Determination of Buprenorphine in Human Plasma by Gas ...

Journal of Analytical Toxicology,Vol. 21, October 1997

Determination of Buprenorphinein Human Plasmaby Gas Chromatography-PositiveIon Chemical Ionization Mass Spectrometryand Liquid ChromatographyTandem Mass Spectrometry David E. Moody*, John D. Laycock, Alan C. Spanbauer, Dennis J. Crouch, and Rodger L. Foltz Center for Human Toxicology, University of Utah, Salt Lake City, Utah 84112

Jonathan L. Josephs~ Finnigan MAT, SanJos~, California 95134

Leslie Amass* and Warren K. Bickel Behavioral Pharmacology Laboratory, Department of Psychiatry, University of Vermont, Burlington, Vermont05401

Abstract I Buprenorphine is used for the management of pain and has been advocated for the treatment of opioid addiction. Therapeutic doses result in low plasma concentrations of buprenorphine. In order to assessthe safety and efficacy of buprenorphine, sensitive analytical methods are needed. Until recently, gas chromatography-positive ion chemical ionization mass spectrometry (GC-PCI-MS) offered the most sensitive method to selectively quantitate buprenorphine. We have developed and validated a sensitive liquid chromatography-electrospray ionization-tandem mass spectrometry (LC-ESI-MS-MS) method for buprenorphine. The method is described and compared with a GC-PCI-MS method validated in this laboratory. One-milliliter aliquots of plasma are required for the LC-ESI-MS-MS method and 2-mL aliquots for the GC-PCI-MS method. Buprenorphine-d4 is used as internal standard for both methods. Derivatization with pentafluoropropionic acid anhydride is used for the GC-PCI-MS method, in which the derivatized protonated molecular ions after loss of water are monitored at m/z596 and 600. For LC-ESIMS-MS, the parent protonated molecule ions are monitored at m/z468 and 472. A single-step extraction of basic plasma with n-butyl chloride provided recoveries of 70-87%. Although a limit of quantitation (LOQ) of 0.1 ng/mL could be established for LC-ESI-MS-MS, we could only achieve an LOQ of 0.5 ng/mL with the GC-PCI-MS assay. The GC-PCI-MS method has a linear range of 0.5 to 40 ng/mL (mean r 2 = 0.998, n = 7). For quality control samples at 1.0, 2.5, and 12.5 ng/mL, the intra- and interassay coefficients of variation (CV) did not exceed 14%, and percent of targets were within 16%. The LC-ESI-MS-MS method had a linear range of 0.1 to 10 ng/mL (mean r 2 = 0.999, n = 7). * Address correspondence to David E. Moody, Center for Human Toxicology, 8PRB, Rm 490, University of Utah, Salt Lake City, UT 84112. E-mail [email protected]. f Current address: Pharmaceutical Research Institute, Bristol-Meyers Squibb, New Brunswick, NJ, 089O3. r Current address: Department of Psychiatry, University of Colorado School of Medicine, Denver, CO 80262.

406

For quality control samples at 0.25, 2.5 and 7.5 ng/mL, the intraand interassay CVs did not exceed 4%, and percent of targets were within 12%. Stability studies demonstrated buprenorphine was stable for up to 24 h, 125 days, and 55 days when stored at room temperature, 4~ and -20~ respectively. The utility of the lower LOQ was demonstrated in 40 plasma samples collected up to 96 h after a sublingual dose of buprenorphine; 10 were quantitatable using GC-PCI-MS and 38 using LC-ESI-MS-MS.

Introduction Buprenorphine is an oripavine derivative having partial agonist and antagonist opioid activity (1-3). For the treatment of moderate to severe pain, buprenorphine has been used successfullyby intramuscular, intravenous, or sublingual routes at doses ranging from 0.3 to 0.6 mg (4). Clinical studies have shown that buprenorphine, like methadone, can also be used for the treatment of opioid addiction and may be safelyused to withdraw patients from heroin (5-7) (For a review, see Bickel and Amass [8]). When used for treatment of opioid dependence, buprenorphine is usually administered sublingually in doses of 2-32 rag. Clinical pharmacokinetic, safety, and efficacystudies examining the utility of buprenorphine as a treatment of opioid dependence require an accurate and sensitive analytical method. Immunoassay methods have been used to detect buprenorphine and its metabolites in urine and plasma (9-13). Although some of these assays reported good sensitivity (e.g., 10 pg/mL [12]), they were not specific for buprenorphine because they cross-reacted with norbuprenorphine and/or conjugated buprenorphine metabolites. Reversed-phase highperformance liquid chromatographic (HPLC) methods with UV (14), fluorescence (15,16), and electrochemical (17,18)

Reproduction(photocopyin8)of editorialcontentof thisjourna[is prohibitedwithoutpublisher'spermission.

Journal of Analytical Toxicology, Vol. 21, October 1997

detection have been reported for analysis of buprenorphine in plasma and urine. These methods lacked the sensitivity to detect buprenorphine in plasma at concentrations less than 1 ng/mL. Other published methods for the analysis of buprenorphine in biological fluids include gas chromatography (GC)with electron capture detection (19) and quantitative thin-layer chromatography (20). These methods were limited to analysis of buprenorphine in urine. Quantitation of buprenorphine has also been examined by mass spectrometric methods. Biota et al. (21) used gas chromatography-mass spectrometry (GC-MS) with electron ionization to measure buprenorphine in human plasma and urine. The method consisted of extraction at pH 9.4, back extraction into dilute sulfuric acid, and heating at 110~ which caused the buprenorphine to undergo cyclization with the loss of methanol. The cyclizedproduct was then extracted and derivatized with pentafluoropropionicanhydride (PFPA).The limit of detection of this method was 0.15 ng/mL. A gas chromatography-positive ion chemical ionization-mass spectrometry (GC-PCI-MS) method was reported by Ohtani et al. (22). In that method, samples were made acidic and extracted with an organic solvent to remove potential interferences. The pH was adjusted to 10.5, and the samples were extractedwith a second organic solvent and derivatized with PFPA. These authors reported a limit of quantitation (LOQ)of 0.2 ng/mL. Recently, Kuhlman et al. (23) used solid-phase extraction and derivatization with heptafluorobutyric anhydride to prepare samples for GC coupled with negative ion chemical ionization and tandem mass spectrometry. Selected reaction monitoring of a prominent product ion at m/z 464 formed by collision-induced dissociation of the molecular anion permitted an LOQ of 0.2 ng/mL.Twoother recent publications have employedthe combination of HPLC with single-stage MS for determination of buprenorphine in biologicalsamples. Tracqui et al. (24) used a single-stepliquid-liquid extraction of biologicalfluids and hair samples, whereas Hoja et al. (25) used acetonitrile to deproteinize whole blood then solid-phase extraction and liquidliquid backextraction.Both of these research groups monitored the protonatedmoleculesfor buprenorphine(m/z 468) formedby electrosprayionization(ESI).The more extensiveextractionprocedure by Hoja et al. (25) permitted a lower LOQ of 0.1 ng/mL. We reporttwo validatedMS methods for the determination of buprenorphine in human plasma. One method employs a single-step liquid-liquid extraction followedby derivatizatioff with PFPAand GC-PCI-MSanalysiswith an LOQof 0.5 ng/mL. The other method consists of a verysimilar liquid-liquidextraction followeddirectly by liquid chromatography-electrospray ionization-tandem mass spectrometry (LC-ESI-MS-MS)with an LOQ of 0.1 ng/mL. Both methods use a deuterium-labeled isotopomer of buprenorphine as the internal standard.

Experimental Materials Drug reference standards of buprenorphine (100 lag/mLin methanol) and buprenorphine-d4 (100 l~g/mLin methanol)

were obtained from Radian (Austin,TX). Sodium hydroxidewas purchased from Sigma (St. Louis, MO). High-purity grade n-butyl chloride, acetonitrile, and toluene were obtained from Baxter Diagnostics (McGrawPark, IL), and PFPAwas obtained from Regis (Morton Grove, IL). Formic acid (88%) was purchased from J.T. Baker (Phillipsburg, NJ). The GC column was a DB-1 fused-silica capillary column (15 m x 0.32-mm i.d., 1-1~m film thickness, J&W Scientific, Folsom, CA). The LC column was an Alltech (Deerfield,IL) Solvent Miser| C8 150 x 2.l-ram column with a 5-1~mparticle size.

Specimens Plasma specimens were collected as part of a larger clinical trial of buprenorphine using dosing schedules previously described in a preliminary communication (26). Three dosing schedules were employed: an 8-mg/70-kg dose was administered every 24 h, a 16-mg/70-kgdose was administered every 48 h, and a 24-mg/70-kg dose was administered every 72 h. Four opioid-dependent outpatients abstinent from illicit opioids who had been maintained on 8 mg/70 kg sublingual buprenorphine were exposed to the dosing schedules in a random sequence. A dosing schedule lasted 17 days and consisted of dailyadministration of a sublingual dose of buprenorphine (8, 16, or 24 mg/70 kg) or saline placebo. During each dosing schedule, laboratorysessions (each separated by at least 72 h) were randomly scheduled to occur 24 and 48 h after an 8-mg/70-kg dose; 24, 48, and 72 h after a 16-mg/70-kg dose; and 24, 48, 72, and 96 h after a 24-mg/70-kg dose. Therefore, an additional placebo was interposed before the laboratory session with the longest time since dosing. A control session was interposed that consistedof three consecutive24-h dosings with 2 mg/70 kg buprenorphine. During laboratory sessions, subjects were also administered 0 (saline), 6, and 12 rag/70 kg of subcutaneous hydromorphone at 90-rain intervals using a cumulative-dosing paradigm. Blood was collected in 10-mL heparinized Venoject| tubes (Becton Dickinson, Franklin Lakes, NJ) at the beginning of a laboratory session (30 rain before first subcutaneous injection).The blood was centrifuged immediately,and the plasma was stored at-20~ until analysis. Calibrator and control preparation All buprenorphine stock solutions used for calibrator and quality control (QC) sample preparation were prepared by diluting the purchased reference solutions in methanol (1 to 10). Stock solutions were diluted (1 to 10) in distilled water to obtain concentrations of 1.0 pg/mL for analysis by GC-PCI-MS and 0.1 pg/mL for analysis by LC-ESI-MS--MS. Stock solutions for calibratorsand QC sampleswere preparedby different individuals.Internal standard spiking solutions were prepared by successive 1 to 10 dilutions of methanolic stock solutions of buprenorphine-d4to obtain concentrations of 1.0 I~g/mL for GC-PCI-MS analysis and 0.1 pg/mL for LC-ESIMS-MS analysis. GC-PCI-MSanalytical method Extraction. Two-milliliteraliquotsof calibrators (0, 0.5,1.0, 2.5, 5,10, 20, 30, and 40 ng/mL),QCsamples(1.0,2.5 and 12.5ng/mL) and subject plasma samples were pipetted into 16 x 100-ram 407


tubes. Fifty microliters of internal standard (buprenorphine-d4 silanized tubes. Twenty microliters of internal standard at 0.1 ng/pL) was added to each tube; the tubes were mixed (buprenorphine-d4 at 1.0 ng/pL) was added to each tube; the brieflyand allowedto equilibrate for approximately 1 h. The pH tubes were mixed briefly, allowed to equilibrate for approximately 1 h, and the pH adjusted to approximately 10.5 with of each sample was adjusted to approximately 10.5 with 25 pL of 2N sodium hydroxide. Four milliliters of n-butyl chlo40 pL of 2N sodium hydroxide.Four milliliters of n-butyl chloride/acetonitrile (4:1, v/v) was added to each tube. The tubes ride/acetonitrile (4:1, v/v) was added to each tube. The tubes were mixed on a rocker for 30 rain and centrifuged for 10 rain were mixed on a rocker for 30 rain and centrifuged for 10 rain at 2250 rpm. The organic layer was transferred to clean 13 x at 2250 rpm. The organic layer was transferred to clean 13 x 100-ram silanized tubes and evaporatedat 40~ under a stream 100-ram silanized tubes, and the solvent was evaporated at of air. Seventy-fivemicroliters of H20/acetonitrile (2:1) con40~ under a stream of air. One-hundred microliters of PFPA taining 0.1% formic acid was added to each tube. The tubes and 100 pL of toluene were then added to each tube, and the tubes were allowed to sit at room temperature for 30 rain. The were vortex mixed, centrifuged at 2250 rpm for 5 rain, and the samples were evaporated under a stream of air at room temliquid transferred to labeled autosampler vials for injection perature until completely dry (i.e., no odor of PFPAremained). into the LC-ESI-MS-MS. Forty microliters of n-butyl chloride was added to each tube, LC-ESI-MS-MS conditions. The analysis was performed using a Finnigan-MATTSQ 7000 triple-stage quadrupole MS the tubes vortex mixed, and the liquid transferred to labeled with an ESI interface. A CTC A200 LC autosampler was used to autosampler vials for injection into the GC-MS. GC-PCI-MS conditions. The analysis was performed using a inject 20 pL of the extracts onto the Alltech Cs LC column. A LDC Analytical(RivieraBeach, FL) constaMetric| 4100 MS LC Finnigan-MAT4500 MS equipped with a 9610 GC, INCOSsoftpump was operated isocratically to deliver a flow rate of ware, and a CTC A200S autosampler (San Jos~, CA). A DB-1 fused-silica capillary column was used with hydrogen as the carrier gas. A typical temperature program began at 160~ for 0.1 A (MH*-H20) 100 " rain, and the temperature was increased at a CF3CF2C-O"~ rn/z = 596 rate of 20~ to 310~ The final tern= perature was held for 0.1 min. All injections "~ = were made in the splitless mode of opera~ so CH3 7, ~ N'cH (MH*) tion. The GC to MS interface temperature ndz [] 614 HOG-CH3 m/z = 4 6 7 was maintained at 285~ and the injector C(CH3)3 temperature was held at 270~ The MS was I I l I I I I I / I operated in the positive ion detection mode. o 2O0 600 300 400 500 A reagent-gas mixture of methane and m/z ammonia at a ratio of approximately4:1 was ,,.=B introduced into the MS until an ion source m/z = 596 pressure of approximately 1 torr was reached. The ion source was maintained at 140~ The electron multiplier was operated at 1600 volts and the conversion dynode at -3 kV. Selected ion monitoring centroid data were collected at m/z 596 and 600 for I I I .=_ Z211416. buprenorphine and buprenorphine-d4, C .~.~ I~,0" respectively. Peak-height ratios (d0/d4) of the calibrators were used to generate a calm/z = 600 c~ ibration curve, and a least-square equation was calculated and applied to the peak height ratios of QC and subject samples to determine their buprenorphine concentration.

2"~4MH+_CF3CF2CO )

t==

LC-ESI-MS-MS analytical method Extraction. The extraction was performed as described here previously,except that the method was modified to use 1-mL samples. One-milliliter calibrator samples (0, 0.10, 0.25, 0.50, 0.75, 1.0, 2.5, 5.0, 7.5, and 10 ng/mL), QC samples (0.25, 2.5, and 7.5 ng/mL), and subject plasma samples were pipetted into separate 16 x 100-ramsilanized 408

4121

,=e

4143

~,&

5ll~

l=e

5l2",e

,~

gl48

r&

61|0

Scan time (min)

Figure 1. Fragmentation and chromatography of buprenorphine during GC-PCI-MS. (A) Spectraof buprenorphine derivatized with PFPA and ionized with methane-ammonia under positive ion chemical ionization conditions. The full molecular structure of derivatized buprenorphine is shown. Ion current profile of plasma that was fortified with buprenorphine at 0.5 ng/mL (limit of quantitation) (B) and buprenorphine-d4 at 10 ng/mL (C), extracted, derivatized with PFPA,and analyzed by GC-PCI-MS.


0.25 mL/min. The solvent was H20/MeOH/acetonitrile (25:30:45, v/v/v) containing 0.1% formic acid. The tube lens and capillaryvoltages were optimized for maximum buprenorphine signal. The ESI spray voltage was 5 kV,and the capillary temperature was 225~ The sheath gas pressure was 80 psi nitrogen, and the auxiliary flow was 10 units. The MS was operated in the MS-MS mode with a collision energy of-25 eVand 2.5 retort argon collision gas pressure. Both Q1 and Q3 were set to monitor ions at m/z 468 and lOO U 472. Peak-area ratios (do/d4) of the calibrators were used to generate a calibration curve; a least-square equation was calculated and applied to the peak-area ratios of 50.>_ QC samples and subjects to determine 107 buprenorphine concentration. nr i

0

Results and Discussion

396 and 414 were monitored over a range of collision energies. The plot in Figure 2B shows that the buprenorphine MH§ ion intensity (m/z 468) remains relativelyconstant up to a collision energy of -20 eV and then drops sharply above -30 eV. The intensities for the two most intense product ions (m/z 396 and 414) remain small throughout the range of collision ener-

A m/z =

152 187

/ ,.t II

' 100

267

I

I

312

I II

,

200

I

396

414

\/I

I

I

300

468

,

I

I

400

m/z

B

Chromatography and spectrometry

A positive ion chemical ionization mass spectrum of the pentafluoropropionyl derivative of buprenorphine is shown in Figure 1A.The relativelylow intensity protonate molecule [MH+] at m/z 614 undergoes loss of H20 to give the intense peak at m/z 596. The only other significant peak above m/z 200 is the low intensity peak at m/z 467 that is formedby loss of CF3CF~CO from the MH+. The corresponding ions of derivatizedbuprenorphine-d4 occur at m/z 618, 600, and 471. For sample analysis, the ion currents at m/z 596 and 600 are monitored. Figures 1B and C show the ion current profiles resulting from analysis of a plasma sample containing 0.5 ng/mL (LOQ) of buprenorphine and 10 ng/mL of the deuterated internal standard. Triple quadrupole MS are commonly used in pharmaceutical analysis for the measurement of trace drugs and metabolites. Most often, the technique of selected reaction monitoring is used in which an abundant analyte ion is selectively transmitted through Q1, collisionally dissociated to product ions in Q2, and only selected product ions are transmitted through Q3. This process greatly reduces the chemical noise reaching the detector and, therefore, results in a higher signalto-noise ratio than that achievable by single-stage mass analysis. However, the buprenorphine MH+ ion formed by ESI proved difficult to fragment, even under rigorous collision cell conditions (Figure 2A). The most intense product ions at m/z

I

500

2e+6

I e+ 6

Oe+O i

0

10

20

30

40

50

60

-eV

Figure2. Collision-induced dissociation of buprenorphine after LC separation and ESI.(A) Spectra of fragments formed in Q2 at a collision energy of 30 eV after selection of m/z 468 by Q1. (B) Changes in intensity of m/z 468 ( 9 ), 414 ( 9 and 396 (o) with changes in collision energy.

Source

91

92

03

n d z = 468

Buprenorphlne

m / z = 468 lm

~

D

Background

Figure3. Illustration of the MS-MS mode of operation. The MS-MS has three quadrupoles, Qt, Q2, and Q3. The parent ion, along with some background material, is selected by Q1 set at m/z 468. The collision reactions (i.e., fragmentation) in Q2 do not produce significant fragment ions of the parent but do fragment background ions. Selection of parent ion (m/z 468) by Q3 increasesthe signal-to-noise ratio.

409


gies. At collision energies above -40 eV the buprenorphine MH§ is shattered and forms many low intensity product ions. Consequently, the best signal-to-noise was achieved when the MH§ ions for buprenorphine (m/z 468) and buprenorphine-d4 (m/z 472) were selected and transmitted through both Q1 and Q3. In this case, the improvement in signal-to-noise is due to the fact that ions at m/z 468 and 472 from chemical background undergo collision-induced fragmentation in Q2 and are Table I. Recovery of Burpenorphine from Human Plasma after Single-Step Liquid-Liquid Extractions GC-PCI-MS

LC-ESI-MS-MS

QC (ng/mt)

Recovery

QC

Recovery

(%)

(rig/rot)

(%)

5.00 25.0 75.0

87.2 81.2 97.2

0.25 2.50 7.50

82.9 74.2 69.5

3.0 E3 ~-~ ~

m/z:468

not transmitted through Q3 as a result. The process is depicted schematically in Figure 3. Buprenorphine was monitored as the surviving parent ion, which provided symmetrical ion current profiles with a minimum of interference even at the LOQ of 0.1 ng/mL (Figure 4). Recovery

Recovery of buprenorphine was determined by comparing the peak-area ratios calculated when buprenorphine-d0 and buprenorphine-d4 were extracted together (Batch A) to the peak-area ratios obtained when buprenorphine-d0 was extracted and buprenorphine-d4 added just before derivatization (GC-PCI-MS) or before reconstitution (LC-ESI-MS-MS) (Batch B). Recovery was determined at three concentrations with five replicates per batch at each concentration. Recovery was expressed as (mean peak-area ratio for Batch A) / (mean peak-area ratio for Batch B) x 100% (TableI). Recoveryvaried from 81 to 97% for the GC-PCI-MS method at concentrations ranging from 5 to 75 ng/mL and from 70 to 83% for the LC-ESI-MS-MS method at concentrations ranging from 0.25 to 7.5 ng/mL. Although the extraction methods were fairly similar, the greater volume of organic or the 2-mL sample A volume used in the GC-PCI-MS method resulted in greater extraction efficiency.

tinearity

2.0 E3

1.0 E3

B

rn/z = 472

1.0 E5

0.6 E5

0.2 E5 .

.

.

,

.

.

6:00

.

.

.

.

.

.

.

,

_

.

6:50

,

.

.

.

.

,

.

7:40

.

.

.

.

.

.

8:30

9:20

Scan time (rain) Figure 4. Ion current profiles of (A) 0.10 ng/mL buprenorphine monitored at m/z 468 and (B) 5 nglml_buprenorphine-d4monitored at m/z 472 after LC-ESI-MS-MS.

410

During the GC-PCI-MS method development, a linear range of 0.5 to 40 ng/mL was established. The calibration curves were reproducible and linear through the entire concentration range. The average correlation coefficients (r 2) were 0.998 (Table II). The LC-ESI-MS-MS calibration curve was reproducible and linear from 0.1 to 10 ng/mL with a mean correlation coefficient of 0.999 (TableII). During analysis of clinical specimens, we found that buprenorphine concentrations rarely exceeded 10 ng/mL. Therefore, during LC-ESIMS-MS method development, we limited the upper range of the calibration curve to 10 ng/mL. The practicality of this upper limit is consistent with other studies that have specifically measured buprenorphine. With the exception of the first 10-15 rain following intravenous injection of buprenorphine, no other plasma concentrations have been reported that exceeded 10 ng/mL (16,21-23,27).

Accuracy and precision The accuracy and precision of the methods were determined from replicate analysis of the LOQ calibrator and the QC samples at three concentrations. Preliminary experiments demonstrated that accurate quantitation could not be achieved at

Journal of Analytical Toxicology, Vol, 2 I, October 1997

concentrations below 0.5 and 0.1 ng/mL for the GC-PCI-MS and LC-ESI-MS-MS methods, respectively.These concentrations were then confirmed as the LOQs from the subsequently described precision and accuracy experiments. Accuracy was expressed as the percent of measured concentration relative to fortified (target) concentration. Precision was expressed as the percent coefficient of variance (%CV). Intra-assay accuracy and precision were determined from the analysis of five replicates at each concentration within a single analyticalbatch. For the GC-PCI-MSmethod, the LOQ accuracy was within 20% of target concentration, and the three QC samples were all within 14% of their targets. These determinations were precise with %CVs ranging from 4 to 6% (Table III). The LC-ESI-MS-MS method demonstrated even better intra-assay accuracy (within 12% of target) and similar precision with %CVsranging from 4 to 10%. Interassay values were determined from seven analytical runs performed on separate days. The average value for 3-5 replicates of the LOQ and QC samples within the run were used. Interassay accuracy for the GC-PCI-MS method at the LOQ was within 20% of target and within 8% of target for the QC samples. The interassay %CVs ranged from 7 to 14%. The LC-ESI-MS-MS interassay accuracy for the LOQ and the QC samples was within 8% of target, while the interassay %CVfor the LOQwas 10%, those for the QC samples did not exceed4% (Table III). Both methods demonstrated accuracy and precision that was acceptablefor clinical pharmacokinetic, safety and efficacy studies (28). An LOQ of 0.1 ng/mL of buprenorphine was established for the LC-ESI-MS-MS method. Although other methods have reported LOQs or limits of detection that approach 0.1 ng/mL (21-24), only Hoja et al. (25) have reported accuracy and precision at this concentration. The lowest concentrations at which accuracy and/or precision were reported by others were 0.5 (22), 0.5 (23), 1.0 (24), and 5.0 ng/mL (21). Stability Buprenorphine undergoes a chemical rearrangement when heated under acidic conditions. This has been described in detail for studies performed in aqueous solutions (15,29). The stability of buprenorphine in biological specimens, however, has only been addressed in a single study (16). That study demonstrated that buprenorphine was stable in frozen plasma for up to 28 days. The bench-top and freeze-thaw stability of buprenorphine was determined for both of the methods pre-

sented here. At concentrations ranging from 0.25 to 40 ng/mL, we found that buprenorphine was stable in plasma stored at room temperature for up to 24 h. Two freeze-thaw cycles did not result in any appreciable loss of buprenorphine at concentrations ranging from 0.25 to 12.5 ng/mL (Figure 5). In addition, QC samples routinely analyzed by the GC-PCI-MS method were stored at 4~ These QC samples were analyzed over a 124-day period without any loss of buprenorphine (Figure 6A). QC samples routinely used in the LC-ESI-MS-MS method were aliquotted and stored at-20~ These QC samples were followed for 55 days without any noticeable loss of buprenorphine (Figure 6B). These data set lower limits for refrigerated and frozen storage of plasma samples that extend beyond stability limits of 28 days of frozen storage as reported by Ho (16). Clinical applications

Forty specimens were collected from four subjects who received sublingual doses of buprenorphine (2, 8, 16, and 24 mg/70 kg) as described in Experimental. Collection times were not successive; one collection followed one individual dose. The results of LC-ESI-MS-MS analysis of these specimens are presented in Table IV.Averageplasma buprenorphine concentrations were dose and postdose time-dependent. The utility of the method was demonstrated because buprenorphine could be detected in the plasma of all four subjects 24-h after administration of a 2-mg/70-kg sublingual dose. A previous study using a method with an LOQ of 0.2 ng/mL could only follow plasma buprenorphine in all subjects for 7 h after a 4-mg sublingual dose. At 24 h after their 4-mg dose, buprenorphine was not quantitatable in two of the six subjects (27). This difference between the two studies may be explained by the lower LOQ of our method. Another possibility is the interindividual differences in buprenorphine pharmacokinetics. There were interindividual differences in buprenorphine concentrations between individuals in our study. For example, subject V008 had dose-dependent concentrations at 24 h postdose of 0.4, 0.5, 1.0 and 1.9 ng/mL following 2, 8, 16, and 24mg/70-kg doses, whereas subject V001 had 24-h postdose concentrations of 0.2, 0.4, 0.5, and 0.3 ng/mL following the same respective doses (Table IV). This individual difference ranged from 1.5-fold in the samples taken 24 h after the 8-mg/70-kg dose to 10-foldin the samples taken 48 h following the 24-mg/70-kg dose (Table IV). This individual variation is

Table II. Characteristics of Calibration Curves* FullCurve

Linear range

Intercept

Method

n

(ng/mL)

(ng/mL)

GC-PCI-MS LC-ESI-MS-MS

7 7

0.5 - 40 0.1 - I 0

-0.108 • 0.155 0.001 • 0.003

Slope height ratio)]

[(ng/mC)/(peak

11.48 • 1.42 4.82 • 0.32

LowCurve

Correlation coefficient

Intercept

(r z)

(ng/mC)

0.998 • 0.001 0.999 • 0.001

0.026 • 0.033 0.001 • 0.000

* Note: Calibration curves were calculated with concentration on the y-axis and peak height ratios on the x-axis. Split low/high curves were used. The characteristics of the full curve are shown to demonstrate overall linearity. The intercept for the low curves is also shown to demonstratethat averageoffset of the curve at the low end did not exceed 5 and 1% of the GC-PCI-MS and LC-ESI-MS-MS LOQs, respectively. The slopesdiffer since the GC-PCI-MS method employed twice the concentration of internal standard.

411


consistent with that reported by Kuhlman et al. (27) who found up to 21.6-, 8.6-, and 4.8-fold differences among six subjects after buccal, sublingual, and intravenous administration of buprenorphine, respectively. Comparison of methods

The forty clinical specimens were also analyzed by the GC-PCI-MS method. The utility of decreasing the LOQ from 0.5 to 0.1 ng/mL was demonstrated by these comparative analyses. With the LOQ of 0.5 ng/mL, only 10 of the 40 (25%) specimens had quantitatable buprenorphine, which was also true for 38 (95%) of the specimens when the LOQ was 0.1 ng/mL. As only 10 specimens were quantitatable by both methods, only a limited comparison of quantitation could be made. Regression analysis gave an r 2 of 0.96. This provides limited evidence that the two methods provide agreeable results. Because of the long postdose collection times used in this study, it provides a relatively rigorous test of the sensitivity of the methods. It should be noted that the GC-PCI-MS method was used for a large multi-center study that collected 24-h postdose specimens, and buprenorphine was quantitatable in approximately two-thirds of these specimens.

administration is 27.7 h (5.2 to 49.1 h) (27). Therefore, detection of buprenorphine out to 96-h postdose, or longer, is important for accurate determination of pharmacokinetic parameters, such as areas under the curve. Reducing the LOQ from 0.5 to 0.1 ng/mL is also important to follow the pharmacodynamic effects of buprenorphine. For the subjects described in Table I~ Amass et al. (26) performed additional experiments to assess the opioid antagonist effects of buprenorphine. They found that buprenorphine blocked the effects of hydromorphone on observer- and subject-related measures of opioids for up to 48 h. As shown in Table IV,this antagonism occurred in several subjects with buprenorphine plasma concentrations of less than 0.5 ng/mL. This study has also presented long-term 1201008060" 40-

Conclusion

20C~

Of the two assays described here, the LC-MS-MS method provides the best sensitivity. It also requires less time to perform; there is no derivatization, and the cycle time between injections is approximately two-thirds that of the GC-PCI-MS method. The major disadvantage of the LC-MS-MS method is the high cost of the instrumentation. We have found that LC-MS-MS can achieve a precise and accurate LOQ of 0.1 ng/mL for buprenorphine. Although a similar LOQ was achieved with LC-MS, a more rigorous extraction was required to reach this sensitivity with the single-stage MS (25). With this LOQ, we were able to demonstrate plasma buprenorphine concentrations out to 96-h postdose. The reported mean half-life of buprenorphine after sublingual Table III. Intra- and Interassay Accuracy and Precision*

Method GC-PCI-MS

LC-ESI-MS-MS

Target concentration (ng/mL) 0.5 (LOQ) 1.0 2.5 12.5 0.1 (LOQ) 0.25 2.5 7.5

O-

0.25

2.5

7.5

12.5

40

2.5

7.5

12.5

40

120

~

100

0

~

80

e~

60

40 20

Accuracy and Precision Intra-assay Interassay (% target _+% CV) 120 + 4 116+5 112+6 114__.4 100_+10 108_+4 88 _+4 99 _+4

120 + 10 101+12 102+14 108_+7 100+10 104_+4 92 _+4 96 _+1

* Intra-assay values are from replicates of 5 within a single run. Interassay values are from means of seven analytical runs performed on separate days with each samples at an n of 3-5.

412

0

0.25

Sample concentration (ng/mL) Figure 5. Bench-top and freeze-thaw stability of buprenorphine in human plasma. Blank plasmas were fortified with buprenorphine at the concentrations noted. (A) One set of plasma at each concentration was thawed and stored at room temperature 24 h before controls; samples were extracted and analyzed once the controls thawed. (B) One set of plasma at each concentration was thawed and then refrozen at -20~ These and control sampleswere then thawed a day later and analyzed. Values are the mean plus or minus standard error of the mean of three replicatesfor each concentration.Solid bars: analysisperformed by LC-ESI-MS-MS; hatched bars: analysis performed by GC-PCI-MS.


stability data on refrigerated and frozen plasma samples containing buprenorphine. Recently, we were asked to modify the LC-MS-MS assay in order to permit simultaneous measurement of buprenorphine and naloxone. This is because naloxone is being evaluated in a coformulation with buprenorphine for use in treatment of opioid addiction. It is anticipated that the dosage of naloxone can be adjusted in the buprenorphine-naloxone formulation such that the naloxone will have no antagonistic effect when administered sublingually, where it undergoes extensive firstpass metabolism, but will act as an opiate-receptor antagonist

when the medication is injected intravenously. This coformulation would thereby discourage diversion of the medication for purposes of drug abuse. The modified LC-MS-MS assay has been validated and has a LOQ of 0.1 ng/mL for both buprenorphine and naloxone. A preliminary communication describing the assay has been accepted (30).

Acknowledgment

This work was supported in part by NIDA contract N01DA-1-9205and NIDAgrant DA 06969.

Table IV. Plasma Buprednorphine Concentrations Versus Doses of Sublingual Buprenorphine Plasmabuprendorphine(ng/mL) Dose (mg/70 kg)

Subject

Postdose (h)

V001

V002

V006

V008

2

24

0.2

0.2

0.3

0.4

0.3•

8

24 48

0.4 0.2

0.6 0.3

0.4 0.5

0.5 0.6

0.5• 0.4•

16

24 48 72

0.5 0.2 0.2

1.2