The Detection of Ag-Tetrahydrocannabinol (THC) and. in Whole Blood Using Two-Dimensional Gas Chromatography and El-Mass Spectrometry

Transcription

1 ournal of Analytical Toxicology, Vol. 30, May 2006 The Detection of Ag-Tetrahydrocannabinol (THC) and ll-nor-9-carboxy-ag-tetrahydrocannabinol (THCA) in Whole Blood Using Two-Dimensional Gas Chromatography and El-Mass Spectrometry Rodger D. Scurlock*, Greg B. Ohlson, and David K. Worthen AZ DPS Central Regional Crime Lab, Toxicology Unit, 2102 W. Encanto Blvd., Phoenix, Arizona Abstract] A method is described for the simultaneous analysis of A 9- tetrahydrocannabinol (THC) and its carboxylic acid metabolite, 11 -nor-9-carboxy-ag-tetrahydrocannabinol (THCA) as their trimethylsiyl derivatives using 2-dimensional chromatography and electron ionization-mass spectrometric detection. The addition of a Deans switch to a standard GC oven allows the use of two chromatographic columns of differing stationary phase to greatly reduce matrix interference. The analytes are extracted from 1 ml of whole blood by first precipitating the blood proteins with the addition of acetonitrile followed by solid-phase extraction. The limit of quantitation for both THC and THCA was determined to be 1.0 ng/ml. The between-run precision at 1.0 ng/ml (N = 30) was 7.7% and 7.4% for THC and THCA, respectively. The method is linear from 1 to 100 ng/ml. Introduction Following a single use of marijuana by an infrequent user, the plasma concentration of parent Ag-tetrahydrocannabinol (THC) peaks at about 10 rain, falls below 5 ng/ml (-2.75 ng/ml whole blood) within 3 h (1,2), and to below 1 ng/ml (~0.55 ng/ml whole blood) within 6 h (1). The carboxylic acidmetabolite, ll-nor-9-carboxy-a9-tetrahydrocannabinol (THCA), peaks at about 2.5 h after a single use and can persist above 2.0 ng/ml (plasma) for more than 72 h (1). In chronic marijuana users THC and THCA can persist longer. Although the plasma concentration of THC also falls below 5 ng/ml within 3 h (the same as with the infrequent user), THC can remain at or above i ng/ml (~0.55 ng/ml, whole blood) for 24 h (3), and THCA can remain in excess of 20 ng/ml for up 72 h after use (~11 ng/ml, whole blood) (3) [Blood concentrations of THC and THCA were calculated from plasma concentrations using a blood/plasma ratio of 0.55 (4)]. In laboratories that support law enforcement efforts to re- * Author to whom correspondence should be addressed. E-maih rscurlock@azdps.gov. duce drug-impaired driving, it may be important to have a limit of quantitation (LOQ) for the parent THC as low as 1.0 ng/ml (0.55 ng/ml whole blood) for the purpose of narrowing the probable time-of-use window. By contrast, a low LOQ for the metabolite THCA is not as critical in narrowing the probable time-of-use window because it can persist at a comparatively high concentration for an extended time after use. This article reports an improved system for analysis of both THC and THCA; however, the greatest benefit is for THC. As mentioned previously, it may be important for a laboratory to have a LOQ for THC that is lower than the LOQ for THCA. In our laboratory the chromatographic peak for THC had been more frequently burdened by matrix interference than the metabolite, THCA. Various gas chromatography-mass spectrometry (GC-MS) approaches to improve the LOQ for THC have been reported by other workers. Felgate and Dinan (5) focused their efforts on improving specimen cleanup by using multiple-stage extractions, liquid-liquid followed by solid-phase extraction (SPE) in advance of analysis with electron ionization (EI)-GC-MS detection. Others took advantage of the better detector specificity of MS detection with negative (6) and positive (7) chemical ionization (CI) or tandem MS detection (8,9) with standard SPE. Although CI-MS provides impressive detection limits, the various derivatives of THC do not provide adequate fragment ions for the purposes of many forensic laboratories-- often, a primary ion and two qualifier ions are required. In this article we report a 2-dimensional (2D) chromatography system with an EI-MS detector (2D-GC-MS) that has been successful in routine measurements of THC and THCA down to 1.0 ng/ml in whole blood. This system consists of a comparatively inexpensive addition to common GC ovens, which allows the use of two chromatographic columns of differing phases to separate the analyte from matrix interference. A Deans switch connects the output of the primary column to either a flame-ionization detector (FID) or the entrance of the second column, which terminates at the MS detector. This is illustrated in Figure 1. The Deans switch is composed of an electronic pressure control (EPC) module, a mechanical valve, and a man- 262 Reproduction (photocopying) of editorial content of this journal is prohibited without publisher's permission.

2 ournal of Analytical Toxicology, Vol. 30, May 2006 ifold to connect the capillary columns. The mechanical valve and the EPC module are outside the oven so that there are no moving parts inside. In practice, the analyst first injects a high-concentration standard and directs the whole output of the primary column to the FID with the Deans switch. The resulting chromatogram allows the determination of the retention times of both the analytes on the primary column. The Deans switch is then programmed to direct the output of the primary column to the secondary column for only a small window surrounding the retention time of each analyte peak. In effect, the components in those time windows are "cut" from the primary chromatogram and switched onto the secondary column for further chromatographic separation and MS detection. Any compound that might coelute with the analyte on the first column will separate from the analyte on the second column (given the proper choice of the two columns). The result has been extraordinarily clean chromatography regardless of the degree of degradation of the blood. Materials and Methods The drug standards for THC, THC-d3, THCA, and THCA-d3 were purchased from Cerilliant (Round Rock, TX). Human blood was from United Blood Services (Scottsdale, AZ). BSTFA + 1% TMCS, MTBSTFA + 1% TBDMSC, and trifluoroacetic anhydride (TFAA) were from United Chemical Technologies (Bristol, PA). Hexafluror-isopropanol (HFIP) was from Campbell Science (Rockton, IL). SPE columns were the Cerex polychrome THC columns from SPEware (San Pedro, CA). Acetonitrile, hexane, ethyl acetate, and methanol were optima grade from Fisher Scientific (Pittsburgh, PA). No effort was made to silanize the glassware before use. Sample preparation Appropriate mixtures of THC and THCA in methanol were prepared beforehand so that the desired calibrators and quality control (QC) samples could be made immediately before the extraction by adding 20 ll of the appropriate mixture to I ml of blank blood. Each batch contained calibrators at 1.0, 2.0, 5.0, 20, and 50 ng/ml. The QC samples in each batch were at 0, 4, 16, and 40 ng/ml. A 100 ng/ml limit-of-linearity sample was in- I Injection port ' Deans Switch using EPC-2 Figure 1. Diagram of a 2-dimensional chromatography system using a Deans switch. cluded. To all samples in the batch (calibrators, controls, and unknown samples), 10 ial of internal standard (1.0 ng/ll THCd3 and THCA-d3 in methanol) was added. To each sample, 2.0 ml of cold (-20~ acetonitrile was added and immediately vortex mixed for 30 s before proceeding to the next sample. The cold acetonitrile produced a more finely divided protein precipitate than did the room temperature acetonitrile. The samples were briefly centrifuged, the supernate was decanted to a clean tube, and the solids were discarded. Two milliliters of deionized water was added to each sample before it was poured onto the SPE column. The columns were washed with a 1.0 ml mixture of water/acetonitrile/nh4oh (85:15:1, prepared daily) and dried for 10 rain by a forced flow of nitrogen. The THC fraction was eluted into a tube with 2.0 ml of ethyl acetate. The SPE column was dried for 10 rain before the THCA fraction was eluted into the same tube with 2.0 ml of hexane/ethyl acetate/acetic acid mix (90:10:3, prepared daily). All flows through the column were ~ 1 drop/s controlled by positive pressure. The samples were evaporated to dryness at 40~ under a stream of nitrogen. Fifty microliters BSTFA + 1%TMCS and 50 ll toluene were added to each tube; the tube was vortex mixed, and the contents were transferred to a GC-MS vial. The vials were crimp-capped and heated at 70~ for 20 rain. Instrumentation The GC was an Agilent 6890 equipped with an FID and a Deans switch. The Deans switch, from Agilent Technologies, consists of a second EPC module, a solenoid switch that is outside the oven, and a manifold inside the oven to connect the GC columns. The mass spectrum detector (MSD) was an Agilent The carrier gas was helium. The injection port temperature was 250~ and the transfer line was at 300~ The MSD was operated at 200 V above the tune in the selected ion monitoring (SIM) mode. The SIM ions were m/z 386.3, 371.3, and for THE (the 315 ion was not included for THC because it is also an ion of the internal standard, THE-d3); rn/z and for THC-d3; m/z 371.3, 473.3, and for THCA; and m/z and for THCA-d3. The dwell time for all ions was 25 ms and "high" resolution was selected. The precise ion mass was determined using a separate GC-MS program for each ion in which 10 ions were monitored (5 on either side of the target ion) in steps of 0.1 m/z mass unit. The oven temperature program was as follows: initial temperature 120~ increased at 20~ to 200~ further increased at 10~ to 250~ increased again at 25~ to 300~ and held for 0.5 rain. The injection port liner was a deactivated 4-ram splitless gooseneck with glass wool from Restek. The injection volume was 2 IL. Deans switch parameters Column I was an RTX-200 (20 m, 0.18-mm i.d., 0.20-lm df). Column 2 was a DB-17 (15 m, 0.25-ram i.d., 0.254am do. The pressures for the injection port and the Deans switch were calculated using the Deans Switch Calculator software (Agilent Technologies) to achieve I ml/min flow through the primary column and 2.0 ml/min flow through the secondary column at an oven temperature of 200~ The injection port was set for constant pressure at 41.0 psi, and the Deans switch 263

3 ournal of Analytical Toxicology, Vol. 30, May 2006 pressure was 17.0 psi. The post-run program was set for 1 min with the oven temperature at 320~ and the inlet pressure at 1 psi. With the inlet pressure at 1 psi in the post-run, the carrier gas flow through the primary column was reversed. This ability to back-flush the primary column is an advantage of the Deans switch system and reduces the maintenance frequency of the primary column. The secondary column stays remarkably clean because it is exposed to only a small fraction of the injection volume for each chromatographic run. Selection of derivative Three derivatives were compared: trimethylsilyl (TMS), tertbutyldimethyl-silyl (TBDMS), and the mixed derivative using HFIP and TFAA. A mass of 0.5 ng of THC and THCA in methanol was pipetted into a GC-MS vial, dried, and derivatized using either 1. BSTFA to form the TMS derivatives, 2. MTBSTFA to form the TBDbiS derivative, or 3. HFIP/TFAA to form the hexafluoropropyi ester/trifluoroacetate mixed derivative (excess TFAA was removed by drying after derivatization). Toluene was added so that the final volume after derivatization was 100 lal. One microliter was injected into the GC-MS, and the intensity of the weakest ion for each THC derivative was compared. For our purpose, the intensities of the THC ions were more critical than the intensities of the THCA ions because we needed a lower detection limit for THC. Although none of the derivatives were dramatically better than the others, the TMS derivative was chosen because it gave the most intense signal of its least intense ion (m/z 303). We focused on the absolute intensity of the least intense ion and proceeded assuming that we would eventually find a combination of chromatographic columns that would give 100% chromatographic resolution from the matrix interference. Figure 2 shows the mass spectra of the TMS derivatives of THC and THCA, respectively. Selection of the GC column We investigated GC columns for 2D-GC-MS from the inventory of our laboratory in the combinations listed in Table I. Among the columns tested, only the final combination (#6) was satisfactory, and no further effort was made to investigate other combinations. Of the columns listed in Table I, the retention times for THC and THCA were the shortest on the RTX-200 and the longest on the DB-17. This might be a possible clue to a strategy for choosing columns for a 2D system. The analyte should exhibit as short as possible retention time on the pri- 7oooo~ '1 a~co~ ~ ioo~o THC T H C A 9 i.~o' " " i./,o' " " k~d " " ~,o' " " i.~o' " ~.~o' " " i.~0' " ' ~o' " " ~., T~e (m~) 9 ~.~o,..., Figure 3. FID chromatogram of the primary column. The sample was extracted from 1 ml of blood spiked with 2.0 ng of THC and THCA. The elution times for THC and THCA are indicated. 4000~ 2~00~ 900~ s,~.~: r162 )t,,,,ctl A ff.K~) 350~OO -o 3tX~OO Z ~ 151)000 THC - TMS 3~4 I Time 0n~n). TH[ ]co-elu r g pe~ks from ~e ~'~ [ p~rnary ~Iurnn ~ / I B 51X~0( f iX) ~) 390,'1]/7. TIICA- TMS Thnr (mln) C 2r~ Time (mln) Figure 2. Mass spectra of the TMS derivatives of THC and THCA. Figure 4. Example of 2D GC-MS: THC and THCA (2.0 ng) extracted from 1 me of blood. FID signal from the primary column showing the cut windows for THC and THCA (A), MS-full scan signal of the secondary column (B), and MS-SIM signal of the secondary column (C). 264

4 ournal of Analytical Toxicology, Vol. 30, May 2006 Table I. Column Combinations Tested for 2D Chromatography* Primary Column Secondary Column mary column and as long as possible retention on the secondary column for best results. DB-1 (15 m, 0.25-mm i.d., 0.25-pro df) Ultra-2 (12 m, 0.18-mm i.d., 0.33-pm df) DB-17 (15 m, 0.25-mm i.d., m df) DB-17 (15 m, 0.25-mm i.d., 0.25-pm df) RTX-200 (20 m, 0.18 mm i.d., 0.20-lam df) RTX-200 (20 m, 0.18 mm i.d., 0.20-pm df) * Only combination #6 gave adequate resolution from the interfering matrix peaks. ~oo ~o 500 4oo 330 3OO IOO ]~s 303,371, 3 o ~;..;~:~.. g~.. L;" ';,.~;;' g.~" L~' "~.~"':,.'~o' ~.~ ';.~ '-}g'-}.~ ~.~ -}.~,~:~ -}.~ Time (mb~) Figure 5. Signal-to-noise example. SIM-chromatogram of an extracted blood sample spiked with 0.5 ng/ml of THC. The 303 ion of THC exhibits signal-to-noise = 10. Table II. Precision at 1.0 ng/ml Overall Mean Within-Run CV Between-Run CV THC THCA g loo o g 0 0 2OO loo ~ so THC Response ratio = 0.991x B 2 = THCA Concentration (ng/ml) Response ratio = 0.982x R 2 = , Concentration (ng/rnl) 200 Figure 6. Linear least-squares fit of standards for THC and THCA. The points at 100 and 200 ng/ml are not included in the fits. DB-17 (15 m, 0.25-mm i.d., 0.25-pm df) DB-17 (15m, 0.25-mm i.d., m df) DB-1 (15 m, 0.25-mm i.d., 0.25-pm dr) D8-5 (30 m, 0.25-mm i.d., 0.25-pm df) DB-1 (15 m, 0.25-mm i.d., 0.25-pm df) DB-17 (15 m, 0.25-ram i.d., 0.25-pm df) Results and Discussion The difficulty that is encountered by anyone who wishes to measure THC and THCA in blood is the potential interference from the matrix. Figure 3 shows the scope of the problem; it is the FID chromatogram of a sample extracted from blank blood containing 2.0 ng/ml of THC and THCA and 10 ng of internal standard. The intense peaks are unidentified components of the matrix that remain after the SPE workup. The retention times for THC and THCA on the primary column, 4.95 and 7.65 min, respectively, were determined separately using a clean un-extracted sample. The chromatogram was scaled to contrast the intensity of the THC and THCA signals (invisible on this scale) to the intense peaks from the matrix. The number and intensity of the peaks from the matrix varies with the source and age of the blood specimen. Figure 4 shows the "clean-up" potential of 2D-GC-MS using the same sample as in Figure 3. Figure 4A shows again the FID chromatogram from the primary column, but with two windows "cut" from the primary chromatogram by the Deans switch. The cut windows were set to capture output of the primary column at the retention times of THC and THCA and direct it to the secondary column. Figure 4B is the resultant MS full scan chromatogram of the secondary column. It shows the components that had been present in the two "cut windows". The matrix components that had co-eluted with THC and THCA on the primary column were separated on the second column. Figure 4C is the MS-SIM chromatogram from the secondary column showing the final results for analysis. Although the m/z 303 ion used to monitor THC was also common to a matrix component, the chromatographic peaks were well resolved. Recovery Extraction recovery was evaluated by preparing two sets of standards prepared using blank blood. Set "one" included calibrators with the internal standard added to the blood before the extraction as usual. Set "two" consisted of standards fortified at concentrations of 16 and 40 ng/ml prepared as the samples in set "one" but with the internal standard added after the extraction. The samples from set "two" were evaluated using the calibration curve from set "one". The percent recovery was the resulting value of the standard from set "two" divided by its target value. The average percent recovered was 73% and 68% for THC and THCA, respectively. Sensitivity Figure 5 shows the THC signal-to-noise ratio (S/N) at 0.5 ng/ml. At this concentration, the S/N of the least intense ion (m/z 303) was approximately 10. The extraction efficiency was -70%, the final extract volume was 100 ~L (50 ~L BSTFA, 265

5 ournal of Analytical Toxicology, Vol. 30, May ll toluene), and the injection volume was 2 IL. Thus, ~7 pg was injected onto the column. The LOD is usually defined as the concentration where the S/N is greater than 3. Although the S/N of the 0.5 ng/ml sample exceeded this criterion, upon repeat measurements in 11 different batches, the qualifying ion ratios (QI) concentration failed approximately 30% of the time (the acceptable Q! range was + 20% of the calibrator value). In contrast, the 1.0 ng/ml sample was failure free in every batch. Precision at 1 ng/ml In six different batches, five separate samples (N = 30) were prepared by spiking human blood with 1.0 ng/ml THC and THCA. For each value, the variance from the batch mean was calculated as well as the overall mean. A standard analysis of variance (ANOVA) gave the within-run and between-run coefficient of variation (CV) listed in Table II. Linearity Figure 6 shows the unweighted linear least-square fits for THC and THCA with concentrations: 0, 1, 2, 4, 5, 18, 20, 45, and 50 ng/ml. The points at 100 and 200 ng/ml are not included in the calculated fit but are in the figure to demonstrate deviation from linearity. Our acceptance range was + 10% from linearity. Both THC and THCA demonstrated similar nonlinear behavior; the 100 ng/ml standard was about 6% low (within acceptance range), whereas the 200 ng/ml standard was approximately 17% low. Interference Potential interference from 63 different drugs 1 was tested in the following way: to a vial that contained 2 ng of THC and THCA and 10 ng internal standard (THC-dg and THCA-d3), a methanol solution of the drug was added so that the drug mass was 2000 ng. The vial was dried, and the contents were derivatized with BSTFA in toluene before injection into the GC-MS. None of the drugs thus tested interfered with the quantitation of THC or THCA, nor did they cause any qualifying ion failure. The robustness of this assay against interference from unidentified substances in the blood matrix has been demonstrated over time. This assay was used as a confirmation test for more than 300 whole blood specimens that had tested positive for cannabinoids by ELISA using a cutoff of 2 ng/ml. Using 2D- GC-MS as the confirmation assay, none of the specimens that had a THe or THCA concentration greater than 1.0 ng/ml exhibited a QI failure. The acceptable QI range was _+ 20% of the average value of the calibrators. 1 The following drugs were tested for their potential to interfere with the assay: cocaine, benzoylecgonine, ecgoninemethylester, cocaethylene, oxymorphone, codeine, morphine, hydrocodone, hydromorphone, oxycodone, nordiazepam, temazepam, hydroxyalprazolam, alprazolam, zolpidem, amitriptyline, bupropion, carbamazepine, carisoprodol, chlordiazepoxide, chlorpheniramine, cydobenzaprine, dextromethorphan, diphenhydramine, doxepin, fenfluoramine, fluoxetine, hydroxyzine, imipramine, iprindol, ketamine, MDA, MDMA, meperidine, meprobamate, mescaline, mesoridazine, methadone, EDDP, valproic acid, methocarbamol, methylphenidate, porpoxyphene, norpropoxyphene, PCP, pseudoephedrine, gabapentin, risperidone, propranolol, mirtrazepine, citalopram, sertraline, fentanyl, olanzepine, trazodone, tramadol, thioridazine, thiopental, phentermine, phenytoin, promethazine, nortriptyline, and p-methoxyamphetamine. Conclusions The 2D-GC-MS system described here is a relatively inexpensive addition to common GC ovens that enables an improvement in detection limits of cannabinoids by greatly reducing matrix interference. Acknowledgments Special thanks are due to Bart Gray and Erin Boone for their efforts in preparing samples for the method validation studies and also to ulie Olander for her suggestions and careful edit of the manuscript. References 1. M.A. Huestis,.E. Henningfield, and E.. Cone. Blood cannabinoids I. Absorption of THC and formation of 11-OH-THC and THCCOOH during and after smoking marijuana.. Anal. ToxicoL 16: (1992). 2. M.A. Huestis,.E. Henningfield, and E.. Cone. Blood cannabinoids II. Models for the prediction of time of marijuana exposure from plasma concentrations of Ag-tetrehydrocannabinol (THC) and 11-nor-9-carboxy-Ag-tetrahydrocannabinol (THCCOOH).. Anal. Toxicol. 16: (1992). 3. M.A. Peat. Advances in Analytical Toxicology, Vol II, R.C. Baselt, Ed. Yearbook Medical Publishers, Chicago, IL, 1989, pp M. Widman, S. Agurell, M. Ehrnebo, and G. ones. Binding of (+)- and (-)-delta-l-tetrahydrocannabinols and (-)-7-hydroxy-delta- 1-tetrahydrocannabinol to blood cells and plasma proteins in man.. Pharm. Pharmacol. 26: (1974). 5. P.D. Felgate and A.C. Dinan. The determination of A 9- tetrahydrocannabinol and 11-nor-9-carboxy-A9-tetrahydro - cannabinol in whole blood using solvent extraction combined with polar solid-phase extraction.. Anal. Toxicol. 24: (2000). 6. W. Huang, D.E. Moody, D.M. Andrenyak, E.K. Smith, R.L. Foltz, M.A. Huestis, and.f. Newton. Simultaneous determination ofa 9- tetrahydrocannabinol and 11-nor-9-varboxy-A9-tetrahydrocannabinol in human plasma by solid-phase extraction and gas chromatography-negative ion chemical ionization-mass spectrometry.. Anal. Toxicol. 25: (2001). 7. R.A. Gustafson, E.T. Moolchan, A. Barns, B. Levine, and M.A. Huestis. Validated method for the simultaneous determination of delta 9-tetrahydrocannabinol (THC), 11-hydroxy- THC and 11-nor-9-carboxy-THC in human plasma using solid phase extraction and gas chromatography-mass spectrometry with positive chemical ionization.. Chromatogr. B 798(1): (2003). 8. S. Niedbala, K. Kardos, S. Salamone, D. Fritch, M. Bronsgeest, and E.. Cone. Passive cannabis smoke exposure and oral fluid testing.. Anal. Toxicol. 28: (2004). 9..P. Weller, M. Wolf, and S. Szidat. Enhanced selectivity in the determination of,59-tetrahydrocannabinol and two major metabolites in serum using ion-trap GC-MS-MS.. Anal. Toxicol. 24: (2000). Manuscript received September 30, 2005; revision received December 29,