Journal of Analytical Toxicology, Vol. 30, May 2006
Quantitative Analysis of Buprenorphine and
Norbuprenorphine in Urine using Liquid
Chromatography Tandem Mass Spectrometry
E.J. Fox1, *, V.A. Tetlow 2, and K.R. Allen 1
1Department of Clinical Biochemistry, Leeds Teaching Hospitals NHS Trust, Britannia House, Britannia Road, Morley,
Leeds, LS27 ODQ, U.K. and 2Department of Clinical Biochemistry, Salford Royal NHS Trust, Hope Hospital, Stott Lane,
Salford, M6 8HD, U.K.
Buprenorphine is an opioid analgesic drug that is used as an
alternative to methadone to treat heroin addiction. Established
methods for the analysis of buprenorphine and its metabolites in
urine such as gas chromatography-mass spectrometry (GC-MS)
involve complicated sample extraction procedures. The aim of the
present study was to develop a sensitive yet straightforward
method for the simultaneous analysis of buprenorphine and
norbuprenorphine in urine using liquid chromatography-MS-MS.
The method comprised an enzymatic hydrolysis using Patella
vulgata ~-glucuronidase, followed by centrifugation and direct
analysis of the supernatant. The limits of detection and quantitation
were < 1 pg/L for buprenorphine and < 1 and 4 pg/L, respectively,
for norbuprenorphine. Assay coefficients of variation (CVs) were <
15%, with the exception of concentrations close to the limit of
quantitation, where CVs were below 20%. In direct comparison
with an established GC-MS protocol, the method showed minimal
negative bias (8.7% for buprenorphine and 1.8% for
norbuprenorphine) and was less susceptible to sample carryover.
The extent of conjugation in unhydrolyzed urine was investigated
and found to be highly variable, with proportions of unconjugated
buprenorphine and norbuprenorphine of 6.4% [range 0% to 67%;
standard deviation (SD) 9.7%] and 34%
(range 0% to 100%; SD 23.8%), respectively.
Buprenorphine is a synthetic opioid derivative of the naturally
occurring poppy alkaloid thebaine. It has partial agonist activity
at the lJ opioid receptors and antagonist activity at the ~r re-
ceptors. Through binding to the opioid receptors, it is able to
mimic the analgesic and euphoric effects of the full agonists
heroin, morphine, and methadone, while simultaneously
blocking their pharmacological activities. At high buprenor-
phine doses, certain unwanted side-effects such as respiratory
* Author to whom correspondence should be addressed. E-mail: email@example.com.
depression are less pronounced, making it a safer alternative to
methadone for the treatment of opiate addiction (1).
Buprenorphine is metabolized primarily in the liver, where it
is N-dealkylated by cytochrome P450 3,44 to form its major
metabolite norbuprenorphine (2,3). Drug and metabolite are
then conjugated to form buprenorphine-3-glucuronide and
norbuprenorphine-3-glucuronide, respectively. The majority
of a given dose is excreted via the biliary system, with the re-
mainder being excreted in the urine (4). A small amount can
also be found in sweat, hair, and oral fluid (5-7). It has been es-
timated that 10-20% of the buprenorphine and norbuprenor-
phine excreted in the urine is unconjugated, with the
remainder being the glucuronide conjugates (8).
Because of its favorable pharmacological properties, the use
of buprenorphine for the treatment of opiate abuse is on the in-
crease. The abuse potential is relatively low, but misuse is a
problem in some areas (9,10). As a consequence it is important
to regularly assess those on treatment regimes to prevent the
diversion of prescription medication to illegal use. This, in
turn, has created a need for a robust assay for compliance
testing. Gas chromatography-mass spectrometry (GC-MS) is
currently the "gold standard" assay, but it involves a multi-
step sample derivatization and extraction procedure (11,12).
Immunoassay is rapid and amenable to automation or point of
care use; however, antibody-based kits are expensive and only
detect the parent drug. These assays would not be able to iden-
tify samples that had been adulterated by addition of prescrip-
tion medication to the urine to generate a positive result. One
such assay has also recently been reported to suffer from sig-
nificant cross-reactivity with opiate drugs (13).
To date, only one liquid chromatography (LC)-MS-MS
method for the analysis of buprenorphine and norbuprenor-
phine in urine has been reported in the literature. This involves
an initial qualitative screen for the glucuronide conjugates in
diluted urine, followed by hydrolysis and quantitation of all
glucuronide-positive samples with the inclusion of a solid-
phase extraction step for low concentration samples (14). The
aim of the present study was to develop a quantitative assay for
buprenorphine and its metabolite norbuprenorphine that is
238 Reproduction (photocopying) of editorial content of this journal is prohibited without publisher's permission.
Journal of Analytical Toxicology, Vol. 30, May 2006
straightforward and sensitive enough to use on all samples.
The method was intended for high-throughput screening for
compliance in the clinical laboratory. Therefore, in order to
minimize sample preparation, the work-up was limited to an
enzymatic hydrolysis. The potential for simplifying the assay
further by eliminating the hydrolysis step was also explored by
quantifying the amount of unconjugated buprenorphine and
norbuprenorphine in neat untreated urine.
Materials and Methods
Chemicals and reagents
Drug standards buprenorphine and norbuprenorphine and
the deuterated internal standards buprenorphine-d4 and nor-
buprenorphine-d3 were purchased from LGC Promochem
(Teddington, Middlesex, U.K.) at a concentration of 0.1 g/L
in methanol. ~-Glucuronidase (Patella vulgata and Helix
pomatia), trifluoroacetic acid anhydride (TFAA), N-methyt-N-
(trimethylsilyl)trifluoroacetamide (MSTFA), and high-perfor-
mance liquid chromatography (HPLC)-grade methanol were
purchased from Sigma (Poole, U.K.). Ammonium acetate was
purchased from BDH (Lutterworth, U.K.); and formic acid
and propan-2-ol were purchased from Scientific Laboratory
Supplies (Nottingham, U.K.).
Clinical samples and design of study
Urine samples (n = 148) were obtained from drug treatment
clinics in the Greater Manchester area; these had been submitted
for routine analysis of buprenorphine and norbuprenorphine
and were anonymized prior to involvement in this study. Sam-
ples were collected in polystyrene or polypropylene tubes
without preservative and analyzed within three weeks of the
date of collection. This study was the result of a collaboration be-
tween the Department of Clinical Biochemistry, Hope Hospital,
Salford and the Department of Clinical Biochemistry, Leeds
Teaching Hospitals, Britannia House, Leeds, U.K. The
LC-MS-MS method was developed, validated, and executed in
Leeds; the GC-MS analyses were carried out in Salford.
Calibration standards, internal standards, and quality con-
trol samples. Drug-free urine from a healthy volunteer was
spiked with buprenorphine and norbuprenorphine to a con-
centration of 1000 l~g/L, and a set of 11 calibration standards
were prepared by serial dilutions to the lowest standard (0.98
lJg/L). A working stock of the deuterated internal standards
buprenorphine-d4 and norbuprenorphine-d3 was prepared by
dilution in methanol to a final concentration of 1000 lJg/L.
Quality control (QC) material was purchased from LGC Pro-
mochem and comprised an exogenously spiked urine sample
containing 8.9 ]~g/L buprenorphine and 25.1 lJg/L nor-
buprenorphine. Additional QC samples were prepared in-house
by spiking drug-free urine to concentrations of 5, 50, 100, 200,
and 400 lag/L. Two QC samples to assess the imprecision of the
hydrolysis protocoI were prepared by making pools of
anonymized patient samples. All drug stocks, standards, and QC
samples were stored at -20~ Calibration standards and QC
samples were stored in single-use aliquots.
The hydrolysis protocol was adapted from Combie et al. (15).
Four hundred microliters of patient sample were treated with
2000 units Patella vulgata [3-glucuronidase (50,000 Units/mL)
in the presence of 40 IaL 2M sodium acetate buffer pH 4.4 and
50 lJL internal standard for 1 h at 56~ Samples were cooled to
room temperature and centrifuged at 13,000 xg for 5 rain. The
supernatant was transferred to autosampler vials, and 50 I~L was
injected directly into the LC system. Urine calibrators were
treated in an identical fashion, except that enzyme solution
was substituted for an equal volume of deionized water. During
the optimization of hydrolysis using the Patella vulgata en-
zyme, it was noted that the reaction with both analytes reached
completion within the 1 h incubation period.
For the analysis of non-hydrolyzed samples, neat urine/cali-
brator/QC samples were centrifuged at 13,000 x g for 5 min.
Fifty microliters of internal standard was added to 450 IJL
sample and mixed immediately prior to analysis. Fifty micro-
liters was injected into the LC system.
LC was performed using a Shimadzu system (Milton Keynes,
U.K.) comprised of an SIL-HT autosampler, 2 x LC10AC pumps,
a DGU-14A degasser, and a CTO-10AS column oven. The
column effluent was directed either to the tandem MS or to
waste via a Va]co switching valve. The setup was controlled by
Analyst 1.4 software (Applied Biosystems, Warrington, Cheshire,
U.K.), which was also used for data collection and interpreta-
tion. Chromatographic separation was performed using a
50- mm• 3-ram HyPURITY C8 column (Thermo Electron Cor-
poration, Runcorn, Cheshire, U.K.) fitted with a security guard
column of the same packing material (Phenomenex, Maccles-
field, Cheshire, U.K.). The column was maintained at 35~
Mobile phase A was 4ram ammonium acetate, 5% (v/v)
methanol in HPLC-grade H20; mobile phase B was 1% (v/v)
propan-2-ol and 0.05% (v/v) formic acid in HPLC-grade
methanol. Gradient elution was performed at a constant flow-
rate of 0.8 mUmin as follows: 0% B was maintained for 1 rain,
followed by a linear increase to 95% over 3 rain; 95% B was
Table I. Molecular Transitions and Retention Times of
Buprenorphine, Norbuprenorphine, and Their
Transition (m/z) Average
Time (rain) Analyte
Norbuprenorphine 3.72 3.54-3.80
* Transitions shown in italics were used for qualitative confirmation.
Journal of Analytical T(~xi(()logy, Vol. 30. May 2006
maintained for 1.5 min, then reduced to 0% over 0.1 min; and
0% B was maintained for 0.9 rain. Column eluate was directed
to waste for the first 2 rain of the program, then to the MS for
the remaining time. Total run time per sample was 6.5 rain.
The drugs were detected using an Applied Biosystems/MDS
Sciex AP13000 Triple-Quad MS equipped with a TurbolonSpray
source (Applied Biosystems, Warrington, Cheshire, U.K.). Ana-
lytes were detected in positive ion mode using multiple reaction
monitoring of the molecular ion transitions listed in Table I.
The transitions shown in italics were used for qualitative con-
firmation only. Nitrogen was used for the nebulizer, curtain,
and collision gase with flow rates of 10 LImin, 10 L/min, and 9
L/min, respectively. The ion spray voltage was 5 kV, and the
temperature was 475~ Data were collected and quantified
using Analyst software version 1.4.
LC-MS-MS method validation
The lower limit of detection (LLOD) was defined as the lowest
urine calibrator that gave a signal-to-noise ratio (S/N) of at
least 3. The lower limit of quantitation (LLOQ) was defined as
the lowest urine calibrator that gave a S/N of at least 10 and a
Table II. Between-Batch Imprecision of the GC-MS Assay
Conc. (pg/t) n CV (%)
Conc. (pg/L) n CV (%)
+MRM 468,31396.0 (Buprenorphlne)
30 000 ]
~20 000 ]
+MRM 414.3/101.1 (Norbuprenorphine)
o ..... 'LO" :i.o' ~:d" ~i:d' :~:0':' '
I. lon current traces for buprenorphine and norbuprenorphine detected in hydrolyzed pa-
tient sample. Ions were deteded in positive multiple reaction monitoring (+MRM) mode using
the following molecular transitions: m/z 468.3 --+ 396.0 (buprenorphine) (A) m/z 414.3 -->
101.1 (norbuprenorphine) (B).
o ......................... 1.o 2.0 ~":"::'3.o ...........
coefficient of variation not greater than 20% (16). S/Ns were
calculated using Analyst 1.4. Quantitation was achieved by
adding deuterated internal standards to the set of calibration
standards and plotting a graph of the calibrator concentration
against the analyte/intemal standard peak-area ratios.
Assay imprecision was assessed both within-batch and be-
tween-batch by repeat analysis of QC samples. Within-batch co-
efficients of variation (CV) were calculated from 10 identical QC
samples in a single batch. Between-batch CVs were calculated
from a minimum of five identical QC samples prepared and
analyzed in different batches over the course of 6 months.
Accuracy was assessed by comparison of the observed concen-
tration with the spiked concentration of the spiked QC samples.
Sample stability was assessed for 7 days at room temperature,
7 days at 4~ and 3 months at -20~ Analyte recovery at 10
IJg/L and 100 lag/L was calculated for both calibrators and sam-
ples using spiked blank urine (17). Differences in recovery be-
tween calibrators and samples were assessed by analysis of
variance (ANOVA). Matrix effect and analytical selectivity were
assessed by analysis of spiked (10 lJg/L) blank urines from six in-
dependent sources. Differences in observed drug concentra-
tion were evaluated using ANOVA analysis. Selectivity was
assessed by the analysis of unspiked blank urine at each molec-
ular transition; chromatograms were visually inspected for in-
terfering peaks at the relevant retention times. The sample
carryover was estimated by analysis of spiked blank urine
(10,000 IJg/L buprenorphine and 4000 lag/L norbuprenorphine)
followed by 3 blank samples. The chosen values correspond to
typical adulterated and strong positive samples, respectively.
GC-MS was performed using an Autosystem XL GC linked to
a Turbosystem Gold MS (PerkinEImer, Beaconsfield, U.K.). The
flow rate was 0.8 mL/min: the injection port and sample line
were at 320~ the MS was at 200~ One-milliliter aliquots
of sample were treated with 5000 units Helix pomatia
[3-glucuronidase in the presence of 0.5 mL internal standard
(250 IJg/L each buprenorphine-d4 and nor-
buprenorphine-d:0 and 0.1 mL 1M sodium ac-
B etate buffer (pH 5.0) for 16 h at 37~ Samples
were extracted with 1-chlorobutane/propan-
2-ol (9:1) at pH 9.5 and evaporated to dryness.
The extract was reconstituted in 1 mL toluene
and 1001JL TFAA, mixed, and allowed to react
for 10 rain at room temperature and then
evaporated to dryness. This was dissolved in
100 IJL MSTFA and heated at 57~ for 30 rain.
Two microliters of derivatized extract was in-
jected in splitless mode onto a 30-ram x 0.25-
mm x 0.25-tJm ZB-5MS capillary column
(Phenomenex, Macclesfield, Cheshire, U.K.).
The temperature was held at 120~ for 2 rain,
followed by a ramp of 40~
which was held for 5 rain. Total run time was
12 rain per sample. Drugs were detected by
selective ion recording of the following molec-
ular ions: buprenorphine, m/z 450; nor-
buprenorphine, m/z 524; buprenorphine-d4,
Journal of Analytical Toxicology, Vol. 30, May 2006
m/z 454; and norbuprenorphine-d3, m/z 527. The LODs were 1.5
]Jg/L for buprenorphine and 3.5 lJg/L for norbuprenorphine. The
assay imprecision data are shown in Table II. Carryover was es-
timated to be 1% to 2% for buprenorphine and 2% to 3% for
Calibration graphs were plotted using linear regression
through zero. Linearity was consistently observed (n = 10) be-
tween 0 and 500 lJg/L (mean r = 0.9991; CV = 0.1%) for
buprenorphine and between 0 and 1000 IJg/L (mean r = 0.9994,
CV = 0.1%) for norbuprenorphine. The LLOD and LLOQ were
found to be < 1 IJg/L for buprenorphine and < 1 lJg/L and
4 pg/L, respectively, for norbuprenorphine. Representative chro-
matograms obtained from a patient urine sample are shown in
The between-batch and within-batch imprecision data for
the assay are shown in Table III. CVs are < 15% for both analytes
with the exception of values close to the LLOQ, where the CVs
are < 20% (16). Accuracy data are also shown in Table III. The
majority are within the recommended 15% of the target value.
Data regarding the stability of samples during routine han-
dling and storage are shown in Figure 2. The mean (n = 4)
changes in observed analyte concentration after storage at 4~
for 7 days were found to be -8% (buprenorphine) and + 5.5%
(norbuprenorphine). After storage at room temperature for
seven days, mean (n = 4) changes in concentration were --4%
(buprenorphine) and + 12% (norbuprenorphine). Serial mea-
surements of two pools of patient samples stored at -20~ for
3 months demonstrated that all values lay within + 2 standard
Table III. Imprecision Data for the LC-MS-MS Assay
Within-Batch (n = 10)
Conc. (pg/t) n CV (%) Accuracy (%) CV (%) Accuracy (%)
* Pool of patient samples.
deviation (SD) of the mean value.
The mean (n = 4) analyte recoveries were 75% (10 lag/L) and
105% (100 ]Jg/L) for buprenorphine and 114% (10 ]Jg/L) and
89% (100 lJg/L) for norbuprenorphine. No significant difference
between recovery of calibrators and samples was observed for ei-
ther analyte (p > 0.5).
The assay is specific and selective for the analytes and their
deuterated analogues as confirmed by the absence of peaks at
the relevant retention times in six independent blank samples.
No significant difference was found between the values ob-
served for either analyte in 10 lJg/L spiked blank urines from six
independent sources (p > 0.3). Carryover between samples was
found to be no greater than 0.2% for either analyte.
~= 6oo l
Low Pool; -20~ C
~ . . ~ _ /
....... .-. . . . . . . . . . . . . .
High Pool; -20~
500 1 .....................................
200 ~ ......... ~..-.'~.- ............... " -2SD
............. -= - 2SD
~ 0~ .....
N ~ N 9 ~
Figure 2. Stability of urine samples. Four independent anonyrnized
patient urine samples were assayed for buprenorphine and norbupre-
norphine before and after storage for 7 days at room temperature (A) and
4~ (B). Data points represent the mean of four replicate samples. Pooled
patient samples of low (C) and high (D) buprenorphine and nor-
buprenorphine concentration stored at-20~ and assayed at various in-
tervals over the course of approximately 3 months. Data points represent
single measurements. Dotted lines represent _+ 2 SD from the mean
Journal of Analytical Toxicology, Vol. 30, May 2006
Comparison of LC-MS-MS with GC-MS
To test the utility of the LC-MS-MS method, a set of samples
from patients participating in buprenorphine substitution pro-
grams was analyzed and the results compared with those ob-
tained using the GC-MS assay. A wide distribution of analyte
y = 0.9153x + 5.882
r = 0.95
100 200 300 400 500 600
y = 0.9466x + 13.617
r = 0.92
o ~ / ~ - ?9
200 300 400 500 600
GC-MS (plffL norbuprenorphine)
~:'~' ~" < " ," ?9
?9 ?9 el
Mean + 1.96SD
Mean (-8.7 %)
Mean - 1.96SD
-50 50 150 250 350 450 550 650
Mean of methods (pg/l. buprenorphine)
Mean + 1.96SD
(5z8 %) ....... ;,~, ..........................................
?9 ee ?9 ee ?9
m ?9 ?9 ?9
?9 "i * ?9 " el
* " *
Mean (-1.8 %)
?9 " , ?9
(-61.3 %) ................. ?9 ................................
100 200 300 400 500 600 700
Mean of methods (pg/L norhuprenorphine)
Figure 3. Linear regression and Bland-Altman difference plots of 148
urine samples analyzed by GC-MS and LC-MS-MS. Urine samples
were analyzed and scatter plots of GC-MS data versus LC-MS-MS data
for buprenorphine (A) and norbuprenorphine (B) are shown. The linear
regression lines and correlation coefficients (r) are shown. Percentage dif-
ference plots of the data in A and B (C) and (D). The solid line represents
the mean value and the dotted lines represent the upper and lower 95%
limits of agreement.
concentrations was observed ranging from below the LOD to
greater than 1000 pg/L Three samples were below the
LC-MS-MS LOD for both analytes but detectable by GC-MS.
The buprenorphine and norbuprenorphine concentrations of
these samples were 3.9 and 8.7, 3.3 and 1.8, and 5.1 and 6.7
The data were analyzed first by linear regression, plotting
the GC-MS data against the LC-MS-MS data as shown in Fig-
ures 3A and 3B. There is significant correlation, but an in-
creasing degree of scatter at higher concentration values. To
examine the correlation further, the data were compared using
Altman and Bland analysis. The mean of the two values gener-
ated for each sample [(LC-MS-MS + GC-MS)/2] was plotted
against the difference between these values for each sample
expressed as a percentage of the mean [(LC-MS-MS -
GC-MS)* 100/mean] (Figure 3C and 3D). From these plots, the
relative method bias and 95% limits of agreement were calcu-
lated. The LC-MS-MS method showed a negative mean bias of
-8.7% for buprenorphine with 95% limits of agreement of +
73.8%. For norbuprenorphine, the LC-MS-MS method showed
a negative mean bias of-1.8% with 95% limits of agreement of
_ _ _ 59.6%.
Analysis of unconjugated versus total post-hydrolysis
buprenorphine and norbuprenorphine by LC-MS-MS
In order to investigate the measurement of the unconjugated
drug and metabolite in untreated urine, each sample was ana-
lyzed by LC-MS-MS for buprenorphine and norbuprenorphine
both before and after treatment with 13-glucuronidase. A small
"oE 10 /
".,.~.0". ?9 eee ?9 ?9
0 100 200
Total buprenorphine (pglL)
300 400 0 600
,.~ ~ 300
=o ~. 250
?9 ?9 #
?9 oO o ?9
oOOo ?9 eoOo ?9 ?9
?9 ?9 ?9 ?9 ?9
0 200 400 600 800 1000
Total norbuprenorphine (tJg/L)
Figure 4, Comparison of the amount of unconjugated drug detectable be-
fore hydrolysis with the total amount of drug detectable after hydrolysis.
Urine samples were analyzed for buprenorphine and norbuprenorphine
by LC-MS-MS before and after glucuronide hydrolysis. Scatter plot of
total concentration measured after hydrolysis against the concentration
of unconjugated drug measured before hydrolysis for buprenorphine
(A) and norbuprenorphine (B).