Journal of Chromatography B 1184 (2021) 122971

Contents lists available at ScienceDirect

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

HPLC-UV assay of tramadol and O-desmethyltramadol in human plasma

containing other drugs potentially co-administered to participants in a
paediatric population pharmacokinetic study
O. Yoo a, E.K.Y. Tang a, M.N. Nguyen a, S. Salman b, c, A.J. Hua a, B.S. von Ungern Sternberg c, d, e,
L.Y. Lim a, *
a
Division of Pharmacy, School of Allied Health, University of Western Australia, Perth, Australia
b
Clinical Pharmacology and Toxicology Unit, PathWest, Perth, Australia
c
Division of of Emergency Medicine, Anaesthesia and Pain Medicine, Medical School, University of Western Australia, Perth, Australia
d
Department of Anaesthesia and Pain Management, Perth’s Children Hospital, Australia
e
Perioperative Medicine Team, Telethon Kids Institute, Perth, Australia

A R T I C L E I N F O A B S T R A C T

Keywords: Multimodal analgesia is employed in paediatric pain management to maximise analgesia and minimise side
Tramadol effects. Tramadol is dosed at 1–1.5 mg/kg to treat severe pain in children but the assay for tramadol in plasma
HPLC assay samples for pharmacokinetic and toxicology studies does not often consider concurrently administered medi­
Paediatrics pharmacokinetics
cations. In this study we developed and validated an HPLC–UV method to quantify tramadol and its main
Plasma extraction
O-desmethyltramadol
metabolite (O-desmethyltramadol) in human plasma in the presence of seven potentially interfering drugs.
Population pharmacokinetics Sample preparation method was developed by combining liquid–liquid extraction and protein precipitation.
Chromatographic separation was achieved on a BDS-Hypersil-C18 column (5 µm, 250 × 4.6 mm) using a double
gradient method. The limit of quantification was 6.7 ng/ml for both tramadol and ODT. The precision and ac­
curacy were in compliance with ICH guidelines. This method was successfully employed to analyse the blood
samples of 137 paediatric participants in a tramadol pharmacokinetic trial.

1. Introduction tramadol [5].

Multimodal analgesia is employed for pain management to maximise
Tramadol (Fig. 1) is a centrally acting analgesic whose effect has analgesia and minimise side effects [6]. Patients scheduled for surgery
been ascribed to multiple mechanisms, including weak agonistic activity may therefore receive tramadol as pre-medication together with other
on the opioid receptor, and inhibition of neuronal noradrenaline and analgesics. These commonly include paracetamol and non-steroidal
serotonin uptake [1]. The multipronged mechanisms allow tramadol to anti-inflammatory agents, such as ibuprofen, diclofenac, and par­
provide analgesia with fewer side effects than the opioids. Tramadol is ecoxib. To manage postoperative nausea and vomiting, ondansetron and
widely prescribed for adults and is also used to treat severe pain in dexamethasone are also often co-administered with tramadol during the
children at the dose of 1–1.5 mg/kg. procedure. Published assays of tramadol in plasma samples have
Upon administration, tramadol is eliminated mainly by hepatic considered the presence of anticancer drugs [7], drugs used in palliative
transformation, with 30% of the drug dose excreted unchanged in the care [8], opioid analgesics [9], sildenafil [10] and pregabalin [11]. To
kidney. There are 23 metabolites of tramadol identified in humans [2], date there is no tramadol and ODT assay that considers potential
but only the O-desmethyltramadol (ODT) metabolite is active. ODT is interference from a number of other medications concurrently admin­
generated by CYP2D6-mediated O-demethylation of tramadol [3]. The istered with tramadol for surgical procedures.
plasma concentration of ODT is typically less than 25% that of tramadol Tramadol in biological matrices has been quantified by Liquid
[4], however, ODT is 200 times more potent than tramadol at the Chromatography-Mass Spectrometry (LC/MS) [12] and High Perfor­
µ-opioid receptor and is the main contributor of the analgesic effect of mance Liquid Chromatography (HPLC) with ultraviolet (HPLC- UV)

* Corresponding author.
E-mail address: [email protected] (L.Y. Lim).

https://doi.org/10.1016/j.jchromb.2021.122971
Received 28 May 2021; Received in revised form 28 September 2021; Accepted 30 September 2021
Available online 7 October 2021
1570-0232/© 2021 Elsevier B.V. All rights reserved.
O. Yoo et al. Journal of Chromatography B 1184 (2021) 122971

[10] or fluorescence (HPLC-FL) spectroscopy [13–15]. LC/MS offers − 80 ◦ C until analysis was performed by spiking with methanol or
high sensitivity and selectivity, and is the prevalent method in high in­ analytes.
come countries for assay of drug in plasma samples [16]. However,
HPLC-UV is a more accessible method, as it has a lower cost barrier than 2.2. Chromatographic conditions
LC/MS or HPLC with other detectors, and it has been applied success­
fully for tramadol pharmacokinetic study [17]. Chromatographic analysis was performed on an integrated Agilent
The aim of our study was to develop and validate a HPLC-UV method 1260 Infinity HPLC instrument (Agilent Technologies Australia, NSW,
to quantify tramadol and ODT in plasma in the presence of drugs that Australia) with an Agilent 1260 Infinity diode-array detector. Chro­
may be co-administered with tramadol in patients undergoing surgery. matographic separations were carried out at 18 ◦ C on a reversed phase
Tramadol has a weak chromophore [13] and there is limited information C18 column (Hypersil BDS, 250 mm × 4.6 mm, 5 µm particles size,
on the quantification of tramadol in plasma using a UV detector. We Thermo Fisher Scientific Australia, WA, Australia). The column was
applied a liquid–liquid extraction followed by protein precipitation coupled to a BDS-Hypersil-C18 Guard Cartridge (5 µm, 10 × 4 mm). The
method to pre-concentrate the tramadol samples. This increased the chromatographic data were collected and processed using an Agilent
efficiency of tramadol extraction adequately to allow its quantification ChemStation software (OpenLAB C01.07, version 02.05.021, Knauer,
by UV absorption. The validated HPLC-UV method is applicable for the Berlin, Germany).
analysis of tramadol and ODT in pharmacokinetic and toxicology studies The mobile phase consisted of 20 mM potassium phosphate buffer
where other potentially interfering drugs are present, and has been (pH 3.0) and acetonitrile. A gradient elution method applied is shown in
successfully applied to analyse the plasma samples of 137 children who Table 1.
participated in a tramadol pharmacokinetic study at the Perth Children’s The total analysis time was 40 min. The flow rate was 0.8 ml/min,
Hospital (manuscript under review). injection volume was 30 µl at ambient temperature, and the eluent was
monitored at 270 nm and 220 nm.
2. Materials and methods

2.1. Materials 2.3. Preparation of calibration standard and quality control (QC)
samples
Unless specified, all chemicals and reagents were of analytical grade.
Tramadol (TRM) hydrochloride, propranolol and ibuprofen were pur­ Stock solutions of TRM, ODT, and propranolol were prepared in
chased from PCCA (NSW, Australia), while O-desmethyltramadol hy­ methanol at concentrations of 50, 25 and 50 µg/ml, respectively, and
drochloride (ODT), dexamethasone and paracetamol were from Sigma- stored at − 20 ◦ C until use. TRM and ODT working solutions (0.1–5 µg/
Aldrich (NSW, Australia). Parecoxib was purchased from Pfizer
Australia (Dynastat®, NSW, Australia), ondansetron from Mayne Table 1
Pharma International Pty Ltd. (SA, Australia), and midazolam from The double gradient elution method employed in the HPLC-UV assay of tra­
Cambrex Profarmaco (Milano, Italy). Methanol (HPLC grade) and 0.1 M madol and O-desmethyltramadol using a mobile phase comprising of potassium
HCl were from Scharlau (Barcelona, Spain), acetonitrile from RCI Lab­ phosphate buffer (20 mM, pH 3.0) and acetonitrile.
scan (Bangkok, Thailand), and diethyl ether (HPLC grade) and ethyl Time 20 mM potassium phosphate buffer (pH 3.0) (% Acetonitrile (% v/
acetate from Ajax Finechem (Cheltenham, Australia). Potassium dihy­ (min) v/v) v)

drogen phosphate was purchased from Unilab (NSW, Australia). 0 86 14

Human plasma applied to optimise analyte extraction and to validate 9 65 35
14 65 35
the HPLC-UV assay was voluntarily donated by members of the research
19 20 80
team. Blood samples were collected into EDTA tubes from healthy adults 25 20 80
not on any medications and were centrifuged at 4226 rpm at 4 ◦ C for 10 27 86 14
min before the plasma was separated into Eppendorf tubes for storage at 40 86 14

Fig. 1. Chemical structures of Tramadol (a) and propranolol (internal standard, (b)).

2
O. Yoo et al. Journal of Chromatography B 1184 (2021) 122971

Peak area of the TRM extraced in plasma

Extraction efficiency(%) of TRM = x100
Peak area of TRM in methanol

ml) were prepared from the respective stock solutions using methanol as Parallel experiments were conducted to determine the extraction
diluent. The internal standard (IS) was prepared by diluting the pro­ efficiency for ODT. Carryover effects were evaluated by analysing blank
pranolol stock solution to 15 µg/ml with methanol. Plasma calibration plasma samples injected immediately following the elution of a standard
standards were prepared by spiking 450 µl of drug-free human blank solution of the highest concentration applied in the calibration curve.
plasma with 30 µl of working standard solutions of TRM, 30 µl of
working standard solutions of ODT, and 30 µl of IS solution (final vol­
ume 540 µl) to prepare concentrations ranging from 6.7 − 333.3 ng/ml 2.6. Validation methodology
for TRM and from 6.7 − 83.3 ng/ml for ODT. Standard curves were
freshly constructed for each day of plasma analysis. QC samples were Validation of the HPLC-UV method was performed according to the
similarly prepared by spiking the plasma samples with the relevant ICH guidelines [18].
standard solutions to obtain the specified concentrations.
2.6.1. Linearity
2.4. Extraction protocol Linearity was evaluated by constructing 6 calibration curves on
different days. Each calibration curve was constructed at 6 concentra­
Plasma samples were stored in microfuge tubes at − 80 ◦ C until tion levels for ODT (6.7, 16.7, 33.3, 50, 66.7 and 83.3 ng/ml) and TRM
analysis. The samples were thawed and vortexed prior to use. The (6.7, 66.7, 133.3, 200, 266.7 and 333.3 ng/ml), both including the limit
samples were centrifuged at 12,000 rpm for 3 min to remove insoluble of quantification (LOQ). Linearity was assessed with and without IS. The
protein. Aliquots of 450 μl were transferred to 10 ml glass centrifuge correlation between response (peak area) and analyte concentration was
tubes with screw-capped lids. The IS at 30 µl, together with 60 µl of evaluated using the correlation coefficient of the weighted linear
methanol were added to the patient samples, to give a final volume of regression line (Adjusted R2 > 0.99). The square of standard error (n =
540 µl, same as the plasma calibration standards. The same extraction 6) at each concentration level was used as a weighting factor.
protocol was applied to the plasma calibration standards and patient LOQ, the limit of quantification, is the minimum quantity of analyte
plasma samples. To extract TRM and ODT, 4 ml of diethyl ether, and 30 that can be measured accurately and precisely in an assay. This was
µl of 0.5 M NaOH, were added with vortexing for 10 min to each 540 µl defined as the lowest of a series of measured concentrations where
sample. The addition of base facilitated the transition of TRM and ODT precision and accuracy of measurement were within 20% [17,19]. LOQ
to the non-ionised states to increase extraction efficiency into the was determined for both TRM and ODT.
organic phase. The samples were centrifuged at 4000 rpm for 5 min
(Sigma 2-16PK, Osterode, Germany) at 10 ◦ C to separate the organic and 2.6.2. Precision and accuracy
aqueous phases, then stored at − 80 ◦ C for 30 min to freeze the aqueous Intraday precision analyses were determined using QC samples
phase. The organic phase was decanted into 10-ml disposable glass representing LOQ, low, intermediate and high concentrations (6.7, 33.3,
culture tubes (Livingstone, NSW, Australia) and the diethyl ether was 166.7 and 400 ng/ml for TRM and 6.7, 13.3, 40 and 100 ng/ml for ODT).
evaporated off under blowing nitrogen gas. The residue was recon­ Each sample was prepared from a separate stock solution which was not
stituted in 500 µl of methanol with vortexing for 30 s, and the solution confounded with calibration standard. Interday precision analysis was
was syringe filtered (0.2 μm × 13 mm nylon filter, Phenomenex, NSW, performed over two separate days using QC samples at 3 concentration
Australia) into a 1.5 ml microfuge tube. The glass tube was rinsed with levels (33.3, 166.7 and 400 ng/ml for TRM, and 13.3, 40 and 100 ng/ml
400 μl methanol with vortexing, and the washing after syringe filtration for ODT). Analysis was performed in triplicates, and analysed by
was also added into the microfuge tube. The solution was dried under calculating the relative standard deviation (RSD (%)) of predicted con­
nitrogen gas to complete dryness, reconstituted with 80 µl of solvent centrations. Precision between two analysts was also determined where
consisting of 14% v/v methanol in 20 mM phosphate buffer (pH 3) with both analysts performed the analyses in triplicates using QC sample at
vortexing, and centrifuged at 1200 rpm for 20 min. The clear superna­ the intermediate concentration, at 166.7 and 40 ng/ml for TRM and
tant (extract) was transferred to a 200 μl flat bottom HPLC vial insert ODT, respectively.
within an amber 1.5 ml HPLC vial and frozen at − 20 ◦ C until analysed. Accuracy analysis was conducted in triplicates using QC samples
Samples were thawed at ambient temperature before injection into the representing LOQ, low, intermediate and high concentrations (6.7, 33.3,
HPLC for analysis. 166.7 and 400 ng/ml for TRM and 6.7, 13.3, 40 and 100 ng/ml for ODT)
and expressed as follows:
2.5. Extraction efficiency Predicted concentration
Recovery(%) = × 100
Nominal concentration
Stability of TRM and ODT during the extraction and HPLC analysis
processes was assessed by determination of extraction efficiency. 2.6.3. Selectivity/specificity
Extraction of TRM and ODT was optimised by using different solvent Selectivity refers to the target analyte being sufficiently separated
ratios and duration of vortex. Triplicate QC samples at three concen­ from endogenous and exogenous matrix components. Aside from
tration levels (16.7, 200, and 333.3 ng/ml for TRM, and 6.7, 33.3, and assessing non-interference of peaks from endogenous plasma compo­
83.3 ng/ml for ODT) were used. The extraction efficiency was deter­ nents, we also sought to assess the sufficient resolution of peaks resulting
mined by comparing the peak area of TRM in the extracts of spiked from 7 drugs (exogenous components) present in the plasma sample.
plasma samples with the peak area of equivalent concentrations of TRM These 7 drugs - paracetamol, midazolam, propranolol, dexamethasone,
dissolved in methanol (non-extracted samples) using the following ondansetron, ibuprofen and parecoxib – were assessed because they,
equation: excluding propranolol (IS), are commonly co-administered with tra­
madol in children scheduled for surgery at the Perth Children’s Hospital
and may therefore interfere with a tramadol assay developed for the

3
O. Yoo et al. Journal of Chromatography B 1184 (2021) 122971

pharmacokinetic trial.
The 7 drugs, as well as TRM and ODT, were separately dissolved in
methanol to prepare 0.1 mg/ml solutions, and 1 ml from each of the
solutions were mixed to provide a combined 9 analytes solution. Sam­
ples of blank plasma, and plasma spiked with 30 μl of the combined drug
solution were extracted using a procedure similar to that described in
Section 2.4 and analysed. Potential back conversion of the ODT
metabolite was assessed by evaluating the precision and accuracy of
measurement of the ODT and TRM concentrations.

2.7. Ethics approval

The HPLC-UV assay was developed for the simultaneous quantifi­

cation of TRM and ODT in human plasma, and was integral to a pilot
paediatric pharmacokinetic study of TRM conducted at the Perth Chil­
Fig. 2. The addition of 0.5 M NaOH for the liquid–liquid extraction of plasma
dren’s Hospital. Approval for the clinical study was obtained from the
samples was necessary to improve extraction efficiency. Addition of 20 µl of 0.5
Human Research Ethics Committees of the Western Australian Child and
M NaOH resulted in a thicker interfacial layer (a) than the addition of 30 µl of
Adolescent Health Services (RGS0000000205) and the University of 0.5 M NaOH (b). A thinner interfacial layer resulted in more efficient analyte
Western Australia (RA/4/20/4049). recovery from the organic phase.

3. Results and discussion drug in diethyl ether was dried under nitrogen gas at ambient temper­
ature, and the residue reconstituted with methanol to precipitate re­
TRM absorption maxima were at 220 nm and 278 nm, while ODT sidual protein. Methanol was applied as it has been reported to be more
absorption maxima were at 220 nm and 276 nm. The signal at 220 nm effective than acetonitrile for protein precipitation from plasma samples
was the higher of the two for both TRM and ODT, however, noise [22]. The residue after protein precipitation with methanol was recon­
contributed by the endogenous plasma components was also higher at stituted into a solution. The solvent for reconstitution was selected to
220 nm, which adversely affected the assay sensitivity and the LOQ and have a similar eluent selectivity as the initial mobile phase used in the
limit of detection (LOD). For this reason, absorption at 270 nm was reverse phase HPLC-UV assay, as this was found to further improve the
selected for the analysis of both TRM and ODT. resolution of target peaks. Solvent selectivity refers to the ability of
solvent to dissolve a solute selectively in the presence of solutes of
3.1. Sample extraction similar polarity [23,24]. The initial mobile phase comprised of 14% v/v
acetonitrile in 20 mM potassium phosphate buffer (pH 3.0). Solvent
Liquid-liquid extraction, double extractions and protein precipita­ selectivity matching was performed based on the following equation
tion were examined as sample preparation methods to minimise the [25]: Ø1P’1=Ø2P’2, where Ø was the volume fraction of a solvent and P’
chromatographic noise contributed by the endogenous compounds in its reversed-phase polarity. The P’ values for methanol and acetonitrile
human plasma. In liquid–liquid extraction, we first applied ethyl acetate were 3.0 and 3.1, respectively, and the volume fraction of methanol for
and diethyl ether in a 1:1 v/v ratio as it was reported to be the most the reconstitution solvent was calculated as (Ømethanol=Øacetonitrile ×
efficient solvent for extracting TRM and ODT from human plasma [20]. P’acetonitrile/P’methanol). Thus, the residue was reconstituted with 14% v/
However, extraction with diethyl ether alone was preferred for the faster v methanol in 20 mM potassium phosphate buffer (pH 3.0). The low pH
evaporation and the extraction efficiency was comparable to that ach­ of this solvent further aided in the precipitation of residual endogenous
ieved with the solvent combination using ethyl acetate. Additionally, the proteins, which were removed by centrifugation, and the supernatant
chromatographic noise of samples extracted with diethyl ether alone was used for the HPLC analysis. The stepwise protein precipitation
was also lower than that extracted with the solvent combination. techniques, together with the matching of the reconstitution solvent to
Double extraction was applied but was not favoured as it prolonged the eluent selectivity of the initial mobile phase of the HPLC analysis,
sample processing time without improving recovery compared to single was found to greatly enhance the resolution of target peaks.
extraction. Protein precipitation with acetonitrile was also found not to
provide sufficient noise reduction, even though the application of
acetonitrile in a 2:1 v/v ratio to human plasma has been reported to 3.2. Extraction efficiency
remove more than 96% of endogenous plasma proteins [21]. In our
study, protein precipitation mediated by acetonitrile alone resulted in Mean extraction efficiency (n = 3) of TRM at 16.7, 200 and 333.3 ng/
higher chromatographic noise level than the liquid–liquid extraction ml were 115.0, 105.4, and 105.1%, respectively. Mean extraction effi­
method. ciency (n = 3) of ODT at 6.7, 33.3 and 83.3 ng/ml were 108.5, 102.1,
The addition of 0.5 M NaOH was a necessary step in the liquid–liquid and 106.0%, respectively. The extraction efficiency for both analytes
extraction process to improve extraction efficiency. The strong base showed slightly positive bias, which may be attributed to solvent dif­
converts the normally ionised TRM and ODT to nonionised forms, ferences. The solvent of analyte extracted from plasma samples was 14%
causing a more favourable transfer of the analytes from plasma (aqueous v/v methanol in 20 mM phosphate buffer (pH 3), whereas the solvent of
phase) to extraction solvent (organic phase). Increasing the volume of analyte in the non-extracted sample was 100% methanol. There are no
0.5 M NaOH from 20 µl to 30 µl led to a reduced interfacial layer guidelines for setting acceptable criteria of extraction efficiency, how­
thickness between the organic and aqueous phases (Fig. 2) that then ever, considering the extraction efficiency (%) for both analytes regis­
allowed the organic phase to be more efficiently separated from the tered less than RSD (%) of 10%, the extractions for TRM and ODT from
aqueous phase for sample recovery. Partitioning of the analytes into the the plasma samples may be regarded to be highly efficient and
organic phase was further facilitated by vortexing. Duration of vortexing reproducible.
was examined at 5, 10 and 20 min, with 10 min found optimal for
efficient extraction without unnecessarily prolonging the processing 3.3. Specificity
time.
Following liquid–liquid extraction, the organic phase containing the Chromatograms of the extracted plasma spiked with a combination

4
O. Yoo et al. Journal of Chromatography B 1184 (2021) 122971

the ionisation status of both analytes in the changing solvent composi­

tion employed in the gradient system for this assay was not straight
forward, as pKa value can be influenced by solvent dielectric constant
[28]. To improve peak symmetry, we added 0.1% v/v triethylamine
(TEA) to the mobile phase at pH 3. TEA yielded slightly better peak
symmetry for ODT, but peak deterioration was noted after a few runs
with routine care of column, making it impractical to implement. Peak
separation and peak symmetry were further improved by changing from
an Xbridge column (150 mm × 4.6 mm) to the longer BDS column (250
mm × 4.6 mm), and by optimising the gradient of the mobile phase with
acetonitrile and 20 mM phosphate buffer (pH 3). The final outcome and
retention times of the 9 analytes are shown in Fig. 3. Retention times for
TRM and ODT were 11.6 and 8.0 min, respectively, and none of the
Label RT (min) Drug coadministered drugs nor IS interfered with the TRM and ODT peaks.
P1 5.8 paracetamol
Additionally, none of the peaks of the other drugs interfered with each
P2 8.0 ODT
P3 10.7 other.
midazolam
17.2
P4 11.6 tramadol 3.4. Linearity and LOQ
P5 13.3 ondansetron
P6 16.0 propranolol
P7 18.6 dexamethasone For both TRM and ODT, the absolute error for peak AUC increased
P8 22.3 parecoxib with concentration, which resulted in relatively constant relative error
P9 23.9 ibuprofen values, within 15% for accuracy and precision, across the concentration
levels employed in the standard curve (Tables 2 and 3). This suggests
Fig. 3. HPLC-UV chromatograms showing tramadol and ODT peaks separated
that the data was heteroscedastic, where the proportionate increase in
from those of seven other drugs and internal standard (propranolol) for a
absolute error with concentration rendered a relatively constant value
human plasma sample spiked with the 9 analytes. Analytes were detected at the
wavelength of 270 nm. for the relative error across the calibration concentration levels [29]. To
correct for the bias, as the larger concentrations would have greater
impact on the standard regression line, the absolute error was weighted
of TRM, ODT, paracetamol, midazolam, ondansetron, propranolol,
to allow each concentration point to equally contribute to the regression
dexamethasone, parecoxib and ibuprofen, showed baseline separation
line. The weighting factor applied was the square of standard error, and
of all peaks without interferences (Fig. 3). Examples of HPLC chro­
concentration was calculated based on the weighted regression line.
matograms of blank plasma spiked with IS, plasma from a control, and a
Percent RSD (n = 6) at all concentrations was within 15%, with the
patient plasma sample are shown in Fig. 4. To achieve this, we trialled
exception of the RSD (%) for LOQ of ODT (Table 3), which while higher
using mobile phases at pH 3 and pH 6, and found the AUC of the ODT
at 17.47% was still within the acceptable ICH range of 20% [18].
peak to be improved at the lower pH. The pH difference did not affect
The inclusion of IS did not improve linearity; the calibration curves
the retention times of TRM and ODT. While TRM (pKa 9.41) [26] and
obtained with and without IS both showed good correlation coefficients
ODT (pKa 9.62) [27] might be considered to be ionised at pH 3 and pH 6,
(>0.99) (Table 2). Analysis with IS, however, resulted in the RSD (%) of

Fig. 4. Representative HPLC chromatograms of: (a)

blank plasma spiked with IS (1 µg/ml); (b) control
blank plasma spiked with tramadol (33.3 ng/ml), ODT
(100 ng/ml) and IS propranolol (1 µg/ml); (c) a plasma
sample (tramadol 47.7 ng/ml and ODT 31.8 ng/ml)
obtained from a child participant in a pharmacokientic
trial at 0.27 h after oral administration of tramadol at
dose of 1 mg/kg and spiked with 1 µg/ml IS. The
retention times of peaks 1 (ODT), 2 (tramadol), and 3
(propranolol) are identified at the highlighted
positions.

5
O. Yoo et al. Journal of Chromatography B 1184 (2021) 122971

Table 2
Calibration parameters and standard errors of weighted regression lines for tramadol (TRM) and its metabolite (ODT) assayed with and without propranolol, the
internal standard (IS). Calibration was performed using 6 levels of concentrations for ODT and TRM.
Analyte Conc Range (ng/ml) Weighted Regression Equation Adjusted R2 Standard error × 103 for

Slope Intercept

Without IS TRM 6.7–333.3 y = 0.075x-0.14 0.9995 0.78 24

ODT 6.7–83.3 y = 0.060x + 0.21 0.9972 1.4 29
With IS TRM 6.7–333.3 y = 0.0011x + 0.0026 0.9978 0.023 0.95
ODT 6.7–83.3 y = 0.0011x + 0.0024 0.9977 0.027 0.48

A limitation of the assay is that it did not take into account the me­
Table 3
tabolites and degradation products of all the drugs analysed. In practice,
Statistical evaluation of the tramadol (TRM) and its metabolite (ODT) calibra­
the analysis of all potential interfering molecules resulting from co-
tion curves obtained over 6 days, without propranolol as internal standard.
administered drugs is difficult to achieve due to cost, as well as the
Nominal Predicted Precision Accuracy
lack of knowledge and availability of these molecules. Notwithstanding
concentration concentration (RSD (%)) (Recovery
(ng/ml) (ng/ml) (%))
this limitation, the developed method had been successfully applied to
analyse the blood samples of 137 paediatric patients enrolled in a tra­
Tramadol 6.7 6.6 ± 0.8 12.4 98.9
madol pharmacokinetic study. Another limitation was the use of adult
(n = 6) 66.7 70.5 ± 9.9 14.0 105.8
133.3 137.2 ± 10.3 7.5 102.9 plasma for the method development when the final assay was applied to
200.0 204.3 ± 12.0 5.9 102.1 the analysis of paediatric plasma samples. It was observed that the adult
266.7 263.2 ± 11.7 4.4 98.7 plasma samples were visually more turbid and presented a larger
333.3 329.2 ± 16.1 4.9 98.8
organic phase after centrifugation than the paediatric plasma samples.
ODT (n = 6.7 6.2 ± 1.1 17.5 93.4
6) 16.7 16.2 ± 2.1 12.8 97.1
This was unavoidable as it would be unethical and impractical to recruit
33.3 36.4 ± 5.3 14.6 109.3 paediatric patients to donate plasma for the assay development and
50.0 49.1 ± 4.1 8.3 98.2 validation. Our method appeared to have mitigated the compositional
66.7 68.2 ± 6.6 9.7 102.3
83.3 80.8 ± 11.1 13.7 97.0
Table 4
Intraday precision and accuracy for the assays of tramadol (TRM) and its
nominal concentration being 2–5% higher than analysis without IS, metabolite (ODT).
where the RSD (%) remained within 15% of nominal value (Table 3).
Nominal concentration Precision (RSD Accuracy
The accuracy could be confounded by errors introduced through the (ng/ml) (%)) (Recovery (%))
additional sample processing steps to include the IS in the assay.
Tramadol (n 6.7 7.9 91.4
Random measuring error (noise) in this study could be associated with ¼ 3) 33.3 2.4 94.6
weight and volume measurements in the preparation of standard and IS 166.7 3.9 100.9
stock solutions, as well as volume measurements of samples for the 400 2.9 94.5
HPLC analysis. The noise associated with each step had a cumulative ODT (n ¼ 3) 6.7 9.5 103.0
13.3 2.9 101.0
effect to confound the final HPLC data. Thus, the accuracy of the assay
40 2.5 96.1
could be compounded by the additional sample processing steps 100 1.8 96.4
required to add IS to the samples. In addition, while propranolol has
been successfully applied as IS for TRM and ODT assays [12,30], pro­
pranolol (Fig. 1) showed relatively larger absorbance at 270 nm than
TRM and ODT in our study. A slight change in propranolol AUC there­ Table 5
fore had a relatively significant impact on the TRM (or ODT) to pro­ Interday precision and accuracy for the assays of tramadol (TRM) and its
pranolol AUC ratio, and consequently the TRM (or ODT) predicted metabolite (ODT).
concentration. These errors associated with the use of IS in HPLC anal­ Nominal concentration (ng/ml) Precision Accuracy
ysis are not unusual [31,32], and the high extraction efficiency obtained (RSD (%)) (Recovery (%))
in this study indicated that the assay was satisfactory without using IS.
Tramadol 33.3 2.4 100.7
For these reasons, the validation of the HPLC-UV assay was subsequently
(n ¼ 6) 166.7 4.4 98.5
proceeded without inclusion of propranolol as IS. 400 3.2 94.4
LOQ was 6.67 ng/ml for both TRM and ODT (RSD (%) < 20%). This ODT 13.3 3.1 97.1
was lower than the value of 10 ng/ml obtained by Gan et al., who (n ¼ 6) 40 2.3 99.1
assessed the LOQ of TRM using a similar method [17], as well as the 100 2.6 94.3

value of 60 ng/ml based on a signal to noise ratio obtained in another

HPLC-UV assay [10].
Table 6
3.5. Precision and accuracy Precision and accuracy for the assays of tramadol (TRM) and its metabolite
(ODT) performed by two different analysts.
ICH guidelines for assay validation require both the precision and ODT (Nominal concentration 40 TRM (Nominal concentration
accuracy at LOQ to be within 20%, and the precision and accuracy at ng/ml plasma, n = 3) 166.7 ng/ml plasma, n = 3)

other analyte concentrations to be within 15% [18,19]. The RSD (%) of Precision Accuracy Precision Accuracy
interday, intraday and analyst precision ranged from 1.8 to 9.5 for ODT, (RSD (%)) (Recovery (%)) (RSD (%)) (Recovery (%))
and 0.6 to 7.9 for TRM, while the accuracy ranged from 94.3 to 108.6 for Analyst 6.3 100.8 0.6 97.3
ODT, and from 91.4 to 106.4 for TRM (Tables 4–6). Precision and ac­ 1
curacy errors for both the TRM and ODT assays were within the ICH Analyst 2.5 108.6 2.1 106.4
2
guidelines, suggesting that the HPLC-UV assay is robust.

6
O. Yoo et al. Journal of Chromatography B 1184 (2021) 122971

differences between adult and paediatric plasma as it was able to be References

applied successfully to the analysis of the plasma samples of paediatric
participants in the tramadol pharmacokinetic trial. [1] M.D. Vickers, D.O’ Flaherty, S.M. Szekely, M. Read, J. Yoshizumi, Tramadol: pain
relief by an opioid without depression of respiration, Anaesthesia 47 (1992)
291–296.
4. Conclusion [2] W.N. Wu, L.A. McKown, S. Liao, Metabolism of the analgesic drug ULTRAM ®
(tramadol hydrochloride) in humans: API-MS and MS/MS characterization of
metabolites, Xenobiotica 32 (2002) 411–425.
An HPLC-UV method to simultaneously quantify TRM and its main [3] S. Grond, A. Sablotzki, Clinical pharmacology of tramadol, Clin. Pharmacokinet. 43
metabolite (ODT) in the presence of paracetamol, midazolam, ondan­ (2004) 879–923.
[4] K.A. Payne, J.A. Roelofse, E.A. Shipton, Pharmacokinetics of oral tramadol drops
setron, propranolol (IS), dexamethasone, parecoxib and ibuprofen was for postoperative pain relief in children aged 4 to 7 years–a pilot study, Anesthesia
successfully developed and validated. The assay did not require pro­ progress 49 (2002) 109–112.
pranolol as internal standard, as the addition of the internal standard [5] M. Vazzana, T. Andreani, J. Fangueiro, C. Faggio, C. Silva, A. Santini, M.L. Garcia,
A.M. Silva, E.B. Souto, Tramadol hydrochloride: Pharmacokinetics,
was associated with higher potential errors due to additional sample
pharmacodynamics, adverse side effects, co-administration of drugs and new drug
processing steps. Protein precipitation by acetonitrile alone did not delivery systems, Biomed. Pharmacother. 70 (2015) 234–238.
provide for sufficient noise reduction. We have instead developed a [6] South Australian Paediatric Clinical Guidelines Reference Committee, Clinical
liquid–liquid extraction coupled with stepwise protein extraction Guideline: South Australian Paediatric Practice Guidelines – Pain Management and
Opioid Safety, Adelaide, South Australia, 2020.
method that provided for effective reduction of chromatographic noise [7] M.M. Tolba, M.M. Salim, Inclusive study for segregation of two commonly used
attributable to endogenous plasma proteins. The increased resolution of anticancer drugs with tramadol: Applying a green fluorimetric strategy to
target peaks led to higher sensitivity, and a LOQ level that is lower than pharmaceutical dosage forms and human plasma, Microchem. J. 162 (2021),
105859.
those reported for HPLC-UV assay of TRM and ODT. As the HPLC-UV [8] M. Taghizade, M. Ebrahimi, E. Fooladi, M. Yoosefian, Simultaneous
equipment is widely used in clinical laboratories due to its relatively spectrophotometric determination of the residual of ciprofloxacin, famotidine, and
lower cost compared to LC/MS, the assay developed in this study offers a tramadol using magnetic solid phase extraction coupled with multivariate
calibration methods, Microchem. J. 160 (2021), 105627.
cost-effective, precise and accurate alternative for the analysis of TRM [9] N. Fernández, N.M. Olivera, G.A. Keller, R.A. Diez, G. Di Girolamo, P.N. Quiroga,
and its main metabolite in human pharmacokinetics and toxicology Simultaneous quantitation of meperidine, normeperidine, tramadol, propoxyphene
studies. and norpropoxyphene in human plasma using solid-phase extraction and gas
chromatography/mass spectrometry: Method validation and application to
cardiovascular safety of therape, Rapid Commun. Mass Spectrom. 31 (2017)
CRediT authorship contribution statement 1519–1533.
[10] H.E. Dahshan, M.A. Helal, S.M. Mostafa, M.S. Elgawish, Development and
validation of an HPLC-UV method for simultaneous determination of sildenafil and
Ms Yoo drafted the manuscript, participated in the conceptualiza­ tramadol in biological fluids: Application to drug-drug interaction study, J. Pharm.
tion, data curation and formal analysis. She also participated in the Biomed. Anal. 168 (2019) 201–208.
validation of the assay, data visualization and revised the manuscript. Dr [11] A.H. Almalki, N.A. Ali, F.A. Elroby, M.R. El Ghobashy, A.A. Emam, I.A. Naguib,
ESI–LC–MS/MS for Therapeutic Drug Monitoring of Binary Mixture of Pregabalin
Tang participated in the conceptualization, data curation and formal and Tramadol: Human Plasma and Urine Applications, Separations 8 (2021) 21.
analysis. She also participated in validation of the assay, data visuali­ [12] B.N. Patel, N. Sharma, M. Sanyal, P.S. Shrivastav, An accurate, rapid and sensitive
zation and revised the manuscript. Ms Nguyen participated in validation determination of tramadol and its active metabolite O-desmethyltramadol in
human plasma by LC-MS/MS, J. Pharm. Biomed. Anal. 49 (2009) 354–366.
of the assay, and the development of methodology. Dr Salman super­ [13] D. Baconi, M. Stan, Z.A.J. Ebrahim, C. Tuchila, C. Balalau, Determination of
vised methodology of the assay and reviewed the manuscript. Mr Hua Tramadol in human plasma by HPLC with fluorescence detection, J. Mind Med. Sci.
participated in formal data analysis. Prof Regli-von Ungern-Sternberg 3 (2016) 55–64.
[14] S. Miriana, E.Z.A. Jalil, C. Negrei, C. Balalau, B. Daniela, HPLC-FL method for
participated in funding acquisition and project administration. She tramadol quantification in human plasma, Toxicol. Lett. 238 (2015) S378–S379.
provided resources, supervised the plasma collection and reviewed the [15] M. Nobilis, J. Kopecky, J. Kvetina, J. Chladek, Z. Svoboda, V. Vorisek, F. Perlik,
manuscript. Prof Lim conceptualized the study, coordinated and su­ M. Pour, J. Kunes, High-performance liquid chromatographic determination of
tramadol and its O-desmethylated metabolite in blood plasma - Application to a
pervised data curation, and reviewed and edited the manuscript. Prof
bioequivalence study in humans, J. Chromatogr. A 949 (2002) 11–22.
Lim participated in funding acquisition, the development of methodol­ [16] M.M. Moein, A. El Beqqali, M. Abdel-Rehim, Bioanalytical method development
ogy, project administration and provision of resources. All authors and validation: Critical concepts and strategies, J. Chromatogr. B 1043 (2017)
3–11.
approved the final manuscript as submitted and agree to be accountable
[17] S.H. Gan, R. Ismail, W.A.W. Adnan, Z. Wan, Method development and validation of
for all aspects of the work. a high-performance liquid chromatographic method for tramadol in human plasma
using liquid-liquid extraction, J. Chromatogr. B 772 (2002) 123–129.
[18] European Medicines Agency, ICH Guideline M10 on Bioanalytical Method
Validation, 2019.
Declaration of Competing Interest [19] Food and Drug Administration Center for Drug Evaluation Research, Bioanalytical
Method Validation; Guidance for Industry, FDA Mayland, 2018.
The authors declare that they have no known competing financial [20] B. Yilmaz, A.F. Erdem, Simultaneous Determination of Tramadol and its Metabolite
in Human Plasma by GC/MS, J. AOAC Int. 98 (2015) 56–61.
interests or personal relationships that could have appeared to influence [21] C. Polson, P. Sarkar, B. Incledon, V. Raguvaran, R. Grant, Optimization of protein
the work reported in this paper. precipitation based upon effectiveness of protein removal and ionization effect in
liquid chromatography–tandem mass spectrometry, J. Chromatogr. B 785 (2003)
263–275.
Acknowledgements [22] T.M. Alshammari, A.A. Al-Hassan, T.B. Hadda, M. Aljofan, Comparison of different
serum sample extraction methods and their suitability for mass spectrometry
analysis, Saudi Pharm J 23 (2015) 689–697.
Funding from the Australian and New Zealand College of Anaes­
[23] A.R. Johnson, M.F. Vitha, Chromatographic selectivity triangles, J. Chromatogr. A
thetists (ANZCA) John Boyd Craig Research Award, Telethon-Perth 1218 (2011) 556–586.
Children’s Hospital Research Fund, University of Western Australia [24] L.R. Snyder, Classification of the solvent properties of common liquids,
Pathfinder Fund and Perth Children’s Hospital Foundation (HPLC J. Chromatogr. A 92 (1974) 223–230.

equipment) are gratefully acknowledged.

7
O. Yoo et al. Journal of Chromatography B 1184 (2021) 122971

[25] V. Meyer, Practical high-performance liquid chromatography, fourth ed., John regression of HPLC convoluted peak responses using internal standard method in
Wiley and Sons, Chichester, 2004. drug bioavailability studies, Anal. Bioanal. Chem. 405 (2013) 4835–4848.
[26] World Health Organization, Tramadol Update Review Report, Geneva, 2014. [30] M.A. Hilal, K.M. Mohamed, Simultaneous Determination of Tramadol and O-
[27] D.S. Wishart, C. Knox, A.C. Guo, S. Shrivastava, M. Hassanali, P. Stothard, Desmethyltramadol in Human Plasma Using HPLC-DAD, J. Chromatogr. Sci. 52
Z. Chang, J. Woolsey, DrugBank: a comprehensive resource for in silico drug (2014) 1186–1192.
discovery and exploration, Nucleic Acids Res. 34 (2006) D668–D672. [31] S. Imre, A. Tero-Vescan, M.T. Dogaru, L. Kelemen, D.-L. Muntean, A. Curticăpean,
[28] J. Reijenga, A. Van Hoof, A. Van Loon, B. Teunissen, Development of Methods for N. Szegedi, C.-E. Vari, With or Without Internal Standard in HPLC Bioanalysis. A
the Determination of pKa Values, Anal. Chem. Insights 8 (2013) ACI.S12304. Case Study, J. Chromatogr. Sci. 57 (2019) 243–248.
[29] M.A. Korany, H.M. Maher, S.M. Galal, M.A.A. Ragab, Comparative study of some [32] P. Haefelfinger, Limits of the internal standard technique in chromatography,
robust statistical methods: weighted, parametric, and nonparametric linear J. Chromatogr. A 218 (1981) 73–81.