Tris buffer (25 mM, pH 10.3) and 30% ACN (v/v) to
afford electrokinetical sample solutions. For hydrodynamical sample solutions, stock solutions were diluted with phosphate buffer
(30 mM, pH 2.5) to analytical concentrations.

228

H.-L. Cheng et al. / J. Chromatogr. A 1426 (2015) 226232

Fig. 2. Mechanism of the on-line focusing by LVSI-ASEI-sweeping.

2.3. Pretreatment and extraction of urine samples

3. Results and discussion

Blank urine samples were obtained from healthy volunteers

in our laboratory and were stored at 20 C until analysis. A
40 L stock solution of each analyte was added to a drug-free
urine sample (880 L). The spiked urine samples were treated
with 50 L of 10 M NaOH and incubated for 15 min at 60 C for
the hydrolysis of THC-COOH-glucuronide. Afterwards, 67 L of
acetic acid and 85% phosphoric acid (20 L) were added for acidication. The extraction of urine samples was performed in the
following way: a urine sample (300 L) was added with 1 mL
n-hexane/EA (8:2), then vortexed for 3 min, and centrifuged at
10,000 rpm for 10 min. The supernatant (900 L) was evaporated
by a centrifugal vaporizer (EYELA CVE-200D, Japan). The residue
was reconstituted with either Tris or phosphate buffer before
analysis.

3.1. LVSI-ASEI-sweeping MEKC model

2.4. CE system
A Beckman P/ACE MDQ System equipped with a photo diode
array detector was used. This CE method was performed in an
uncoated fused-silica capillary (Polymicro Technologies, Phoenix,
AZ, USA) of 50 m i.d. and 30 cm effective length (40 cm, total
length). The UV absorbance wavelength was set at 214 nm. New
capillaries were preconditioned prior to use with MeOH for 10 min,
water for 10 min, 1 M HCl solution for 10 min, water for 10 min,
1 M NaOH solution and water for 10 min. Before daily analysis
and between each run, the capillary was rinsed with 0.1 M NaOH
for 5 min, water for 5 min, and running buffer for 5 min. A pressure of 20 psi was used for all the preconditioning and rinsing
steps.
General electrophoresis procedures are described as follows.
The background buffer consisting of 30 mM phosphate (NaH2 PO4 ,
pH 2.5), 100 mM SDS and 40% MeOH (v/v) was used to rinse the
capillary. Next, high conductivity buffer (HCB, phosphate buffer
200 mM, pH 2.5) was injected at 0.5 psi for 1 min. After the HCB,
a large volume sample was hydrodynamically introduced (0.5 psi,
3 min), followed by electrokinetically loading of additional analyte
for 6 min at a voltage of 2.5 kV. Finally, the inlet of capillary was
switched into the background buffer and a voltage of 15 kV was
applied for separation.

The concept of the LVSI-ASEI-sweeping MEKC model is illustrated in Fig. 2. Initially, the capillary was rinsed with acidic
background buffer (pH 2.5) to suppress the electroosmotic ow
(EOF). A plug of the high-conductivity buffer (HCB) was then hydrodynamically injected into the capillary (Fig. 2A). In this study,
samples were introduced by combining two injection modes. First,
a large volume sample prepared in phosphate buffer at pH 2.5 was
loaded with pressure (at 0.5 psi, 3 min) after the HCB, as seen in
Fig. 2B. Then, a negative voltage (2.5 kV) was applied for 6 min,
and an additional amount of anionic analytes prepared in Tris buffer
(pH 10.3) were electrokinetically injected. Due to the different dissociation states in both adjoining electrolytes, the mobility levels
of the anionic analytes gradually decreased when entering the capillary and further accumulated in the acidic sample zone (Fig. 2C).
After sample injection, the inlet of the capillary was switched to the
background buffer and a negative voltage was applied. The anionic
micelles entered the capillary from the cathodic end and swept
and stacked the sample into a narrow band (Fig. 2D). Finally, the
separation was performed by MEKC mode, as shown in Fig. 2E.
3.2. Optimization of LVSI-ASEI-sweeping MEKC model
Several parameters were evaluated before achieving higher resolution and optimal conditions, including the pH and concentration
of the phosphate buffer, modier amounts and the concentration
of SDS in the background buffer, as well as the optimization of the
HCB plug, sample matrix composition, sample injection time and
separation voltage.
3.2.1. Optimization of sample injection
Relative to THC-COOH (pKa 4.5), THC and THC-OH are weaker
acids with pKa values of approximately 9.6, and thus, it is hard
for them to compete with THC-COOH while using electrokinetic
injection. In contrast, pressure injection provides equal probability for all analytes. This study combines LVSI and ASEI modes to
increase sample loading for improved sensitivity. In order to avoid
the dissociation of analytes and to suppress EOF, samples used for

H.-L. Cheng et al. / J. Chromatogr. A 1426 (2015) 226232

2
0.02 AU

1
3

Abs.

(D) 6 min

(C) 5 min

(B) 4 min

(A) 0 min

10 11 12 13 14 15

Time (min)
Fig. 3. Electropherograms of various sample electrokinetic injection times at
2.5 kV. Sample concentrations: THC, THC-OH 1 g/mL; THC-COOH 0.4 g/mL. CE
Conditions: background buffer, 30 mM phosphate (pH 2.5) with 40% methanol and
100 mM SDS; HCB, 200 mM phosphate, 0.5 psi for 1 min; dynamic injection, 0.5 psi
for 3 min; separation voltage, 15 kV. Migration order, 1: THC, 2: THC-COOH, 3:
THC-OH.

hydrodynamic injection were diluted with 30 mM phosphate buffer

(pH 2.5). All analytes should be neutral at this pH. The sample injection time of LVSI at 0.5 psi was investigated from 0 to 4 min, with
an estimated sample length of approximately 018 cm. The peak
heights of analytes increased with increasing injection times up
to 3 min; however, peak broadening was observed as the injection
time was increased to 4 min. With longer sample injection times,
more analyte can be introduced; however, the consequences of
additional sample volume may be band broadening and decreased
resolution, rather than increased intensity. Therefore, a sample
injection time of LVSI was set for 3 min at 0.5 psi for further studies.
In ASEI, factors such as sample matrix composition, injection
voltage, and injection time were studied. The properties of the sample matrix such as alkalinity and conductivity have notable effects
on the sensitivity of ASEI [22]. The amount of analyte injected
depends on the dissociation state of the compound. To investigate
the effect of different matrices on sampling, the analytes were prepared with 25 mM sodium borate, 0.1 M NaOH or 25 mM Tris. The

229

highest signal response was obtained when Tris was used as the
sample matrix. As a result, 25 mM Tris was chosen as the buffer
for the sample matrix in our subsequent analyses. Previous studies have shown that ACN has a signicant effect on the sample
stacking efciency [2325]. Thus, the addition of ACN in the range
of 040% (v/v) to the sample matrix was also examined in our
system. The result indicated that the presence of ACN increased
the stacking ability, with 30% ACN providing the best improvement. Therefore, we added 30% ACN (v/v) to the sample matrix
in the following experiments. When considering the inuence of
injection voltage and injection time on ASEI, we found that variations in injection voltages (2.5, 3, 4, and 5 kV) did not result
in a signicant change in signal intensity, yet the analyte peak
response increased with increasing electrokinetic injection time
(Fig. 3). However, peak broadening was also observed when the
injection time was increased to 7 min. Taken together, based on
the detection sensitivity and resolution, samples used for ASEI were
electrokinetically injected into the capillary at 2.5 kV for 6 min in
the LVSI-ASEI MEKC mode.
3.2.2. Optimization of sweeping and separation conditions
The proposed use of a preinjection plug was reported in 1998
[26]. The HCB has a high conductivity and low electronic eld that
reduced the electrophoretic velocities of analytes and micelles.
Therefore, it allows for the injection of sample ions for a long period
of time and creates a narrow stacked zone [21]. The inuence of HCB
(200 mM phosphate buffer with pH 2.5, 1 min by 0.5 psi) on separations was evaluated. It was found that, the peak intensity appeared
to increase when an HCB plug was applied; therefore, an HCB plug
was used in the subsequent experiment. According to the formula
derived from the sweeping model Isweep = Iinj [1/(1 + k)],where k is
the retention factor, and Isweep and Iinj are the length of zones after
sweeping and sample injection [27], higher concentrations of SDS
may result in larger retention factors and shorter sweep zones. SDS
also plays an important role in the separation. The effects of the
concentration of SDS were studied in the range of 75150 mM. As
anticipated, the total analysis time was shortened, and peak heights
were raised as the concentration of SDS increased. Nevertheless,
poor resolution resulted from higher SDS concentrations. Hence,
100 mM SDS was chosen for use in further experiments.
In addition, the separation of analytes was strongly affected by
the composition of background buffer. Several parameters such as
the concentration (20100 mM) and pH (2, 2.5 and 3) of phosphate
buffer, the identity and amount of added organic modier, were
studied. Organic solvents showed a considerable effect on separations in these experiments. The purpose of adding organic solvent
to background buffer is to adjust the polarity of the buffer in order
to affect the partition coefcient and selectivity of micelles as well
as to exert EOF inhibition. After the stacking process, the analytes
were well concentrated but migrated together in the absence of an
organic modier (Fig. 4). Various organic solvents, such as MeOH,
ACN and THF, were added to the background buffer to improve resolution. Improved resolution and sensitivity were obtained when
MeOH was added as the buffer modier. Moreover, various concentrations of MeOH (040%, v/v) were tested in phosphate buffer
(30 mM, pH 2.5, containing 100 mM SDS). As shown in Fig. 4, the resolution improved progressively with increasing amounts of MeOH.
When methanol concentration was greater than 40%, baseline separation could be achieved. Taken together, the optimum separation
conditions for our experiments were dened as 30 mM phosphate
buffer (pH 2.5) containing 40% MeOH and 100 mM SDS.
3.3. Method validation and sensitivity enhancement
To evaluate the quantitative applicability of this method, validation was conducted. Calibration curves were generated by

230

H.-L. Cheng et al. / J. Chromatogr. A 1426 (2015) 226232

0.14

(A) 0%

0.12

(B) 30%
1+2+3

0.10

1+2+3

Abs.

0.08
0.06
0.04
0.02
0.00
-0.02
0.14
0.12

10

(C) 35%

12

14

10

12

(D) 40%

14

0.10

Abs.

0.08
0.06

0.04

1
3

0.02
0.00
-0.02
0

10

12

14

Time (min)

10

12

14

Time (min)

Fig. 4. Electropherograms with methanol at various concentrations in the background buffer. See Fig. 3 for other CE conditions.
Table 1
Regression equations, coefcient of determination, LODs and EF of the method.
Analytes

Regression equation

Coefcient of determination (r2 )a

LODb (ng/mL)

EFc (folds)

THC
THC-OH
THC-COOH

y = (0.1296 0.0013)x (0.0034 0.0109)

y = (0.1429 0.0148)x + (0.0005 0.0085)
y = (1.8073 0.0520)x + (0.0046 0.0028)

0.993
0.995
0.994

10
5
0.5

100
120
2000

a
b
c

Coefcient of determination of inter-batch analysis were calculated from the assay values of prepared standards on ve different days (n = 5).
LOD, limits of detection were calculated from S/N = 3.
EF, enhancement factor were calculated as dividing the LODs obtained from normal MEKC by the LODs obtained from LVSI-ASEI-sweeping.

Table 2
Precision and accuracy for the determination of analytes.
RSD (%)

REa (%)

(2.16 0.07) 102

(1.51 0.08) 103
(4.10 0.07) 103

3.28
5.51
1.66

8.30
0.94
2.43

2.0 102
1.5 103
4.0 103

(1.94 0.19) 102

(1.58 0.15) 103
(4.09 0.43) 103

9.74
9.60
10.58

3.10
5.05
2.20

1.0 101
7.5 101
2.0 102

(1.07 0.04) 101

(7.53 0.83) 101
(1.96 0.14) 102

3.48
11.06
7.15

6.70
0.32
1.80

2.0 102
1.5 103
4.0 103

(1.82 0.11) 102

(1.51 0.08) 103
(4.06 0.23) 103

5.91
5.07
5.66

9.24
0.48
1.59

THC-OH

2.0 102
1.5 103
4.0 103

(1.93 0.15) 102

(1.48 0.13) 103
(3.99 0.36) 103

7.64
8.76
8.99

3.43
1.32
0.23

THC-COOH

1.0 101
7.5 101
2.0 102

(9.88 0.74) 100

(7.64 0.66) 101
(2.06 0.10) 102

7.51
8.61
4.80

1.19
1.41
2.95

Concentration known (ng/mL)

Concentration found (ng/mL)

2.0 102
1.5 103
4.0 103

THC-OH

THC-COOH

Intra-day (n = 3)
THC

Inter-day (n = 5)
THC

RE (%) = (concentration found concentration known)/(concentration known) 100.

H.-L. Cheng et al. / J. Chromatogr. A 1426 (2015) 226232

3.4. Application in urine sample analysis

0.008

0.006

Abs.

0.004

To demonstrate the applicability of the developed method, urine

samples were tested. It is necessary to perform sample extraction
before CE analysis. In addition to eliminating interferences, the
composition and pH of the sample must be adjusted. The pretreatment procedure was stated in Section 2.3. Briey, the urine samples
were alkaline hydrolyzed before liquidliquid extraction. After the
solvent was evaporated, the residue was reconstituted with either
Tris or phosphate buffer and analyzed by the LVSI-ASEI-sweeping
method. The electropherogram is displayed in Fig. 5, showing that
three analytes could be identied without interference under optimized separation conditions. The recoveries for THC, THC-COOH
and THC-OH were 89.1%, 69% and 87.1%, respectively. The application of LVSI-ASEI-sweeping to urine analysis was found to be
satisfactory.

(B) spiked urine

0.002

3
0.000

-0.002

231

4. Conclusions

(A) blank urine

10

11

12

13

Time (min)
Fig. 5. Electropherograms of urine sample after extraction. (A) Blank urine (dotted
line), (B) blank urine spiked with analytes (THC, THC-OH 250 ng/mL; THC-COOH
125 ng/mL). The CE conditions are the same as those in Fig. 3.

plotting ve known concentrations of analyte on the X-axis against

the peak area ratios on the Y-axis. The peak-area ratios were calculated by dividing the corrected peak areas of each analyte by that
of the IS. The data of the regression equations, coefcient of determination (r2 ), limits of detection (LODs) and enhancement factor
(EF) under the optimal conditions are listed in Table 1. The linearity ranges in the calibration curves were 0.046 g/mL for THC and
THC-OH, and 2300 ng/mL for THC-COOH. The coefcient of determination (r2 0.993) indicated that high linearity between Y and
X was achieved over the ranges studied. For greater precision and
accuracy in evaluation, the relative standard deviation (RSD) and
relative error (RE) of the method were studied based on statistical
determinations (n = 3) of each analyte at three different concentrations. The results were as shown in Table 2. The RSD values were
below 11.06%, and the RE values were below 9.24% indicating good
precision and accuracy for this method. The lower detection limits
(LODs, S/N = 3) were 10 ng/mL for THC, 5 ng/mL for THC-OH, and
0.5 ng/mL of THC-COOH detected at 214 nm.
To obtain the enhancement factor of this method, we further
compared MEKC (sampling at 0.5 psi, 6 s) with this LVSI-ASEIsweeping method, LVSI-sweeping-MEKC method (hydrodynamic
injection, 0.5 psi for 3 min) and ASEI-sweeping-MEKC method
(electrokinetically injection at 2.5 kV for 6 min). The EF for LVSIsweeping-MEKC for the detection of THC, THC-OH and THC-COOH
were 70, 96, and 129, respectively. The EF for ASEI-sweeping-MEKC
for the detection of THC, THC-OH and THC-COOH were 70, 30, and
1700, respectively. Furthermore, the EF of the proposed method
could reach a 2000 fold increase for THC-COOH. The results indicated that all stacking approaches contribute to the overall EF, and
the integration of LVSI and ASEI techniques provides the best sensitivity. In addition, more than a 2000-fold increase in sensitivity
was achieved for THC-COOH, conrming that electrokinetic injection (EKI) can improve the LODs more than hydrodynamic injection
because the quantity of the injected sample is not restricted by the
length of the capillary.

A highly sensitive stacking method called LVSEP-ASEI-sweeping

combining the advantages of LVSI, ASEI and sweeping modes was
developed for the analysis of THC and its metabolites. In comparison to conventional MEKC method, an enhancement of up to
2000-fold sensitivity was achieved. There is no need to address the
complex derivatization processes; moreover, the detection sensitivity was comparable to MS detection. In addition, the total
analysis time, including sample pretreatments, of this method
was approximately 2 h, which is much shorter than LCMS (4 h)
or GCMS (5 h) analysis [13]. This method has the potential
application for the analysis of clinical samples with ppb level
analytes and for rapid urine screening of cannabinoids as a complementary method to GCMS or LCMS.
Conict of interest
The authors have declared no conicts of interest.
Acknowledgements
We gratefully acknowledge the support of the Ministry of Science and Technology of Taiwan (NSC 100-2113-M-039-002-MY2)
and China Medical University (CMU97-286) in funding this work.
References
[1] United Nations Ofce on Drugs and Crime, World Drug Report 2014. Available
from: http://www.unodc.org/documents/wdr2014/Cannabis 2014 web.pdf.
[2] T.L. Burns, J.R. Ineck, Cannabinoid analgesia as a potential new therapeutic
option in the treatment of chronic pain, Ann. Pharmacother. 40 (2006) 251260.
[3] N.M. Kogan, R. Mechoulam, Cannabinoids in health and disease, Dialogues Clin.
Neurosci. 9 (2007) 413430.
[4] H. Kalant, Adverse effects of cannabis on health: an update of the literature
since 1996, Prog. Neuropsychopharmacol. Biol. Psychiatry 28 (2004) 849863.
[5] M.E. Wall, B.M. Sadler, D. Brine, H. Taylor, M. Perez-Reyes, Metabolism, disposition, and kinetics of -9-tetrahydrocannabinol in men and women, Clin.
Pharmacol. Ther. 34 (1983) 352363.
[6] M.A. Huestis, Pharmacokinetics and metabolism of the plant cannabinoids, 9tetrahydrocannabinol, cannabidiol and cannabinol, Handb. Exp. Pharmacol.
(2005) 657690.
[7] F. Grotenhermen, Pharmacokinetics and pharmacodynamics of cannabinoids,
Clin. Pharmacokinet. 42 (2003) 327360.
[8] S.H. Burstein, The cannabinoid acids: nonpsychoactive derivatives with therapeutic potential, Pharmacol. Ther. 82 (1999) 8796.
[9] S.W. Cotten, D.L. Duncan, E.A. Burch, C.J. Seashore, C.A. Hammett-Stabler,
Unexpected interference of baby wash products with a cannabinoid (THC)
immunoassay, Clin. Biochem. 45 (2012) 605609.
[10] J.Y. Moon, J.Y. Kim, M.H. Moon, B.C. Chung, M.K. In, M.H. Choi, Validated
gas chromatographic-mass spectrometric analysis of urinary cannabinoids
puried with a calcium-hardened beta-cyclodextrin polymer, J. Chromatogr.
A 1204 (2008) 8792.
[11] S.C.W.D. Eller, L.G. Flaiban, B.A.P.B. Paranhos, J.L. da Costa, F.R. Lourenco,
M. Yonamine, Analysis of 11-nor-9-carboxy-delta(9)-tetrahydrocannabinol

232

[12]

[13]

[14]
[15]

[16]

[17]

[18]

H.-L. Cheng et al. / J. Chromatogr. A 1426 (2015) 226232

in urine samples by hollow ber-liquid phase microextraction and gas
chromatographymass spectrometry in consideration of measurement uncertainty, Forensic Toxicol. 32 (2014) 282291.
U. Backofen, F.M. Matysik, C.E. Lunte, Determination of cannabinoids in
hair using high-pH* non-aqueous electrolytes and electrochemical detection. Some aspects of sensitivity and selectivity, J. Chromatogr. A 942 (2002)
259269.
Y. Iwamuro, R. Iio-Ishimaru, S. Chinaka, N. Takayama, K. Hayakawa, Analysis
of 11-nor-9 -tetrahydrocannabinol-9-carboxylic acid and its glucuronide in
urine by capillary electrophoresis/mass spectrometry, Biomed. Chromatogr. 26
(2012) 14521456.
J.P. Quirino, S. Terabe, Exceeding 5000-fold concentration of dilute analytes in
micellar electrokinetic chromatography, Science 282 (1998) 465468.
P. Britz-McKibbin, K. Otsuka, S. Terabe, On-line focusing of avin derivatives
using dynamic pH junctionsweeping capillary electrophoresis with laserinduced uorescence detection, Anal. Chem. 74 (2002) 37363743.
T.C. Chiu, Recent advances in on-line concentration and separation of
amino acids using capillary electrophoresis, Anal. Bioanal. Chem. 405 (2013)
79197930.
L. Zhu, C. Tu, H.K. Lee, On-line concentration of acidic compounds by
anion-selective exhaustive injectionsweeping-micellar electrokinetic chromatography, Anal. Chem. 74 (2002) 58205825.
J.P. Quirino, S. Terabe, Approaching a million-fold sensitivity increase
in capillary electrophoresis with direct ultraviolet detection: cationselective exhaustive injection and sweeping, Anal. Chem. 72 (2000)
10231030.

[19] K. Sueyoshi, F. Kitagawa, K. Otsuka, On-line sample preconcentration and

separation technique based on transient trapping in microchip micellar electrokinetic chromatography, Anal. Chem. 80 (2008) 12551262.
[20] J.P. Quirino, P.R. Haddad, Online sample preconcentration in capillary electrophoresis using analyte focusing by micelle collapse, Anal. Chem. 80 (2008)
68246829.
[21] A.T. Aranas, A.M. Guidote Jr., J.P. Quirino, Sweeping and new on-line sample
preconcentration techniques in capillary electrophoresis, Anal. Bioanal. Chem.
394 (2009) 175185.
[22] I. Maijo, F. Borrull, C. Aguilar, M. Calull, On-column preconcentration
of anti-inammatory drugs in river water by anion-selective exhaustive
injection-sweeping-MEKC, Chromatographia 73 (2011) 8391.
[23] Z.K. Shihabi, Sample matrix effects in capillary electrophoresis. 2. Acetonitrile
deproteinization, J. Chromatogr. A 652 (1993) 471475.
[24] H.Y. Huang, W.C. Lien, I.Y. Huang, Anion-selective exhaustive injectionsweeping microemulsion electrokinetic chromatography, Electrophoresis 27
(2006) 32023209.
[25] K. Vitkova, J. Petr, V. Maier, J. Znaleziona, J. Sevcik, Study of electromigration
effects on a pH boundary during the on-line electrokinetic preconcentration by
capillary electrophoresis, Electrophoresis 31 (2010) 27712777.
[26] C.X. Zhang, W. Thormann, Head-column eld-amplied sample stacking in
binary system capillary electrophoresis. 2. Optimization with a preinjection
plug and application to micellar electrokinetic chromatography, Anal. Chem.
70 (1998) 540548.
[27] J.P. Quirino, S. Terabe, Sweeping of analyte zones in electrokinetic chromatography, Anal. Chem. 71 (1999) 16381644.