Journal of Chromatography B, 947–948 (2014) 83–95

Contents lists available at ScienceDirect

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

Screening and quantitative determination of drugs of abuse in diluted

urine by UPLC–MS/MS
Solfrid Hegstad a,∗ , Sigurd Hermansson b , Ingvar Betnér b ,
Olav Spigset a,c , Berit Margrethe Hasle Falch a
a
Department of Clinical Pharmacology, St. Olav University Hospital, Trondheim, Norway
b
Waters Corporation Sweden AB, Sollentuna, Sweden
c
Department of Laboratory Medicine, Women’s and Children’s Health, Norwegian University of Science and Technology, Trondheim, Norway

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

Article history: The purpose of this work was to develop and evaluate a fast, robust and specific UPLC–MS/MS screening
Received 1 July 2013 platform for the determination and quantification of a variety of commonly used drugs of abuse in
Accepted 14 December 2013 urine, i.e. a high-throughput quantitative analysis. Substances in the drug classes opioids, central nervous
Available online 25 December 2013
system stimulants and benzodiazepines and related agents were included in addition to cannabis and
pregabalin, a total of 35 different analytes. Based on the concentrations and the physico-chemical proper-
Keywords:
ties of the substances, three UPLC–MS/MS methods were developed in parallel. Prior to analysis, sample
UPLC–MS/MS
preparation consisted of two different simple dilutions with 60 and 100 ␮L urine, respectively, using a
Opioids
Benzodiazepines
Tecan Freedom Evo pipetting robot platform. A Waters Xevo TQ-S tandem quadrupole mass spectrometer
Central nervous system stimulants coupled to a Waters I-class UPLC was used for quantitative analysis of one quantitative and one qualifying
Drugs of abuse MRM transition for each analyte, except for tramadol for which the metabolite O-desmethyl-tramadol
Screening was included in the MRM method to confirm tramadol identity. Deuterated analogs were included as
internal standards. The between-assay relative standard deviations varied from 2% to 11% and the limits
of quantification were in the range 1–200 ng/mL for the various analytes. After development and initial
testing, the method has been successfully implemented and routinely used at our hospital for quantitative
screening of drugs of abuse in more than 35,000 urinary samples.
© 2013 Elsevier B.V. All rights reserved.

1. Introduction chromatographic peaks are very narrow. UPLC–MS/MS provides

The use of liquid chromatography–tandem mass spectrometry
(LC–MS/MS) is increasing in the clinical laboratory for the analysis
of drugs of abuse [1]. Improved sensitivity and specificity makes
LC–MS/MS useful for simultaneous analysis of a wide variety of
drugs, thus being ideal for drug screening with quantification.
Several methods have previously been published for the
screening of drugs and medications in urine with approaches
for sample preparation ranging from traditional liquid–liquid and
solid phase extraction to online extraction and direct injection
of diluted samples [1–5]. Ultra-performance liquid chromatogra-
phy (UPLC)–MS/MS technology allows a high resolution separation
of multiple analytes, requiring relatively short run-times. The
fast data acquisition rate of the MS/MS detector enables a suf-
ficient number of data points per peak even though the UPLC
∗ Corresponding author at: Department of Clinical Pharmacology, St. Olav Univer-
sity Hospital, 7006 Trondheim, Norway. Tel.: +47 72 57 30 00.
E-mail address: solfrid.hegstad@stolav.no (S. Hegstad).
sufficient selectivity and sensitivity to perform direct injections
of diluted urine samples without any time-consuming sample
preparation in advance. However, matrix effects induced by the
LC–MS/MS method can adversely affect the results when direct
injection of a urine sample is applied [6], either by reducing
the analytical response (ion suppression) or by increasing it (ion
enhancement). As ion suppression in the worst case can cause false
negative results, it is of crucial importance to minimize this effect.
Several approaches can be used with this respect. The more effi-
cient chromatographic separation with UPLC compared to HPLC
and the use of stable isotope labeled internal standards will reduce
the matrix effect [7,8]. Moreover, the increased sensitivity of the
MS/MS detection allows a higher degree of sample dilution, thereby
reducing the matrix effect.
Multi-component screening using UPLC–MS/MS has been
described in the literature for the determination of drugs of
abuse and medications in diluted urine. For example, Thorngren
et al. have described a UPLC–MS/MS method of 130 analytes with
only a 1:1 dilution before analysis [9], whereas Eichhorst et al.
have presented a method of 40 analytes using 1:10 dilution and

1570-0232/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.jchromb.2013.12.014
84 S. Hegstad et al. / J. Chromatogr. B 947–948 (2014) 83–95
enzymatic hydrolysis of the urine before analysis [10]. Recently Al-
Saffar et al. published an analytical method of 26 new psychoactive
drugs within the group of legal highs, with a 1:5 dilution [11].
9 -Tetrahydrocannabinol carboxylic acid (THCA) and
buprenorphine are particularly challenging analytes due to
more than 10-fold–100-fold lower concentrations than most other
drugs and medications in urine. In addition, buprenorphine has an
unfavorable fragmentation in the ion source [12]. Therefore, these
drugs are often not included in methods aiming to identify several
drugs in parallel. For example, only THCA and not buprenor-
phine was included in method presented by Eichhorst et al. [10].
Although direct LC–MS/MS methods of THCA as the only analyte
in urine have been reported [13,14], we have not identified any
previous studies describing a method for the detection of THCA
and buprenorphine in addition to other common drugs of abuse
using simple sample dilution and LC–MS/MS.
Another challenge in the development of an analytical urinary
method is the presence of glucuronide metabolites. Often, this
issue has been approached by applying glucuronide hydrolysis to
the sample, either enzymatically, alkaline or by acid hydrolysis.
However, as these procedures are often incomplete, variable con-
centrations of the parent substance and transformation to other
analytes might be the result, with additional uncertainty in the clin-
ical interpretation of the results [15–18]. During the last decade,
glucuronide metabolites are more regularly included in the analy-
ses, such as morphine 3-glucuronide and morphine 6-glucuronide
[19,20]. One advantage by including glucuronides is that the pres-
ence of them represents additional evidence for intake of the parent
substance. On the other hand, glucuronide standards are often
expensive and the supply situation is not always stable.
The Department of Clinical Pharmacology at St. Olav Univer-
sity Hospital analyses about 50,000 urine samples per year. Most
of these samples are from opioid addicted individuals treated
with buprenorphine or methadone in a rehabilitation program.
The samples are analyzed to confirm the use of buprenorphine
or methadone and to monitor possible intake of other drugs of
abuse, either illegally consumed or in some cases prescribed. The
quantitative results are (after adjustment based on the concentra-
tion of creatinine in the samples) used to differentiate new intake
from residual drug excretion caused by a former intake [21,22].
Based upon these prerequisites, the aim of the present study was
to develop a simple, robust and specific UPLC–MS/MS screening
platform for the determination and quantification of a variety of
commonly used drugs of abuse in urine, suitable for implementa-
tion as a high-throughput routine method. Due to the wide range of
physico-chemical properties of the analytes, the differences in uri-
nary concentrations between analytes and the desire of performing
glucuronide hydrolysis for some analytes but not for others, we
have developed two methods for sample preparation with three
different UPLC–MS/MS methods (Fig. 1).
2. Materials and methods
2.1. Chemicals and reagents
The reference substances were purchased from the fol-
lowing companies: amphetamine, benzoylecgonine (BECG),
buprenorphine, codeine, codeine 6-glucuronide (C6G), ethylmor-
phine, ␣-hydroxyalprazolam (OH-alprazolam), 3,4-methylene-
dioxyamphetamine (MDA), 3,4-methylenedioxymethampheta-
mine, (MDMA), 6-monoacetylmorphine (6-MAM), morphine,
morphine 3-glucuronide (M3G), paramethoxyamphetamine
(PMA) and paramethoxymethamphetamine (PMMA) were from
Lipomed AG (Arlesheim, Switzerland); 7-aminoclonazepam (7-
AC), 7-aminoflunitrazepam (7-AF), 7-aminonitrazepam (7-AN),
Method 1 Methods 2 and 3
60 µL urine 100 µL urine
Total dilution 1:250 Enzymatic hydrolysis
Total dilution 1:10
UPLC method 1 UPLC method 2 UPLC method 3
Initial conditions: Initial conditions: (THCA)
98% formic acid (0.1 %) 98% formic acid (0.1%) Initial conditions:
in water in water 40% formic acid (0.1%)
2% methanol 2% acetonitrile in water
60% acetonitrile
MS/MS method 1 MS/MS method 2 MS/MS method 3
ESI + ESI + (THCA) ESI –
Fig. 1. Sample preparation and UPLC–MS/MS conditions.
morphine 6-glucuronide (M6G), oxazepam, ritalinic acid and
9 -tetrahydrocannabinol carboxylic acid (THCA) were from Chi-
ron (Trondheim, Norway); fentanyl, hydromorphone, ketamine,
methamphetamine, oxycodone, pethidine, tramadol and zolpi-
dem were from Sigma-Aldrich (Saint Louis, MO); pregabalin
and zopiclone were from Sequoia (Berkshire, UK); desmethyl-
diazepam (DMD) was from Synfine Research (Ontario, Canada);
ephedrine was from Norsk Medisinaldepot AS (Oslo, Norway); O-
desmethyltramadol (O-DM-tramadol) was from Grunenthal GmbH
(Aachen, Germany) and methadone was from St. Olav University
Hospital Pharmacy (Trondheim, Norway). The internal standards
7-AC-d4 , amphetamine-d3 , BECG-d3 , codeine-d3 , methadone-d3 ,
methamphetamine-d5 , MDMA-d5 , 6-MAM-d3 , morphine-d3 ,
M3G-d3 and oxazepam-d5 and were from Lipomed AG (Arlesheim,
Switzerland); 7-AF-d3 , DMD-d5 , ephedrine-d6 , ethylmorphine-d5 ,
fentanyl-d5 , hydromorphone-d3 , MDA-d5 , OH-alprazolam-d5 ,
oxycodone-d6 , ritalinic acid-d9 , zolpidem-d6 and zopiclone-d8
were from Chiron (Trondheim, Norway); buprenorphine-d4 was
from Synfine Research (Ontario, Canada); pregabalin-d4 was from
C/D/N Isotopes Inc. (Quebec, Canada); tramadol-d6 was from
Toronto Research Chemical Inc. (Ontario, Canada) and THCA-d3
was from Cerilliant (Round Rock, TX). The enzyme ß-glucuronidase
(Type 2 HP-2 from Helix pomatia, ≥100,000 units/mL) was obtained
from Sigma-Aldrich (Saint Louis, MO). External quality control
samples (Liquichek Urine Toxicology Control, Level C4) were
purchased from Biorad, (Irvine, CA).
LC–MS grade methanol and acetonitrile were purchased from
Merck (Darmstadt, Germany) and formic acid 98%, ammonium
acetate 99% and acetic acid 100% were from VWR (Leuven,
Belgium). All water used was provided from a Millipore A10 Synthe-
sis filtering system (Billerica, MA). 96-well sample collection plates
(1 mL) and polypropylene cap mat round wells for 96-well plates
were purchased from Waters (Milford, MA).
2.2. Preparation of solutions
The stock solutions were dissolved in methanol, with the follow-
ing exceptions: C6G, M3G and M6G stock solutions were dissolved
in methanol/water (50:50, v/v); zopiclone stock solutions were
dissolved in acetonitrile and amphetamine, hydromorphone, and
pregabalin stock solutions were prepared in water. The stock solu-
tions were fortified with methanol, except for the glucuronides
(C6G, M3G and M6G) and zopiclone, which were prepared in
methanol/water (50:50, v/v) and acetonitrile, respectively. The
stock solutions were stored at 4 ◦ C, except for 6-MAM and rital-
inic acid, which were stored at −20 ◦ C. The concentrations of stock
solutions were in the range of 0.5–10 mg/mL.
Method 1: Analytes included in method 1 are presented in
Table 1. Due to instability of tramadol and zopiclone around pH
 S. Hegstad et al. / J. Chromatogr. B 947–948 (2014) 83–95 85

Table 1
Analyte and internal standard transition ions and associated mass spectrometric parameters (cone voltage, collision energy) for method 1 (ESI+).

Analyte Transition (m/z) Cone voltage (V) Collision energy (eV) Internal standard

7-AC 286 > 121 50 30 7-AC-d4

286 > 222 50 25
7-AF 284 > 135 50 25 7-AF-d3
284 > 227 50 25
7-AN 252 > 121 50 25 7-AC-d4
252 > 146 50 26
Amphetamine 136 > 119 20 8 Amphetamine-d3
136 > 91 20 15
BECG 290 > 168 70 25 BECG-d3
290 > 105 70 28
Codeine 300 > 215 50 25 Codeine-d3
300 > 165 50 40
C6G 476 > 300 65 28 Codeine-d3
476 > 215 65 38
Ephedrine 166 > 117 20 10 Ephedrine-d6
166 > 148 20 18
Ketamine 238 > 125 20 50 Zolpidem-d6
238 > 189 20 18
Methadone 310 > 223 50 18 Methadone-d3 ,
310 > 105 50 60
Methamphetamine 150 > 119 20 15 Methamphetamine-d5
150 > 91 20 40
MDMA 194 > 163 20 12 MDMA-d5
194 > 105 20 22
Morphine 286 > 201 50 25 Morphine-d3
286 > 165 50 40
M3G 462 > 286 65 30 M3G-d3
462 > 201 65 40
M6G 462 > 286 65 30 M3G-d3
462 > 201 65 40
O-DM-tramadol 250 > 58 30 14 Tramadol-d6
Oxycodone 316 > 241 50 25 Oxycodone-d6
316 > 256 50 25
Pethidine 248 > 174 90 30 Zolpidem-d6
248 > 70 90 40
Pregabalin 160 > 97 20 12 Pregabalin-d4
160 > 142 20 17
Ritalinic acid 220 > 174 40 20 Ritalinic acid-d9
220 > 56 40 40
Tramadol 264 > 58 50 40 Tramadol-d6
Zolpidem 308 > 92 90 70 Zolpidem-d6
308 > 248 60 35
Zopiclone 389 > 245 20 18 Zopiclone-d8
389 > 217 20 33
7-AC-d4 290 > 226 50 25
7-AF-d3 287 > 138 50 25
Amphetamine-d3 139 > 92 20 16
BECG-d3 293 > 171 40 18
Codeine-d3 303 > 215 40 24
Ephedrine-d6 172 > 154 20 20
Methadone-d3 , 313 > 268 24 14
Methamphetamine-d5 155 > 92 30 15
MDMA-d5 199 > 165 20 12
Morphine-d3 289 > 201 50 24
M3G-d3 465 > 289 65 30
Oxycodone-d6 322 > 262 26 24
Pregabalin-d4 164 > 101 20 12
Ritalinic acid-d9 229 > 93 50 25
Tramadol-d6 270 > 64 42 15
Zolpidem-d6 314 > 263 60 27
Zopiclone-d8 397 > 245 30 16

7, the calibrators and quality control (QC) samples were pre-
pared in three different standard solutions. For tramadol and
zopiclone the calibrators and QC samples were prepared in blank
urine, adjusted to pH 8.0 and pH 5.0, respectively. For practi-
cal purposes ketamine, O-DM-tramadol, pethidine and zolpidem
were prepared in the same calibrators and QC samples as tra-
madol. For the remaining compounds in method 1, calibrators
and QC samples were prepared in blank urine (pH 6.2–6.7). The
three different standards solutions were prepared in four sets of
calibrators and three sets of QC. The three different standards solu-
tions (20 ␮L of each) were mixed by a Tecan Freedom Evo 200
pipetting robot before each analysis. The solutions used for calibra-
tor and QC samples were stored at −20 ◦ C. The internal standard
was diluted with methanol/water (50:50, v/v) and contained
150 ng/mL 7-AC-d4 , 75 ng/mL 7-AF-d3 , 50 ng/mL amphetamine-d3 ,
50 ng/mL BECG-d3 , 100 ng/mL codeine-d3 , 100 ng/mL ephedrine-
d6 , 25 ng/mL methadone-d3 , 15 ng/mL methamphetamine-d5 ,
50 ng/mL MDMA-d5 , 150 ng/mL morphine-d3 , 150 ng/mL M3G-
d3 , 150 ng/mL oxycodone-d6 , 2500 ng/mL pregabalin-d4 , 75 ng/mL
ritalinic acid-d9 , 100 ng/mL tramadol-d6 , 100 ng/mL zolpidem-d6
and 100 ng/mL zopiclone-d8. The solution was stored at 4 ◦ C until
use.
86 S. Hegstad et al. / J. Chromatogr. B 947–948 (2014) 83–95

Table 2
Analyte and internal standard transition ions and associated mass spectrometric parameters (cone voltage, collision energy and) for method 2 (ESI− for THCA/THCA-d3, ESI+
for all other analytes).

Analyte Transition (m/z) Cone voltage (V) Collision energy (eV) Internal standard

Buprenorphine 468 > 396 60 42 Buprenorphine-d4

468 > 414 60 36
DMD 272 > 141 50 28 DMD-d5
272 > 166 50 28
Ethylmorphine 315 > 166 50 40 Ethylmorphine-d5
315 > 230 50 25
Fentanyl 338 > 106 54 55 Fentanyl-d5
338 > 189 54 35
Hydromorphone 287 > 158 50 30 Hydromorphone-d3
287 > 228 50 30
MDA 180 > 133 20 17 MDA-d5
181 > 164 20 10
6-MAM 328 > 165 50 35 6-MAM-d3
328 > 211 50 25
OH-alprazolam 326 > 298 45 28 OH-alprazolam-d5
326 > 206 45 42
Oxazepam 289 > 165 50 40 Oxazepam-d5
288 > 242 70 40
PMA 166 > 121 20 16 Ethylmorphine-d5
166 > 149 20 16
PMMA 181 > 122 25 20 Ethylmorphine-d5
181 > 150 25 9
THCA 343 > 299 45 20 THCA-d3
343 > 245 45 30
Buprenorphine-d4 472 > 400 60 42
DMD-d5 276 > 140 50 28
Ethylmorphine-d5 319 > 234 50 25
Fentanyl-d5 343 > 188 54 22
Hydromorphone-d3 289 > 185 50 30
OH-alprazolam-d5 330 > 302 45 28
MDA-d5 185 > 168 20 10
6-MAM-d3 331 > 211 60 23
Oxazepam-d5 292 > 246 50 20
THCA-d3 346 > 302 45 20

Methods 2 and 3: Analytes included in methods 2 and 3 are pre-
sented in Table 2. All analytes were included in one set of calibrators
and QC samples (buprenorphine, DMD, ethylmorphine, fentanyl,
MDA, 6-MAM, OH-alprazolam, oxazepam, PMA, PMMA, and THCA).
The solutions used for calibrator and QC samples were stored at
−20 ◦ C.
The internal standard was diluted with methanol/water (50:50,
v/v) and contained 50 ng/mL buprenorphine-d4 , 300 ng/mL DMD-
d5 , 300 ng/mL ethylmorphine-d5 , 20 ng/mL fentanyl-d5 , 300 ng/mL
hydromorphone-d3 , 300 ng/mL OH-alprazolam-d5 , 300 ng/mL
MDA-d5 , 150 ng/mL 6-MAM-d3 , 300 ng/mL oxazepam-d5 and
150 ng/mL THCA-d3 . The solution was stored at 4 ◦ C until use.
2.4. UPLC conditions
A Waters Acquity UPLC I-Class FTN system (Waters, Mil-
ford, MA) was used for separation, applying an Acquity HSS
T3-column (2.1 mm ID × 100 mm, 1.8 ␮m) maintained at 50 ◦ C.
The mobile phases and their gradient profiles are shown in
Table 3. The total cycle time for all three methods combined
was 7.9 min. The pre-inject and post-inject washes were per-
formed with methanol/acetonitrile/isopropanol/water/formic acid
(25:25:25:24:1, v/v) for 3 s and 12 s, respectively. The injection vol-
ume was 5 ␮L for all three methods. A column filter (in-line filter,
0.2 ␮m, Waters) was used in the front of the analytical column in
methods 2 and 3.

2.3. Preparation of samples

Method 1: Calibrator, QC sample or urine sample (60 ␮L) was
mixed with 240 ␮L internal standard and then serially diluted 50
times in two steps (50 + 450 ␮L and 100 + 400 ␮L, respectively) with
methanol/water (10:90, v/v). The samples were mixed and diluted
in 96-well plates using a Tecan pipetting robot (Tecan Nordic, Möl-
ndal, Sweden).
Methods 2 and 3: Calibrator, QC sample or urine sample (100 ␮L)
was mixed with 20 ␮L internal standard, 60 ␮L 0.2 M ammonium
acetate buffer (pH 4.8) and 20 ␮L ß-glucuronidase (25,000 U/mL),
and incubated at 65 ◦ C for 1 h, thereby hydrolyzing glucuronide
conjugated analytes. An aliquot of 50 ␮L was diluted with 200 ␮L
methanol/water (60:40, v/v), and centrifuged at 1800 × g (Rotanta
460, Hettich Lab Technology, Tuttlingen, Germany) for 5 min. All
dilution steps were done in 96-well plates using a Tecan pipetting
robot.
2.5. MS/MS
A Xevo TQ-S tandem-quadrupole mass spectrometer (Waters,
Manchester, UK) equipped with a Z-spray electrospray interface
was used. Positive and negative electrospray ionization (ESI) was
performed in the multiple reaction monitoring (MRM) mode. The
capillary voltage was set to 1.0 kV in the positive mode and to
−3.0 kV in the negative mode. The source block temperature was
120 ◦ C, and the desolvation gas (nitrogen) was heated to 650 ◦ C and
delivered at a flow rate of 1000 L/h. The cone gas (nitrogen) was set
to 150 L/h, and the collision gas (argon) pressure was maintained
at 0.005 mbar in the collision cell. Manual tuning for optimal cone
voltage and collision energy was performed using the instrument’s
built-in fluidics system. A 10 ␮L/min flow of 100 ng/mL tuning
solution was introduced to the mass spectrometer in combination
with a LC flow of 0.2 mL/min and composition 20/80 mobile phase
 S. Hegstad et al. / J. Chromatogr. B 947–948 (2014) 83–95
Table 3
Mobile phases, total cycle times and flow rates used in the UPLC separation of the analytes.
Method 1 Method 2
Flow rate 0.6 mL/min Flow rate 0.7 mL/min
Cycle time 3.2 min Cycle time 2.7 min
Time (min) 0.1% formic Methanol (%) Time (min)
acid in water
(%)
0 98 2 0
0.5 75 25 0.8
1.9 20 80 1.5
1.91 2 98 1.9
2.4 2 98 1.91
2.41 98 2 2.0
2.5 98 2
A/mobile phase B. This was done for each of the analytes individ-
ually. The dwells times for the MRM channels were automatically
calculated by the software, given the premises that the peaks were
expected to elute over 2–3 s, and a desired number of 20 data points
determining the peak. This resulted in a MS method cycle time of
0.10–0.15 s, and dwell times ranging from 4 to 63 ms. The inter-
channel delay time was set to 3 ms and the inter-scan delay time
was 3 ms at all times.
The signal of amphetamine was unexpectedly low at the lowest
calibration level compared to methamphetamine due to fragmen-
tation of amphetamine in the ion source to the fragment ion m/z 91.
Reducing the ion source fragmentation of amphetamine increased
the signal of the lowest calibration level. Due to the high sensi-
tivity of the system it was for several compounds necessary to
detune the analytical signal to obtain linear calibration curves. This
was done by increasing cone voltage or collision energy, or mak-
ing use of a second transition with less intensity or using the C13
isotope as precursor ion of the analyte. MRM transitions, cone volt-
age and collision energy for the analytes and the internal standards
used in the presented methods are shown in Tables 1 and 2. The
optimized/detuned MS conditions are presented in the supplemen-
tary Tables 1 and 2. System operation and data acquisition were
controlled using Mass Lynx 4.1 software (Waters). All data were
processed with the Target Lynx quantification program (Waters).
The analytes were identified by comparing the retention time and
the ratio between the two MRM transitions of the corresponding
calibrator and QC samples. The retention times deviated less than
±1% from the average ratio, and the maximum deviation was set
to 20%. In addition, the identity of tramadol was verified by the
presence of O-DM-tramadol.
Supplementary material related to this article can be found,
in the online version, at http://dx.doi.org/10.1016/j.jchromb.
2013.12.014.
2.6. Method validation
The validation was done according to guidelines given by Peters
et al. [23] and the Food and Drug Administration [24].
2.6.1. Calibration curves
The four-point calibration curves (three replicates of each
standard) were based on peak area ratios of the analyte relative
to the internal standard. For amphetamine the peak height was
used instead of the area due to interfering peaks at the transition
m/z 135.9 > 91 in some of the authentic specimens tested during
method development. Quadratic calibration curves were used with
linear weighting (1/x). Origin was included for all analytes except
amphetamine, methamphetamine and methadone, for which ori-
gin was excluded.
87
Method 3
Flow rate 0.6 mL/min
Cycle time 2.0 min
0.1% formic Acetonitrile (%) Time (min) 0.1% formic Acetonitrile (%)
acid in water acid in water
(%) (%)
98 2 0 40 60
65 35 0.8 2 98
2 98 1.3 2 98
2 98 1.31 40 60
98 2
98 2
2.6.2. Limits of quantification and detection
The limit of quantification (LOQ) was determined by diluting
standard samples of different concentrations, denoted LOQ sam-
ples. LOQ samples were run in one replicate on ten different days.
A standard sample with a concentration identical to the LOQ sample
was included in the calibration curve. The signal-to-noise criteria
(using the peak-to-peak algorithm) for the LOQ samples were ≥10
and precision and bias of calculated concentrations were within
±20%. The limit of detection (LOD) was determined by further dilu-
tion until the signal-to-noise ratio reached 3 (S/N ≥ 3).
2.6.3. Precision and bias
Within-assay precision was estimated by analysis of ten sep-
arate replicates of QC samples at three concentrations in a single
assay. Between-assay precision and accuracy were determined by
analysis of one replicate at three QC concentrations on ten different
days.
2.6.4. Matrix effects
Matrix effects (ME) were evaluated by the method pro-
posed by Matuszewski et al. [25]. The analyte signal in the
spiked methanol/water solution (10:90, v/v; method 1 or 60:40,
v/v; method 2 and 3) was compared with the analyte sig-
nal in the matrix. Six replicates of urine from six different
individuals were analyzed. The concentrations corresponded to
the lowest QC level of the analytes. Spiked QC samples in
methanol/water or urine (methods 2 and 3) were prepared as
described in Section 2.3, with internal standards and analytes
added after the hydrolysis step. The ME (in percent) was calculated
as ME% = (Peak intensitymatrix /Peak intensitymethanol/water ) × 100.
Relative ME (CV%) expresses the precision of peak intensity
in matrix. ME% corrected with internal standard was cal-
culated as = [(Peak intensitymatrix /Peak IS intensitymatrix )/(Peak
intensitymethanol/water /Peak IS intensitymethanol/water )] × 100. Rela-
tive ME corrected to IS (CV %) expresses the precision of Peak
intensity/Peak IS intensity in matrix.
2.6.5. Specificity
Specificity of the methods was investigated by selecting
drugs often detected in urinary samples at our laboratory with
almost the same MH+ and MH− ions (±2 atomic mass units)
as the analytes. Substances meeting the molecular weight (MW)
criteria for the analytes included in method 1 were analyzed
by methods 2 and 3, and vice versa. The drugs/metabolites
tested were alimemazine (MW 298), carbamazepine (MW 236),
carbamazepine-10, 11-epoxide (MW 252), chlorprothixene (MW
316), citalopram (MW 324), clomipramine (MW 315), clon-
azepam (MW 316), clozapine (MW 327), desmethylclomipramine
(MW 301), desmethyldoxepin (MW 265), desmethylflunitrazepam
(MW 299), desmethylmianserin (MW 250), diazepam (MW 284),
88 S. Hegstad et al. / J. Chromatogr. B 947–948 (2014) 83–95
duloxetine (MW 297), felbamate (MW 238), flunitrazepam (MW
313), fluoxetine (MW 309), ketobemidone (MW 247), levome-
promazine (MW 329), meprobamate (MW 218), mianserin (MW
264), midazolam (MW 326), mirtazapine (MW 265), moclobemide
(MW 269), N-desmethylcodeine (MW 285), nortriptyline (MW
263), O-desmethylvenlafaxine (MW 263), olanzapine (MW 312),
paroxetine (MW 329), pentazocine (MW 285), pimozide (MW 462),
pipothiazine (MW 476), promethazine (MW 284), reboxetine (MW
313), sertraline (MW 306), tetrahydrocannabinol (MW 314) and
warfarin (MW 308). The concentrations tested ranged from 30 to
15,000 ng/mL for the various analytes.
2.6.6. Carry-over effects
To evaluate possible carry-over effects by UPLC–MS/MS,
authentic specimens containing amphetamine (>90,000 ng/mL),
buprenorphine (>2000 ng/mL), codeine (>10,000 ng/mL), DMD
(>3500 ng/mL), ethylmorphine (>2000 ng/mL), methadone
(>90,000 ng/mL), methamphetamine (>100,000 ng/mL), 6-MAM
(>500 ng/mL), morphine (>100,000 ng/mL), OH-alprazolam
(>2000 ng/mL), oxazepam (>100,000 ng/mL) and THCA
(>10,000 ng/mL) were analyzed. A standard with double con-
centration of the highest calibrator was used for the other analytes.
The specimens or standards were run followed by one diluted
matrix blank.
Carry-over (%) was determined by comparing the found concen-
tration in the blank samples versus the concentration of the lowest
calibrator.
2.6.7. Stability
To evaluate stability, QC samples (1 mL) were prepared in urine
and kept frozen at −20 ◦ C for four months in ScreenMates tubes,
1.4 mL (Thermo Scientific, Hudson, NH). Stability of THCA in urine
(10 mL) was evaluated after storage at 4 ◦ C for 1 month in Pyrex
tubes, 11.5 mL (Corning, Lowell, MA). The external quality control
C4 (145 ng/mL) was used to verify stability of THCA. The stabil-
ity of diluted urine samples stored at 4 ◦ C in the autosampler was
evaluated by re-injecting QC samples after three days.
2.6.8. Application of methods
In order to identify possible methodological problems such as
false positive/negative results, variation of internal signal, stability
of retention time, disturbing peaks, approximately 600 consecutive
routine samples were analyzed with the new screening methods
and compared to the results found with our previous routine LC–MS
screening methods [26].
3. Results and discussion
3.1. Method development
Originally, the aim of this work was to develop one method for
all analytes. Due to various considerations and sometimes mutually
exclusive requirements, such as the strength of the analytical signal
related to the LOQ for each substance, the width and level of the
calibration ranges, the desire to hydrolyze glucuronides for some
analytes but not for others, and the availability and cost of reference
substances, two pre-treatment methods and three UPLC–MS/MS
methods were developed (Fig. 1).
The yield of enzymatic glucuronide hydrolysis is depend-
ing on various parameters like the chemistry of the specific
glucuronidated metabolite, the enzyme strain used, incubation
temperature, time and pH. As hydrolysis of C6G by H. pomatia is
low (22% at 60 ◦ C and 4 h [18]), measuring glucuronidated codeine
instead of hydrolyzing the sample will enhance the reproducibility
and extend the time of detection, thereby facilitating the inter-
pretation of which opiate has been ingested. Although M3G and
M6G are hydrolyzed to a higher degree [18] we also decided to
analyze these metabolites in urine not undergoing hydrolysis. In
contrast to C6G, benzodiazepine glucuronides such as oxazepam
glucuronide are efficiently hydrolyzed by H. pomatia [27]. There-
fore, and due to the fact that of most glucuronidated standards were
very expensive and the supply situation could not be guaranteed
by the manufacturers, we decided to include buprenorphine, DMD,
ethylmorphine, hydromorphone, OH-alprazolam, oxazepam and
THCA in the method with hydrolysis. The remaining compounds
were included in method 2 due to limitation of sensitivity with the
sample dilution of 1/250 used in method 1.
Conditions for the enzymatic hydrolysis of buprenorphine,
DMD, ethylmorphine, OH-alprazolam, oxazepam and THCA
were systematically examined in authentic urine specimens.
Parameters tested were enzyme concentration (10,000, 25,000 and
50,000 U/mL), incubation time (40, 60, 90 and 120 min), pH in buffer
(pH 4.3, 4.8 and 5.3) and incubation temperature (60, 65 and 70 ◦ C).
The most satisfactory results were achieved when 100 ␮L of urine
was buffered to pH 4.8, incubated with 2500 U of enzyme at 65 ◦ C
for 60 min (data not shown).
Another objective was to dilute the samples as much as possible
with a satisfactory signal/noise ratio (>40–50) at the lowest cali-
bration level. A high sample dilution will decrease ME and extend
column life time. In method 1 the samples were diluted 1/250,
while the analytical signals of THCA, 6-MAM and buprenorphine
in methods 2 and 3 limited the sample dilution to 1/10. For sev-
eral analytes the signal of the highest standard was saturated. It
was therefore necessary to decrease the signal by increasing cone
voltage and collision energy, and using either another transition
with less intensity or the C13 -transition. Supplementary Tables 1
and 2 show optimized and detuned parameters/transitions for the
analytes.
In method 1, carry-over was seen after injections of
highly concentrated samples or standards (∼100,000 ng/mL) for
amphetamine, methadone and methamphetamine when a high pH
mobile phase and a Waters Acquity BEH C18 column was used. By
changing the mobile phase to 0.1% formic acid in water, carry-over
in the first blank was reduced to <20% of calibrator 1 response. In
addition, to obtain satisfactory chromatography for the most polar
compounds, the column was changed to a Waters Acquity HSS T3-
column. Further experiments confirmed that the carry-over was
found to be on-column and not in the autosampler. Thus, we con-
cluded that formic acid minimized the interactions between basic
drugs, being positively charged, and the column used.
In the analysis of the remaining substances, carry-over was
shown for THCA, and another approach had to be used to reduce
this effect. Starting the UPLC gradient at 2% acetonitrile gave a carry-
over in the first blank equal to 200% of calibrator 1 after injection
of a high THCA standard (10,000 ng/mL). By increasing the initial
acetonitrile concentration from 2% to 60%, carry-over was reduced
to 10% of calibrator 1. However, an initial concentration of 60%
acetonitrile was not suitable for the remaining compounds, due
to high elution strength. The remaining substances were therefore
analyzed with two separate UPLC methods (methods 2 and 3), as
described in Fig. 1. The pre-column filter used in methods 2 and
3 was replaced after approximately every 600 injections and the
life-time of the analytical column was around 2000 injections.
3.2. Method validation
3.2.1. Calibration curves
Calibration curves were made for each analyte in the concen-
tration ranges listed in Tables 4 and 5. As analytes with high
as well as with low concentrations were included in the same
assay, not all compounds had linear calibration curves. A weighted
(1/x) second order regression line was therefore applied for each
 S. Hegstad et al. / J. Chromatogr. B 947–948 (2014) 83–95 89

Table 4
Calibration concentrations, correlation coefficient (R2 ), limit of detection (LOD), limit of quantification (LOQ), within-assay precision, between-assay precision and bias for
analytes included in method 1.

Analyte Calibration R2 (n = 4) LOD (ng/mL) LOQ (ng/mL) Theoretical Within-assay Between-assay Between-day
concentrations concentration precision, precision, bias (%)
(ng/mL) (ng/mL) (%CV) (n = 10) (%CV) (n = 10)

7-AC 100, 250, 500, 2000 0.9963 6 20 150 6.0 7.6 4.9
500 5.4 6.3 3.7
1600 4.1 6.0 1.9

7-AF 50, 100, 500, 2000 0.9993 2 10 75 2.5 4.0 4.4

300 3.2 4.1 3.2
1600 4.3 4.6 0.6

7-AN 100, 250, 500, 2000 0.9948 6 20 150 6.4 7.9 3.9
500 5.5 6.4 0.8
1600 4.6 7.2 3.0

Amphetamine 100, 500, 2000, 10,000 0.9982 6 50 150 3.2 3.2 8.2
1500 2.2 4.6 5.3
8000 5.2 5.9 0.7

BECG 100, 500, 2000, 10,000 0.9995 1 10 150 3.1 5.1 9.5
1500 4.5 6.8 3.4
8000 4.7 5.7 −0.4

Codeine 100, 500, 2000, 10,000 0.9978 12 20 150 5.5 5.4 8.9
1500 4.6 6.3 6.0
8000 6.0 6.1 3.1

C6G 500, 1000, 2000, 10,000 0.9975 8 50 750 4.1 4.3 7.6
1500 4.6 4.4 9.9
8000 4.3 5.6 2.9

Ephedrine 100, 500, 1000, 2000 0.9985 6 20 150 3.9 4.6 11.0
300 3.3 4.9 7.3
1600 4.1 4.7 0.1

Ketamine 100, 500, 2000, 10,000 0.9981 6 20 150 6.5 5.6 9.0
1500 3.0 6.8 5.4
8000 3.0 7.1 1.0

Methadone 200, 500, 2000, 10,000 0.9991 25 40 300 2.5 8.3 8.2
1500 2.1 4.7 5.1
8000 4.4 4.0 1.4

Methamphetamine 100, 500, 2000, 10,000 0.9992 3 10 150 3.6 3.6 11.4
1500 3.7 4.3 7.1
8000 5.7 3.5 2.7

MDMA 100, 250, 500, 1000 0.9985 3 10 150 3.1 4.4 9.4
300 2.7 3.3 5.7
800 3.8 3.7 1.1

Morphine 100, 500, 2000, 10,000 0.9992 6 20 150 4.5 6.2 14.0
1500 4.8 6.1 5.3
8000 5.8 7.4 2.7

M3G 500, 2000, 10,000, 50,000 0.9985 8 50 750 3.6 2.3 4.9
8000 3.3 4.5 3.2
40,000 4.8 4.5 0.6

M6G 100, 500, 2000, 10,000 0.9986 25 50 150 3.9 2.6 4.4
1500 3.3 3.4 3.3
8000 4.5 4.6 −1.4

O-DM-tramadol 100, 250, 500, 1000 0.9846 1 10 150 3.2 4.9 8.0
300 3.2 3.6 7.8
800 2.0 2.5 2.6

Oxycodone 100, 500, 2000, 10,000 0.9986 3 20 150 4.6 3.8 8.1
1500 4.8 7.1 5.2
8000 4.0 4.6 2.5

Pethidine 100, 500, 2000, 10,000 0.9982 6 20 150 7.1 7.7 6.8
1500 2.8 7.6 5.6
8000 3.7 7.9 2.3

Pregabalin 500, 1000, 5000, 10,000 0.9987 30 100 750 3.3 5.7 7.8
3000 2.5 4.7 5.0
8000 4.2 4.1 2.5

Ritalinic acid 1000, 5000, 20,000, 50,000 0.9994 60 200 1500 4.4 8.1 6.7
15,000 5.0 5.6 0.02
40,000 4.8 6.5 0.2
90 S. Hegstad et al. / J. Chromatogr. B 947–948 (2014) 83–95

Table 4 (Continued)

Analyte Calibration R2 (n = 4) LOD (ng/mL) LOQ (ng/mL) Theoretical Within-assay Between-assay Between-day
concentrations concentration precision, precision, bias (%)
(ng/mL) (ng/mL) (%CV) (n = 10) (%CV) (n = 10)

Tramadol 100, 500, 2000, 10,000 0.9979 3 10 150 3.9 3.8 5.5
1500 3.9 4.7 2.6
8000 3.2 3.3 3.1

Zolpidem 100, 500, 2000, 10,000 0.9981 6 20 150 6.2 6.6 9.1
1500 3.9 7.1 5.0
8000 3.8 5.4 1.0

Zopiclone 100, 500, 2000, 10,000 0.9991 3 50 150 4.1 3.4 10.6
1500 3.2 3.2 6.2
8000 3.6 3.2 5.6

compound to better fit the observed data points. The methods
were initially developed with inclusion of the origin in the regres-
sion line for the calibration curves. Due to response in blank
matrix, exclusion of the origin caused a slightly higher inter-
cept of the calibration curve with the y axis for amphetamine,
methamphetamine and methadone, as compared to the inclusion
of the origin. Therefore, origin was excluded for these sub-
stances. For all other compounds, background responses were
negligible. Consequently, for these analytes, the intercepts with
the y axis were almost identical irrespective of whether the
point of origin was included or not. We therefore in these cases
chose to keep the origin in the regression line for the calibra-
tion curve. The correlation coefficients for the regression line
were above 0.99 for all the compounds (Tables 4 and 5). The
MRM chromatograms of the lowest calibration level are shown in
Figs. 2 and 3.

Table 5
Calibration concentrations, correlation coefficient (R2 ), limit of detection (LOD), limit of quantification (LOQ), within-assay precision, between-assay precision and bias for
analytes included in methods 2 and 3.

Analyte Calibration R2 (n = 4) LOD (ng/mL) LOQ (ng/mL) Theoretical Within-assay Between-assay Between-day
concentrations concentration precision, precision, bias (%)
(ng/mL) (ng/mL) (%CV) (n = 10) (%CV) (n = 10)

Buprenorphine 1, 10, 100, 1000 1.000 0.3 1 1.5 3.2 4.9 2.3
150 5.4 4.8 0.2
800 4.6 3.1 −0.9

DMD 50, 100, 500, 2000 1.000 2 5 75 3.2 4.6 0.5

300 2.8 3.5 0.5
1600 4.0 3.0 −1.1

Ethylmorphine 100, 500, 1000, 2000 0.999 2 10 150 1.9 4.7 1.3
300 4.4 6.1 −0.1
1600 4.4 3.1 −3.2

Fentanyl 10, 50, 200, 1000 1.000 0.6 2 15 2.1 4.4 2.0
200 3.5 3.7 −0.6
800 4.8 2.4 −0.5

Hydromorphone 100, 500, 2000, 10000 1.000 15 50 150 1.7 3.3 1.3
1500 2.6 3.7 −0.3
8000 4.8 1.5 0.4

MDA 100, 250, 500, 1000 0.999 3 20 150 2.2 4.4 0.7
300 3.7 4.4 −0.6
800 5.6 3.4 −0.5

6-MAM 5, 50, 200, 500 1.000 0.2 5 10 4.0 6.8 0.9

150 4.3 4.7 −0.5
400 6.9 2.7 1.4

OH-alprazolam 50, 100, 500, 2000 1.000 1 5 75 2.3 4.2 2.2

300 3.4 4.7 −1.3
1600 3.7 3.8 0.6

Oxazepam 100, 500, 2000, 10,000 1.000 3 20 150 2.6 4.6 2.1
1500 3.5 5.4 −2.0
8000 6.2 3.6 3.2

PMA 100, 250, 500, 1000 0.998 6 50 150 2.6 8.6 −8.6
300 5.0 8.2 −10.8
800 6.9 5.4 −3.1

PMMA 100, 250, 500, 1000 0.999 2 10 150 3.0 11.0 −2.7
300 4.7 8.2 −6.3
800 5.6 4.0 0.1

THCA 10, 50, 500, 1500 0.999 1 2 15 4.6 8.0 3.8

150 7.5 7.2 2.9
1200 4.9 4.7 −0.5
 S. Hegstad et al. / J. Chromatogr. B 947–948 (2014) 83–95 91

Fig. 2. Method 1; MRM chromatograms of the analytes with quantitative transitions from the lowest concentration calibrator.
92 S. Hegstad et al. / J. Chromatogr. B 947–948 (2014) 83–95
Fig. 3. Methods 2 and 3; MRM chromatograms of the analytes with quantitative transitions from the lowest concentration calibrator.
3.2.2. Limit of detection and quantification, precision and bias
The LOD, LOQ, within-assay precision, between-assay precision
and bias for the analytes are presented in Tables 4 and 5. The pre-
cision was <15% and bias was ±15% at LOQ for all analytes. The
within-assay CVs were 2–8%, the between-assay CVs were 2–11%,
and bias of the QC samples varied between −11% and +14%. Accord-
ing to the Bioanalytical method validation guidelines published
by the U.S. Food and Drug Administration [24], precision and bias
below 15% (20% at the LOQ level) are acceptable. There were no
indications that precision and bias were affected for the compounds
included when using detuned MS parameters. In fact, we have also
previously demonstrated high reproducibility and robustness for
oxazepam and 3-hydroxy-diazepam using detuned MS parameters
in the routine setting [28].
3.2.3. Matrix effects
In method 1, ME % ranged from 78% to 115% (Table 6). Due to
less sample dilution, ion suppression was more extensive (<85%)
for several of the analytes in method 2 (Table 7). However, when
corrected with internal standard the observed matrix effects were
reduced significantly for all analytes except PMA (67%) and PMMA
(59%), for which ethylmorphine-d5 was used as internal standard.
The use of MDA-d5 as internal standard for PMA and PMMA was also
evaluated. Due to less variability of intensity in authentic specimens
and a retention time closer to PMA and PMMA, ethylmorphine-d5
was preferred instead of MDA-d5 . Obviously, the use of deuterated
PMA and PMMA as internal standards could have reduced the ME.
Since our method is used for screening, only, the observed matrix
effects were nevertheless considered acceptable.
3.2.4. Specificity
Hydromorphone, morphine and N-desmethylcodeine were sep-
arated chromatographically using both methods. In addition,
O-desmethylvenlafaxine and tramadol were baseline separated in
method 1. Desmethylclomipramine and desmethylflunitrazepam
gave an interfering peak for codeine. This was not investigated
any further, since these metabolites are excreted in urine at low
concentrations [29,30].
3.2.5. Carry-over effects
Neither authentic specimens nor spiked standards caused false
positive results due to carry-over. The highest concentration in a
sample subject to carry-over was 40% of the lowest calibrator for
methadone and methamphetamine, 26% for amphetamine and 15%
for THCA. In the routine setting, samples with concentrations lower
than approx. 20% above the lowest calibrator for amphetamine,
methadone, methamphetamine and THCA are reanalyzed if the
 S. Hegstad et al. / J. Chromatogr. B 947–948 (2014) 83–95 93

Table 6
Matrix effect (ME) for analytes included in method 1.

Analyte Theoretical concentration (ng/mL) ME (%) Relative ME (% CV) ME Corrected with ISa (%) Relative ME corrected with IS (% CV)

7-AC 150 96 6 101 6

7-AF 75 95 5 100 6
7-AN 150 97 6 89 5
Amphetamine 150 97 1 106 4
BECG 150 104 2 107 4
Codeine 150 98 4 104 4
C6G 750 100 3 108 5
Ephedrine 150 97 3 105 3
Ketamine 150 78 9 83 1
Methadone 300 115 13 119 5
Methamphetamine 150 99 3 104 4
MDMA 150 99 1 106 4
Morphine 150 100 8 114 7
M3G 750 109 7 110 6
M6G 150 106 5 106 5
O-DM-tramadol 150 99 3 105 7
Oxycodone 150 106 6 104 6
Pethidine 150 104 4 110 12
Pregabalin 750 99 3 103 5
Ritalinic acid 1500 100 2 107 6
Tramadol 150 108 2 115 6
Zolpidem 150 119 6 127 12
Zopiclone 150 116 7 124 9
a
IS = internal standard.

sample has been analyzed directly after a highly concentrated sam- analytes were found to be stable in the autosampler for three days
ple (>10,000 ng/mL) of the same analyte. at 4 ◦ C.

3.2.6. Stability 3.2.7. Applications of methods

The QC samples were found to be stable in urine for four months
at −20 ◦ C for all analytes except THCA. The loss of THCA (20–30%)
was observed after overnight storage at −20 ◦ C (pH ∼6.4). As the
concentration thereafter remained stable for several months of
storage, the mechanism is possibly adsorptive. These results are
similar to those reported by others [31], testing the stability of
THCA in frozen urine samples. In the present method, aliquots
of urine were pipetted from plastic tubes containing 1 mL urine,
which can explain the adsorptive loss [4,32]. A study by Jamerson
et al. [33] demonstrated that adsorptive loss of THCA in urine kept
at room temperature was significant in acidic urine but relatively
absent in urine close to neutral or basic. However, our experi-
ments showed that the loss of THCA in urine was not reduced at
basic conditions (pH 8.4) when stored at −20 ◦ C. Further investiga-
tions demonstrated that the adsorptive loss of THCA was negligible
in glass tubes (pH ∼6.4) stored at 4 ◦ C for up to one month. For
routine use the calibrators and QC are separated in two sets of
solutions, one for THCA (stored at 4 ◦ C) and one for the remaining
drugs in method 2 (stored at −20 ◦ C). Diluted QC samples for all
To detect any interference or possibly false positive or false
negative results, approximately 600 urine samples were ana-
lyzed by the new screening methods and compared to the results
found with the previously used routine LC–MS screening meth-
ods. Approximately 60% of the samples were positive for one or
several substances, and since these samples are mainly from opi-
oid addicted patients participating in a rehabilitation program,
the analytes most frequently detected were buprenorphine and
methadone. Somewhat higher values compared to the routine
LC–MS methods were seen for amphetamine, methamphetamine,
oxazepam and THCA (results not shown), which are the most fre-
quently detected drugs. No false negative samples were found with
the new methods compared to the former methods. However, the
new methods gave a few more positive samples of amphetamine
(0.5%), oxazepam (1.5%) and THCA (4%), and all available evidence
indicated that these were true positives. For example, in the new
method the cut-off for THCA was lowered from 25 to 10 ng/mL.
The variation of the signal of the internal standard was more
pronounced in authentic specimens in method 2 than in method

Table 7
Matrix effect (ME) for analytes included in methods 2 and 3.

Analyte Theoretical concentration (ng/mL) ME (%) Relative ME (% CV) ME Corrected with ISa (%) Relative ME corrected with IS (% CV)

Buprenorphine 1.5 60 21 93 7
DMD 75 88 8 100 5
Ethylmorphine 150 70 16 90 8
Fentanyl 15 76 10 94 6
Hydromorphone 150 68 8 117 6
MDA 150 50 25 94 8
6-MAM 10 95 14 115 10
OH-alprazolam 75 100 4 100 5
Oxazepam 150 89 6 93 4
PMA 150 52 20 67 12
PMMA 150 46 22 59 14
THCA 15 94 15 101 14
a
IS = internal standard.
94 S. Hegstad et al. / J. Chromatogr. B 947–948 (2014) 83–95
1. The most obvious reason for this finding is the difference in
sample dilution between the methods. However, by using another
LC-gradient and negative ESI mode, the internal standard of THCA
was not affected in the same manner as in method 2. Major ion sup-
pression reduces sensitivity of the method and may lead to false
negative results. A minimum recovery level of signal of internal
standards was therefore set to 20%.
In a few samples (<1% of all samples positive for amphetamine),
an interfering peak was observed in the MRM chromatogram for
amphetamine. In order to investigate this issue further, a few urine
samples were analyzed by accurate mass MS/MS using a Xevo-
G2-S Q-TOF (Waters). The disturbing compound was identified
to origin from gabapentin and its ion-source fragmentation (data
not shown). To separate amphetamine and gabapentin chromato-
graphically, an alternative UPLC method involving a Waters Acquity
BEH C18 column, and ammonium formate pH 10.1/methanol as
mobile phases can be used if needed.
In addition, a few false positive PMA results at low concen-
trations (100–300 ng/mL) were observed in some samples. There
was no presence of PMMA in these samples whereas retention
time and ion ratio of PMA corresponded to those of the calibra-
tors/QCs. We suspected these positive samples to be false because
PMMA was not present in them, as PMA is usually produced
from PMMA in the body after intake of PMMA. Due to lack of
experience with these compounds we decided to use the same
alternative UPLC method as for amphetamine, to further elucidate
this matter. In addition, the cut-off value for reporting PMA as pos-
itive was increased from 100 to 300 ng/mL, thereby solving the
issue.
For a number of compounds the urinary concentrations in
several samples were above the highest calibration level. The
analytes with the highest proportions of samples above the high-
est calibration level were methadone (40%), methamphetamine
(10%), codeine (10%), amphetamine (6%), oxazepam (6%) and
morphine (6%). Although the analytical results are used to dis-
tinguish between a previous and a recent intake based upon a
decrease in the (creatinine-adjusted) concentration compared to
the previous sample, it is not necessary to quantify concentra-
tions precisely at these high levels. As far as any two samples
from the same individual are obtained with at least one or a
few days’ interval (depending on the specific drug), such high
concentrations could always be evaluated as a new intake. The
same practice is described by Berg et al. [34], where only low
concentrations (within the linear range) were used for interpre-
tation.
The work of Eichhorst et al. [10] included a similar number
and types of analytes and the samples were injected once instead
of thrice, as we did. However, our methods included morphine
and codeine glucuronides in addition to buprenorphine, and were
validated for both lower (for some compounds) and higher concen-
tration ranges. In contrast to Eichhorst et al. we use two transitions
for each compound to confirm identity except tramadol, for which
the metabolite O-desmethyl-tramadol was included for confirma-
tion. In addition, the within-assay and between-assay CVs were
substantially lower with our methods, indicating a more reliable
quantitative determination of the analytes.
3.3. Experience from routine use
The methods have been run daily for eight months and both
the simple pre-treatment of the samples and the robust data
processing ensure consistent results between many analytes. Two
separate analytical columns are routinely used; in addition meth-
ods 2 and 3 samples are injected twice on the same column.
Due to a daily number of routine samples of about 250, the fil-
ter in front of the columns for both methods was replaced by a
pre-column. For method 1 the pre-column and analytical column
are changed simultaneously after approx. 10,000 injections. The
lifetime of the column used for methods 2 and 3 was significantly
increased by using a pre-column (up to 10,000 injections). How-
ever, the pre-column must be replaced after 400–600 injections
to prevent peak tailing. Carry-over of buprenorphine (in the con-
centration range 1–3 ng/mL) was gradually observed in samples
analyzed after specimens containing > 1000 ng/mL buprenorphine.
These samples are routinely reanalyzed in order to avoid false pos-
itive test results of buprenorphine. The methods were accredited
according to ISO 17,025 by the Norwegian Accreditation Services
(http://www.akkreditert.no/en/) in April 2013.
4. Conclusions
In conclusion, the presented methods have facilitated screening
of a large number of drugs of abuse with simple sample prepa-
ration, good sensitivity and specificity, and short run-times. The
methods have been routinely used at our hospital for eight months,
analyzing more than 35,000 samples. The implementation of the
methods has eliminated or highly reduced the need of reagents
and toxic extraction solvents, significantly improved the efficiency,
and considerably reduced the manual workload at our routine
laboratory.
Acknowledgements
The authors would like to thank Hilde Havnen, Audhild Naavik
and Lisa Sylten for technical assistance during method develop-
ment and validation.
References
[1] F.T. Peters, Clin. Biochem. 44 (2011) 54.
[2] H.H. Maurer, Ther. Drug Monit. 32 (2010) 324.
[3] J.J. Pitt, Clin. Biochem. Rev. 30 (2009) 19.
[4] A.D. de Jager, N.L. Bailey, J. Chromatogr. B: Anal. Technol. Biomed. Life Sci. 879
(2011) 2642.
[5] C. Yuan, C. Heideloff, M. Kozak, S. Wang, Clin. Chem. Lab. Med. 50 (2012) 95.
[6] E. Gustavsson, M. Andersson, N. Stephanson, O. Beck, J. Mass Spectrom. 42
(2007) 881.
[7] J.C. Van De Steene, W.E. Lambert, J. Am. Soc. Mass Spectrom. 19 (2008) 713.
[8] F.T. Peters, D. Remane, Anal. Bioanal. Chem. 403 (2012) 2155.
[9] J.O. Thorngren, F. Ostervall, M. Garle, J. Mass Spectrom. 43 (2008) 980.
[10] J.C. Eichhorst, M.L. Etter, N. Rousseaux, D.C. Lehotay, Clin. Biochem. 42 (2009)
1531.
[11] Y. Al-Saffar, N.N. Stephanson, O. Beck, J. Chromatogr. B: Anal. Technol. Biomed.
Life Sci. 930 (2013) 112.
[12] D. Thieme, H. Sachs, M. Thevis, J. Mass Spectrom. 43 (2008) 974.
[13] N. Stephanson, H. Dahl, A. Helander, O. Beck, Ther. Drug Monit. 24 (2002) 645.
[14] C. Chebbah, O.J. Pozo, K. Deventer, P. Van Eenoo, F.T. Delbeke, Rapid Commun.
Mass Spectrom. 24 (2010) 1133.
[15] P. Wang, J.A. Stone, K.H. Chen, S.F. Gross, C.A. Haller, A.H. Wu, J. Anal. Toxicol.
30 (2006) 570.
[16] W.C. Duer, S. McFarland, J. Anal. Toxicol. 31 (2007) 419.
[17] S. Fu, J. Lewis, H. Wang, J. Keegan, M. Dawson, J. Anal. Toxicol. 34 (2010) 243.
[18] L.P. Hackett, L.J. Dusci, K.F. Ilett, G.M. Chiswell, Ther. Drug Monit. 24 (2002)
652.
[19] J.O. Svensson, M. Andersson, E. Gustavsson, O. Beck, J. Anal. Toxicol. 31 (2007)
81.
[20] D.M. Shakleya, R. Dams, R.E. Choo, H. Jones, M.A. Huestis, J. Anal. Toxicol. 34
(2010) 17.
[21] A.A. Westin, M.A. Huestis, K. Aarstad, O. Spigset, J. Anal. Toxicol. 33 (2009) 610.
[22] M.A. Huestis, E.J. Cone, J. Anal. Toxicol. 22 (1998) 445.
[23] F.T. Peters, O.H. Drummer, F. Musshoff, Forensic Sci. Int. 165 (2007) 216.
[24] Guidance for industry, Bioanalytical Method Validation, May 2001, http://
www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/
Guidances/ucm070107.pdf
[25] B.K. Matuszewski, M.L. Constanzer, C.M. Chavez-Eng, Anal. Chem. 75 (2003)
3019.
[26] G. Bagoien, G. Morken, K. Zahlsen, T. Aamo, O. Spigset, J. Clin. Psychopharmacol.
29 (2009) 248.
[27] R. Meatherall, J. Anal. Toxicol. 18 (1994) 382.
[28] S. Hegstad, E.L. Oiestad, U. Johansen, A.S. Christophersen, J. Anal. Toxicol. 30
(2006) 31.
 S. Hegstad et al. / J. Chromatogr. B 947–948 (2014) 83–95 95

[29] A. Wohlfarth, N. Toepfner, M. Hermanns-Clausen, V. Auwarter, Anal. Bioanal.
Chem. 400 (2011) 737.
[30] M. Forsman, I. Nystrom, M. Roman, L. Berglund, J. Ahlner, R. Kronstrand, J. Anal.
Toxicol. 33 (2009) 491.
[31] S. Golding Fraga, J. Diaz-Flores Estevez, C. Diaz Romero, Ann. Clin. Lab. Sci. 28
(1998) 160.
[32] D.E. Moody, K.M. Monti, A.C. Spanbauer, J. Anal. Toxicol. 23 (1999) 535.
[33] M.H. Jamerson, J.J. McCue, K.L. Klette, J. Anal. Toxicol. 29 (2005) 627.
[34] T. Berg, E. Lundanes, A.S. Christophersen, D.H. Strand, J. Chromatogr. B: Anal.
Technol. Biomed. Life Sci. 877 (2009) 421.