Journal of Pharmaceutical and Biomedical Analysis 145 (2017) 616–620

Contents lists available at ScienceDirect

Journal of Pharmaceutical and Biomedical Analysis

journal homepage: www.elsevier.com/locate/jpba

Short communication

Quantification of paracetamol and 5-oxoproline in serum by capillary

electrophoresis: Implication for clinical toxicology
Tomáš Hložek a,b , Tomáš Křížek b , Petr Tůma c , Miroslava Bursová a,b ,
Pavel Coufal b , Radomír Čabala a,b,∗
a
Charles University and General University Hospital, First Faculty of Medicine, Institute of Forensic Medicine and Toxicology, Ke Karlovu 2, 121 08, Prague
2, Czech Republic
b
Charles University, Faculty of Science, Department of Analytical Chemistry, Albertov 6, 128 43, Prague 2, Czech Republic
c
Charles University, Third Faculty of Medicine, Department of Biochemistry, Cell and Molecular Biology, Ruská 87, 100 00, Prague 10, Czech Republic

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

Article history: High anion gap metabolic acidosis frequently complicates acute paracetamol overdose and is generally
Received 15 May 2017 attributed to lactic acidosis or compromised hepatic function. However, metabolic acidosis can also be
Received in revised form 4 July 2017 caused by organic acid 5-oxoproline (pyroglutamic acid). Paracetamol’s toxic intermediate, N-acetyl-p-
Accepted 20 July 2017
benzoquinoneimine irreversibly binds to glutathione and its depletion leads to subsequent disruption
Available online 26 July 2017
of the gamma glutamyl cycle and an excessive 5-oxoproline generation. This is undoubtedly an under-
diagnosed condition because measurement of serum 5-oxoproline level is not readily available. A simple,
Keyword:
cost effective, and fast capillary electrophoresis method with diode array detection (DAD) for simulta-
Paracetamol
5-oxoproline neous measurement of both paracetamol (acetaminophen) and 5-oxoproline in serum was developed
Capillary electrophoresis and validated. This method is highly suitable for clinical toxicology laboratory diagnostic, allowing rapid
Metabolic acidosis quantification of acidosis inducing organic acid 5-oxoproline present in cases of paracetamol overdose.
The calibration dependence of the method was proved to be linear in the range of 1.3–250 ␮g mL−1 , with
adequate accuracy (96.4–107.8%) and precision (12.3%). LOQ equaled 1.3 ␮g mL−1 for paracetamol and
4.9 ␮g mL−1 for 5-oxoproline.
© 2017 Elsevier B.V. All rights reserved.

1. Introduction of gamma glutamylcysteine (Fig. 1). It has been demonstrated that

Paracetamol (acetaminophen) supratherapeutic ingestion,
either accidental or intentional, is the most common pharma-
ceutical overdose reported across the world [1]. High anion gap
metabolic acidosis frequently complicates paracetamol poisoning
and is generally attributed to lactic acidosis, renal failure or
compromised hepatic function [2]. However, metabolic acidosis
can also be caused by the accumulation of the organic acid 5-
oxoproline, an intermediate of gamma glutamyl cycle, which is
responsible for glutathione production. This pathology is apparent
when glutathione levels are diminished either via inherited or
acquired defects, since normal glutathione levels are necessary
for feedback inhibition on the enzyme gamma glutamylcysteine
synthase. The loss of feedback inhibition causes an overproduction
∗ Corresponding author at: Charles University and General University Hospital,
First Faculty of Medicine, Institute of Forensic Medicine and Toxicology, Ke Karlovu
2, 121 08, Prague 2, Czech Republic.
E-mail addresses: radomir.cabala@vfn.cz, cabala@natur.cuni.cz (R. Čabala).
this situation occurs more frequently in women and during sepsis
or treatment with paracetamol, antiepileptic drug vigabatrin or
certain antibiotics [3,4]. Glycine deficiency during pregnancy
or malnutrition is another contributing factor [5]. Nevertheless,
accumulation of gamma glutamylcysteine is uncommon, because
it is partially metabolized to 5-oxoproline by gamma glutamylcy-
clotransferase [6]. On the other hand, 5-oxoproline is converted to
glutamate by 5-oxoprolinase, but this enzyme is quickly saturated
and that is the rate limiting step leading to increase in 5-oxoproline
levels [4].
Therefore, 5-oxoprolinemia should be added to the differential
diagnosis for patients presenting with clinically significant acidosis
after acute acetaminophen overdose [7,8] as proposed by Mehta
in GOLD MARK (glycols, oxoproline, L- and D-lactate, methanol,
aspirin, renal failure and ketoacidosis) mnemonic [9].
In addition, given the prevalence of the risk factors, that
may lead to a baseline glutathione deficiency among hospital
patients, this diagnosis should also be considered in any case
with unexplained metabolic acidosis and a history of therapeutic
acetaminophen use [4,10,11].

http://dx.doi.org/10.1016/j.jpba.2017.07.024
0731-7085/© 2017 Elsevier B.V. All rights reserved.
 T. Hložek et al. / Journal of Pharmaceutical and Biomedical Analysis 145 (2017) 616–620
Fig. 1. The gamma glutamyl cycle.
Many different diagnostic methods based on spectrophotome-
try [11,12], immunoassay detection [13] or GC–MS [14] and LC–MS
[15] techniques for paracetamol serum levels are readily available.
However, there is lack of routine laboratory methods for deter-
mination of 5-oxoproline, although the recognition of its elevated
presence is crucial for proper diagnosis and effective treatment.
The general method for determination of 5-oxoproline in clini-
cal practice was gas chromatography mass spectrometry GC–MS
[4,7,8]. However, this separation method requires extensive sam-
ple pretreatment resulting in inadequate clinical turnaround time
and may not be readily available in toxicology laboratories on
24 h basis. On the other hand, recently published liquid chro-
matography mass spectrometry LC–MS methods for 5-oxoproline
serum determination were designated for either evaluation of 5-
oxoproline as hepatotoxicity marker following the administration
of glutathione-depleting hepatotoxic xenobiotics in rats [16] or elu-
cidating metabolic changes associated with certain types of cancer
in humans [17]. Capillary electrophoresis with MS or UV detection
was used as a metabolomic tool for non-targeted fingerprinting,
including among others 5-oxoproline as an analyte in children urine
samples [18]. Nevertheless, these methods are highly specialized
research tools, and hence, our aim was to develop a simple, fast
CE method for simultaneous determination of paracetamol and 5-
oxoproline in human serum with minimal sample pretreatment.
This newly developed method was successfully used for the mon-
itoring of both paracetamol and 5-oxoproline levels in series of
patients after either intentional paracetamol ingestion or therapeu-
tic misadventures, since cases reporting the incidence and severity
of 5-oxoprolinemia after an acute paracetamol overdose are lim-
ited.
2. Materials and methods
2.1. Chemicals and materials
Acetonitrile (HPLC gradient grade, Fisher Scientific, UK), parac-
etamol (≥99%, Sigma-Aldrich, USA), L-pyroglutamic acid (≥99%,
Aldrich, France), N-cyclohexyl-2-aminoethane sulfonic acid (≥99%,
CHES, pKA 9.55, Fluka, Switzerland), 3-(cyclohexylamino)-1-
propane sulfonic acid (≥99%, CAPS, pKA 10.40, Fluka, Switzerland),
sodium hydroxide (≥98%, Fluka, Switzerland) were used as pur-
chased. Deionized water prepared by a water purification system
617
(Premier MFG’D, Phoenix, AZ, USA) was used for preparation of all
aqueous solutions.
2.2. Electrophoretic conditions
All the electrophoretic measurements were carried out using
the Agilent 7100 Capillary Electrophoretic System (Agilent Tech-
nologies, Waldbronn, Germany) equipped with a built-in diode
array detector (DAD), which cell is placed in the electrophoretic
cassette at a constant temperature of 25 ◦ C. All CE separations
were performed in a fused-silica capillary (Composite Metal Ser-
vices, UK), 25 ␮m inner diameter (id), 363 ␮m outer diameter (od),
with minimal total length of 31.5 cm; 23.0 cm to the DAD. A new
capillary was washed stepwise with NaOH 0.1 M (30 min), water
(15 min), and background electrolyte (BGE, 20 min) at a pressure
of 1000 mbar. Two min washing with the BGE was used between
individual analyses. All the separations were carried out in the opti-
mized CHES/NaOH 0.04 M (pH 10.2) BGE at a separation voltage of
+30 kV with ramp from 0 to +30 kV within 0.2 min.
2.3. Preparation of serum samples, calibrators
Blank human serum for method validation was purchased from
ACQ Science GmbH (Rottenburg-Haifinger, Germany). The six con-
centration levels of the calibrators in blank human serum were
prepared for paracetamol: 1.3, 5, 10, 50, 100, 250 ␮g mL−1 and 5-
oxoproline: 4.9, 10, 30, 50, 150, 250 ␮g mL−1 . Samples with either
paracetamol or 5-oxoproline concentrations outside the calibration
range were diluted as appropriate. Every day method calibration is
necessary to obtain precise values of serum concentrations.
Human whole blood samples were delivered from various
hospitals with suspected paracetamol intoxications, mainly for
emergency purposes. For the research purposes the samples were
anonymized and do not fall under the approval by the Hospital
Ethics Commission. Whole blood was centrifuged at 10,000 rpm
for 30 s to obtain serum sample for analysis.
2.4. Sample preparation for CE analysis
Prior to the analysis, the serum samples were deproteinised
by mixing 33.3 ␮L of serum with 100 ␮L of ACN with addition of
NH4 OH 0.01 M; ACN serves as a deproteining medium and NH4 OH
guarantees deprotonization of weak acids in sample zone. Depro-
teinisation was performed in an Eppendorf tube by shaking for
618 T. Hložek et al. / Journal of Pharmaceutical and Biomedical Analysis 145 (2017) 616–620

Fig. 2. Electropherogram of blank human serum with native concentration 39.4 ␮g mL−1 of 5-oxoproline (A), the same serum sample spiked with 36.4 ␮g mL−1 of paracetamol
and 36.4 ␮g mL−1 of 5-oxoproline (total 75.8 ␮g mL−1 ) (B), and patient sample number 14 (C) under optimized conditions recorded at 200 nm. Peak identification: 1. EOF, 2.
paracetamol, 3. 5-oxoproline.

Table 1
Within- and between-day method precision and accuracy of paracetamol and 5-oxoproline in serum (six days, six replicates).

Analyte Within-day Between-day

cnominal cmeasured Precision Accuracy (%) cnominal cmeasured Precision Accuracy (%)
(␮g mL−1 ) (mean ± SD) (RSD, %) (␮g mL−1 ) (mean ± SD) (RSD, %)

paracetamol 7 7.0 ± 0.6 8.7 99.8 7 6.8 ± 0.6 9.2 97.6

90 95.0 ± 7.2 7.6 105.6 90 96.3 ± 9.2 9.6 107.0
200 203.5 ± 11.8 5.8 101.8 200 203.5 ± 18.0 8.8 101.8

5- 7 6.8 ± 0.8 11.9 96.4 7 6.9 ± 0.7 10.3 99.0

oxoproline 90 95.5 ± 8.9 9.3 106.1 90 97.0 ± 11.9 12.3 107.8
200 205.5 ± 13.1 6.4 102.8 200 208.7 ± 19.7 9.4 104.3

120 s. Then the plasma samples were centrifuged at 10 000 rpm for
120 s; 50 ␮L of the obtained supernatant were taken for CE analy-
sis. The final ACN content in treated clinical samples is 75% v/v. The
samples were hydrodynamically injected from the sampling end of
the capillary using a pressure of 50 mbar for 20 s.
3. Results and discussion
3.1. Analytical performance
Several analytical parameters were optimized with respect to
high separation efficiency and low limit of detection (LOD) of CE
method. The high separation efficiency is connected with perform-
ing separation in capillaries with thin 25 ␮m id to achieve the
baseline separation of both tested compounds from other plasma
components; 50 ␮m id capillary was not successful for this pur-
pose (data not shown). Application of high intensity of driving
electric field in combination with the use of short separation path
has diminished the peak broadening. Minimal effective length of
23.0 cm and maximal separation voltage of 30 kV were used for
Agilent CE instrument.
The choice of appropriate BGE including its pH represents a crit-
ical point to achieve the successful CE separation of weak acids,
as is paracetamol with pKA of 9.4 for its phenolic group and 5-
oxoproline with pKA of 3.5 for its carboxyl group. For accomplishing
the simultaneous determination of both compounds in one CE run,
it is necessary to perform separation at pH value higher than is pKA
of paracetamol to ensure sufficient deprotonization of its phenolic
group. The range of BGEs based on boric acid/NaOH, CAPS/NaOH
and CHES/NaOH at pH from 9.5 to 11.0 was tested for this pur-
pose and the optimum BGE composed of CHES/NaOH 0.04 mol L−1
at pH 10.2, see Fig. 2 [19]. Under these conditions, a fast cationic
electroosmotic flow (EOF) moves through the capillary in opposite
direction of both dissociated acids migration. Because paracetamol
is only partially dissociated at optimized pH, its peak is recorded
tightly after the zone of electroneutral compounds in time 70 s. On
the other hand, 5-oxoproline with pKA of 3.5 is completely dis-
sociated at pH 10.2; its anionic electrophoretic mobility is high
under these experimental conditions and 5-oxoproline reaches the
detector later in time 105 s compared to paracetamol.
3.2. Method validation
Because of the natural occurrence of 5-oxoproline in the human
serum, the calibration was based on standard addition method.
The six concentration levels (added) of the calibrators in blank
human serum were prepared to yield the following final serum con-
centrations for paracetamol: 1.3, 5, 10, 50, 100, 250 ␮g mL−1 and
5-oxoproline: 4.9, 10, 30, 50, 150, 250 ␮g mL−1 (five replicates for
each point). The slope of the linear dependence of the peak area on
the concentration equaled 0.117 (SD 0.003) mAU• min mL• ␮g−1 for
paracetamol with a correlation coefficient (R) of 0.999; and 0.032
 T. Hložek et al. / Journal of Pharmaceutical and Biomedical Analysis 145 (2017) 616–620 619

Table 2
Concentrations of paracetamol and 5-oxoproline in serum of patients with paracetamol overdose.

Patient Gender Dose (g) Elapsed CEDIA (paracetamol) CE (paracetamol) CE (5-oxoproline)

time (h) (␮g mL−1 ) (␮g mL−1 ) (␮g mL−1 )

1 F 4.0 4 22.1 30 21
2 F 10.0 18 15.2 17 27
3 F 6.3 4 96.1 88 25
4 F X 5 91.1 92 27
5 F 5.5 4 58.6 72 7.7
6 F X 3–4 0.3 <LOQ 33
7 F X 10 89.7 75 10
8 M 50.0 4–8 36.5 31 14
9 F X 2–3 56.0 54 25
9a F X 5 21.2 21 16
10 F 6.0 6.5 148.2 139 27
10 a F X X 0.7 <LOQ 29
11 F 6.0 5 105.6 102 19
11 a F X X 2.1 3.0 19
12 F X 4–8 66.7 76 180
13 M 4.5 7 109.8 125 34
13 a M X 14 26.3 28 32
14 F 12 X 38.4 38 371
15 F 6 4 4.1 4.1 18
16 F X 8 9.9 9.9 9.2
17 M 6.0 5.5 35.4 46 20

a. . .treatment control samples.

X. . .unknown detail.

(SD 0.001) mAU• min mL• ␮g−1 for 5-oxoproline with R 0.999, both
at 200 nm.
The employed treatment of the serum samples is sufficient for
eliminating the effect of the matrix and thus for attaining high reli-
ability of the determination. The obtained LOD calculated as the
peak height corresponding to three times the noise level equaled
0.3 ␮g mL−1 for paracetamol (2.0 ␮M; Mr 151) and 1.3 ␮g mL−1
for 5-oxoproline (10.0 ␮M; Mr 129). Five times better LOD value
for paracetamol in comparison to 5-oxoproline is caused by the
presence of benzene ring in its structure. LOQ calculated as the
peak height corresponding to ten times the noise level equaled
1.3 ␮g mL−1 for paracetamol and 4.9 ␮g mL−1 for 5-oxoproline.
The precision of the developed method (Table 1), expressed as
relative standard deviation (RSD), was evaluated by analyzing six
individual blank serum samples containing both analytes at the
concentration of 7, 90, 200 ␮g mL−1 on the same day (within-day
precision) as well as on 6 different days (between-day precision).
The RSD obtained was found to be satisfactory with value lower
than 15% in all cases and the method performed well with adequate
accuracy (96.4–107.8%). The repeatability of the migration time
for ten consecutive analyses of serum sample with 6.0 ␮g mL−1
of paracetamol and 37.0 ␮g mL−1 of 5-oxoproline equals RSD 0.2%
for paracetamol and 0.3% for 5-oxoproline; the repeatability of the
peak area is 2.5% for both analytes. The repeatability of the migra-
tion time for three consecutive days (10 analyses each day) equals
RSD 1.5% for paracetamol and 1.4% for 5-oxoproline.
3.3. Analysis of patient samples
The newly developed method performance was tested on a
series of serum samples from patients with suspected acute parac-
etamol overdose, which had been previously confirmed with
routine immunoassay kit for paracetamol (CEDIA) and all the
results are summarized in Table 2. Paracetamol levels measured
with both methods were evaluated using orthogonal regression
(Deming regression) to test whether two different methods provide
comparable measurements (R2 0.9663). There is no statistically sig-
nificant difference between compared methods at 95 % confidence
level. Given the range of paracetamol levels, ingested paracetamol
dose and corresponding estimated time since ingestion, none of
the presented cases was classified as a severe poisoning accord-
ing to Rumack-Matthew nomogram [20]. Normal physiological
serum levels of 5-oxoproline are given in a range 1–30 ␮g mL−1 ,
however this range may slightly vary due to actual state of organ-
ism, i. e. malnutrition, pregnancy or induced oxidative stress [4,5].
In documented cases of acquired 5-oxoprolinemia, either after
acute or chronic paracetamol use, the reported serum levels of 5-
oxoproline are significantly higher and these values are in a range
300–1200 ␮g mL−1 [4,8]. Within our set of patients, most of the 5-
oxoproline levels were in physiological range. Only in two cases
the measured level was significantly higher, namely in the patient
12 (180 ␮g mL−1 ) and patient 14 (371 ␮g mL−1 ) respectively. In
the first case, the serum level approaches the lower limit of the
published values associated with acquired 5-oxoprolinemia, while
the second one slightly exceeds this limit, but neither of these
cases has a pH value out of a reference range of 7.35–7.45. These
cases illustrate an increase in 5-oxoproline levels out of physi-
ological range in patients that possess one or more risk factors
associated with 5-oxoprolinemia, particularly female gender and
malnutrition. However, since the measurement of 5-oxoproline is
not routine toxicological/biochemical practice, and the presenting
acidosis is generally attributed solely to lactic acid acidosis, the inci-
dence of 5-oxoprolinemia and its role in the severity of paracetamol
overdose may be greatly underappreciated.
4. Conclusion
In this paper, the simple, effective and fast method for simulta-
neous determination of paracetamol and 5-oxoproline in human
serum using capillary electrophoresis with spectrophotometric
detection was developed for clinical toxicology, since obtaining 5-
oxoproline concentrations is not part of the standard practice in
evaluating patients that present either after acute or chronic parac-
etamol (mis)use or in those who have high anion gap metabolic
acidosis. A universal and accessible method for many toxicological
laboratories could contribute to determination of the true incidence
of 5-oxoprolinemia and its clinical significance and relationship
to acute or chronic paracetamol ingestion, since the occurrence of
that kind of acidosis is generally underestimated and omitted with
respect to differential laboratory diagnosis of the usual suspects, i.e.
lactic acidosis, ketoacidosis, and ingestion of methanol, salicylates
or ethylene glycol.
620 T. Hložek et al. / Journal of Pharmaceutical and Biomedical Analysis 145 (2017) 616–620
Acknowledgements
This work was supported by Charles University (project GAUK
968216) and Specific University Research (SVV).
References
[1] L.R. Khan, G.C. Oniscu, J.J. Powell, Long-term outcome following liver
transplantation for paracetamol overdose, Transpl. Int. 23 (2010) 524–529.
[2] P.V. McBride, B.H. Rumack, Acetaminophen intoxication, Semin. Dial. 5 (1992)
292–298.
[3] A.C. Croal, C.J. Glen, Transient 5-oxoprolinuria (pyroglutamic aciduria) with
systemic acidosis in an adult patient receiving antibiotic therapy, Clin. Chem.
44 (1998) 336–340.
[4] D.B. Liss, M.S. Paden, E.S. Schwarz, M.E. Mullins, What is the clinical
significance of 5-oxoproline (pyroglutamic acid) in high anion gap metabolic
acidosis following paracetamol (acetaminophen) exposure? Clin. Toxicol. 51
(2013) 817–827.
[5] R.W. Friesen, E.M. Novak, D. Hasman, S.M. Innis, Relationship of
dimethylglycine choline, and betaine with oxoproline in plasma of pregnant
women and their newborn infants, J. Nutr. 137 (2007) 2641–2646.
[6] M. Emmett, Acetaminophen toxicity and 5-Oxoproline (Pyroglutamic acid): a
tale of two cycles, one an ATP-depleting futile cycle and the other a useful
cycle, Clin. J. Am. Nephrol. 9 (2014) 191–200.
[7] D.T. Lawrence, L.K. Bechtel, N.P. Charlton, C.P. Holstege, 5-oxoproline
−induced anion gap metabolic acidosis after an acute acetaminophen
overdose, J. Am. Osteopath. Assoc. 110 (2010) 545–551.
[8] P.S.C. Leung, S.H. Tsui, T.S. Siu, S. Tam, Acquired 5-oxoprolinuria (Pyroglutamic
acidaemia) as a cause of early high anion gap metabolic acidosis in acute
massive paracetamol overdose, Hong Kong J. Emerg. Med. 18 (2011) 264–270.
[9] A.N. Mehta, J.B. Emmett, M. Emmett, GOLD MARK. an anion gap mnemonic for
the 21 st century, Lancet 372 (2008) 892.
[10] A.Z. Fenves, H.M. Kirkpatrick, V.V. Patel, L. Sweetman, M. Emmett, Increased
anion gap metabolic acidosis as a result of 5-oxoproline (pyroglutamic acid):
A role for acetaminophen, Clin. J. Am. Soc. Nephrol. 1 (2006) 441–447.
[11] C. Lanoy, Y. Bouckaert, Metabolic acidosis and 5-oxoprolinuria induced by
flucloxacillin and acetaminophen: a case report, J. Med. Case Rep. 10 (2016)
184–188.
[12] F. Shihana, D. Dissanayake, P. Dargan, A. Dawson, A modified low-cost
colorimetric method for paracetamol (acetaminophen) measurement in
plasma, Clin. Toxicol. 48 (2010) 42–46.
[13] J. Polson, F.H. Wians Jr., P. Orsulak, D. Fuller, N.G. Murray, J.M. Koff, A.I. Khan,
J.A. Balko, L.S. Hynan, W.M. Lee, False positive acetaminophen concentrations
in patients with liver injury, Clin. Chim. Acta 391 (2008) 24–30.
[14] A. Trettin, A.A. Zoerner, A. Böhmer, F.M. Gutzki, D.O. Stichtenoth, J. Jordan, D.
Tsikas, Quantification of acetaminophen (paracetamol) in human plasma and
urine by stable isotope-dilution GC–MS and GC–MS/MS as pentafluorobenzyl
ether derivative, J. Chromatogr. B 879 (2011) 2274–2280.
[15] D. Tonoli, E. Varesio, G. Hopfgartner, Quantification of acetaminophen and
two of its metabolites in human plasma by ultra-high performance liquid
chromatography–low and high resolution tandem mass spectrometry, J.
Chromatogr. B 904 (2012) 42–50.
[16] S. Geenen, C. Guallar-Hoyas, F. Michopoulos, J.G. Kenna, K.L. Kolaja, H.V.
Westerhoff, P. Thomas, I.D. Wilson, HPLC–MS/MS methods for the
quantitative analysis of 5-oxoproline (pyroglutamate) in rat plasma and
hepatic cell line culture medium, J. Pharm. Biomed Anal. 56 (2011) 655–663.
[17] A. Klupczynska, S. Plewa, W. Dyszkiewicz, M. Kasprzyk, N. Sytek, Z.J. Kokot,
Determination of low-molecular-weight organic acids in non-small cell lung
cancer with a new liquid chromatography–tandem mass spectrometry
method, J. Pharm. Biomed. Anal. 129 (2016) 299–309.
[18] C. Barbas, E.P. Moraes, A. Villaseñor, Capillary electrophoresis as a
metabolomics tool for non-targeted fingerprinting of biological samples, J.
Pharm. Biomed. Anal. 55 (2011) 823–831.
[19] B. Vochyánová, F. Opekar, P. Tůma, Simultaneous and rapid determination of
caffeine and taurine in energy drinks by MEKC in a short capillary with dual
contactless conductivity/photometry detection, Electrophoresis 35 (2014)
1660–1665.
[20] B.H. Rumack, Acetaminophen hepatotoxicity: the first 35 years, J. Toxicol.
Clin. Toxicol. 40 (2002) 3–20.