CLIN.CHEM.37/12,2062-2068 (1991)

2062 CLINICAL CHEMISTRY, Vol. 37, No. 12, 1991

Ultrasensitive Measurement of t-9-Tetrahydrocannabinol with a High Energy DynodeDetector and Electron-Capture Negative Chemical-Ionization Mass SpectrometryLeslie M. Shaw,’ Judith Edling-Owens,’ and Richard Mattes2

Plasma concentrations of -9-tetrahydrocannabinol, theprincipal psychoactive cannabinoid in marijuana, declineto values substantially <1 pg/L within a few hours after asubject has smoked a marijuana cigarette. Using a single-quadrupole gas chromatograph-mass spectrometer (GC/MS) operated in the negative chemical-ionization modeand retrofitted with a High Energy Dynode detector sys-tem, we measured -9-tetrahydrocannabinol and a pri-mary metabolite, 11-nor-i-9-tetrahydrocannabinol-9-COOH. Using a trifluoroacetic anhydride derivatizationprocedure and the High Energy Dynode detector system,we improved by 6.25-fold the limit of detection for -9-tetrahydrocannabinol in plasma over that obtained withthe same GC/MS system without the new detector (80 vs500 ng/L). The new detector system will thus permitfurther investigation of the post-distributionpharmacoki-netics of -9-tetrahydrocannabinol and detection of -9-tetrahydrocannabinol in plasma for a longer time afteringestion of the drug in forensic cases. The High EnergyDynode detector system should be applicable to a widevariety of other GC/MS analyses that require significantlyimproved sensitivity.

AddftlonalKeyphraeee:manjuana drugs . pharma-cokinetics

A recent report provided preliminary evidence that anew detection system, the High Energy Dynode (lIED)detector, could improve detection sensitivity two- to8.6-fold in a quadrupole gas chromatograph-mass spec-trometer (GC/MS) instrument by using the electron-impact, chemical-ionization, and electron-capture ion-ization modes (1). In the drug analysis field an increas-ing number of analyses require greater sensitivity thancan be obtained with quadrupole GCIMS systems, themost common type of mass analyzer systems used inclinical laboratories today. We have investigated thesensitivity for detection of -9-tetrahydrocannabinol(THC), the primary psychoactive cannabinoid in mari-juana, in the plasma of volunteer subjects. Detection ofplasma concentrations considerably <1 g/L is requiredto characterize the post-distribution pharmacokineticsof THC (2-4). In investigations of the pharmacokineticsof THC in normal casual users of marijuana, we found

‘Department ofPathologyand LaboratoryMedicine, Universityof Pennsylvania Medical Center, 3400 Spruce St., Philadelphia,PA 19104.

2Monell Chemical SensesCenter, Philadelphia, PA 19104.3Nonstandard abbreviations: lIED, High Energy Dynode; Gd

MS, gas chromatography/mass spectrometry; THC, -9-tetrahy-drocannabunol; and TFA, trifluoroacetic anhydride.

Received June 18, 1991; accepted October 16, 1991.

that, within the first few hours after smoking a standard(National Institute on Drug Abuse) marijuana ciga-rette, plasma concentrations of THC decreased to <0.5gfL, the limit of detection for a standard detector andan electron-capture negative chemical-ionization iso-tope-dilution GCIMS procedure. Here we describe animprovement of 6.25-fold in the limit of detection forTHC in plasma when the HED detector is used insteadof the standard detector. This will make possible thefurther investigation of the post-distribution pharma-cokinetics of THC in relation to its pharmacologicalactivity. For instance, reliable detection of THC inplasma 6 h or more after smoking may be useful inestablishing recent use in forensic cases; studying therelationship of plasma THC concentration to residualeffects on psychomotor performance (5); and evaluatingthe anti-emetic and appetite-stimulating activity ofTHC (6) taken orally or by other routes. Such a detectorsystem should be applicable to a wide variety of otherGC/MS analyses requiring significantly improved sen-sitivity.

MaterIals and Methods

ReagentsTHC, 11-nor--9-THC-9-COOH (THC-COOH), and

their respective trideuterated analogs were obtainedfrom the National Institute on Drug Abuse, Rockville,MD. Trifiuoroacetic anhydride (TFA) was from SigmaChemical Co., St. Louis, MO. Surfasil was obtained fromPierce Chemical Co., Rockford, IL, and BF3/methanol(120 g/kg) was obtained from Supelco, Bellefonte, PA.All organic solvents were HPLC grade.

Gas Chromatography-MassSpectrometrySystemA Model 5985B Gd/MS system (Hewlett-Packard,

Palo Alto, CA) was used for the analyses. A SupelcoSPB5 capillary column (15 m x 0.25 mm) fitted into aModel 5840A gas chromatograph interfaced directlywith the 5985B mass spectrometer. The standard detec-tor was replaced by retrofitting with an HED detector(Phrasor Scientific, Inc., Duarte, CA).

Procedures

All pipets and other glassware used in the extractionand derivatization of THC and THC-COOH were si-lanized with Surfasil, 20 mLfL in acetone.

Extraction and derivatization. THC and THC-COOHwere extracted from plasma samples prepared fromwhole blood collected in EDTA Vacutainer Tubes (Bee-ton Dickinson and Co., Franklin Lakes, NJ) and thenderivatized essentially as described by Foltz et al. (7).Before the extraction of THC or THC-COOH, the

CLINICALCHEMISTRY, Vol. 37, No. 12, 1991 2063

plasma samples, stored at -20 #{176}Cuntil the day ofanalysis, were thawed and 0.1 mL of ethanol containing1 ng of trideuterated THC (unless otherwise stated) or 5ng of trideuterated THC-COOH was added to 1.0 mL ofplasma specimen. These mixtures were vortex-mixed for10 s and allowed to equilibrate for 1 h at 4#{176}C.For THCanalyses, the initial extraction was achieved by adding2 mL of acetonitrile to each internally standardizedplasma sample. After vortex-mixing the samples for 30s, we centrifuged them for 5 mm at 2000 x g. Thesupernates were decanted into silanized 16 x 100 mmtest tubes and their volume reduced to about 1 mL in aTurboVap LV (Zymark, Hopkinton, MA) at 40#{176}C.Weadded 1 mL of 0.2 mol/L NaOH and 2 mL of hexane/ethyl acetate (9/1 by vol) to the concentrated supernates.After shaking these mixtures for 30 mm with a recip-rocating shaker (60 cycles/mm), we centrifuged them for5 miii at 2000 x g. The organic phase, containing THC,was transferred to clean silanized 16 x 150 mm plastic-stoppered test tubes, then mixed with 2 mL of 0.1 molJLHC1 and shaken for 15 min on a reciprocating shaker.Then the mixtures were centrifuged at 2000 X g for 5mm and we transferred the organic layer to a 12-mLsilanized conical screw-top evaporation tube, takingextreme care to avoid transfer of any water, whichinterferes with the derivatization procedure. After evap-oration of the organic layer to dryness under a stream ofnitrogen at room temperature, 0.1 mL of chloroform and0.1 mL of TFA were added. The tubes were capped,vortex-mixed for 10 s, heated at 70 #{176}Cfor 15 mm, andthe contents evaporated to dryness under nitrogen.Each specimen was reconstituted with 20 L of n-hep-tune before analysis.

For THC-COOH analyses, the initial acetonitrile ex-traction step described for THC was the same. Theliquid phase obtained from this step was treated with0.2 moIJL NaOH and extracted with hexane/ethyl ace-tate as described for the THC extraction procedure. Theresulting aqueous phase, which contained THC-COOHand its deuterated internal standard, was acidified bythe addition of 1 mL of 1 mol/L HC1. This mixture wasextracted with 2 mL of the hexane/ethyl acetate solutionfor 30 mm on the reciprocating shaker. After centrifu-gation for 5 mm at 2000 x g, the organic layer wastransferred to a silanized conical screw-top tube andevaporated to dryness under nitrogen. The first stage ofderivatization was achieved by adding 0.2 mL of theBF3lmethanol solution, vortex-mixing for 10 a, andheating at 70#{176}Cfor 10 mm. After cooling the tubes toroom temperature, we added 1 mL of de-ioni.zed water,then 2 mL of hexane. After shaking the sample for 15mm on the reciprocating shaker and centrifuging at2000 x g for 5 ruin, we transferred the organic layer toa clean silanized evaporation tube, evaporated it todryness, and added 0.1 mL of chloroform and 0.1 mL ofTFA. The tubes were capped and the contents werevortex-mixed for 10 s. The samples were heated, evap-orated to dryness, and reconstituted with 20 L ofn-heptane, as was done after reaction with TFA for the

THC analysis.GC/MS analyses. Electron-capture negative chemical-

ionization isotope-dilution CC/MS analyses were per-formed with a Hewlett-Packard 5985B system with thestandard detector or with the HED detector (1). Thelatter detector featured a dynode to which as much as 10kV could be applied and a modified continuous dynodeelectron multiplier (Model 4761; Gallileo Electro-OpticsCorp., Sturbridge, MA). For detection of negative ions,the potential applied to the conversion dynode is posi-tive. As described in the Discussion, the intensity orpeak area of a negatively charged fragment is increasedby increasing the positive potential applied to the dyn-ode (a measure of relative gain), and the magnitude ofthe increase becomeslarger with increasing mass of thenegatively charged fragments. For example, with ourGd/MS system, the peak intensities for the mlz 414 and633 negatively charged ions of perfluorotri-n-butyl-amine increased linearly by 7.0- and 8.7-fold, respec-tively, as the potential applied to the dynode was in-creased from +2.5 to +8 kV (the maximum appliedvoltage sustainable in our CC/MS system). The tuningprocedure and overall operation of the 5985B Gd/MSsystem were performed according to the protocol of themanufacturer (8). Briefly, the manual tuning procedureensures that the following operating conditions areoptimized: the pressure of the negative-ionization mod-erating gas, methane, adjusted to maximize the m/z 182negative molecular ion peak of benzophenone intro-duced to the mass analyzer through the direct insertionprobe; the voltage settings for the repeller, drawout, ionfocus, and entrance lens; and the adjustment of theatomic mass unit (amu) offset to give a peak width ofabout 0.5 at half the height of the benzophenone peak.Then the mass axis is calibrated with the m/z 414 and633 negative ions of perfiuorotri-n-butylamine intro-duced through the direct insertion probe;the peak widthat half-height of the peaks for the miz 414, 452, and 633ions of the latter compound are readjusted to about 0.5with the amu offset. The sensitivity performance of the5985B Gd/MS in the negative-ionization mode ischecked weekly by iijecting 50 pg of benzophenonesolution (50 pg/pL in methanol) into the mass spectrom-eter through the capillary gas chromatograph. Dwelltime for all selected-ion monitoring analyses was 50 insper ion. The ionization energy was 230 eV. The flow rateof the helium carrier gas was 1.5 mL/min.

For electron-capture negative chemical-ionization iso-tope-dilution Gd/MS analysis of THC, we injected intothe capillary CC/MS system 2 pL of the n-heptanesolution of the TFA derivative of the THC extracts,prepared as described above. The Gd injector was heldat 265 #{176}C,the sample injected in the splitless mode, thesource temperature maintained at 150 #{176}C,and the col-umn temperature ramped from an initial value of100 #{176}Cto a final temperature of 290 #{176}Cat a rate of30 #{176}C/min.The retention time of the TFA derivative ofTHC is about 5 miii under these conditions.

Analysis of the TFA derivative of the methyl ester of

a)0

S

V

E#{149}0

THC.COOH CONCENTRATION (/L)

Fig.2. Calibrationcurve forThC-COOH:ratioof the area of the m/z454 molecular ionpeakof TFA-THC-COOHto thatof the m/z 457molecular Ion peakofTFA-ttideuteratedTHC-COOHforeachstan-dardvstheconcentration of each plasma standard

0 5 10 15 20 25

$00

1200

1000

600

400

200

0 2 4 6 6 10

ITHC CONCENTRATION (pg/U

FIg.1. CalibratIoncurveforThC: ratioof the areaof the m/z 410molecular-ion peakofTFA-ThCtothatofthem/z 413molecular-IonpeakofTFA-thdeuteratedTHCforeachstandardvstheconcentra-tion of each plasma standard

2064 CLINICAL CHEMISTRY, Vol. 37, No. 12, 1991

THC-COOH included the same volume of injected sam-ple in n-heptane, and the same injector and sourcetemperatures as used for analysis of the TFA derivativeof THC. The temperature of the column was rampedlinearly at a rate of 10 #{176}C/minfrom an initial value of200#{176}C,at which it was held for 1 mm, to a final value of260 #{176}C,at which it was also held for 1 mm. Under theseconditions, the retention time of the TFA derivative ofthe methyl ester of THC-000H is about 6.5 mm.

Calculation of THC concentration was based on deter-mination of the ratio of the area under the THC-derivative peak obtained by selected-ion monitoring ofthe molecular ion at mlz 410 to that of the miz 413 peakof the derivative of trideuterated THC. Final calculationwas obtained by comparing this ratio with the ratiosobtained from a series of plasma specimens that con-tained the same concentration of deuterated internalstandard (1 zg/L unless stated otherwise) but a range ofTHC concentrations (see Figure 1 for a typical THCcalibration curve) that were carried through the identi-cal extraction and derivatization procedure describedfor the plasma specimens from the study subjects. TheTHC-COOH concentration was calculated the same wayas for THC, except that the negative molecular ion ofthe TFA derivative of the methyl ester of THC-COOH isn/z 454 and that of the trideuterated internal standard(5 ig/L) is 457 (see Figure 2 for a typical calibration

curve).Limit of detection studies. The limit of detection was

defined in these studies as the lowest THC concentra-tion in plasma that will consistently give a signal-to-noise ratio of at least 3-to-i for the m/z 410 molecularion of TFA-THC prepared from that plasma. The signal-to-noise ratio was defined as the average ratio of theheight, in millimeters, of the mlz 410 TFA-THC molec-ular-ion peak to the height of baseline noise in eightreplicate analyses. When determined in this manner,plasma containing added THC, 80 ng/L, gave a signal-to-noise ratio of 3.0 (SD 0.7). Replicate analyses ofplasma supplemented with lower concentrations of THCfailed to consistently produce a signal-to-noise ratio of3-to-i or better.

Study Subjects

Five volunteer subjects who were casual users ofmarijuana smoked one marijuana cigarette at about0900 h in the Clinical Research Center under thesupervision of a nurse. The cigarettes used were ob-tained from the National Institute on Drug Abuse,which had determined that each cigarette contained19.6 mg of THC. After lighting the cigarette, eachsubject usedthe following procedure for smoking: inhalefor 3 s, hold for 12 s, exhale, wait 15 s, and repeat untilthe cigarette burns down to a line 2 cm from the end.Blood specimens were drawn into lavender-top Vacu-tamer Tubes at the following times: pre-smoking controland 5, 15, 30, 120, 240, 360, and 480 mm after comple-tion of smoking. These subjectswere involved in studiesapproved by our institution’s Committee on StudiesInvolving Human Beings. Plasma prepared from eachspecimenwas stored at -20 #{176}Cuntil the day of analysis.

ResuftsSensitivity and limit of detection. The sensitivity and

limit of detection for analysis of THC in plasma weresubstantially improved by using the HED detector in-stead of the standard detector system (Table 1 andFigure 3). Sensitivity, defined as the slope of the abun-dance (area count) of the molecular ion (mIx 410) ofTFA-THC vs the quantity of injected THC calculated tobe present in the 2 L of injected extract of THC-supplemented plasma samples, was 4.8 area counts/pgof THC by the standard detector and 39 area counts/pgof THC by the HED detector, an increase of 8.1-fold. Thelimit of detection for THC in supplemented plasmasamples was 500 1g/L with the standard detector and 80pgfL with the HED detector, an improvement of 6.25-fold. The reason why different values for sensitivity andlimit of detection were obtained is that the magnitude ofincrease in sensitivity is directly proportional to theincrease in the signal, whereas the limit of detectionchange must also take into account any change inbaseline noise.

The reproducibility obtained for seven plasma sam-ples supplemented with THC, 80 ng/L, was 7.8% (Table

ii’

1%

UmIt of quantlflcatlon

#{163}23

‘343

Fig. 3. Selected-Ionmonltonngtraces of TFA-THC preparedfroma1 tg/L THC plasmastandard containIng tndeuterated THC, 5as obtainedwiththestandarddetector (top) and the HEDdetectorsystem (bottom)The areas of the m’z 410 and 413 molecular-Ionpeaks forTFA-THC andTFA-trldeuteratedmc are indicated to the right ofeachmolecular ion peak.The analysis with the HEDdetectorsystemwasperformedtwodaysafterthatperformed with the standard detector system.M mass spectrometer tuningsettings were the same except that the potential applied to the electronmultiplierwas 1000V for the HEDdetectorsystemand 2400V for the standarddetector system. The TFA-THC derivative was preparedon the day of theanalysiswith the standarddetector.An aliquotof thisderivative inheptanewasstored at -20#{176}CuntIl analysis two days laterwiththe MEDdetector

CLINICALCHEMISTRY, Vol. 37, No. 12, 1991 2065

2). Because the actual volume of derivatized THC in-jected into the GC/MS, 2 1zL, is only one-tenth of thetotal volume (20 tL) of the heptane reconstitution fluid,the limit of detection in terms of the injected quantity ofTHC is actually 8 pg. However, becausethe recovery ofTHC is about 70% (see below), the limit of detection isactually somewhat lower. Figure 4 shows an example ofselected-ion monitoring tracing for the mIx 410 and 413molecular ions of TFA-THC and the TFA derivative oftrideuterated THC, respectively, for a plasma sampleobtained 2 h after completion of smoking a marijuanacigarette.

Although we did not evaluate the limit of detection forTHC-COOH concentrations in plasma as formally as wedid for THC, primarily because the required concentra-tions for pharmacokinetic studies are not nearly as lowas for THC, our general experience is that we canreliably detect concentrations of THC-COOH in plasmaof 100 ng/L or higher.

Reproducibility, accuracy, and analytical recovery.The interday reproducibility and the accuracy obtainedwith the HED detector are summarized in Table 2 for

Table 1. SensitIvity and Limit of Quantification for THCAnalysis (NegatIve ChemIcal Ionlzation)a

Sensltlvfty, Plasma, Extract,eras cts/pgc ng/L pg/2pL

Standard 4.8(4.4)d 50() 50HED (8 kV) 39 (1.9) 80 8

‘THC measuredas theTFA derivative,as described in the textThe potentialapplied to the contlnuotjs-dynodeelectronmultiplier was 1.8

kV forthe standarddetector system.The potential applied to the continuous-dynode electron multiplierwas 1.0 kV for the HED detector.

C Theslope of theabundance(areacount) of the m/z 410 molecular ion ofthe TFA derivativeof THC vs the InjectedquantityofTMC(as TFA-THC).Eachresult Is the average slope obtained from two analytical runs of ThC-supplemented plasma samples (80, 100, 200, 300, 500, 1000, 5000, and10000 ng/L for the HEDdetector and500, 1000, 5000, 10000, 15000, and20000 ng/L for the standarddetector) that were extracted, derlvatlzed withTFA, and analyzed by electron-capturenegative chemical-ionizationGC/MSas described in the text

dMean (SD).The limit of quantification for mc In plasma is the pIasm concentration

requiredto give a signal-to-noiseratio of 3-to-i for the mlz 410 molecular-ionpeakof THC-TFApreparedfromthat plasma.The limit of quantificationofTHCper 2 L of injected extract is one-tenth of that In 1 mL of extracted plasmabecause the total volume of the reconstitutedextractwas20 iL.

Table 2. Precision and Accuracy of THC andThC-COOH Measurement

Concn, Mg/L Precision

Mean SD, CV, Deviation,Added P9/’

mc0.08 0.09 0.007 7.8 12.50.4 0.40 0.05 12.5 01.0 0.99 0.12 12.0 1.07.0 7.35 0.78 10.6 5.0

THC-COOH0.4 0.43 0.08 18.6 7.51.0 0.99 0.12 12.1 1.07.0 6.94 0.37 5.3 0.86

20.0 19.93 1.46 7.3 0.3535.0 35.22 1.53 4.3 0.63

a Interdayvalues (n = 16forTHC, 11 forThC-COOH) except for the 0.08pg/L.THC specimens, which were Intradayvalues (n = 7).

b Deviation of the mean measured concentrationfrom the addedconcentration.

both THC and THC-COOH. Over the concentrationranges investigated, mean measured concentrationswere uniformly close to the respective added amounts.The calibration curves for THC and THC-COOH shownin Figures 1 and 2 demonstrate excellent linearity from0.08 to 10 ,ug/L and from 0.1 to 25 zg/L, respectively.Analytical recovery of THC from plasma was evaluatedby comparing the average abundance of the mIx 410molecular ion of TFA-THC for three replicate analysesof plasma samples supplemented with THC at 0.5 or 2,ag/Lwith the corresponding abundance values obtainedfor three replicate analyses of THC at 0.5 and 2 .tgtL inethanol. An average recovery of 70% was obtained atboth concentrations, comparable with the value re-

,3.r

#{163}4?

I,”

-a

-I

0z000I.-

Fig. 4. Selected-ion monitoringtraces of the m/z 410 and 413molecular Ions of TFA-THC and TFA-trldeuteratedTHC, respec-tively,preparedfroma plasmasampleof a subject 2 h aftersmokln#{231}one marijuanacigarette containing 19.6 mg of ThCThe THC concentration was1.4 g/L measuredwith the HED detector;theconcentrationof trideuteratedmc Internalstandardwas I MO/I-

0 60 120 180 240 300 360 420 480 540

TIME (MIN)

- 10

S

z00C,01xI-

0 ...........

IA-A A A A A

0- o______ -

-

0 60 120 180 240 300 380 420 480 540

2066 CLINICAL CHEMISTRY, Vol. 37, No. 12, 1991

ported by Foltz et al., using this extraction and deriva-tization procedure (7).

No co-eluting interfering peaks at mIx 410,413, 454,or 457 were found in 50 plasma specimens obtained fromcontrol subjects whose urine tested negative for canna-binoids with the EMIT-20 (Syva, Palo Alto, CA) testprocedure. Five of these specimens were obtained fromthe study subjects described here just before theysmoked a marijuana cigarette.

Comparison of THC concentrations measured withand without the HED detector. Serial plasma concentra-tions of THC and THC-COOH in five individuals aftersmoking one marijuana cigarette containing 19.6 mg ofTHC are displayed in Figure 5. The data show the rapiddisappearance of THC from plasma during the first fewhours after smoking that corresponds to the distributionof the drug into tissues, followed by a much slower rateof disappearance, which reflects the elimination phase(9). On the other hand, the THC-COOH concentrationschange at a much slower rate; this probably reflects abalance between its production from THC, as the THC isbeing released from tissue depots and metabolized bythe liver, and its clearance by kidneys. Table 3 summa-

rizes the plasma concentrations of THC measured withand without the lIED detector for four of the studysubjects (insufficient volume of plasma specimen wasavailable for the other subject to analyze by both proce-dures). The data show that for the 8 h of the evaluationperiod the plasma THC concentration reached the limitof quantification of 500 ng/L for the standard GC/MSsystem without the HED detector after 30 mm and 8 hfor four of the study subjects. Four specimens fromsubjects 1, 4, and 5 had THC concentrations that wereundetectable with the standard detection system. How-ever, their THC concentrations were readily measura-ble by the HED detector (190, 250, 420, and 470 ngfL).Subject 3 had unusually low concentrations of both THCand THC-COOH, which probably reflected poor absorp-tion in this one subject. In this individual, THC concen-trations reached the limit of quantification of 80 ng/L by

TIME (NIN)

Fig. 5. (Top) PlasmaThC concentrationand (bottom) plasmaTHC-COOHconcentrationvs time after smoidng one manjuanacIgarette (19.6 mg of THC)

6 h after smoking. Studies of longer-term pharmacoki-netics of THC are under way in our laboratory. Thesestudies show that the lIED limit of detection of 80 ngILis not reached for as long as 48 h after smoking amarijuana cigarette containing 19.6 mg of THC,whereas the standard detection system limit of detec-tion concentration of 500 ng/L is reached 4 to 6 h aftercompletion of smoking. Student’s paired t-test analysisof the 19 pairs of THC concentrations that were 500ng/L or greater showed that THC concentrations ob-tained with the HED detection system were not statis-tically different from those obtained with the standarddetection system (t = 1.36).

Discussion

The study of THC pharmacokinetics requires sensi-tive and specific analytical methodology capable of mea-suring very low plasma concentrations, from 1000 to atleast 100 ng/L. These are the concentrations typicallyreached in the elimination pharmacokinetic phasewithin hours after ingestion of the drug (7,9). Electron-capture gas-liquid chromatography (10), metastable-ion monitoring in a double-focusing mass spectrometer(11), electron ionization and electron-capture negativechemical-ionization GCIMS (7, 12), electrochemical de-tection-HPLC (13), and radioimmunoassay (14) meth-ods have been described for measuring THC in biologi-

Table 3. ThC Concentrations Measured with andwithout the HED Detector

urns utter mc conan, Mg/I..moatng,

mm Without HED With HED

5 108 8815 37 3230 18 19

120 4.6 3.6240 0.78 0.82360 0.50 0.54480 <0.50 0.47

3 5 2.3 2.315 0.98 1.230 0.58 0.65

120 <0.5 0.17240 <0.5 <0.08360 <0.5 <0.08480 <0.5 <0.08

4 51530

120240360480

20.811.09.71.70.53

<0.5<0.5

17.58.47.51.4ND0.250.19

5 51530

120240360480

38.418.610.41.60.740.50

<0.50

41.519.210.81.80.910.460.42

in measuring TFA-THC (3). Using a Hewlett-Packard5985B single-quadrupole GC/MS system with its stan-dard detector operated in the electron-capture negativechemical-ionization mode, we attained a limit of detec-tion of 500 ng/L of plasma when we used 1 mL ofspecimen. The reason for the 2.5-fold higher limit ofdetection is not clear to us but could be due to inherentdifferences in the design of the mass spectrometer sys-tems or possibly to differences in definition of the limit ofdetection. In the present work the limit of detection wasdefined as the lowest concentration giving a signal-to-noise ratio of 3 or greater. In the comparison study (7)the investigators defined 200 ng/L as their limit forreliable measurement of THC but did not state the basisfor this.

In the studies described here, we improved the limit ofdetection 6.2-fold (from 500 to 80 ng/L) by replacing thestandard detector of our CC/MS with a lIED detectorwhile maintaining all other CC/MS settings the sameexcept for the voltage on the continuous-dynode electronmultiplier, which was lower. A modified version of thelIED is available for use in benchtop CC/MS quadrupolesystems, the most widely used CC/MS systems in clini-cal toxicology laboratories. The added sensitivity could,for example, make possible the use of full-spectrumanalysis for drugs of abuse for more definitive identifi-cation at the required limits of detection, thereby pro-viding greater reliability in substance identificationthan the characterization of the relative abundances ofonly three fragment ions, the most widely used ap-proach in forensic urine drug-testing programs (15).

Figure 6 is a diagram of the HED detector. In theelectron-capture negative chemical-ionization mode, ahigh voltage of +8 kV is applied to a stainless steelconversion dynode. Although the schematic diagram ofthe standard detection system is similar, the primarydifference lies in the dynode (called the x-ray deflector inthe standard detection system). In the standard detec-tor, the voltage applied to the x-ray deflector is +2.5 kV;higher voltages cannot be applied. It is the highervoltage applied to the dynode of the HED detector that

Iris

IDEM (- -1KV)

I <JIignaI

IonsPie

I F -

Dynode (+8KV)

Fig. 6. DIagramof the High EnergyDynodedetectorsystemoper-atingin thenegative-iondetectionmodeCOEM,continuous-dynodeelectnn multiplier

cal specimens. Some of these methods, e.g., electrochem-ical detection-HPLC (13) and the radioimmuno-assays (14), have limited sensitivities, about 5 and 2.5g/L, respectively, that preclude their use in the evalu-ation of post-distribution pharmacokinetics of THC.Furthermore, the radioimmunoassays lack specificityfor the parent drug (14). On the other hand, metastable-ion detection CC/MS reportedly affords the greatestsensitivity of all reported methods, with a limit ofdetection of 10 ng/L in a 5-mL plasma specimen (11).Such an analytical system is very costly, however, andis therefore restricted to a few research laboratories.Ohlsaon et al. (12) developed an electron-ionizationCC/MS method with a reported detection limit of 100ng/L, but this required 3-mL plasma specimens. Foltz etal. (7) described the extraction and derivatization pro-cedure we applied in the present work. Using a Finni-gan/MAT 4023 single-quadrupole CC/MS system in theelectron-capture negative chemical-ionization mode,they reported a limit of detection of 200 ng/L in 1 mL ofplasma specimen (7). Electron-capture negative chemi-cal ionization reportedly affords greater sensitivity thaneither electron ionization or positive chemical ionization

CLINICAL CHEMISTRY, Vol. 37, No. 12, 1991 2067

2068 CLINICAL CHEMISTRY, Vol. 37, No. 12, 1991

makes possible the greater sensitivity of this system incomparison with the standard detector (16).

For organic molecules such as TFA-THC, collision ofthe negatively charged molecular ions with the posi-tively charged dynode has been proposed to cause for-mation of positive secondary ions by as many as threemechanisms, including sputtering, fragmentation, andcharge stripping (17). For molecular ions >200 amu,fragmentation is thought to be the primary mechanismresponsible for negative to positive conversion at thesurface of the dynode (17). According to our currentunderstanding of this process,negatively charged mo-lecular ions collide with the positively charged dynodesurface, where they undergo extensive fragmentation.The fragments consist of neutral species,negative ions,and positive ions. Only the positive ions will be repelledaway from the dynode surface toward the negativelycharged continuous-dynode electron multiplier. For ev-ery negative ion making an initial collision, the numberof positive ions leaving the dynode surface increaseswith increasing positive voltage applied to the dynode,presumably because this further increases the velocityof the negatively charged molecular ions, which in turnincreases the degree of fragmentation at the dyziodesurface (17). The mass of the colliding charged molecu-lar ion is an important contributing factor to determin-ing the degree of increase in the conversion dynodesignal. The relative increase in velocity of a chargedmolecular ion produced by a given voltage on the con-version dynode is greater in proportion to the mass ofthe colliding species (18).

For applications requiring the analysis of positivelycharged fragment ions, produced (e.g.) by electron ion-ization or positive-ion chemical ionization, a high neg-ative potential is applied to the lIED and a positivepotential is applied to the electron multiplier. In thesecases, negatively charged fragments and probably elec-trons released by charge stripping upon impact of posi-tively charged ions with the HED surface are repelledaway from the dynode toward the positively chargedelectron multiplier (19). Preliminary studies have alsoindicated that increases in detection sensitivities can beattained by using other sample introduction modes suchas particle beam and thermospray (18).

Partial support for this work from NIH grants 3 MOl RR0004Oand 5P50-DC00214 is gratefully acknowledged.

References1.Faull KF, Pascoe N, Mahoney .J, Perel J. Investigation ofGCIMS detection limits using a high sensitivity detector on aquadrupole mass spectrometer [Abstract]. Proc.,38th ASMS Conf.on Mass Spectrometry and Allied Topics, 1990:914-5.

2. Agurell S, Haildin M, Lindgren J-E, et al. Pharmacokineticsand metabolism of delta-1-tetrahydrocannabinoland othercan-nabinoids with emphasis on man. Pharmacol Rev 1986;38:21-41.3. Foltz EL. Analysis of cannabinoidsin physiological specimensby gas chromatography/mass spectrometry. 1n Baselt RC, edAdvances in analytical toxicology, Vol. 1. Foster City, CA: Bio-medicalPublications,1984,Chap. 6.4. Harvey DJ, Leuschner JTA, Wing DR.,Paton WDM. Pharma-cokinetics of delta-9-tetrahydrocannabinol in rabbits and micewith measurements to the low picogram range by combined gaschromatography/mass spectrometry using metastable ion detec-tion. In: Agurell S, Dewey WL, Willette RE, eds. The cannabinoids:chemical, pharmacologic, and therapeutic aspects.New York:Academic Press, 1984:291-308.5 HeishmanSi, Huestis MA, Henningfleld JE, Cone EJ. Acuteand residual effects of marijuana: profiles of plasma THC levels,physiological, subjective, and performance measures. PharmacolBiochem Behavior 1990;37:561-5.6. Doblin RE, Kleiman MAR. Marijuana as an antiemetic medi-cine: a survey of oncologista’experience and attitudes. J Cliii Oncol1991;9:1314-9.7. Foltz EL, McGinnis KM, Chinn DM. Quantitative measure-ment of delta-9-tetrahydrocannabunoland two major metabolitesin physiological specimens using capillary column gas chromatog-raphy negative ion chemical ionization mass spectrometry.Biomed Mass Spectrom 1983;10:316-23.8. Mass spectrometer adjusting procedure for chemical ionization(positive and negative). Palo Alto, Ck Hewlett-Packard, 1981:1-22.

9. Agurell S, Lindgren J-E, Ohlsson A, et al. Recent studies onthepharmacokinetics of delta-1-tetrahydrocannabinol in man. Op.cit.(ref. 7):165-83.10. Fenimore DC, Davis CM, Horn All. Determination of delta-9-tetrahydrocannabinol in human bloodserum by electron capturegas chromatography. In: Willette RE, ed. Cannabinoids in hu-mans. NIDA Res Monog 7. Washington, DC: U.S. Govt PrintingOffice, 1976:42-7.11 Harvey DJ, Leuschner JTA, Paton WTM. Measurement ofdelta-1-tetrahydrocannabinol in plasma to the lowpicogram rangeby gas chromatography-mass spectrometry using metastable iondetection. J Chromatogr1980;202:83-92.12 Ohisson A, Lindgren JE, Wahlen A, et al. Singledose kineticsof deuterium labelled delta-1-tetrahydrocannabinol in heavy andlight users. Biomed Mass Spectrom1982;9:6-.1O.13. Shepard EM, Milne GM. The use of high-pressure liquidchromatography with electrocheimcaldetection for the assay ofnantradol and its application to delta-9-tetrahydrocannabinol. In:Hawks EL, ed. The analysis of cannabinoidsin biological fluids.NIDA Res Monog 42. Washington, DC: U.S. Govt Printing Office,1982:7-17.14 Cook CE, Seltzman HH, Schindler CR, et al. Radioimmunoas-says for cannabinoids.Ibid., pp. 19-32.15. DeCresceR, Mazura A, Lifshitz M, Tilson J. Drug testing inthe workplace. Chicago:Am SocClin Pathol Press, 1989:94.16. Perel J. Advancesin achieving greater sensitivity and highermass detection in mass spectrometry. Am Lab/NewsEd August1989:18.17. Stafford GC. Instrumental aspects of positive andnegativeionchemical ionization mass spectrometry. Environ Health Perspect1980;36:85-96.18. Perel J, Mahoney JF, Speiser RC. Mechanisms for improvedhigh mass ion detection [Abstract]. Op.cit. (ref. 1):510-1.19 Perel J, Mahoney JF, Lee TD, LegesseK. High sensitivity iondetectorparameterstudy [Abstract]. Proc., 37th ASMS Conf. onMass Spectrometryand Allied Topics, 1989:1075-6.