CANNABINOID ASSAYSIN HUMANSEditor

ROBERT E. WILLETTE, PH.D.Division of ResearchNational Institute on Drug Abuse

May 1976

NIDA Research Monograph 7National Institute on Drug Abuse

11400 Rockville Pike

Rockville, Maryland 20852

U.S. Department of Health, Education, and Welfare

Public Health Service

Alcohol, Drug Abuse, and Mental Health Administration

The NIDA Research Monograph series is prepared by theResearch Division of the National Institute on Drug Abuse. Itsprimary objective is to provide critical reviews of research problemareas and techniques, the content of state-of-the-art conferences,integrative research reviews and significant original research. Itsdual publication emphasis is rapid and targeted dissemination tothe scientific and professional comunity.

EDITORIAL ADVISORY BOARD

Avram Goldstein, M.D.

Jerome Jaffe, M.D.

Reese T Jones, M.D.

William McGlothlin, Ph.D.

Jack Mendelson, M.D.

Helen Nowlis, Ph.D.

Lee Robins, Ph.D.

Addiction Research FoundationPalo Alto, California

College of Physicians and SurgeonsColumbia University, New York

Langley Porter NeuropsychiatricInstituteUniversity of CaliforniaSan Francisco, California

Department of Psychology, UCLALos Angeles, California

Alcohol and Drug Abuse ResearchCenterHarvard Medical SchoolMcLean HospitalBelmont, Massachusetts

Office of Drug Education, DHEWWashington, D.C.

Washington University School ofMedicineSt. Louis, Missouri

NIDA RESEARCH MONOGRAPH series

Robert DuPont, M.D. DIRECTOR, NIDA

William Pollin, M.D. DIRECTOR, NIDA RESEARCH DIVISION

Robert C. Petersen, Ph.D. EDITOR-IN-CHIEF

Eunice L. Corfman, M.A. MANAGING EDITOR

Rockwall Building 11400 Rockville Pike Rockville, Maryland, 20852

CANNABINOID ASSAYS

IN HUMANS

Thanks are due the people of Macro Systems, Inc.,who capably organized and smoothly coordinatedthe conference, held February 24th and 25th, 1976,under NIDA contract #271-75-1139, from which thepapers in this monograph are derived.

DHEW publication number (ADM) 76-339

Library of Congress catalog card number 76-15843

This document is for sale by National Technical Information ServiceSpringfield Va. 22161

Stock Order #PB 251 905; Papercopy: $6.00; Microfiche: $2.25

i v

FOREWORDThis monograph describes ways of determiningthe levels of cannabinoids in the human bodyafter smoking marihuana. Investigations of anumber of serious social and health problemshave had to wait for the refinements of thesetechniques.

Chief among these concerns is the effect ofmarihuana smoking on driving. That investi-gation along with others can now be undertakenwith an expectation of far more precise find-ings than was formerly possible. We now havethe means not only of detecting cannabinoids,but also of determining in what quantity theyare present. Thus, we can begin to establishspecific correlations between cannabinoid levelsand driving impairment. This information isnecessary in order to build marihuana into thehighway safety campaign now largely restrictedto alcohol.

These assay techniques will be useful toolsfor a range of other problems including simplescreening procedures, epidemiological studies,forensic toxicology, as well as for more funda-mental pharmacokinetic and pharmacological re-search. The developments in method and instru-mentation described here occur at a crucialjuncture for drug abuse research, as we striveto assess the impact of marihuana on our culture.

Robert L. DuPont, M.D.DirectorNational Institute on Drug Abuse

v

PREFACEFor the past several years, there has been anincreasing demand for qualitative and quantitativeassays for identifying and measuring the consti-tuents of marihuana in the human body. This demandis prompted by the need for such assays in severalresearch investigations that are attempting to im-prove our understanding of how this complex drugaffects the body. In addition to these morefundamental issues, there is a growing feeling thatsuitable analytical methods will be required fordetermining the presence of cannabinoids in driverssuspected of being under the influence of the drug.

We have now reached a stage in the search for anddevelopment of such methods that many of them cannow be employed in a routine manner. As we gain inexperience and confidence with these methods, theirvalidity will become increasingly better establishedand their applicability to critical decisions moreacceptable. It will be apparent upon reviewing theprocedures described herein that some were designedfor or are by their complexity only suited for re-search purposes or in validating other methods.Others described are more amenable to routinescreening or survey applications.

The road to acceptable methods has been long andarduous. Early attempts continued to suffer fromlack of adequate sensitivity. It was eventuallylearned, as the studies on the composition ofmarihuana and the metabolism of its constituentsprogressed, that the problem of detecting anyspecific cannabinoid in the body after use wouldbe an extremely difficult task. The primary activeconstituent, delta-9-tetrahydrocannabinol (THC) israpidly distributed and metabolized in the body,making its quantification a major challenge. It isnow very gratifying to be able to present a collec-tion of manuscripts that delineate the tremendousprogress that has been made over the past few years.

v i i

The methods described are grouped together underthree major headings. The first are based on theincreasingly used immunoassay techniques. In gen-eral, immunoassays offer speed and sensitivity andare very amenable to the screening of large numbersof samples. They often suffer from lack of speci-ficity, but this is often acceptable if they crossreact only with metabolites of the target drug andno other drug. The four methods described are atvarious stages of development and refinement, andsome are being employed in various research studies.

Methods of the second group are based on the oldertechnology of chromatography, but are applied inrigorous and innovative ways to provide the degreeof sensitivity required to measure the low levelsof drugs and metabolites. Two different approacheswere taken to reduce the background interference.Using a conventional gas chromatograph, Dr. Garrettemployed a high pressure liquid chromatograph (HPLC)to "clean-up" the sample. The dual-column instru-ment designed by Dr. Fenimore reduces the amount ofeffort required for sample preparation. As thetechnology in HPLC progresses, this method is be-coming extremely popular, offering the ability toseparate difficult mixtures at low temperatures.The effort described here is moving forward andparallels similar efforts that are being carriedout by Dr. Valentine, in an outgrowth of his pro-ject's mass spectroscopic method.

The last group of methods all employ the massspectrometer as the detector for identifying andquantifying the cannabinoids. Rapid advances inthe development of mass spectroscopy have made itthe method of choice in terms of sensitivity andspecificity. Because of its present size, cost,and complexity, it is not ideal for routine appli-cation to routine screening. It is, however, beingused for the routine validation of other methodsand to confirm the presence of cannabinoids insamples found positive by less specific screeningmethods, like the immunoassays. The six papersincluded represent some of the most outstandingwork done in the field of mass spectroscopicanalysis. Since Dr. Agurell first published hismethod in 1973, tremendous strides have been madeby him and the others included in this monograph.

v i i i

We now feel very confident in our ability to get onwith many of the critical studies that have awaitedthese methods.

This collection of papers does not signal the endof the road in the development of suitable methodsfor quantifying cannabinoids in the body. As ourunderstanding of the effects of marihuana pro-gresses, so must the methods used in studying it.We now have methods to study the effects of mari-huana on driving, and if the evidence indicatesthat it poses a significant hazard, then a simple-perhaps roadside- test may be required. Otherexamples could be cited, but it is sufficient toend with the recognition of a notable achievementin this difficult area of research and a sense ofsatisfaction that our perseverance is paying off.

Robert E. Willette, Ph.D.Uivision of ResearchNational Institute on Drug Abuse

April 1976

i x

CONTENTS

FOREWORD

PREFACE

QUANTITATION OF CANNABINOIDS IN BIOLOGICAL FLUIDS BY RADIOIMMUNOASSAY 1Arleen R. Chase, Paul R. Kelley, Alison Taunton-Rigby, Reese T. Jones, Theresa Harwood

SEPARATE RADIOIMMUNE MEASUREMENTS OF BODY FLUID ∆9-THC AND 11-NOR-9-CARBOXY-∆ 9-THCStanley J. Gross, M.D. and James R. Soares, Ph.D.

10

RADIOIMMUNOASSAY OF ∆ 9-TETRAHYDROCANNABINOLClarence E. Cook, Ph.D., Mary L. Hawes, B.A., Ellen W. Amerson, B.A.,Colin G. Pitt, Ph.D., and David Williams, B.A.

15

DETERMINATION OF THC AND ITS METABOLITES BY EMIT® HOMOGENEOUS ENZYME IOIMMUNOASSAY:A SUMMARY REPORT

G.L. Rowley, Ph.D., T.A. Armstrong, C.P. CrowL, W.M. Eimstad, W.M. Hu, Ph.D.,J.K. Kam, R. Rodgers, Ph.D., R.C. Ronald, Ph.D., K.E. Rubenstein, Ph.D.,B.G. SkeLdon, and E.F. Ullman, Ph.D.

SEPARATION AND SENSITIVE ANALYSIS OF TETRAHYDROCANNABINOL IN BIOLOGICAL FLUIDSBY HPLC AND GLC

Edward R. Garrett and C. Anthony Hurt

DETERMINATION OF ∆ 9-TETRAHYDROCANNABINOL IN HUMAN BLOOD SERUM BY ELECTRON CAPTUREGAS CHROMATOGRAP'HY

David C. Fenimore, Ph.D., Chester M. Davis, Ph.D., and Alec H. Horn

DETECTION AND QUANTIFICATION OF TETRAHYDROCANNABINOL IN IN BLOOD PLASMAAgneta Ohlsson, Jan-Erik Lindgren, Kurt Leander, Stig Agurell

A METHOD FOR THE IDENTIFICATION OF ACID METABOLITES OF TETRAHYDROCANNABINOL (THC)BY MASS FRAGMENTOGRAPHY

Marianne Nordqvist, Jan-Erik Lindgren, Stig Agurell

QUANTITATION OF CANNABINOIDS IN BIOLOGICAL SPECIMENS USING PROBABILITY BASED MATCHINGGAS CHROMATOGRAP'HY/MASS SPECTROMlETRY

Donald E. Green, Ph.D.

QUANTITATION OF ∆9-TETRAHYDROCANNABINOL IN BODY FLUIDS BY GAS CHROMATOGRAPHY/CHEMICAL IONIZATION-MASS SPECTROMETRY

Ruthanne Detrick and Rodger L. FoItz

HPLC-MS DETERMINATION OF ∆ 9-TETROHYDROCANNABINOL IN HUMAN BODY SAMPLESJimmie L. Valentine, Ph.D., Paul J. Bryant, Ph.D., Paul L. Gutshall, M.S.,Owen H.M. Gan, B.S., Everett D. Thompson, B.S., Hsien Chi Niu, Ph.D.

ANALYTICAL METHODS FOR THE DETERMINATION OF CANNABINOIDS IN BIOLOGICAL MATERIALSM.E. Wall, Ph.D., T.M. Harvey, Ph.D., J.T. Bursey, Ph.D., D.R. Brine, B.S.,and D. Rosenthal, Ph.D.

28

33

42

48

64

70

88

96

107

LIST OF CONTRIBUTORS 118

LIST OF MONOGRAPHSx i 120

QUANTITATION OF CANNABINOIDSIN BIOLOGICAL FLUIDSBY RADIOIMMUNOASSAY

Arleen R. Chase, Paul R. Kelley, Alison Taunton-RigbyCollaborative Research, Inc.Waltham, Mass.

Reese T. JonesLangley Porter Neuropsychiatric InstituteUniversity of California, San Francisco

Theresa HarwoodDrug Enforcement AdministrationWashington, D. C.

INTRODUCTION

There is an expanding need to assay tetra-hydrocannabinol ( 9THC) and its metabo-lites in biological fluids, particularlyin the area of medical research. Methodsused for the quantitation of cannabinoidsinclude gas chromatography coupled withmass spectroscopy (GC/MS) , liquid chrom-atography and radioimmunoassay. Allmethods are of value, especially when usedto confirm the results of another pro-cedure, so that the method of choice de-pends on the particular application.Radioimnunoassay is a common clinicalmethod which is highly specific, sensitiveand particularly well suited to theroutine, simultaneous analysis of multiplesamples, often without prior purification.The equipment required is routinely usedin university, research, industrial, andclinical laboratories and hospitals.Results can be obtained rapidly and areeasy to interpret.

THE PRINCIPLES OF RADIOIMMMUNOASSAY

Radioimmmunoassay depends on the affinityof a biological molecule, the antibody, for

the antigen in question, in this case9THC. Sensitivity is achieved through the

use of a radioactive tracer molecule ofhigh specific activity, called the labelledantigen. The extent to which the unla-belled antigen ( 9THC) competes with theradioactive antigen for a limited number ofreceptor sites on the antibody serves asthe basis for quantitation in the radio-immunoassay. The assays are simple toperform. Mixtures containing the labelledantigen, the antibody and the sample areincubated, free labelled antigen isseparated from antibody-bound labelledantigen and the exent of binding is deter-mined by counting the disintegrations perminute of the radiolabel. Unknown samplesare quantitated accurately by comparisonwith the binding levels achieved with knownsamples.

Antibodies generally show a remarkableability to bind selectively the antigenthat stimulated their production. Thisspecificity is comparable to that of anenzyme for its substrate. The ability ofan antibody to discriminate between the

1

antigen and the myriad of other compoundsof widely diverse structure, which arefound in biological fluids, is of funda-mental importance in its use as an analyti-cal tool. Macromolecules, such as proteinsnucleic acids and polysaccharides, usuallyelicit an immune response when injecteddirectly into an animal. However, lowmolecular weight compounds, such as 9THC,cannot elicit an immune response unlessthey are bound covalently to an antigenicmacromolecule such as a protein or poly-peptide. The development of a radioimmuno-assay for a molecule, such as 9THC, in-volves (i) the synthesis of suitable cov-alent conjugates for immunization, (ii) theproduction of antisera, (iii) the prepara-tion of a radioactive antigen, and (iv)the establishment of the assay based on theantigen-antibody reaction.

SYNTHESIS OF COVALENT CONJUGATESFOR IMMUNIZATION

Molecules containing an amino or carboxyfunction can be coupled directly to theamino or carboxy groups of amino acid resi-dues in proteins or polypeptides by forma-tion of amide bonds. Since the specificityof an antibody is usually directed towardthose structures on the hapten that aredistal to the linkage group, the haptenshould be coupled to the carrier so thatcharacteristic functional groups are ex-posed to the antibody synthesizing cells.

In the case of 9THC several derivativescontaining carboxy groups have been synthe-sized. and coupled to macromolecules. Theuse of a hemisuccinate ester of 9THC hasbeen reported (Teale et al, 1975) as wellas an azobenzoic acid-derivative (Grant etal 1972 and Gross et al, 1974) . We haveused O-carboxymethyl-THC prepared by re-action of 9THC with iodoethylacetatefollowed by basic hydrolysis as shown inFigure 1.

Figure 1

Synthesis of O-Carboxymethyl-THC

2

PRODUCTION OF ANTISERUM

O-carboxymethyl-THC was coupled to bovineserum albumin (BSA) using two differentdehydrating agents. In the first prepara-tion the water soluble carbodiimide,1-ethyl-3-(3-dimethyl aminopropyl) -carbodiimide (EDC), was used and in thesecond preparation, coupling was effectedby Woodward’s reagent, N-ethyl-5-phenylisooxazolium-3’-sulphonate. These reactionsare surmmarized in Figure 2. UncoupledCBM-THC and other small molecules were re-moved from the two preparations ofO-carboxymethyl-THC-BSA (CBM-THC-BSA) bydialysis and ethanol precipitation. Thenumber of THC residues incorporated wasestimated by spiking the preparations with1 4 C- 9THC.

The two preparations of the conjugate,CBM-THC-BSA, were both used to immunizerabbits. While the response varied fromanimal to animal? antisera were generatedto both preparations. The presence ofantibody was demonstrated by the fact thatthe antisera would bind radiolabelled

9THC and, moreover, this binding couldbe inhibited by unlabelled 9THC.

The response of two of the rabbits issummarized in Figure 3. These plots showthe titer of antiserum required to giveapproximately 50% binding of 3H- 9THC.It can be seen that antisera were generatedto both preparations. Most of the datadescribed in the rest of this paper wereobtained using bleeding H of rabbit 56.

Figure 2

Synthesis of O-Carboxymethyl-THC-Bovine Serum Albumin

3

Figure 3

Titer of Antiserum Giving Approxomately 50% Binding of3 H - 9THC Following Immunization

Note: Rabbit 56 was immunized with CBM-THC-BSA prepared using EDC.

Rabbit 67 was injected with CBM-THC-BSA coupled usingWoodward's Reagent.

RADIOLABELLED ANTIGEN, BINDINGAND INHIBITION STUDIES

High specific activity, radiolabelled com-pounds are used to develop sensitive radio-immunoassays. We have used both 3H and14C- 9THC in our work.1 The 14C- 9THC

9THC, is shown in Figure 4. The 50%inhibition point occurred at 330 nanograms

had a specific activity of 0.07 µCi/µg. At o f 9THC. As expected, this system wasa final titer of 1:3 the antibody gave a too insensitive to be used as a workingmaximum binding of this 14C- 9THC of radioimmunoassay.

4

19.3%. A standard curve, prepared byinhibiting this binding with unlabelled

Figure 4Inhibition of the Binding of 14C- 9 THC

and Antibody of 9 THC

The insolubility of 9THC in aqueous sys-tems is well established. Correspondingly,to provide an internal check on each stand-ard solution, 14C- 9THC is used as astandard. The exact number of nanograms of

IHC in each standard solution is thenverified by counting an aliquot in a liquidscintillation in the 14C channel. The 14Clabel does not interfere with the countingof the 3H label in the 3H-channel.

The assay conditions described above wereused to obtain the standard curve shown inFigure 5. The system is capable of assay-ing as little as 0.25 nanograms of

9THC reproducibly.

CROSS REACTIVITY STUDIES

The cross reactivity of various cannabin-oids, drugs and other compounds was estab-lished for the assay system. The data issummarized in Table 1, and shows that theantiserum apparently reacts exclusively withcannabinoids. None of the non-cannabinoiddrugs, hormones or other compounds cross

Figure 5Inhibition of the Binding of 3H- 9 THC

and Antibody by 9 THC

Practical sensitivity could be achieved,however, using 3H-labelled 9THC with aspecific activity of 41 µCi/µg. Bindinglevels of 30-50% were routinely achievedat a final titer of 1:750. The binding wasinhibited by nanogram levels of unlabelled

9THC as shown in Figure 5. The 50%inhibition point occurred at 0.7 nanograms

9THC. This system was used as the basisof a working radioimmunoassay.

THE TRITIUM BASED RADIOMMUNOASSAY SYSTEM

A working radioimnunoassay has been develop-ed based on the use of 3H- 9THC and anti-serum generated to CBM-THC-BSA. The assaysystem which is described below is availableto interested investigators. The assay iscarried out in a buffer system of 0.1Mphosphate pH 7.0, 0.1% Triton X-405, 0.2%sheep gamma globulin, fraction II. Tubescontaining antibody, 3H- 9THC and eitherthe standard or unknown THC sample areincubated for four hours at 4°C. Antibodybound and free labelled THC are separatedby dextran coated charcoal. The percent ofthe radiolabel bound by the antibody isdetermined by counting samples of thesupernatant in a liquid scintillationcounter.

5

Table 1

Gross Reactivity of Certain Cannabinoids and Other Compounds With

-9THC Antiserum

Cannabinoids50% InhibitionIn Nanograms

9-THC8-THC

11-OH 9-THC11-OH 8-THC

OH 9-THC11-Nor 9-THC-9-COOH11-Nor CBN-9-COOHCannabinol (CBN)Cannabidiol (CBD)Cannabidiolic acid (CBD acid)CannabicyclolCannabichromene

0.70.70.40.72.0

0.251.53.5

100408

4 0

Other Level of Detectioncompounds In Nanograms

Caffeine“Aldactone" or SpironolactoneLSD tartrateSecobarbitol (Technam spiked urine)Gibberellic acidNicotineMorphine SulfateCocaine (Technam spiked urine) Drugs of AbuseMethadone HClPhenobarbitol, sodiumTetracyclineD,L-EphedrineSalicylic AcidCarvone(+) Limonene terpenes

- IhujoneAldosterone-21-hemisuccinateProgesteroneEstradiol steroids

-1 cortisoneTestosteroneCholesterol hydrogen succinate

* Highest level available

>100,000>5,000

>32,000>100*

>100,000>1,000,000

>100,000>60*

>100,000>10,000

>100,000>10,000

>100,000>95,000>80,000>90,000>10,000>10,000>10,000>10,000>10,000>20,000

reacted to any significant extent. Theresults obtained with the various cannabin-

The specificity of the antiserum for can-nabinoids without absolute specificity for

oids show that the antiserum reacts aswell with 9THC as with some of its metabo-

THC is advantageous since native THC is notexcreted in the urine and an assay capable

lites. The antibody recognizes 9THC,8THC and several hydroxy metabolites. It

of detecting THC only would be of littlevalue. The assay system described here can

can differ ntiate to a limited extentbetween 9THC and cannabinol, cannabidiol,

detect 9THC and its hydroxy metabolites,and therefore can be used to detect canna-

cannabidiolic acid and cannabichromene. bis use.6

APPLICATION TO BIOLOGICAL SAMPLES known amounts of 9THC and assayed. Theresults are shown in Table 2, and it can beconcluded that (i) urine can be assayeddirectly with no pre-treatment required,(ii) urine does not interfere with thequantitation of 9THC, and (iii) 20microliter samples can be quantitatedaccurately.

Before applying the assay to the detectionof cannabinoids in urine samples, normalurines were assayed to determine any non-specific interference. It was found thatnormal urine was negative in the assay sys-tern. A pool of normal urine was spiked with

Table 2Assay of a Spiked Normal Urine Pool

Nanograms 9 THCAdded to Sample

2.2 2.0

2.0

5 0.8 0.7

1 11.2 11.2

2 5.6 5.2

3

4 1.6

Sample No.Nanograms 9 THCBy Radioimmunoassay

ASSAY OF BIOLOGICAL SAMPLES

Clinical studies were undertaken to estab-lish the validity of the assay under exper-imental conditions. All samples wereassayed blind, with no knowledge of thekey to the sample number code.

The first study involved analysis of 24hour urine speciments taken from heavy potusers. Prior to receiving oral doses of

THC, the subjects were maintained free ofany drugs for nine days. They were thenmaintained for 12 days on a dose of 120 mg/day of hashish oil in ethanol. Urinespecimens were assayed for cannabinoids onday 9 and day 21. From an examination ofthe data, summarized in Table 3, it canbe concluded that the assay did detect thepresence of large amounts of cannabinoidsin the urine samples obtained on day 21.

Table 3Cannabinoid Levels Found in Urine Samples From

Heavy Pot Smokers, Before and After Oral Administration

of Hashish Oil

No drug for 9 days Level Found

Patient A 30 ng/ml

Patient B 40 ng/ml

Hashish Oil for 12 days

Patient A 7000 ng/ml

Patient B 4000 ng/ml

Note: At the time of these assays, the minimum detectablelevel of cannabinoids was 12.5 ng/ml.

7

A second series of 24 hour urine samplesfrom a similar experiment were alsoassayed. The patients were maintainedfor seven days without any drugs and werethen treated with oral doses of THC.Subject 2 was a fairly heavy cannabis userbefore admission to the hospital, and theinitial urine cannabinoid level was con-sistent with this fact. At 8:00 a.m. onday 8 treatment was initiated on a scheduleof 10 mg of THC in sesame oil every 4hours, This schedule was maintained until8:00 a.m. on day 12 when the dose was in-creased to 30 mg every 4 hours. The treat-

On days 13 and 22 each subject was alsogiven two cigarettes each of which contained20 mg of THC.

sults.

As can be seen from the results in Table 4,inhaled doses of THC can be detected in theurine for 5-7 days after the last exposureto the drug. The administration of oraldoses of THC results in increased urinelevels of cannabinoids and the level fallsfollowing removal of the drug. The urinelevels also reflect an increase in the dos-age level. These studies indicate that theradioimmunoassay will give meaningful re-

ment was discontinued on day 24 at 4:00 p.m.

Day Dose Schedule Patient 1 Patient 2

123456789

10111213141516171819202122232425262728

No drug

10 mg/4 hr.

170 ng/ml764738

No sample17

30 mg/4 hr.

2401718185015822768372541757375333869929413560065634938

1025410722

50885513355016001662

808

1498 ng/ml168106

81692875

992372129944255542578193675608899506313

1576710813132759556

1118113258775

No sample1075

442223

Table 4

Cannabinoid Levels in 24 Hour Urine Samples

No drug

*Two cigarettes each containing 20 mg. of 9THC smoked inaddition to regular dose.

8

SUMMARY

A tritium based radioimmunoassay for 9THCand its metabolites has been developed forthe use of investigators studying the epi-demiological, medical, clinical, and re-search aspects of cannabis use. The assayis sufficiently sensitive to detect can-nabinoids in the urine of marijuana smokersfor several days after their last exposureto the drug. The results obtained from a28 day study indicate that the assay re-flects the administration and removal oforal doses of THC. The specificity of theantisera, as determined in cross reactivitystudies, allows not only the assay ofmetabolites in biological samples withoutinterference from other drugs, but also theevaluation of extracts of other kinds ofsamples which may contain unmetabolized

9 THC.

The technique of radioimmunoassay has manyadvantages over other methods of analysis.It is simple to perform and can be readilyapplied to the rapid analysis of largenumbers of samples, It can be used in thedirect analysis of physiological fluids andother biological samples which ordinarilyhave to be processed before other techniques

ACKNOWLEDGMENTS

can be applied. The method is non-destruct-ive and obviates the need to use radio-labelled drugs in man during metabolic andother studies.

This radioimnunoassay has been designed withparticular emphasis on ease of use by otherinvestigators. We anticipate that it willprove useful to investigators and scientistsfor determining the absence, or presence andamount, of THC metabolite in a biologicalspecimen, for epidemiologists in determin-ing the full extent of cannabis use and tothe medical/clinical community for establish-ing the minimum effective dose of 9THC foreach patient. The widespread applicationof a single method of analysis should alsoremove a great deal of the controversy sur-rounding marihuana studies performed to date.

1 14C- 9THC=(-) 1-Tetrahydro (3',5'-14C) cann-abinol, 31 mCi/mmole, was purchased from Amer-sham Searle. The radiolabel purity was foundto be 97-98% by radioscan in two different TLCsystems capable of separating 8 and 9 THC.

3H THC= 1(G-3H) tetrahydrocannabinol. 13 Ci/mmole was purchased from Amersham Searle. Theradiolabel purity was found to be 98% by radio-scan in three different TLC systems capable ofseparating 8 and 9 THC.

REFERENCES

This work was carried out under the sponsor-ship of the Department of Justice, DrugEnforcement Administration, under ContractNo. DEA-75-4, the Department of Justice,Bureau of Narcotics and Dangerous Drugsunder Contract No. J-70-24 and independ-ently by Collaborative Research, Inc.

Grant, J. D., Gross, S. J., Lomax, P. andWong, R., 1972, Nature New Biol., 236,216.

Gross, S.J., Soares, J. R., Wong, R. andSchuster, R. E., 1974, Nature, 252, 581.

Teale, J. D., Forman, E.J., King, L.J.,Piall, E. M., and Hanks, V., 1975,J. Pharm. Pharmac., 27, 465

9

SEPARATE RADIOIMMUNE MEASUREMENTSOF BODY FLUID THC AND11-NOR-9-CARBOXY T H C

Stanley J. Gross, M.D., James R. Soares, Ph.D.Department of AnatomySchool of MedicineUniversity of CaliforniaLos Angeles, California

INTRODUCTION

Pharmacologic and metabolic studies with can-nabinoids have remained largely unquantita-tive due to lack of simple assay techniques.Initial gas chromatographic and mass spectro-scopic methods were tedious (Fenimore 1973;Garrett 1973; Agurell 1973). Our immuneapproach led to early development of a simpleradioimmune assay for 9-THC in plasma andurine (Gross 1974). This has now been ex-tended to measure separately in body fluids11-nor-9-carboxy- 9-THC (C-THC) (Soares) amajor cannabinoid metabolite (Bernstein1972).

ASSAY METHODS

Institute of Drug Abuse. The generation ofantisera for 9-THC and C-THC has been de-scribed (Gross and Soares 1974; Soares andGross 1976). Plasma 9-THC and C-THC wereextracted, the extracts reconstituted in 50%ethanol or phosphate buffer and assayed.Urine specimens were assayed directly.Plasma and urine samples were obtained from

3H- 9-THC (50 Ci/mM) and C-THC were obtainedfrom Research Triangle Institute, North Caro-l i n a . 9-THC was obtained from the National

subjects at various intervals after smokingone to three marihuana cigarettes (Containing19.8 mg 9-THC) and subsequently assayed forboth 9-THC and C-THC. Pooled plasma fromnon-users of marihuana spiked with variousknown amounts of 9-THC and C-THC was used toestablish the standard inhibition curves.The urine standard curves were similarly de-termined.

Plasma 9-THC and C-THC: Plasma samples1

were obtained from five subjects after smok-ing a single cigarette and from three sub-jects after smoking three consecutive cigar-ettes. None of the first group and only oneof the second group had measurable preintoxi-cation levels of 9-THC. However, four of theeight subjects had measurable preintoxicationlevels of C-THC.

In five single smoke subjects definitive in-crements (10-130 ng/ml) of 9-THC occurred 15minutes after a single cigarette becoming al-most undetectable by 2 hours (Table 1). C-THCpeaked at 30-60 minutes; significant but lowlevels remained at least three hours after ex-posure.

10

Table 1Plasma 9-THC and C-THC levels in occasional THC

users 15-240 minutes after smoking a single 900 mg THC cigarette(2.2% 9-THC/cigarette)

180 240Post intoxication (minutes)

5 15

31144

130- - -

9 0- - -

10- - -

30

3 1224

92 38 0116 24 59

60 120

21 11116 71

97

77 6

79 2 5

0 012 75

14 165 9

0 6036 34

08 23

0 00 40

0 00 0

T. S. 9-THC 0 73 32 1 4 0 0C-THC 0 278 228 216 108 115

50

00

00

019

S. W. 9-THCC-THC

Z. W. 9-THCC-THC

B. W. 9-THCC-THC

P. G. 9-THCC-THC

Preintoxi-cation

0Subject

7 -

5 5

The rapid shifts of plasma 9-THC and C-THC15 minutes to 48 hours following repetitive

tiple consecutive cigarettes than a singleone, permitting detection of 9-THC 2-3 hours

exposure are shown in Figure 1 and summarized9-THC peaks

after completion of the last cigarette. Sig-in Table 2. Expectedly plasma nificant amounts of this metabolite were mea-(100-260 ng/ml) were much higher after mul- sured in plasma 48 hours later.

Table 2

Plasma 9-THC and C-THC equivalent levels in occasional marihuanausers 15 min - 2 days after consecutively smoking three 900 mg

marihuana cigarettes (2.2% 9-THC/cigarette)

Pre

0 15' 30' 60' 120' 180'

8 185 78 123 39 727 285 112 292 211 251

0 260 - - 37 15 7 040 68 - - 158 43 111 140

100 63 87 37 5375 460

B. W. 9-THCC-THC

T. T. 9-THCC-THC

W. Z. 9-THCC-THC

Post intoxication ng/ml2 4 0 ' 2 4 h r s 4 8 h r s

Subject cationintoxi-

12 41 10296 97 75

7 5 8428 60 24

2194

971

046 401 413 363

11

Figure 1

Figure 1: Post-intoxication plasma 9-THC and C-THC (ng/ml) in a representative subject.= 9-THC; = C-THC.

Figure 2

Figure 2: Post-intoxication urine C-THC (ng/ml) in chronic subjects. Arrows indicate anadditional smoke.

12

Table 3

No.

1

2

3

4

5

6

7

8

9

10

11

Urine C-THC levels in chronic THC users 1-12 hrs afterconsecutive smoking 900 mg THC cigarettes

(2.2% 9-THC/cigarette)

Sub-ject

119

123

124

126

127

120

121

121

132

128

128

C-THC level (ng/ml)Postintoxication (hrs)

2 4 6Preintox-ication

72

28

79

61

40

123

13

19

17

17

9

1

176

55

134

90

14

4

31

11

144*

106*

107

57

125

21

18

28

17

12

101*

148*

105

126

93

325*

97*

91*

*

166* 205

10 8

10* 23

23

UD UD

UD UD

8

129

196

48

196*

78

215

5

23

31

UD

U D

*Repeat cigarettes on subject demand

UD = undetectable

Table 4Urine C-THC levels in occasional THC users 1-48 hours after smoking

900 mg THC cigarettes (2.2% 9-THC/cigarette)

Subject

B. W.

K. G.

R. B.

W. Z.

L. G.

T. T.

Prein-toxi-cation

5

UD

5

UD

2

4

Urine C-THC level (ng/ml)

Postintoxication (hrs)41

33

3

9

7

6

20

2

18

17

10

18

7

26

3

17

2

12

21

4

44

33

16

21

9

7

59

6

39

2

7

54

15

--

8

23

3

7

31

3

29

12

15

--

4

--

- -

14

24 4 8

3 2

5 3

4 3

5 U D

7 5

-- 9

12

164

290

UD = undetectable

13

Urine 9-THC: None of four subjects had mea-surable urinary 9-THC. This is consistentwith previous work (Hollister 1974) de-scribing urine 9-THC levels to be far belowpresent immune assay sensitivity,

Urine C-THC: Assays of urine from 9 chronicusers 1, 2, 4, 6 and 8 hours after completionof the first standard cigarette are summar-ized in Table 3 and Figure 2. Preintoxica-tion levels varied from 28-123 ng/ml. LargeC-THC increases occurred (Table 3) peaking at2-4 hours after initial exposure. Unfortun-ately the rigid clinical protocol permittedadditional cigarettes during the 8 hour studyperiod, C-THC continuing to rise after eachadditional cigarette. Urine C-THC levels inoccasional smokers were considerably lower(Table 4) than in chronic users (Table 3).

Though relatively few subjects were studied apattern of relative 9-THC and C-THC levelsdid emerge. In occasional smoker subjectsthere was a 15-30 minute plasma 9-THC peakand a 30-60 minute C-THC peak after use of asingle standardized marihuana cigarette. Im-portantly, 9-THC became almost undetectablein plasma 1-2 hours after exposure while sig-nificant amounts of C-THC persisted in circu-lation for several hours. This critical di-vergence of 9-THC and C-THC was even moreobvious in subjects who had smoked 3 consecu-tive marihuana cigarettes. 9-THC was almostundetectable in all plasma samples four hoursafter the last cigarette had been consumed

REFERENCES

Agurell, S., Gustafsson, B., Holmstedt, B.,Leander, K., Lindgreen, J. E., Nilsson, I,,Sandberg, F. and Asberg, M., J. Pharm. Phar-mac. 25, 554 (1973).

Bernstein, S., Rosenfeld, J. and Wittstruck,T., Science 176, 422 (1972).

Fenimore, D. C., Freeman, R. R. and Loy, P.R., Anal. Chem. 45, 2331 (1975).

Garrett, E. R. and Hunt, C. A., J. Pharm.Sci. 62, 1211 (1973).

despite the prolonged (48 hrs) persistence ofsignificant levels of C-THC.

Urine from occasional marihuana subjects wasnegative or marginal for 9-THC while C-THC(60 ng/ml or less) was detected 1-48 hoursafter one standard cigarette. The pre-smokevalues were zero ( 9-THC and C-THC) for occa-sional smokers. However chronic smokers hadsignificant pre-smoke levels (17-23 ng/ml) inaddition to a vastly greater rise of urineC-THC 2-4 hours after consumption of a singlecigarette.

Clearly, a single THC metabolite level or useof a significantly crossreacting antiserum(Marks, 1975; Teale, 1975) cannot be used toascertain “post-intoxication” intervals.Such a result could reflect a single exposurejust prior to an examination or multiple ex-posures several days earlier. The failure todetect plasma 9-THC indicates unambiguouslythat marihuana was not smoked within the pre-ceeding hour, whereas detection of plasmaC-THC in the absence of 9-THC indicates dis-tant exposure. A large controlled experi-mental population is now essential to corre-late 9-THC and C-THC (ratios) with individ-ual metabolic variants for behavior differen-ces.1Samples were provided by the Department ofPsychiatry, UCLA School of Medicine.

Gross, S. J. and Soares, J. R., Nature 252,581 (1974).

Hollister, L. E., Kanter, S. L., Board, R. D.and Green, D. E., Res. Comm. Chem. Pathol.Pharmacol. 8, 579 (1974).

Marks, V. , Teale, D. and Fry, D., BritishMed. J. 3, 348 (1975).

Soares, J. R. and Gross, S. J. (Submitted forpub1ication).

Teale, J. D., Clough, J. M., Pialli, E. M.,King, L. R. and Marks, V., Res. Comm. Chem.Pathol. Pharmacol. 11, 339 (1975).

14

RADIOIMMUNOASSAY OF9-TETRAHYDROCANNABINOL

Clarence E. Cook, Ph.D., Mary L. Hawes, B. A., Ellen W. Amerson, B. A.,

Colin G. Pitt, Ph.D., David Williams, B.A.

Research Triangle Institute, North Carolina

INTRODUCTION

Since its introduction by Yalow and Berson(see Berson and Yalow, 1971), radioimmuno-assay (RIA) has played an ever-increasingrole in the quantitative analysis of drugsand hormones. A number of review articlesand books are available (inter alia, Skelleyet al., 1973; Spector et al., 1973; Abraham,1974). In particular a dramatic improvementin assay methodology for many steroid hor-mones has resulted from the application ofradioimmunoassay to these substances.

Due to its extreme sensitivity, RIA can per-mit the quantitation of substances presentin concentrations as low as a few pg/ml ofa biological fluid. The inherent selecti-vity of antibodies introduces a further ad-vantage since it often makes sample prepara-tion requirements minimal and assay method-ology simple and thus permits the analysisof large numbers of samples.

These obvious assets of RIA have naturallyled to its consideration as a means for mea-suring blood levels of cannabinoid compounds.A valid RIA procedure for 9-THC would beuseful in studying the pharmacokinetics ofthe drug. Furthermore it is conceivable that

RIA methods could be used for the identifica-tion of THC and/or its metabolites, thusfacilitating forensic investigations. There-fore a number of investigators have beeninterested in the development of RIA proce-dures for the cannabinoids.

We believe it is fair to state that develop-ment of useful and simple RIA procedures for

9-THC has proven to be a relatively diffi-cult research problem. The difficulties in-volved in THC RIA development may be sum-marized as follows:

(1) The highly lipophilic nature of themolecule causes problems in working with theaqueous systems in which RIA is carried out.The compound adheres well to glass and plas-tic--generally in preference to dissolving inan aqueous medium.

(2) In plasma the compound is tightlybound, principally to a lipoprotein fraction(Klausner, et al., 1975).

(3) It has proven difficult to obtainantibodies highly selective for 9-THC vs itsmetabolites and various analogs.

15

(4) The plasma levels of interest (ca.1-100 ng/ml, although not low by the usualstandards of RIA, are sufficiently low thatdilution of plasma samples and direct analy-sis, a technique which has proven highlysuccessful with certain other drugs (Spectoret al., 1973; Cook et al., 1973, 1975b;Christensen et al.,1974) is not readilyfeasible for the lower levels.

(5) Highly concentrated (i.e., hightiter) antisera have been difficult to ob-tain.

(6) High specific activity radioligandfor competitive binding studies has been dif-ficult to obtain and often unstable.

It is the purpose of this paper to illustratethese problems and to discuss progress madein overcoming them.

MATERIALS AND METHODS

Definitions

Titer - The dilution of original antiserumwhich must be added to the assay to obtainX% binding. If one adds 0.1 ml of antiserumwhich has been diluted 1:500 and obtains 50%binding of radioligand, the 50% titer (ini-tial dilution) is 1:500 If in the aboveexample the total assay volume was 0.5 ml the50% titer (final dilution) would be 1:2500.In this paper we will quote initial dilutiontiters unless otherwise noted.

T, B, and N-tubes - Designation for differ-ent assay tubes. At end of assay, volumesin all tubes are equal. T tubes measuretotal radioactivity; N tubes measure non-specific radioactivity (activity not ad-sorbed by charcoal) and B tubes measure theamount of labeled drug bound to antibody inthe absence (Bo) or presence (Bi) of addedunlabeled compound (see Table 1).

Table 1

DEFINITION OF ASSAY TUBES

TubeDesignation Tube Contents

UnlabeledRadioligand Compound Buffer Antiserum Charcoal

T + - + ± -

N + - + - +

B o + - + + +

Bi+ + + + +

Table 1. Definition of abbreviations for assay tubes

Materials

Chemicals used in the work were reagent gradeobtained from commercial sources. Bovineserum albumin was obtained from Sigma Chemi-cal Co., St. Louis, MO (crystallized and lyo-philized, No. A4378; or Fraction V, fattyacid free, No. A6003). Norit A (C-176) wasobtained from Fisher Scientific, Pittsburgh,PA. Triton X-100 was obtained from PalmettoChemical Co., Monroe, NC. Antiserum was ob-tained from Dr. James Soares of the UCLASchool of Medicine (Lot No. G2532; Gross etal., 1974) and from Dr. V. Marks thru the-National Institute for Drug Abuse (Teale etal., 1974, 1975; S133Y/30/9). Additionalantiserum was prepared at Research TriangleInstitute (see below).

RIA buffer (0.1 M, pH 6.8) contained 16.35 gNa2HPO4•7H2O, 5.38 g NaH2PO4•H2O, 9.0 g NaCl,1.0 g NaN3, 1.0 g bovine serum albumin (BSA),and 1000 ml double distilled water adjustedto pH 6.8 and was stored in a refrigeratorfor no more than one month.

Fine particles were removed from charcoal byseveral decantations from a suspension indistilled water. The charcoal was dried at200°C and 25 g was suspended in 1000 ml ofbuffer.

Scintillation fluid contained 2 liters tolu-ene, 1 liter Triton X-100 [purified by stir-ring for 30 min with 606 Tell-Tale SilicaGel (25 g/ ) and filtering] and 18 gramsOmnifluorR (New England Nuclear).

Undiluted serum was stored in a freezer (-20°C).Serum diluted in RIA buffer could be storedin a refrigerator for several months.

The radioligand ( 9-THC-3H; Pitt et al.,1975) stock solution was kept in benzene/10%ethanol and refrigerated. Solutions forassays were made in 50% ethanol:50% water(double distilled). The radioligand was di-luted for assays to 10,000 cpm/10 µl whichgives ca. 100 pg 8-THC/10 µl. Dilutions inEtOH/H2O may be kept refrigerated for no morethan 7-10 days. Radioligand decompositionbecomes significant after that time. Stocksolutions were tested frequently for decompo-sition by thin layer chromatography. Silicagel plates were developed in 100% benzene andthe purity of the radioligand determined byradioscan.

9-THC for standard curves and metabolitesfor cross reaction studies were prepared atRTI. For radioimmunoassay, 1 mg/ml stocksolutions (prepared from solutions used asglc standards and obtained from K. H. Davisof this laboratory) were stored in 100% etha-nol and refrigerated. Dilutions were pre-

16

pared in 50% ethanol/50% water. These dilu-tions may be kept refrigerated for approxi-mately one month. Periodic analyses by thinlayer chromatography are required to testthe purity and stability of the compounds.

RESULTS AND DISCUSSION

Antigen synthesis

In the first part of this century, Land-Steiner (see Landsteiner, 1962) showed thatcovalent bonding of a small molecule (hap-tenic component) to a protein gave an anti-genic substance and that antibodies formed tothis antigen were capable of selectivelybinding the small molecule. He further es-tablished that the antibody selectivity isinfluenced primarily by those portions of thehapten which are removed from the linkage tothe protein carrier. Thus to achieve anantibody which will be selective for theparent drug in the presence of its metabo-lites, proper synthetic design of the antigenis essential. Those portions of the moleculewhich are subject to metabolic alterationsmust be left free to influence antibody selec-tivity.

The very fact that the hapten must beattached to the protein via a covalent bondthen means that it is impossible to devise aconjugate in which all portions of the haptenwill equally influence antibody selectivity.No linkage can be considered completely inert.In addition, in designing the synthesis ofthe antigen one must consider the chemistryinvolved and the ease or difficulty in rela-tion to the expected benefit. Depending upontheir needs, then, it can be expected thatdifferent investigators will synthesize anti-gens in different ways.

Most of the metabolic alterations of 9-THCoccur in the cyclohexene moiety with hydroxy-lation of the 9 compound occurring at the11-and 8-positions along with conversion ofthe 11-carbon to a carboxyl group. More re-cently it has been reported that hydroxyla-tion can also occur in the amyl side chain ofthe phenolic ring (Wall, 1975).

A number of previously reported antigens aresummarized in Figure 1. Teale et al. (1974,1975), formed a hemi-ester linkage with thephenolic hydroxyl group and then coupled thiscompound to bovine serum albumin to form anantigen. Tsui et al. (1974), report forma-tion of a hemisuccinate from the phenolic hy-droxyl, and also formed an ether linkage atthis position to give a carboxymethyl deriva-tive. The products were coupled to a varietyof proteins as shown in Figure 1. The pheno-lic group apparently undergoes no metabolic

Figure 1

a ) CO-NH-BSA (Teale, et al., 1974)

b) CO-CH2CH2CO-NH-BSA (Teale, et al., 1975)

C ) CO-CH2CH2CO-NH-PGG (HSA, SGG, or PLL) (Tsui, et al., 1974)

d ) CH2CO-NH-HSA (or SGG) (Tsui, et al., 1974)

e ) N=N CO- NH-PGG (HSA or SGG) (Tsui, et al., 1974)

f ) N=N CO-NH-KLH (Gross, et al., 1974)

(+4–isomer)

g ) 10-I-9-NH-CO-NH-HSA (or PGG) (Tsui, et al., 1974)

Figure 1. Some positions through which 9-THChas been linked to protein

alterations, but spatially the attachment toprotein is relatively close to the metaboli-cally important cyclohexene ring. To avoidthis problem Tsui et al. also prepared a 2-azophenylcarboxy derivative, a substitutionempioyeh by Gross et al. (1974) as well.This substitution frees the hydroxyl groupfor binding, although at the possible ex-pense of introducing a relatively immunodom-inant azobenzoyl moiety. Finally, Tsui etal. (1974) also prepared a 10-iodo-9-ureidolinked THC.

Our attention was drawn to the amyl sidechain of the aromatic ring as a potentialposition for attachment to the protein.Such a linkage would certainly fulfill therequirement of distance from the metaboli-cally reactive cyclohexene ring (althoughnot from sites of hydroxylation on the amylside chain). Also it was of interest to exam-ine the effectiveness of the flexible hydro-carbon chain in exposing the tricyclic moietyto influence antibody selectivity. Use ofsuch a long flexible chain had proved quitesuccessful for us in synthesis of antigensfor caffeine (Cook et al., 1974) and phenyl-butazone (Cook et al., 1975a). For syntheticreasons our initial attempts dealt with the

8-THC analog as a substrate.

Synthesis of the 8-THC antigen is shown inFigure 2. 5'-Carboxy- 8-THC labeled with

17

Figure 2SYNTHESIS OF – THC ANTIGEN

serum prepared from the amyl-linked antigenis shown in Figure 3. The affinity constants

booster administrations. A third rabbit even-tually achieved a titer (final dilution) of1:2250 at the fourth bleeding. Titers gener-ally declined at the fifth and sixth bleed-ings.

These titers compare quite favorably withthose reported by others (Teale et al., 1975;Gross et al., 1974) and demonstrate that withthis antigen the rabbit is a reasonable ani-mal to use for antibody production. A fur-ther comparison of the azobenzoyl-THC-derivedantiserum obtained from UCLA and the anti-

Figure 3

Figure 2. Synthetic route to antigen based on5'-carboxy- 8-THC

ANTISERUM CHARACTERISTICS

UCLA RTl-RS-81

tracer amounts of carbon-14 in the carboxyl Affinity Constant 1.6 X 109 2.5 X 109

group (Pitt et al., 1975) was converted to itsN-oxysuccinimidoyl ester by reaction with N-hydrbxysuccinimihe and diclohexylcarbodi-imide. This active ester readily coupledwith bovine serum albumin in a mixture of di-oxane and water to yield the desired antigen.Radioactivity measurements indicated the in-corporation of ca. 33 residues of 8-THC/molecule of bovine serum albumin. We have

(for 8 - T H C )

Std. Curve

Characteristics

(Logit - log Plot)

found this mode of coupling to be a veryuseful one in certain circumstances. Where

50% Intercept 0.5 ng 1.4ng

it is applicable, the number of molecules ofdrug moiety incorporated into the bovineserum albumin is relatively easy to control.

Formation of antisera

Rabbits were immunized with the conjugatedissolved in sterile 0.9% NaCl and homoge-nized with an equal volume of Freund's com-plete adjuvant. Immunization with 200 µg ofantigen was carried out by intradermal tech-nique of Vaitukaitus et al. (1971). This wasfollowed after two weeks by another intrader-ma1 immunization (100 µg) and then at fourweek intervals subcutaneous booster injec-tions (100 µg of antigen) were given. Rab-bits were bled on day 52 after the initialdose and every 28 days thereafter. After thethird bleeding the immunization program wasdiscontinued for a three-month period. Anti-gen booster injections were then resumed andthe fourth bleeding was taken 10 days afterimmunization. Reasonable titers (50% bindingof ca. 125 pg of labeled 8-THC at an initialdilution of 240-540 or a final dilution of1200-2700) were achieved in two out of fourrabbits at the first bleeding, but no signifi-cant increases were observed on subsequent

Linear Range 0.3 - 4.1 ng 0.3-6 ng

Figure 3. Characteristics of two antiseraas determined in this laboratory: UCLA(Gross, et al.,paper).

1974); RTI-RS-81-2 (this

are quite close. The slightly higher affi-nity of the 8-antiserum for 8-THC is under-standable. Due to the heterologous nature ofthe systems employed ( 8-radioligand, unla-beled 9 and antiserum generated to either

8- or 9-), the standard curve characteristicsare not similar for the two antisera. Whenstandard curves are converted to the logit-log basis (Rodbard et al., 1969), the UCLAantiserum exhibits a steeper slope with alower 50% intercept and a somewhat narrowerlinear range than does the RTI antiserum.The lower limit in Figure 3 is a conserva-tive number. If one defines sensitivity asthe amount of 9-THC which will reducebinding to a level two standard deviationsbelow the initial value, the limit for theRTI antiserum is ca. 100 pg.

18

Antibody selectivity

The avidity of the antisera for various me-tabolites and analogs of 9-THC was measuredby determining the relative amount of com-pound required for 50% displacement of theinitially bound radioligand. This is a re-latively crude measure of cross-reactivity.It is strictly valid only if the displace-ment curves for the two compounds in ques-tion are completely parallel and it does nottake into account the possibility of smallsubpopulations of antibodies with high avi-dity for one substance in preference to theother. Nevertheless, it offers a useful ifrough guide to antibody selectivity if itslimitations are realized. Figure 4 shows

Figure 4

Figure 4.Structure-binding relationships of variousantisera for cannabinoids. Numbers inparentheses are from the references cited:A (Teale, et al., 1975); B (Gross, et al.,1974)

the cross-reactivity of three types of anti-sera, all measured in our own laboratory.Selectivity can be significantly influencedby the assay conditions. Therefore thevalues reported by the producer of the anti-serum are also given in the figure in paren-theses. The displacement ability of 9-THCis taken as a standard at 100%. For reasonsdiscussed later, the radioligand used was

8-THC-3H.

A change which alters the tricyclic characterof the molecule (see cannabidiol, CBD) es-sentially destroys binding to all of the an-tibodies. However, much more subtle changescan also have significant effects. In thecase of the antibody to 8-THC, shift of thedouble bond from the 8- to the 9-positionresults in a two and one-half fold decreasein cross-reaction. A similar (two-fold) de-crease is seen on comparison of the bindingof 9- and 8-THC to antibody obtained fromthe other two antigens.

Reduction of the cyclohexene double bond hasno greater effect with the 8-antibody thandoes the shift to the 9-position and actu-ally somewhat increases the cross-reactionwith the 9-antiserum prepared from the azo-benzoyl antigen. Aromatization of the cyclo-hexene ring (to CBN) results in a four-folddecrease in cross-reaction as compared with

9-THC in the case of the azobenzoyl- andamyl-linked antigens. It has little effecton antibody to the O-succinoyl antigen.

Oxygenated C-11 metabolites cross-react rela-tively little with the antibody from the amylantigen. Cross-reaction of the 11-hydroxymetabolite with the azobenzoyl antiserum israther significant (49%) but the cross-reac-tion drops markedly when the 11-nor-9-car-boxy metabolite is considered (2%). Both ofthese metabolites cross-react strongly withantibody from the O-succinoyl antigen.

Thus it appears that good, but not outstand-ing, selectivity for 9-THC vs a number ofmetabolites and analogs can be achieved andthat modest but usable antibody titers can beobtained in either the goat, sheep, or rabbit.Let us now consider some of the other prob-lems in the development of an assay.

Assay parameters

General procedure.--The general assay proce-dure finally adopted is shown in Figure 5.Data leading to definition of the variousconditions are presented in suceeding para-graphs.

Radioligand.--The sensitivity of a radioim-munoassay is very much a function of thespecific activity of the bound radioligand.Thus, in general, one would prefer the high-est possible specific activity. Storage ofsuch high specific activity compounds canoften lead to radiation-induced decomposi-tion. The general experience of our labora-tory in the preparation and storage of suchmaterials is that unlabeled substances whichreadily decompose through autoxidation, etc.are often particularly unstable when pre-pared radiolabeled and with high specificactivity. This was found to be the case for

19

Figure 5 Figure 6

Figure 5. General procedure for RIA of 9-THC

9-THC which was prepared at a specific ac-tivity of ca. 50 Ci/mmole (Pitt et al.,1975) but had a very short shelfilife. Onthe other hand, 8-THC-3H, prepared in simi-lar specific activity, has proven to be aquite stable entity. Its binding character-istics with the antisera raised to either

8- or 9-THC make it a useful radioligand.Hexahydrocannabinol also exhibits goodcross-reactivity with the antisera but syn-thetic problems have discouraged our use oftritiated HHC as a radioligand.

Buffer Protein.--The relatively intractablenature of THC in aqueous systems is by nowpresumably well-known to all interested inthis field of research. Because of the smallamounts of material involved, RIA proceduresfor most compounds call for addition of sometype of protein to the buffer in order toenhance solubility of the various componentsand prevent binding to the incubation tubes.Our experience has been that the nature ofthis protein is of crucial import in workwith 9-THC. This is illustrated in Figure6. When ca. 5 ng of tritiated 8-THC in asmall amount of ethanol was added to a normalRIA buffer (phosphate buffered saline) con-taining 0.1% bovine -globulin as the pro-tein, very significant amounts of the THCwere bound to the glass. This was illus-

Figure 6. Illustration of the binding to glassobserved with THC

trated by decanting the buffer into scintil-lation liquid. The glass tube was then bro-ken up, and the pieces were placed in ascintillation vial and treated with a littlemethanol. Scintillation fluid was thenadded. This experiment indicated that only57% of the THC remained in the buffer and 41%adhered to the glass tube (percentages arecorrected for quenching). Even allowing forincomplete removal of buffer from the glassprior to crushing it, this is still a verysignificant loss of the THC.

Experiments on the effect of silanizing glasstubes or washing them with nitric acid gaveno improvement and in most cases the resultswere worse. However we found that in ourlaboratory, the use of bovine serum albuminas the protein component of the buffer re-sulted in much diminished adherence to glassand a better assay system. This is illustra-ted in Figure 7. Labeled THC was incubatedwith buffer containing either bovine serumalbumin (BSA) or bovine -globulin (BGG) plusvarying dilutions of antiserum. The solu-tions were then decanted into scintillationliquid and radioactivity determined. Resultswere not corrected for quenching, which hadan approximately equal effect on all values.

At high concentrations of antiserum, thereis no difference between the two buffer pro-teins. In effect the antiserum itself iskeeping the THC in solution. However, asthe antiserum concentration is decreased(dilution increased) less and less of thelabel can be recovered from the tubes con-taining the BGG and eventually the recovery

20

Figure 7 Figure 8

Figure 7.Variation of THC solubility as a function ofantiserum dilution and buffer protein. Anti-serum of Gross, et al. (1974); BSA = bovineserum albumin; BGG = bovine gamma globulin

falls to essentially the level found in theprevious experiment. The adsorption toglass may be looked upon as competing withthe binding to antiserum. Although the pro-blem may be overcome by higher concentra-tions of antiserum, it was to us preferableto avoid this effect if possible, removeanother source of variability from the assay,and increase the practical titer of the an-tiserum by using bovine serum albumin in thebuffer system. At the concentrations em-ployed, this protein did not decrease theamount of radiolabel which was adsorbed tocharcoal.

Not only was the type of protein important,but the assay characteristics were alsogreatly influenced by the type of bovineserum albumin used and even by different lotsof the same type of BSA. This is illustratedin Figure 8, where the 50% titers for threedifferent antisera (one obtained from UCLAand two prepared at RTI) are compared in thepresence of the same concentration of protein(0.1%) but varying the type or lot of bovineserum albumin used. The first lot of crystal-lized and lyophilized BSA used gave the bestoverall binding characteristics. Subsequentlots gave significantly lower titers with onein particular giving extremely poor results.Somewhat more consistent results were obtainedwith Sigma’s fraction V fatty acid-free BSA.Because this material is less expensive thanthe crystallized type, we have settled on itin practice. A third type (fraction V powder)

Figure 8.Influence of different types and lots ofbovine serum albumin on apparent antiserumtiter (initial dilution). Each verticalline represents a separate lot of BSA:A4378 (crystallized once and lyophilized);A4503 (Fraction V powder); A7511 (crystal-lized and lyophilized; less than 0.005%fatty acids); A6003 (Fraction V; less than0.005% fatty acids)

gave very poor results and a BSA type preparedby Sigma specifically to aid in the RIA of in-sulin proved worst of all.

We were unable to pinpoint the difference inthe various lots which might have influencedthe results. The presence of fatty acids asa contaminant was suggested since batchescharacterized as fatty acid-free gave gener-ally better results. Treatment of one of theless useful batches with charcoal to furtherremove fatty acids (Chen, 1967) gave some im-provement, but did not result in raising thequality to that of some of the better batches.We have, therefore, concluded that (a) eachbatch of BSA used should be checked out be-fore large amounts of antiserum are dilutedin the buffer, (b) large batches should pre-ferably be checked and then used in the assayto attain consistency, (c) the quality of anantiserum in terms of titer may be stronglyinfluenced by the composition of the proteinin the buffer (which may account for inter-

21

Figure 9 Figure 10

Figure 9.

Net binding as a function of BSA concentrationand type: Sigma fraction V (<0.005%fatty acids); different lot of the aboveafter second removal of fatty acids;crystallized and lyophilized

Figure 10.

Effect of pH on binding and non-adsorbedradioactivity. Antisera were obtained fromUCLA (Gross, et al., 1974) and as describedin this paper (RTI-RS-81-4)

laboratory differences in antiserum titers)and (d) some antisera are more sensitive tochanges in the protein than are others. Thisis particularly true of RTI RS-81-3.

As would be expected, the concentration ofprotein had an effect on the net percentbinding and this concentration dependence wasalso somewhat a function of the antiserum andthe type of bovine serum albumin used. Thisis illustrated in Figure 9. At low concen-trations (<0.05% BSA) the net percent binding[100 (B-N)/T] was low, principally due to thefact that the total binding was very low.Non-specific radioactivity (N tubes, radio-activity not adsorbed by charcoal) was alsorelatively low up to 0.2% concentration ofBSA. At concentrations of 0.5-1% it becamea significant factor and resulted in a con-siderable lowering of the net percent binding.

Best results were generally obtained at 0.1-0.2% protein for the antisera and proteinsamples examined. Therefore for reasons ofboth economy and best binding, 0.1% was chosenas the best concentration.

pH Effects.--As shown in Figure 10, pH had adefinite effect on both total binding andradioactivity not adsorbed by charcoal. Thiseffect was again somewhat sensitive to thetype of antibody used. The UCLA antiserumexhibited considerably better binding at pHvalues of 8 or above, but the non-specificradioactivity also increased. Although an in-crease in apparent binding also appeared tooccur at higher pH with the RTI antiserum, thenet binding showed no change. To minimizenonspecific radioactivity pH 6.8 was acceptedas the best compromise. Since relativelylittle THC would be ionized under the condi-tions used, it seems likely that the changein pH acts principally by its effect upon theconformation of the binding globulin and/orby protonation effects in the binding region.

Incubation Time.--The effect of the time ofincubation at 4°C on the percent bindingachieved was also studied (Figure 11). Inthe case of the RTI antiserum a perceptiblerise in percent binding occurred up to about6 hr, although binding appeared to beginleveling off after about 4 hr. Binding equi-

37

Figure 11 Figure 12

Figure 11.Percent of radioligand bound as a functionof time of incubation at 4°C

librium was established more rapidly in thecase of the UCLA antiserum and was completein approximately 2 hr. Incubation times of4 hr or more would be suitable for eitherantiserum and there was no deleteriouseffect if incubation continued for an over-night period.

Charcoal Treatment.--Since the charcoal ad-sorbant removes free radioligand, one mightexpect that exposure to charcoal would even-tually shift the binding equilibrium so as toremove essentially all of the label from theantibody. The rate at which this would hap-pen would be dependent upon the rate of bind-ing reversal, which in turn should be a func-tion of temperature. A finite time is alsorequired to adsorb the free radiolabel and toeffect its removal from weakly binding sub-stances such as protein. The best equilibra-tion time then is often a compromise and mustbe studied for each antiserum.

Figure 12 shows the effect of letting thecharcoal suspension stand with the incubationmixture. Non-adsorbed radioactivity in theabsence of antiserum reached the minimumvalue in approximately 15 min and did notfluctuate for up to 60 min thereafter. Inthe presence of antiserum, a somewhat similarsituation obtained. By 15 min the minimumpercent bound was essentially reached. There-fore, in order to reduce the non-specificbinding to its minimum, a 15 min time forstanding with the charcoal is necessary.

Figure 12.Effect of time on adsorption of radioligandby charcoal at 4°C. One ml of 2.5% charcoalsuspension was added to incubation tubes con-taining 0.52 ml. Contents were mixed for10 sec on a vortex mixer and allowed to stand

Other experiments on the time of mixing of thecharcoal suspension with the incubation mix-ture showed that this also affected the amountof radioactivity left in solution. It wasnecessary to compromise in this regard and itwas found that a 10 sec mixing on a Vortexmixer gave the best results within a reason-able period of time.

Variation of charcoal concentrations (1 ml ofsuspension was added in all cases) from 1-10%showed that a 2.5% suspension gave best re-sults. Addition of dextran to the charcoaldid not prove useful.

Plasma analysis

Having established assay conditions, onewould now like to transfer these to the ana-lysis of samples of biological importance.Ideally, one would like to be able to analyzea diluted sample of plasma directly withoutcarrying out any extraction procedures. Whe-ther this is successful depends upon theamount of plasma which can be added to eachassay tube and therefore upon the concentra-tion of the drug in plasma and the sensiti-vity of the assay itself.

Plasma was added directly to T, B, and Ntubes containing both the RTI and UCLA anti-sera and its effect on each was studied. Asexpected, T tubes showed no change in goingfrom 0 to 60 µl of plasma in a total volumeof 0.52 ml. Addition of 15 or more µl of

23

Figure 13

Figure 13.Procedure used for extraction of 9-THCfrom plasma

plasma resulted in an increase in the N-tubesand a concomitant decrease in the percentbinding so that the net percent bound valuesat 15 µl of plasma were only 60-70% that ofthe tubes containing no plasma. Further in-creases in the amount of plasma resulted inlower levels of net binding. However, if oneassumes an assay sensitivity of 0.3 ng andignores any problems caused by the presenceof metabolites, direct assay of plasma downto a level of 20 ng/ml should be feasible.This technique assumes that one has a sampleof blank plasma from the subject or is will-ing to ignore the possibility of intersubjectvariation in effects on the assay. This lat-ter point will be considered later.

For pharmacokinetic studies, assays at levelsbelow 20 ng/ml are required. Solvent extrac-tion to remove protein and other interferen-ces must apparently be used to analyze lowlevels of 8-THC. Some care must be used inthe choice of solvents since small amounts ofcertain solvents can have a significant in-fluence on the assay. For example, the addi-tion of only 1 µl of isoamyl alcohol (thesolvent commonly used in conjunction withhexane to increase the extractability of THCfrom plasma) reduces the binding in an assay

24

tube to ca. 60% of that obtained in the ab-sence of the isoamyl alcohol. Thereforetraces of this relatively non-volatile sub-stance remaining from an extraction processmay have a very significant effect upon assayresults.

Ethanol has somewhat less of an effect. In-deed we routinely add up to 10 µl of ethanolto each assay tube in preparing the standardcurve. Addition of 20 µl more ethanol redu-ces binding to about 75% of the initial value.Nevertheless, extraction of 9-THC from plasmaby precipitation of proteins with ethanol hasbeen reported to be a useful procedure in RIAof THC (Teale, et al., 1975) and we havetherefore studied this particular extractionsystem with plasma samples. The overall ex-traction process used is shown in Figure 13.Plasma is treated with twice its volume ofethanol and precipitated material is centri-fuged out. Aliquots of the supernatant maybe taken for analysis at this point. Wetested the possibility of utilizing this su-pernatant as has been reported by Teale etal. (1975).

Since the plasma/ethanol extract has a defi-nite influence on the standard curve, it isnecessary to compensate for this effect byrunning the standard curve in the presenceof the same volume of plasma extract as thatto be analyzed. This works reasonably wellwhen it is possible to obtain a blank plasmafrom the same individual whose plasma THCconcentration is being studied. Howeverthis would not be easy to do in the case ofchronic users of THC and would represent adistinct limitation on the applicability ofthe method. It was therefore of interest tocompare standard curves obtained from dif-ferent plasmas.

Four human male plasma samples were spikedwith five concentrations of THC from 5-100ng/ml. Each plasma series was then used togenerate a standard curve. Each standardcurve, in turn, was used as the basis for the“analysis” of the spiked plasma samples. Wethen looked at the percent of the sampleswhich were analvzed with less than 20% error.Results are shown in Figure 14. As expected,when plasma series A was analyzed by standardcurve A (generated from plasma A), all of thesamples fell within ±20% of the actual spikedvalues. Under these-conditions it was possi-ble to add sufficient of the ethanol extractto get down to levels of 5 ng/ml with reason-able accuracy.

With cross plasma comparisons the results arenot so good. Plasma samples B and D gavereasonable accuracy when measured againststandard curves A and B. However, the accu-racy of analysis of samples from plasma C was

Figure 14 Figure 15COMPARISON OF ASSAY RESULTS FROM

SPIKED PLASMA SAMPLES AFTEREtOH PRECIPITATION OF PROTEIN

Percent of Analyses with

< 20% Error

PlasmaSeries

AnalyzedStandard Curve EmployedA B C D

A

B

C

D

Figure 14.Accuracy of analysis when extracts from oneplasma spiked with 9-THC are analyzed bymeans of standard curves generated fromextracts of other plasma samples. Antiserumfrom UCLA (Gross, et al., 1974)

STD. CURVES FROM SPIKED PLASMA SAMPLES

(EtOH PRECIPITATION: Et2O EXTRACTION;RECONSTITUTED IN BUFFER)

INITIAL SLOPE 50%BlNDlNG (LOGlT - LOG) INTERCEPT

Std Curve 35% -3.54 0.59 ng(Buffer)

+ 50µl extract A 32% -2.88 0.81 ng

+ 50µl extract A 32% -2.84 0.77 ng

Std. Curve 45% 4 . 0 6 0.59 ng

+ 50µl extract B 35% -3.50 0.71 ng

+ 50µl extract C 38% -2.36 0.71 ng

+ 50µl extract D 39% -2.49 0.68 ng

Figure 15.

CORRELATIONCOEFFICIENT

0 . 9 9

-0.98

-0.98

-0.99

-0.99

-0.96

-0.99

Comparison of standard curves generated inbuffer alone with those containing plasmaextracts. Results are from two separateexperiments. Duplicate extracts A shouldbe compared with the first standard curve;B, C, and D with the second. Antiserumfrom UCLA (Gross, et al., 1974)

quite poor when the other plasmas were usedas the basis of the standard curve. This isparticularly true in the case of plasma Canalyzed vs standard curve A and plasma Aanalyzed vs standard curve C, where only oneout of five samples fell within the 20%limits. If one requires accuracy within+10%, results are much worse.

We believe this study leads to two conclu-sions: (1) if one sets modest goals foraccuracy, a fair number of plasma samples canbe analyzed down to levels of 5 ng/ml of pure

9-THC. However, (2) the influence of plasmaextracts on the standard curve is not highlyreproducible from one plasma to another andcould, on occasion, lead to severe errors inthe analysis. This study also does not takeinto account the potential problems caused bycross-reaction of other cannabinoid materialswhich may be present in plasma at higher con-centrations than the 9-THC itself.

To further purify the plasma material beforeanalysis, we then followed a procedure whichwas described to us by Soares (private com-munication, 1975) and which consists of thelast part of Figure 13. The ethanol extractis treated with water and ether. After

equilibration, ether is removed and evapora-ted. The residue may be taken up either in50% aqueous ethanol or in RIA buffer. Sinceuse of the latter eliminates the effects ofethanol on the assay and does not result inlower recoveries, it is the procedure ofchoice.

Figure 15 shows the results of generatingstandard curves from plasma samples spikedwith 5-100 ng/ml of 9-THC and carriedthrough this procedure. Adding 50 µl (5% ofthe total extract from one ml) of extractfrom plasma A did result in a change in thestandard curve from one run in buffer alone.The initial binding was lowered, the slope ofthe logit-log plot was reduced, and the 50%binding intercept was increased. However,duplicate samples of the same plasma gaverather reproducible standard curves. Whenthree different plasma extracts were comparedthere were found to be some differences. Thisimplies that caution again should be used inanalyzing plasma samples from one subjectbased upon a standard curve generated fromthe plasma of a second subject. The use of apooled plasma for generation of standardcurves is probably the best compromise atpresent.

25

CONCLUSIONS (3)

The work we have done leads us to be-lieve that there are factors which canbe extracted from normal plasma underthe conditions described which will in-terfere with the assay. Since thesefactors have not yet been identified andmay vary from subject to subject andperhaps from extraction to extraction,extreme caution should be used in eval-uating the data obtained in such a man-ner. Care must also be taken in ob-taining plasma samples and storing themto avoid inadvertent addition of suchpotentially interfering materials asplasticizers, etc. The effect of freez-ing and storage on the stability of theTHC and on the formation of interferingsubstances needs to be further evaluated. (4)

(1)

( 2 ) The sensitivity of the assay to condi-tions such as buffer composition andplasma materials makes interlaboratorycomparisons difficult. Standardizedtechniques and materials are thereforea pressing need.

Radioimmunoassay has not yet lived upto its potential in the analysis ofcannabinoid substances. The sensiti-vity of the assay is not as great asone would expect based on other compoundssuch as the steroid hormones. Nor havethe high titer antisera of the type avail-able for the steroid hormones yet been re-ported for 9-THC. Since factors suchas titer and sensitivity can appar-ently be strongly influenced by theactual assay conditions, this suggeststhat further manipulation of conditions,the synthesis of even higher specificactivity radioligand, and the develop-ment of antisera to a variety of antigensand in a variety of animals are all validapproaches to further work.

The reasonable results obtained by im-munization of animals with a conjugateprepared from 5'-carboxy- 8-THC haveencouraged us to undertake the synthe-sis of similar antigens based on 5'-carboxy- 9-THC. Work along these linesis in progress and can be reported atfuture meetings.

REFERENCES

Abraham, G. E. Radioimmunoassay of steroidsin biological materials. Acta Endocr.,1974, Suppl. 183, 1-42.

Berson, A. S. and Yalow, R. S. Currentknowledge of basic concepts in immuno-logy and their clinical application. InR. A. Good and D. W. Fisher (Eds.),Immunobiology. Sinauer Associates, Stan-ford, CN, 1971: 287-293.

Chen, R. F. Removal of fatty acids fromserum albumin by charcoal treatment.J. Biol. Chem., 1967, 242: 173-181.

Christensen, H. D., Amerson, E., Myers,M. W., and Cook, C. E. Comparison of di-phenylhydantoin and phenobarbital serumand blood levels. Pharmacologist, 1974,16 (2): Abstract 215.

Cook, C. E., Kepler, J. A. and Christensen,H. D. Antiserum to diphenylhydantoin:Preparation and characterization. Chem.Path. Pharmacol., 1973, 5: 767-774.

Cook, C. E., Tallent, C. R., Kepler, J. A.,Amerson, E. W., and Christensen, H. D.Caffeine radioimmunoassay: Synthesis ofantigen and characterization of antibody.26th Southeastern Regional Meeting of theACS, 1974, Abstract No. 84.

Cook, C. E., Tallent, C. R., Amerson, E.,Christensen, H. D., Taylor, G. and Kepler,J. A. Phenylbutazone antiserum. Fed.Proc., 1975a, 34: Abstract 3059.

Cook, C. E., Amerson, E., Poole, W. K.,Lesser, P., and O'Tuama, L. Phenytoinand phenobarbital concentrations in salivaand plasma of man measured by radioimmuno-assay. Clin. Pharmacol. Therap., 1975b,18: 742-747.

Gross, S. J., Soares, J. R., Wong, S. L., andSchuster, R. E. Marihuana metabolitesmeasured by a radioimmnune technique.Nature, 1974, 252: 581-582.

26

Klausner, H. A., Wilcox, H. G., and Dingell,J. V. The use of zonal ultracentrifuga-tion in the investigation of the bindingo f 9-tetrahydrocannabinol by plasma lypo-proteins. Drug Metab. Disp., 1975, 3:314-319.

Landsteiner, K. The specificity of serologi-cal reactions, Revised Edition. DoverPress, Inc., New York, 1962.

Pitt, C. G., Hobbs, D. T., Schran, H., Twine,C. E., Jr., and Williams, D. L. The syn-thesis of deuterium, carbon-14 and car-rier-free tritium-labeled compounds. J.Label. Comp., 1975, : - .

Rodbard, D., Bridson, W. and Rayford, P. L.Rapid calculation of radioimmunoassayresults. J. Lab. Clin. Med., 1969, 74:770-781.

Skelly, D. S., Brown, L. P. and Besch, P. K.Radioimmunoassay. Clin. Chem., 1973, 19:146-186.

Spector, S., Berkowitz, B. Glynn, E. J. andPeskar, B. Antibodies to morphine, bar-biturates, and serotonin. Pharmacol.Revs., 1973, 25: 281-291.

Teale, J. D., Forman, E. J., King, L. J. andMarks, V. Production of antibodies totetrahydrocannabinol as the basis for itsradioimmunoassay. Nature, 1974, 249:154-155.

Teale, J. D., For-man, E. J., King, L. J.,Piall, E. M. and Marks, V. The develop-ment of a radioimmunoassay for cannabi-noids in blood and urine. J. Pharm.Pharmacol., 1975, 27: 465-472.

Tsui, P. T., Kelly, K. A., Ponpipom, M. M.,Strahilevitz, M. and Sehon, A. H. 9 -Tetrahydrocannabinol-protein conjugates.Can. J. Biochem., 1974, 52: 252-253.

Vaitukaitus, J., Robbins, J. B., Meschlay, E.and Ross, G. E. A method for producingspecific antisera with small doses ofimnunogen. J. Clin. Endocr., 1971, 33:988-991.

Wall, M. E. Recent advances in the chemistryand metabolism of the cannabinoids. InV. C. Runeckles (Ed.), Recent Advances inPhytochemistry, Vol. 9, Plenum, New York,1975: pp. 29-59.

2 7

DETERMINATION OF THC AND ITS METABOLITESBY EMIT® HOMOGENEOUS ENZYME IMMUNOASSAY:A SUMMARY REPORT

G.L. Rowley, Ph.D., T. A. Armstrong , C.P. Crowl W. M. Eimstad, W.M. Hu, Ph.D.,

J.K. Kam, R Rogers, Ph.D., R. C. Ronald, Ph.D., K. E. Rubenstein, Ph. D.,

B.G. Sheldon, E. F. Ullman, Ph. O.Syva Research InstitutePalo Alto, California

I N T R O D U C T I O N

We found that certain enzymes couldbe inhibited by antihapten antibodieswhen these enzymes were covalentlyattached to the corresponding hapten.In 1972 we introduced the first EMIT®homogeneous enzyme immunoassay basedon this principle (Rubenstein,Schneider, and Ullman, 1972). Sincethen we have demonstrated the general-ity of the technique and have devel-oped immunoassays based on threeenzymes, lysozyme (Schneider,Linquist, Wong, Rubenstein, andUllman, 1973), malate dehydrogenase(Ullman, Blakemore, Leute, Eimstadand Jaklitsch, 1975), and glucose-6-phosphate dehydrogenase (Chang, Crow1and Schneider, 1975). The changeof enzyme activity of hapten-enzymeconjugates produced upon bindingantihapten antibodies permits directmeasurement of the amount of antibodybound to the conjugates. Enzymeimmunoassays based on this principleare classified as "homogeneous"immunochemical techniques sinceseparation of free from bound hapten

is not required. Separation isrequired when the signal of thelabeled hapten is not changed byantibody binding as is the case inradioimmunoassay and enzyme-linkedimmunosorbent assay (ELISA) which aretermed "heterogeneous" immunochemicaltechniques. The observed changeof activity in homogeneous enzymeimmunoassays is reminescent of inhi-bition or activation produced whencertain antibodies to enzymes bind totheir respective enzymes (Arnon,1973). The mode of action may besimilar. Either the antigenic determ-inant is an integral part of theenzyme surface or it is attachedcovalently to the surface.

Malate dehydrogenase (MDH) was chosenfor development of a THC assay becauseof its high stability, its avail-ability and its ease of detection bya simple highly sensitive assay. Aslittle as 10-11 molar enzyme can bedetected in a one minute spectrophoto-metric measurement. The mechanism of

28

antibody inhibition of morphine con-jugates of MDH had previously beenstudied and an immunoassay for mor-phine was developed (Rowley,Rubenstein, Huisjen, and Ullman, 1975)which permitted detection of 2x10-9

molar morphine in the assay mixture.

SYNTHESIS OF THE THC HAPTEN

Synthesis of the THC hapten, III,that was used for producing the com-ponents of the assay is illustratedbelow:

1) O3,-78°

2) Zn/HOAc

9,11-THC, I, was treated with oneequivalent of ozone at low temperatureto yield an ozonide which was reducedwith zinc in acetic acid to yieldketone II in 55% yield. Compound IIwas treated with 1-C14-O-carboxy-methylhydroxylamine in anhydrousrefluxing methanol to yield thedesired acid, III, in 61% yield.

PREPARATION OF THC ANTIBODY

Acid III was conjugated to bovineglobulin via its N-hydroxysuccinimideester, IV, prepared by condensationwith N-hydroxysuccinimide using thecondensing reagent, N-dimethylamino-propyl-N'-ethyl carbodiimide hydro-chloride in anhydrous dimethylforma-mide.

Subsequently, the dimethylformamidesolution of "active ester" was addedto a buffered solution of the proteincontaining 30% dimethylformamide co-solvent at pH 8.5. The protein wasexhaustively dialyzed to remove non-

covalently bound THC residues. Thenumber of bound residues was deter-mined by scintillation counting of theradiolabeled products to yield conju-gates containing 21 and 32 THC resi-dues per protein molecule in twoseparate preparations. Antibodieswere obtained from sheep by immuniz-ing with the conjugates.

PREPARATION OF THC-MDH CONJUGATES

Acid III was conjugated to MDH via IVunder almost identical conditions. Aseries of THC-MDH conjugates were thusprepared by adding increasing amountsof IV to constant amounts of MDH inseparate reaction vessels. Non-covalently bound THC residues wereremoved either by exhaustive dialysisor by gel chromatography on SephadexG-25. The number of bound radio-labeled THC residues and the residualenzyme activity of each conjugate wasdetermined (Figure 1). Enzymeactivity of the conjugates decreasedsharply with increasing sub-stitution up to about 5 residuesbound (20% activity). Substitutionbeyond 5 residues provided a conjugate(12.2 residues) with only a moderatefurther loss of activity (8% activity).

Figure 1

Addition of excess THC antibody tothe conjugates reduced their enzymaticactivity. The maximum inhibition inthe presence of excess antibody wasonly slightly affected by the animalsource. Maximal inhibition was,however, directly dependent on the

29

number of bound THC residues (Figure2). It increased sharplywith the number of THC groups on theenzyme and reached a maximum when atabout 4.2 residues (51% inhibition).Upon substituting the enzyme moreheavily (12.2 residues) the antibodyinduced inhibition decreased to 35%.

Figure 2

IMMUNOASSAY FOR THC AND ll-NOR- 9-THC-9-CARBOXYLIC ACID

THC-MDH conjugate, which was maximallyinhibited by THC antibodies (4.2 THCresidues), was chosen for the immuno-assay. Stability of the conjugate wasfound to be very sensitive to pH andionic strength. The conditionsselected to store working solutions ofenzyme, 0.50 M potassium phosphate,pH 7.4 , provide room temperaturestability of greater than 34 days.

The conjugate readily adsorbed onglass and plastic measuring devicesthus preventing quantitative andreproducible transfer of dilute solu-tions. Such adsorption is reminis-cent of reported adsorption of THCitself to glass and plastics (Garrettand Hunt, 1974). Triton X-405 wasfound to prevent troublesome adsorp-tion of THC in radioimmunoassays forTHC (Teale, Forman, Ring, Piall, andMarks, 1975). We likewise found thatadsorption of THC-MDH conjugate wasprevented by incorporation of 0.10%Triton X-405 in dilute enzyme solu-tions.

The effect of varying the ratio ofantibody to enzyme is given in Figure3. Upon titration of the conjugate

Figure 3

30

with antibody, the enzyme activity atfirst dropped off rapidly and assymp-totically approached a limiting value.An antibody to enzyme ratio was chosenfor a THC assay where the enzyme wasinhibited 39% (arrow, Figure 3). Adrug-response curve was obtained atthat ratio by adding standard ethan-olic THC solutions to the assaymixture (Figure 4). In the assayprocedure 1 µl ethanolic THC standardsolutions were combined with solutionscontaining antibody and substratesfollowed by addition of the THC-MDHconjugate. Enzyme rate measurementswere made over 3 minute periods. Thelowest detectable level of THC in theassay mixture was 0.5 ng/ml (1.6 x10-9 M). The total useable range was0.5 ng - 10 ng/ml.

Likewise a drug-response curve for11-nor- 9-THC-9-carboxylic acid, Vwas obtained using standard ethanolicsolutions in an identical assay withthe same antibody (Figure 4).

The useable range of detection wasthe same as that for THC, 0.5 ng -10 ng/ml. However, response to V inmid-range was about 1.5 times higherthan response to THC. The cross-reactive behavior of our antibodyboth to THC and to V is differentfrom that exhibited by antibodiesraised by others against O-carboxy-methyl-THC (Van Vunakis and Levine,1974), 2- or 4- (p-carboxyphenylazo)THC, (Grant, Gross, Lomax, and Wong,1972), and O-succinyl-THC (Teale,Forman, King, Piall and Marks, 1975)conjugates. All of these producedantibodies that reacted only poorlywith V. Our antisera, therefore,appear suited for general screening

Figure 4

31

for THC and its metabolites, and Enzyme immunoassays for THC and itscould also be used to assay either metabolites in urine are presentlyTHC or V if these compounds are first being developed and a serum assay isseparated by an extraction procedure. under investigation.

REFERENCES

Arnon, R. (1973) in The Antigens(Sela, M., ed.) Vol. 1 pp. 87-159,Academic, New York, N.Y.

Chang, J. J., Crowl, C. P., andSchneider, R. S. (1975) Clin. Chem.21, 967.

Garrett, E. R., and Hunt, C. A.(1974) J. Pharm. Sci. 63, 1056.

Grant, J. D., Gross, S. J., Lomax, P.,and Wong, R. (1972) Nature, New Biol.236, 216.

Rowley, G. L. Rubenstein, K. E.,Huisjen, J., and Ullman, E. F. (1975)J. Biol. Chem., 250, 3759.

Schneider, R. S., Linquist, P., Wong,E., Rubenstein, K. E. and Ullman,E. F. (1973) Clin. Chem. 19, 821.

Teale, J. D., Forman, E. I., King,L. J., Piall, E. M., and Marks, V.(1975) J, Pharm. Pharmac. 27 465.

Ullman, E. F., Blakemore, J., Leute,R. K., Eimstad, W., and Jaklitsch, A.(1975) Clin. Chem. 21, 1011.

Van Vunakis, H., and Levine, L.(1974) in Immunoassays for DrugsSubject to Abuse (Mule, S. J.,Sunshine, I., Brande, M., andWillette, R. E., eds.) pp. 23-25,CRC Press Inc., Cleveland, Ohio.

Rubenstein, K. E., Schneider, R. S.,and Ullman, E. F. (1972) Biochem.Biophys. Res. Comm. 47, 846.

32

SEPARATION AND SENSlTlVE ASSAY OF THCIN BIOLOGICAL FLUIDS BY HPLC AND GLC

Edward R. Garrett and C. Anthon HuntCollege of Pharmacy, University of FloridaGainesville, Florida

ABSTRACT

HPLC systems were developed to permit quanti-tative separation of 9-tetrahydrocannabinolfrom many of the heptane extractable lipoidaland other endogenous substances in biologicalfluids. These substances interfered with thequantification by flame ionization GLC of un-modified compound and by electron capture GLCof pentafluorobenzoylated compound. Reversephase HPLC elution, with 47% acetonitrile inwater, and normal phase HPLC with 25% chloro-form in heptane, separated tetrahydrocannabinolfrom 11-hydroxy- 9-tetrahydrocannabinol andother monohydroxylated tetrahydrocannabinols.These systems also purified stock solutionsof tetrahydrocannabinol from accompanying con-taminants. The various monohydroxylatedtetrahydrocannabinols were resolved from eachother in normal phase, 80% chloroform in hep-tane. The 8 and 9-tetrahydrocannabinols wereseparable in normal phase with 5% tetrahydro-furan in hexane. The GLC analysis of penta-fluorobenzoylated tetrahydrocannabinol had asensitivity of 1 ng/ml of plasma with an esti-mated 5% standard error of an assay with theextraction and GLC procedures given herein.Radiochemical analysis of the HPLC separatedfraction had a sensitivity of 0.2 ng/ml ofplasma with an estimated 2% standard error ofan assay. There was no significant difference

between the liquid scintillation and electroncapture GLC assays of the HPLC separated 9 -tetrahydrocannabinol obtained from the plasmaof dogs administered the drug. Radiolabelledcompounds can be added to pasma samples asinternal standards to determine the recoveryefficiencies of the several procedures in theanalysis of unlabelled tetrahydrocannabinol.

INTRODUCTION

A GLC analytical procedure for tetrahydro-cannabinol in biological fluids was developedpreviously (Garrett and Hunt 1973) and usedelectron capture detection of the derivedpentafluorobenzoylated tetrahydrocannabinol.It could readily detect 0.5 ng of tetrahydro-cannabinol added to a 5.0 ml blood samplefrom a fasting dog. This sensitivity wasonly obtained when it was realized thattetrahydrocannabinol bound extensively (15%-40%) to glass (Garrett and Hunt 1973;Garrett and Hunt 1974) and rubber stoppers(Garrett and Hunt 1974) and that the time-dependent degree of adsorption could be mini-mized by prior treatment of all glasswarewith an organic solution of a silyl reagent.In the case of organic solutions, the tetra-

33

hydrocannabinol could be reincorporated fromthe glass into solution on vigorous shakingprior to any sampling.

The method’s validity was demonstrated inthe fasting dog with a low fat diet. Plasmalevels down to 1 ng/ml of blood from 5 mlblood samples were monitored for 12 hoursafter the administration of 0.1 mg of puretetrahydrocannabinol per kg (Garrett andHunt 1973). However when the same methodwas applied to non-fasting animals a signif-icant increase in GLC background from inter-facing plasma constituents was observed,particularly within 4 hours of feeding apreviously fasted animal. The resultantminimal detectable quantity unfortunatelyincreased to 5 - 10 ng/ml when a 5 ml bloodsample was taken.

Since phannacokinetic studies were contem-plated in both dogs and humans over longerperiods of time so that fasting would beimpractical, such interferences were antici-pated that would lower the analytical sensi-tivity. Thus it was necessary to devisesuitable separation and clean-up proceduresprior to analysis to improve the sensitivity.

High pressure liquid chromatography (HPLC)provided a powerful method of separation ofdrugs from their potential metabolites andfrom endogenous substances in biologicalfluids. While the classical on-line moni-toring devices such as refractive index orUV spectrophotometry were too insensitive fordirect assay at the plasma levels anticipated,the separated collected pertinent fractionswere analyzed by analytical methods that pro-vided the proper sensitivities. This paperpresents HPLC techniques to separate tetra-hydrocannabinol from various cannabinoidsand endogenous materials of biological fluidswith subsequent analysis by various appro-priate methods.

RESULTS AND DISCUSSION

Purification and Reproducibility of Collectionfrom Injected Ethanolic Solutions of 9 -Tetra-hydrocannabinol on HPLC

The 14C- 9 -tetrahydrocannabinol used assupplied by NIDA was reported low in radio-labelled contaminants by TLC. However,reverse phase HPLC (Fig. 1) revealed thepresence of a labelled contaminant. Themajor peak, I, contained 9 -tetrahydrocan-nabinol ( > 96%) and 8 -tetrahydrocannabi-no1 ( 3%), quantified by flame ionization

Figure 1

Figure 1. Reverse phase HPLC of stock 14C-9-tetrahydrocannabinol. The total radio-

activity per eluate fraction is plotted vsretention volume (curve A). Peak I was 14C-

9 -tetrahydrocannabinol and peak II was an un-known radioactive contaminent. Curve B is aplot of the mean UV detector response (arbit-rary units) vs retention volume and is displac-ed relative to A by the solvent volume betweenthe detector and the collection point. Collec-tion between volumes a and b (peak retentionvolume ± 30%) for curve B recovered 92.5% ofthe radioactivity in this case. Column: Bond-apack C18; eluent: 45% acetonitrile in waterat 1.5 ml/min. Each fraction was correctedfor background CPM.

GLC (Garrett and Hunt 1974; Garrett and Tsau1974). A labelled contaminant was alsofound (Fig. 2) with a normal phase column.The contaminant eluted prior to tetrahydro-cannabinol in both cases. Since the morepolar compounds elute first on reverse phaseHPLC whereas the least polar compounds elutefirst on normal phase HPLC, it suggests thatthe contaminants observed in the two systemswere not the same. The contaminants werenot analyzed further.

34

Figure 2 Figure 3

Figure 2. Normal phase HPLC of stock 14C- 9-tetrahydrocannabinol. The total activity pereluate fraction is plotted vs milliliters ofeluate. Peak I was 14C- 9 - tetrahydrocannabi-nol and peak II was an unknown radioactivecontaminant. Collection between volumes a andb recovered 97.5% of the radioactivity in thiscase. Column: µ-Porsil; eluent: 20% chloro-form in heptane at 2.5 ml/min. Each fractionwas corrected for background CPM.

When the tetrahydrocannabinol under peak I(Figs. 1 and 2) was collected, dried, recon-stituted in ethanol and re-analyzed on thesame HPLC system, only the single peak I wasobserved. All the stock 9-tetrahydrocan-nabinol used herein underwent this purifica-tion procedure. The percent of the totalinjected radioactivity under peak I recoveredwas 97.6 ± 0.7 (standard error of mean) %(Table I) in the reverse phase system for theranges of volumes collected (Figs. 1 and 3)and was 96.3 ± 1.0 (standard error of mean) %(Table II) for 1 to 1000 ng injected in thenormal phase system at a retention volume of9.7 ml within the collection range of 7.3 -16.2 ml (Fig. 2).

Figure 3. The retention volumes for the peakamounts of 9 -tetrahydrocannabinol and11-hydroxy- 9 -tetrahydrocannabinol forreverse phase HPLC are plotted vs soluent com-position. Each point represents the mean peakretention volume for two determinations. Thevertical bars represent the ranges of retentionvolumes which contained approximately 98% ofthe area under the plot of recovered radio-activity vs retention volume.

The fact that the retention volume of peak Iand the appropriate collection ranges for98% recovery of injected labelled tetrahydro-cannabinol were sensitive to solvent composi-tion (Fig. 3) necessitated the prior estab-lishment of an appropriate collection rangefor each newly prepared batch of solvent byradiochemical analysis of the collected HPLCfractions of previously purified 14C- 9

-tetrabydrocannabinol. The retention volumeand the appropriate collection range increasedwith decreasing solvent polarity (less %acetonitrile) due to peak spreading (Fig. 3).

35

Table 1 Table 2

TABLE I - Recovery of Radioactivity of 14C- Terahydrocannabinola from Reverse Phase HPLC

TABLE II - Recovery of Radioactivity of 14C- Terahydrocannabinola

from Normal Phase HPLCb

The quality of the water used in the elutingsolvent acetonitrile-water in reverse phaseHPLC was an important factor in maintainingthe reproducibility of the percent radio-activity recovered for a given collectionrange. The UV detector clearly indicatedthat adsorbed contaminants in impure watercould be eluted from a previously used columnby 100% acetonitrile, a less polar solventthan the used mixed eluent. Additional evi-dence of these contaminants was demonstratedwhen the GLC background significantly variedwhen distilled water from various sourceswas collected after reverse phase chroma-tography, extracted as if 9-tetrahydrocan-nabinol were present, then treated by thederivatization procedure and analyzed byelectron capture GLC.

HPLC Separation of 9 -Tetrahydrocannabinolfrom Selected Cannabinoids and Metabolites

Tetrahydrocannabinol, cannabinol and cannabi-diol were resolved by the normal phase HPLCsystem (Fig. 4). If 98% of the tetrahydro-cannabinol were to be collected after normalphase HPLC in this system, it is apparentthat the chosen range (Fig. 2) would alsocollect cannabinol and cannabidiol. Theretention volumes of cannabinol and cannabi-

diol relative to tetrahydrocannabinol in-creased as the percent of chloroform inheptane increased.

‘The monohydroxylated metabolites had largeretention volumes (> 15 ml) on the normalphase column when 20 - 25% chloroform inheptane was the solvent (Fig. 2) and couldbe completely separated from tetrahydrocan-nabinol on this system. They were resolvedfrom each other with a more polar solvent,80% chloroform in heptane (Fig. 5).

9-Tetrahydrocannabinol and 11-hydroxy-9-tetrahydrocannabinol were quantitatively

separable on the reverse phase HPLC systemat 47% (or less) acetonitrile in water(Fig. 3). The collection efficiencies in theranges given were 98% of the recoverableradioactivities of 3H-11-hydroxy- 9-tetra-hydrocannabinol and 14C- 9-tetrahydrocan-nabinol.

8 and 9 -Tetrahydrocannabinols were notreadily resolvable in any of these systems.However an HPLC system that readily resolvedand separated these two compounds was 5% THFin hexane on the normal phase column at 0.5ml/min with retention volumes of 8.15 and8.45 ml, respectively. The tetrahydrocan-nabinols collected under peak I (Figs. 1 and2) could be further separated by this system.

36

Figure 4 Figure 5

Figure 4. Normal phase HPLC separation ofcannabinoids. The UV detector response (arbi-trary units) is plotted vs retention time afterinjection (inj.) of a mixture of cannabidiol(CBD), 9 -tetrahydrocannabinol (THC) and canna-binol (CBN). The amounts of 9 -tetrahydro-cannabinol and cannabidiol injected were approx-imately twice that of the cannabinol. Column:µ-Porsil; eluent: 20% chloroform in heptaneat 1.5 ml/min.

Figure 5. Normal phase HPLC separation of 9 -tetrahydrocannabinol metabolites. The UV de-tector response (arbitrary units) is plottedvs retention time after injection (inj.) of amixture of cannabinol (CBN), 11-hydroxy- 9 -tetrahydrocannabinol (C), 8 - and 8 -hydroxy-

9-tetrahydrocannabinol (B and A, respectively).Column: µ-Porsil; eluent: 80% chloroform inheptane at 1.5 ml/min.

Table 3TABLE III Percent Recoveriesa of 14C- 9-Tetrahydrocannabinol in the Heptane Extract of Plasma, in

the Collection of the Proper Fraction of the Heptane Extract Separated by HPLC and theOverall Recovery after Both Extraction and HPLC Collection as Monitored by Both ElectronCapture-GLC and Scintillation Analysis

14C- 9-Tetrahydrocannabinolb,ng/2 ml plasma

2.25

Extracted by Heptane, %

Scintillation

90.2 ± 4.7

Extract HPLC Overall Recoveryd, %Collectedc, %

Scintillation GLC Scintillation GLC

93 92 84 ± 8 83 ± 32

22.5 90.8 ± 2.6 88 97 80 ± 6 88 ± 12

225 90.9 ± 8.5 97 89 8.8 ± 4 78 1 ±

Overall Average ± std. error 90.6 ± 0.7 92 93 84 + ± 2 84 ± 5

aGiven as the mean from 4 separate plasma samples ± standard deviation. The scintillation analysis of recovered radioactivitywas performed on a different set of four studies and on a different day than the electron capture GLC analysis of the derivatizedHPLC collection. b142.47 CPM/ng. Two ml of plasma was extracted with 15 ml heptane and 14 ml was analyzed for total 14C.cQuotient divided by the extraction efficiency (0.91). Additional studies were conducted for 26 ng 14C- 9-tetrahydrocannabinolper 2 ml plasma on two other day for 4 samples each and the % recovered on HPLC from the extract were 94 ± 4 and 92 ± 3%(standard error) respectively. dQuotient of amount recovered corrected for volumes of extract used and amount added, which isthe product of the extraction efficiency and the collection efficiency.

37

Effect of HPLC Separation on GLC Analysis of9 Tetrahydrocannabinol in Plasma

An equal amount of 9 tetrahydrocannabinoland 11-hydroxy- 9-tetrahydrocannabinol in2 ml of dog plasma was extracted and sepa-rated from a majority of the extracted com-ponents by reverse phase HPLC. The reduc-tion in potential contaminants from plasmaobservable on GLC was demonstrated by flameionization GLC analysis (Garrett and Hunt1974; Garrett and Tsau 1974) both before andafter HPLC treatment (Fig. 6).

The normal phase HPLC with 20% chloroform inheptane could separate 9-tetrahydrocannabi-nol from monohydroxylated metabolites andfrom 11-hydroxy- 9- t e t r a h y d r o c a n n a b i n o l .

However, a minor overlap could be avoided bycollecting the tetrahydrocannabinol in aslightly narrower volume range. The priorheptane extraction of alkalinized plasmahad separated these non-polar constituentsfrom any acidic metabolite. This separationof plasma extracts and normal phase HPLCcollection of volumes in the appropriaterange resulted in a substantial, reduction inGLC background from plasma components forderivatized tetrahydrocannabinol analyzedwith electron capture detection as shown bythe comparison of curves A and B or A and Cin Fig. 7.

Figure 6

Figure 6. GLC chromatograms (flame ionizationdetection) of an extract of 2 ml of plasmacontaining 9 -tetrahydrocannabinol, I (200ng/ml), before (A) and after (B) reverse phaseHPLC separation of both cannabinoids over arange of predetermined collection volumes(Fig. 1). One µl of 18 µl of extract was in-jected into the GLC prior to HPLC (A). Tenµl of 18 µl of the extract was injected intothe HPLC, the collected fraction was reconsti-tuted in 10 µl of chloroform and 1 µl was in-jected into the GLC (B). Peak III is the sol-vent; peak IV is an unknown from plasma.HPLC: column, Corasil C18; eluent, 52% aceto-nitrile in water at 1.5 ml/min. GLC: column,5' x 2 mm 1.9% OV-225 at 245° with a N2 flowof 24 ml/min. and an attenuation of 8 x 12- 1 2

for both chromatograms. The initial baselines and injection times for both chromato-grams are superimposed for comparison.

Figure 7

Figure 7. GLC (electron capture) analysis ofderivatized samples. The chromatograms repre-sent: A) the injection of 1 of 300 µl from aderivatized extract of 2 ml of blank plasmawithout HPLC purification: B) the injectionof 1 of 200 µl of the final solution of deriv-atized compound which was 70% of the totalamount from 2 ml of dog plasma containing 200ng/ml 9 tetrahydrocannibinol pentafluoroben-zoate (THC) are appropriately labelled. Typi-cal estimated base lines are shown. GLC con-ditions: 6' OV 17 column, 225°; detector,280°; injector, 255°; N2 flow, 45 ml/min;attenuations 4 x 10-9 (A and B) and 8 x 10- l 0

(C).

38

Plasma samples obtained from dogs administer-ed 9-tetrahydrocannabinol solutions intra-venously were analyzed by the electron cap-ture GLC in accordance with the modifiedprocedures described herein that includedextraction, normal phase HPLC separation andderivatization except that no internalstandard was added. These procedures wouldhave included any cannabinol or cannabidiolin the HPLC collection volume range (Fig. 2and 4) used. However, no peaks were seen atthe retention times of cannabinol or cannabi-diol pentafluorbenzoate and no significantamounts of cannabinol or cannabidiol couldbe detected as metabolites of 9-tetra-hydrocannabinol in the dog. Thus, eithercompound, when purified, should serve as anappropriate internal standard in pharmaco-kinetic studies. However, since cannabinolhas been reported as a minor metabolite(McCallum 1973; McCallum et al. 1975;Widman et al. 1974), cannabidiol pentafluro-benzoate was chosen as the internal standard.It must be realized that cannabinol is knownto be a contaminent of degraded 9 tetrahy-drocannabinol (Garrett and Tsau 1974).

The GLC methodology presented herein differedfrom the prior studies (Garrett and Hunt1973) in that the short 30 cm column of 3%OV 225 was supplanted by a longer OV 17packed column to be consistent with the datain the literature accumulated for the reso-I.ution of the cannabinols (Fetter-man andTurner 1972; Turner and Hadley 1973; Turneret al. 1974).

Efficiency, Reproducibility and Sensitivityof the Steps in the Radiochemical and ElectronCapture Assay of l4C- 9 -Tetrahydrocannabinolin Dog Plasma

The heptane extraction efficiency from plasmawas highly reproducible (Table III) over awide range of plasma concentrations, 90.6 ±0.7% standard error of mean.

The recovery of 9-tetrahydrocannabinol fromthe heptane extract of dog plasma by thenormal phase HPLC was reproducible (TableIII) over the range of plasma concentrationsstudied. Equivalent overall recoveries(Table II) were obtained by both radiochem-ical analysis (83.7 ± 1.8% standard error)and by electron capture GLC analysis (84.0 ±4.9% standard error) of the derivatizedtetrahydrocannabinol (Fig. 7). Both methodspermitted estimation of a 92.5% recoveryof the amount in the heptane extract injectedon the normal phase HPLC and collected in the

chosen range. The normal HPLC collectionrange was chosen to be slightly smaller thanin the studies on ethanolic solutions oftetrahydrocannabinol (Fig. 2, Table II) sincethe procedure had to be modified to separatetetrahydrocannabinol from possible monohydro-xylated or 11-hydroxy- 9-tetrahydrocannabi-no1 metabolites in plasma. A slight overlapof their HPLC peak areas with tetrahydrocan-nabinol would have occurred if the largercollection ranges had been used.

A similar study of the reproducibility ofcollection of 14C- 9-tetrahydrocannabinol inplasma assayed by liquid scintillation afterextraction and reverse phase HPLC was alsoconducted. The amounts recovered were pro-portional to the amounts injected (Fig. 8)and the HPLC recovery efficiency of the drugin the heptane extract in this case was 95.7%.

Figure 8

Figure 8. Recovery of 14C- 9 -tetrahydrocanna-binol calculated as concentration in plasmacorrected for an extraction efficiency of 91%against the experimentally prepared concentra-tions. The drug extracted from plasma andcollected over the proper volume range (TableI and Fig. 1) after reverse phase HPLC with a45% acetonitrile in water eluent. The collec-tion was analyzed by liquid scintillation. Theslope, 0.957, is the HPLC collection efficiencyf o r 9 -tetrahydrocannabinol and was typical ofthe values obtained. The vertical bar is therange for ± one standard deviation (n=4).

39

Equivalency of Radiochemical Analyses of 14C-9 -Tetrahydrocannabinol and GLC Electron

Capture Detection of Derivatized Materialafter Normal Phase HPLC in Dog Plasma duringPharmacokinetic Studies

The plasma of a do intravenously administeredsolutions of 14C- 9 - tetrahydrocannabinol wasmonitored with time after heptane extractionby both radiochemical analysis and electroncapture GLC of the derivative of the appro-priately collected eluate fraction fromnormal phase HPLC. Typical plots of thetime course of the results from both methodsare given in Figs. 9 and 10.

Figure 9

Figure 9. Semilogarithmic plots of fractionof the 9 -tetrahydrocannabinol 0.1 mg/kg doseper ml of plasma against time for dog A plottedfrom the liquid scintillation analysis of thetotal 14C collected as 9 -tetrahydrocannabinolon normal phase HPLC and from the electroncapture GLC of the derivatized HPLC collectedfraction . The values were corrected forthe fractions of extracts and total collectionrange used.

Figure 10

Figure 10. Semilogarithmic plots of fractionof the 9 -tetrahydrocannabinol 2.0 mg/kg doseper ml of plasma against time for dog A plottedfrom the liquid scintillation analysis of thetotal 14C collected as 9-tetrahydrocannabinolon normal phase HPLC and from the elctroncapture GLC of the derivatized HPLC collectedfraction . The values were corrected forthe fractions of extracts and total collectionrange used.

The procedures for GLC analysis reportedherein gave a lower limit for quantitativeanalysis of tetrahydrocannabinol in plasmaof approximately 1 ng/ml from twice thestandard deviation (0.32 ng) obtained forthe amount of tetrahydrocannabinol recoveredfrom 2.25 ng in 2 ml plasma (Table II).Similarly, the procedure for radiochemicalanalysis reported herein gave a lower limitof approximately 0.2 ng/ml from twice thestandard deviation (0.084 ng). A statisti-cal analysis of the apparent differencesbetween the tetrahydrocannabinol assays ata given time from both analytical methods(Figs. 9 and 10) showed no significance.

40

This demonstrated that all of the recoveredradioactivity from the HPLC separation pro-cedure could be assigned to the 9-tetra-hydrocannabinol assayed specifically byelectron capture GLC and thus no significantamounts of radiolabelled metabolites werein the collected HRLC fractions.

It can be concluded that 9 -tetrahydrocan-nabinol can be extracted from plasma andother biological fluids, that it can beseparated on HPLC from the simultaneouslyextracted biologically endogenous materialsand metabolites that would interfere with achosen highly sensitive analytical methodsuch as GLC. It is not necessary to collectall of the material to be analyzed; assur-ance that a reproducible or known fractionof the total material injected on HPLC is

REFERENCES

E. R. Garrett and C. A. Hunt, J. Pharm. Sci.,1211 (1973).

E. R. Garrett and C. A. Hunt, J. Pharm. Sci.,63, 1056 (1974).

E. R. Garrett and J. Tsau, J. Pharm. Sci.,63, 1563 (1974).

N. K. McCallum, J. Chrom. Sci., 11, 509(1973).

N. K. McCallum, B. Yagen, S. Levy andR. Mechoulam, Experientia, 31, (5), 520(1975).

M. Widman, M. Nordquist, S. Agurell,J. E. Lindgren and F. Sandberg, Biochem.Pharmacol., 23, 163 (1974).

R. S. Fetterman and C. E. Turner, J. Pharm.Sci., 61, 1476 (1972).

C. E. Turner and K. Hadley, ibid, 62, 251,1083 (1973).

C. E. Turner et al., ibid, 63, 1872 (1974).

recovered is all that is necessary since itis directly proportional to the total drugconcentration. If unlabelled tetrahydrocan-nabinol in a solution of plasma were ana-lyzed, the calculated recovery of knownamounts of labelled tetrahydrocannabinoladded either to plasma prior to extractionor to heptane extract subsequent to extrac-tion would permit calculation of the extrac-tion and/or HPLC collection efficiencies forthat particular biological sample. Theseknown efficiencies would permit the calcula-tion of the original plasma concentrations.If a labelled 14C- 9 -tetrahydrocannabinolwere used in pharmacokinetic studies,extracted and separated on the HPLC, a trit-ium labelled 3H-tetrahydrocannabinol could beused as the appropriate internal standard tomonitor the recovery efficiencies.

ACKNOWLEDGMENTS

Supported in part by Grant DA-00743from the National Institute on Drug Abuse,Rockville, MD 20852

The technical assistance ofMrs. Kathleen Eberst and the contributionsof Hendryk Roseboom are gratefully acknowl-edged.

43

DETERMINATION OF ETRAHYDROCANNABINOLIN HUMAN BLOOD SERUM BYELECTRON CAPTURE GAS CHROMATOGRAPHY

David C. Fenimore Ph.D. Chester M. Davis, Ph.D., Alec H. HornTexas Research Institute of Mental SciencesHouston, Texas

The application of electron capture gaschromatography (ECGC) to the determinationo f 9-tetrahydrocannabinol in blood sampleshas been reported by Garrett and Hunt(1973) and Fenimore, Freeman, and Loy(1973). In these studies laboratory animalswere used as experimental subjects which,as later experience demonstrated to thepresent authors, was a fortunate choice.When attempts were made to extend thetechnique to samples of blood from humansubjects with their more varied diet andenvironment, the presence of interferingcomponents proved to be a serious problem.Had this level of difficulty occurredduring the initial development of themethodology, other less sensitive butmore tractable alternatives would surelyhave been pursued. The initial success withanimal subjects, however, encouraged modi-fications of the method which now permitquantitation of 9-THC in human blood atconcentrations approaching that attainablewith the animal subjects.

Although the cannabinoids do not possesshigh affinities for thermal electrons, itis a relatively simple matter to form poly-halogenated derivatives of these compounds

in quantitative yield to which the electroncapture detector responds with excellentsensitivity. The limit of detection for theheptafluorobutyrate or pentafluoropropio-nate esters of 9-THC is approximately onepicogram (Fenimore, 1973). Because thelevels of 9-THC in the blood are about1000 times this amount per milliliter fora considerable period after a nominallyactive dose of cannabis preparation(Lemberger 1971, Agurell 1973, Garrett 1973),it is obvious that sensitivity of detectionis not the limiting factor. The problemsarise from the fact that the derivatizingreactions are relatively non-specific. Notonly are there a large number of componentspresent in biological samples at parts perbillion concentrations capable of formingderivatives of this type, but also manysimilarly reactive species may be carriedthrough the initial sample preparation inamounts orders of magnitude greater thanthe compound or compounds of interest. Itis apparent, then, that selectivity becomesall important if the sensitivity of thedetermination is to be maintained, and ifthis selectivity cannot be achieved at thedetector, it must be restored through thechromatographic process, the sample prepa-ration, or both.

4 2

One approach toward eliminating the inter-fering components is to achieve a relativelypure isolate prior to gas chromatographicanalysis through sample preparation in-volving other chromatographic separationsor through extensive partitioning processes.Experience has shown, however, that samplerecovery is not favored by manipulationsof this kind. Consequently we have reliedprimarily on improved gas chromatographictechnique to alleviate the problem ofinterference. A dual column-dual oveninstrument described previously (Fenimore,1973) and discussed in some detail in thefollowing section has provided the neces-sary resolution to enable the determinationso f 9-THC in human blood serum.

DESCRIPTION OF THE GAS CHROMATOGRAPHICAPPARATUS

A schematic diagram of the dual column-dualoven gas chronatograph is shown in Figure 1.Oven No 1, containing a conventional packedcolumn, is temperature programmed whileoven No 2, equipped with a capillarycolumn, remains isothermal. A switchingvalve and cold traps are arranged to permittransfer of only that volume of effluentcontaining the compound or compounds beingmeasured from the packed column to thecapillary column. The exact retentionvolume necessary for accurate transfer isdetermined previously by detection ofrelatively large amounts of compound (approx.

Figure 1

1 µg) by flame ionization. The transferredeffluent is further separated by thecapillary column and detected by a micro-volume electron capture detector (Fenimore,1971).

This system provides the means to elimi-nate three major problems encounteredin high resolution ECGC. First, temper-ature programming, which is usually notfeasible with EC procedures because of

exaggerated baseline drift, is accomplishedby raising the temperature of the firstcolumn while holding the second column ata constant temperature. Second, the packedcolumn tolerates large injection volumeswhich would rapidly degrade a small borecapillary column. Third, the two columnsmay contain stationary phases of widelydifferent polarity which increases theprobability of separation of compoundshaving similar partition coefficientson any single chromatographic column.

43

The problem of transferring the solutefrom the high volume mobile phase eluateof the packed column to the much smallervolume of the capillary column is solvedby interposing two water-cooled trapsbetween the columns. The first trap con-sists of 1/8'' ID nickel tubing containinga 3 cm plug of material identical to thepacking of the fore-column, and the secondis short length of 1/16'' ID nickel capillarycoated with the stationary phase used inthe second column. Water is allowed toflow through an annular space surroundingthe traps immediately before and duringthe transfer interval. When the temper-ature of the first trap is restored tooven temperature by interrupting the flowof coolant, the solute is released at thereduced carrier gas flow rate of thecapillary column. The band broadeningintroduced by the volume of the firsttrap is reduced by re-trapping the solutein the capillary tubing and then releasingit to the capillary column in the secondoven. An eight-port switching valve (ValcoInstruments, Houston, Texas) is used totransfer the carrier gas flow betweencolumns and traps. All lines connectingthe various elements of the system arenickel capillary tubing.

The columns and column conditions used inthe present study are as follows:Packed column: 6 ft x 3 mm ID coiledglass containing 10% OV-101 on 100 to120 mesh Gas Chrom Q (Applied Science Lab,State College, Pennsylvania). Nitrogencarrier gas flow rate is 40 cc per minute.Column is temperature programmed from180°C to 240°C at 4°C/min.Capillary column: 100 ft x 0.02'' IDnickel-200 tubing (Handy and HarmonTube Co., Norristown, Pennsylvania)coated with Poly I-110 (Applied ScienceLab, State College, Pennsylvania).Nitrogen carrier gas flow rate is 4 cc/min. Column temperature is 190°C. Ad-ditional gas flow is added prior to theelectron capture detector to provide atotal flow of 10 cc/min.

The electron capture detector containsa 1 Curie scandium tritide foil (U.S.Radium Corp. , Panorama City, California)and is operated at a 1000 µsec. collectionpulse. The signal from the detector isconverted by an analog linearizer (AntekInstruments, Houston, Texas) prior tothe recorder.

SAMPLE PREPARATIONA flow diagram of the blood serum extrac-tion and derivatization procedure isshown in Figure 2. All solvents are

Figure 2

reagent quality or better and are re-distilled and stored in glass beforeuse. All glassware is silylated usingvapor phase silylation. Derivatizationis a standard esterification of thephenolic hydroxyl group of 9-THC usingpentafluoropropionic anhydride (PFPA) inthe presence of trimethylamine. The finalvolume of sample prior to gas chromatog-raphic analysis is 20 µl of which 5 µlis taken for each sample injection.

The internal standard is hexahydrocanna-binol (HHC) prepared by hydrogenation of

8-THC and purified by preparative scalethin layer chromatography. For most de-terminations 4 ng of the HHC are added insolution to the blood serum sample beforeproceeding with the extraction procedure.

RESULTS AND DISCUSSION

As stated previously the amounts of 9-THC which would normally be determinedin a convenient volume of blood serumare some three orders of magnitudegreater than the limit of detectabilityattained by electron capture of the pure,derivatized compound. The primary factorwhich prevents the realization of thisdetection limit is the presence of inter-

44

fering substances originating in the bloodserum or introduced to the sample duringpreparation. This latter source of con-tamination can, at least, be controlledby exercising the usual precautions gener-ally recommended for electron capture de-terminations, i.e. use of glass-distilledsolvents, use of small total solvent vol-ume where evaporative concentration isrequired, re-distillation of derivatizingreagents, scrupulous cleanliness of glass-ware, and avoidance of all plastic mate-rials with the exception of polytetra-fluorethylene type polymers.

Interference from endogenous compounds inthe sample can only be reduced to apractical minimum through selectivesolvent extraction of the cannabinoidand utilization of appropriate chroma-tographic separation. The limit to whichthe extractive isolation of the compoundcan be pursued is very much dependent onhow much loss of that compound can betolerated. The multiplicative nature ofsample loss thus dictates against allbut the very minimal number of parti-tioning manipulations prior to gas chro-matographic separation. On the other hand,loss of sample during gas chromatographycan be all but be eliminated by carefulchoice and preparation of the variouscomponents of the instrument. In thepresent study the individual elementsof the dual oven chromatograph wereexamined for effect on peak area of thedetected 9-THC-PFP and were found tohave no significant contribution tosolute loss even at the limit of de-tectability. It is for these reasonsthat reliance was placed on gas chro-matographic separation rather than ex-tractive techniques for reduction ofinterference.

In even the most meticulous extractionprocedures involving recovery of com-pounds at part per billion concentration,some loss must be anticipated. Inclusionof an internal standard prior to theextraction and derivatization proceduresis therefore necessary to monitor compoundrecovery, and the chemical and physicalproperties of the standard should approxi-mate those of the compounds in questionas closely as possible. The deuterated

9-THC employed by Agurell in mass frag-mentographic assays (Agurell, 1973) cer-tainly approaches the ideal, but, ofcourse. cannot be differentiated fromunlabeled 9-THC by electron captureprocedures. Hexahydrocannabinol was there-fore selected as having similar propertiesto 9-THC while retaining good chromatog-raphic separation.

There are obstacles to relying almostentirely on gas chromatography to isolatea component of a mixture for measurement,because complete separation may requireexcessive analysis time, even with highlyefficient columns. Thus the columns usedand the operating conditions utilized areusually a compromise between resolutionand speed of analysis. The variability ofblood constituents among human subjectscomplicates the selection of operatingconditions. Therefore in this studynumerous blood samples from individualsbelieved to be abstinant from cannabisuse were examined for the purpose ofvarying the chromatographic conditionsuntil extraneous peaks were absent inthe region of the chromatogram wherethe 9-THC and HHC internal standardwere known to appear at the appropriatelevel of sensitivity. Once the conditionshad been achieved which produced areliable baseline, known amounts of 9-THC were added to random sera samples,and the ratio of detected peak heights.o f 9-lHC to the internal standard werecomputed. The regression was linear from30 ng per ml to a limit of sensitivityof 500 pg per ml without dilution ofthe more concentrated trial samples.

Figures 3 and 4 show electron capturechromatograms of human blood serum extractsobtained using this procedure. The bloodsamples were taken before and one hourafter a volunteer had smoked a singlemarihuana cigarette estimated to havecontained 6 mg of 9-THC. Blood serumlevels determined over a four hour periodare shown in Figure 5, and are in closeagreement with those reported by otherinvestigators (Agurell, 1973).The major disadvantage in applying ECGCto determinations of this type is theamount of instrument time involved inperforming a single analysis. For thechromatograms shown in the figures thetotal time from injection to the pointat which the capillary column is clearfor a subsequent determination is approxi-mately forty-five minutes, and this timeperiod would undoubtedly be excessive instudies where large numbers of sampleswere processed daily. On the other hand,this reported time is by no means anirreducible minimum, and improvementsin column technology, operating parameters,and sample preparation could well lowerthis figure substantially. In addition,gas chromatography lends itself readilyto instrument modification. A multiplecolumn system utilizing the principlesdemonstrated successfully in the operationof the dual-oven instrument could beassembled capable of performing analyses

45

Figure 3 Figure 4

Figure 5

46

limited only by the rapidity with whichthe primary temperature-programmedcolumn can be serially operated. Withautomatic sampling and computer controlsuch an instrument could run continuouslywith very little technician supervision.

REFERENCES

Agurell, S., Gustafsson, B., Leander,K., Lindgren, J-E., Nilsson, J.,Saudberg, F., and Ashberg, M., J.Pharm. Pharmacol. 25: 554 (1973).

Fenimore, D.C., Freeman, R.R., andLoy, P.R., Anal. Chem. 45:2331(1973).

Fenimore, D.C., Loy, P.R., andZlatkis, A., Anal. Chem.43:1972(1971).

Garrett, E.R., and Hunt, C.A., J.Pharm. Sci. 62: 1211 (1973).

Lemberger, L., Axelrod, J., Kopin,I.J., Ann. N.Y. Acad. Sci.191:142(1971).

McCallum, N.K., J. Chromatogr. Sci.11:509 (1973).

Rosenfeld, J.J., Bowins, B., Roberts,Perkins J., and Macpherson, A.S.,

Anal. Chem.46:2232 (1974).

47

Although this report concerns the de-termination of 9-THC onlv. the tech-niques and instrumentation'should beapplicable with minor modification toquantitation of other cannabinoids andmetabolites of these compounds in bio-logical samples.

DETECTION AND QUANTIFICATION OFTETRAHYDROCANNABINOL IN BLOOD PLASMA

Agneta Ohlsson1, Jan-Erik Lindgren2,3, Kurt Leander2, Stig Agursll1,3

1Faculty of Pharmacy, University of Uppsala, Uppsala

2Department of Toxicology, Karolinska lnstitutet, Stockholm

3Astra Läkemedel AB, Södertälje, Sweden

INTRODUCTION

During the last few years much effort hasbeen spent on developing methods for theidentification of Cannabis users by ana-lysing different body fluids. Also, tech-niques for the quantitative determinationo f 1-tetrahydrocannabinol ( 1-THC) and re-lated compounds have long been desired.Recent developments in this area have

been summarized in a review by Grli(1974) including some eighty references.This field of detection of cannabinoidsin biological fluids is in a state of rapidexpansion and for the positive identificationand quantification of cannabinoids in bodyfluids several highly sensitive methods areavailable.

It must be emphasized that the definition

of what a "satisfactory method" is will vary

with the goal of the users. For epidemio-

logical studies of Cannabis use, simplescreening procedures - where a high

reliability may be traded for rapidityand expense - will probably be satis-

factory. In forensic toxicology, therequirement for unequivocal identifica-tion hardly allows the use of proceduresyielding ambigious results. Finally,in pharmacokinetic and pharmacologicalstudies it may be necessary to have aspecific as well as accurate quantita-tive procedure. These considerationswould suggest that eventually different

48

procedures will be used for the determi-

nation of cannabinoids in body fluids

(blood, urine) depending upon the aimof the investigator.

What a "complex and sophisticated" technique

is will also depend upon the trendsin analytical chemistry in a specificcountry. In Sweden, where over seventygas chromatography - mass spectrometry

(GC-MS) instruments are available in alimited scientific community, a mass

spectrometric method is obviously less"sophisticated" than in a country wheresuch instruments are scarce.

Radioimmuno assay procedures for cannabi-noids in blood and urine have been ex-

tensively investigated (cf. Gross et al.,

1974; Teale et al., 1975). Both authors

utilized goat antiserum obtained afterimmunization with conjugates of 1-THC.

After reduction of non-specific bindinga sensitivity of ca. 7.5 ng/ml THC inplasma or 1 ng/ml in urine could be achieved

The antibodies were specific for the three

ringed cannabinoid nucleus in THC andcross-reacted with e.g. 7-hydroxy- l -THC and cannabinol (CBN) but not withcannabidiol (CBD). As pointed out byTeale et al. (1975) this cross-reactivityhas disadvantages but also advantages in

e.g. epidemiological studies. Anotherimmunological approach - free radical

immunoassay - has been tested by Cais

et al. (1975).

49

Gas chromatography in combination with

sensitive detection systems has alsobeen widely investigated. In general, theproblems have been to remove lipid materialsinterfering with the gas chromatographicseparation of e.g. THC and increase the

sensitivity in the detection step. Feni-

more et al. (1973) used a dual column

separation, utilizing a capillary columnfor the final resolution, and the electron

capture (EC) sensitive heptafluorobutyratederivative. McCallum (1973) utilized a

flame photometric detection of phosphateesters of l-THC and related compounds.

Garrett and Hunt (1973) described anotherEC-based procedure using the pentafluoro-

benzoate ester. Later, the same authors

(Garrett and Hunt, 1976 a) have discussedthe use of high pressure liquid (HPLC) andgas chromatography (GLC) for separationand analysis of THC in biological fluids.

We have published methods for the quanti-tative determinations of l-THC and 6-THCin the blood of Cannabis smokers (Agurell

et al., 1973, 1974). These methods arebased upon the purification of THC extrac-ted from blood plasma by liquid chroma-

tography on Sephadex LH-20 followed bymass fragmentographic assay using thedeuterated THC-analogue as internalstandard. Rosenfeld and co-workers (1974)also used mass fragmentography for the de-tection of 1-THC in humans after smoking.

The latter authors used trimethylaniliniur

hydroxide for on column methylation of thephenolic group and relied on the trideu-teromethyl ether as internal standard inthe last step.

Many other techniques have been tried as

reviewed by Grlic (1974) but a laterresult utilizing HPLC of dansyl deriva-

tives should be mentioned (Loeffler et al.,

1975). Wall and co-workers in thisvolume present in detail a mass spectro-metric procedure for the determination ofTHC in humans after Cannabis smoking.

Of course, the crucial test for any methodis its performance (sensitivity, speci-

ficity, capability) when applied to theactual problem in man - this is unfortun-ately often lacking in the papers pub-lished.

The purpose of the present paper is to

summarize previously published resultsand to present additional information onthe mass fragmentographic procedures for

THC and other cannabinoids developed inour laboratory.

CBD-d7. Olivetol-d 7 (160 mg) was condensed

with (+)-trans- p -mentha-2,8-dien-l-ol in

benzene using traces of oxalic acid as de-scribed by Petrzilka et al. (1969). After

final purification by TLC (Agurell et al.,1974) on pre-washed silica gel G plates

with methylene chloride-methanol (95:5)

as solvent, 33 mg of 97% pure (GLC)

METHODS

CBD-d7 was obtained. Deuterium contentaccording to mass spectrometry (m/e

314-321): d0 2%, d1 2%, d2 2%, d3 3%,

d4 3%, d5 9%, d6 30%, d7 100%.

to mass spectrometry, m/e (180-187): do 2%,

d1 2%, d2 2%, d3 3%, d4 3%, d5 9%, d6 30%,d7 100%.

Cannabinoids1-THC-d7. Olivetol-d7 was condensed with

(+)-trans-p-mentha-2,8-dien-l-ol using

Synthesis of olivetol-d7. The synthetic

procedure of Pitt et al. (1975) was usedwith the following modifications in orderto increase the amount of deuterium incor-porated into the molecule (Fig. 1). The

ylide was prepared from methyl 4-bromocro-.tonoate and triphenylphosphine, was reactedwith 3,5-dimethoxybenzaldehyde to yield

methyl 5-(3,5-dimethoxyphenyl)penta-2,4-dienoate (1) (Buchta & Andree, 1959, 1960).

1 was then reduced with LiAld4 in ether at

p -toluensulfonic acid (Petrzilka et al,

1969), The preparative purification wascarried out on silica gel G plates elutedthree times with 5% ether in lightpetroleum to separate the 6 and l -

isomers. The purified (94% by GLC, 37 mg)1-THC-d7 showed the following deuterium

content (m/e 314-321): d0 2%, d1 2%,

d2 2%, d3 3%, d4 3%, d5 9%, d6 30%, d7

100%.

room temp. for 2 h. The complex was de-

graded with d20 to insert one more d in CBN-d7. 6-THC-d 7 was treated with two

the side chain. The isolated compound, with equivalents of chloranil and traces ofthe hydrogen atom of the alcohol group ex- p -toluenesulfonic acid in refluxing drychanged for deuterium, and the catalyst toluene for 5 h (Mechoulam et al., 1968).tris(triphenylphosphine)rhodium chloride The solution was chromatographed on a si-was stirred in benzene under d2 atmosphere lica gel column with toluene eluent andfor 16 h at room temp. The 5-(3,5-dimethoxy- further purification by TLC yielded 96%phenyl)pentan-1,1,2,3,4,5- d6-l-ol was then pure CBN-d 7. The isotope distribution wasreacted with PBr3, LiAl d 4 and BBr3 according the same as in l-THC- d7; this in contrastto Pitt et al. (1975) to yield olivetol-d7 to dehydrogenation methods using sulphur(over-all yield from 3,5-dimethoxybenzal- which causes extensive deuterium scramb-dehyde 24%). Deuterium content according ling.

5 0

Figure 1

Fig. 1. Scheme for synthesis of olivetol-d7. Formulas of cannahinoidsand deuterium containing internal stundards.

51

Olivetol- d 3 was prepared as described by CBD- d2, l-THC-d2 CBN-d2. CBD-d2 was

Pitt et al. (1975). Compound 1 (see synthesized as described for the d7-ol ivetol- d7) was hydrogenated in the pre- analogue and crystallized from hexane.sence of Pd and the product reduced with The CBD-d2 was 97% pure by GLC and showedLiAld4 . Subsequent reactions with PBr3,

LiAld4 and BBr3 giving olivetol-d3 werea deuterium content (m/e 314-316): d0 8%,

d1 12%, d2 100%.as described for olivetol-d7 .

1-THC-d2 was prepared as l-THC-d3

and more than 95% pure by GLC: Deuteriuml-THC-d3 and CBN-d3 were synthesized as content (m/e 314-316): do 10%, d1 20%,

described for the d7-analogues. Since larger d2 100%. CBN-d2 was prepared from 6 -amounts were prepared l-THC-d3 was separa- THC-d2 by dehydrogenation with sulphurted from 6-isomer on a column of Florisil at 260°C (Petrzilka et al., 1969). Theusing 2% ether in light petroleum as sol- deuterium content (m/e 310-312): d0 12%,vent. 1-THC-d3 was 95% pure according d1 46%, d2 100%.to GLC and the deuterium content (m/e

314-317: d0 2%, dl 6%, d2 12%, d3 100%.

CBN-d3 eluted with 2-10% ether in lightpetroleum from the Florisil column was

6-THC-d4. The synthesis of this compound

(98.5% pure) was described previously(Agurell et al., 1974); deuterium con-

tent (m/e 314, 318): d0 1.5%, d4 100%.

95% pure according to GLC with a deucerium

content (m/e 310-313): d0 2%, d1 5%, d2

15%, d3 100%.

Olivetol-d2 . The methyl ester of 3,5-

dimethoxybenzoic acid was reduced withLiAld4 in tetrahydrofuran and the re-sulting alcohol was reacted with PBr3.(See olivetol-d7). The bromide crystal-

lized from ether-light petroleum wasmixed with Cu(I)I in ether and butyl

lithium was added drop-wise to themixture (Petrzilka et al., 1969). Theobtained 0,0-dimethylolivetol-1',1'-d2

was distilled at 110° C, 0.15 mm Hg and

Non-labelled cannabinoids. ( 1-THC,6-THC, CBN, CBD) were synthesized accor-

ding to standard procedures and care-

fully purified and dried before use. Stocksolutions were maintained in ethanol(1-5 mg/ml, 4° C, stored in the dark).

GLC was carried out on a Varian 2100FID gas chromatograph using 2 mm (i.d.)

x 180 cm glass columns with 2% SE-30ultraphase on Gas Chrom Q (100-120 mesh)at 250°C.

Analysis of THC in blood plasma

showed the following deuterium content The following procedure was used for 6 -(m/e 180-182): d0 0.3%, d1 4%, d2 100%. THC (Agurell et al., 1974). AnalogousAfter demethylation with hydriodic acid procedures utilizing the proper deutera-(cf. Agurell et al., 1973) olivetol-d2 ted internal standard and mass numberswas purified on a silica column with are applicable for 1-THC, CBN, and CBD.methanol-chloroform of increasing pola- The procedure for l-THC has been pub-rity as eluent (from 3% methanol). lished (Agurell et al., 1973).

52

Blood samples. Blood samples (10 ml) were

collected as desired in heparinized tubesafter smoking of the 6-THC sample.

Plasma is obtained by centrifugation andstored in silanized glass tubes at -20°Cuntil analysed.

Extraction. To a 5.0 ml plasma sample is

added 100 ng (THC levels above 5 ng/ml)or 20 ng (below 5 ng/ml) of deuterated

internal standard ( 6-THC-d4) dissolvedin 50 µl ethanol. The plasma sample is

extracted three times with an equal volumeof light petroleum containing 1.5% iso-

pentanol in a glass stoppered centrifugetube. After centrifugation the lightpetroleum is drawn off and the combined

organic extracts evaporated under nitro-gen at 50°C almost to dryness. This ex-

tract is quantitatively transferred tothe Sephadex LH-20 column using three0.2 ml portions of the elution solvent.

Liquid chromatographic purificationMantled silanized glass columns (1 x 40 cm,void vol. ca. 15 ml) operated at 12°C(const. temp. water cooling system) con-

taining Sephadex LH-20 and eluted withlight petroleum-chloroform-ethanol

(10:10:1) were used for purification ofplasma extracts. Each column was providedwith a 200 ml solvent reservoir and after

each purification the column was washedwith 40 ml solvent, If not in use, thecolumns were washed twice weekly withfresh solvent to maintain constant elution

volumes.

The elution volume for 6-THC was deter-mined by calibration with ng amounts of

6-THC-3H (Fig. 2). The calibration canalso be carried out by GLC analysis of

Figure 2

Fig. 2. Elution pattern for 6 -THC-3H, l-THC,CBD and CBN from a Sephadex: LH-20 column.

the concentrated fractions after appli-cation of 10 µg quantities to the column.The column was run at 0.2 ml/min and the

pertinent 7-8 ml fraction was collectedand evaporated under nitrogen. The

residue was dissolved in ethanol andtransferred to a 50 µl conic vial(Reactivial, Pierce) dried, and finallydissolved in 10-15 µl of ethanol andstored at 4°C in the dark until analysis.This solution was subjected to massfragmentography.

Mass fragmentography was carried out

using an LKB 9000 GC-MS instrument. Thecolumn was a 1.4 m x 2 mm i.d. silanizedglass column containing 3% OV-17 on GasChrom CLP 100/120 mesh. Temperatureswere in the column 180-210°, flash heater

250°, and source 290°. Helium was carriergas (25 ml/min) and typical retentiontimes are: CBD 3.4 min, 6-THC 4.0 min,

1-THC 4.5 min, CBN 6.1 min. For massfragmentography a multiple ion detectorwas added (Elkin et al., 1973) and used

53

at 50 eV. For 6-THC, the mass spectro-

meter was set to continously record theintensities of m/e 314 (molecular ionof non-labelled 6-THC) and m/e 318(internal standard, 6-THC-d4) as well asm/e 299, 303 as shown in Fig. 3.

Figure 3

Fig.3. Mass fragmentation of 6 -THC-d4(8 -THC-d4)

as internal standard (m/e 318, 303) and 6-THC(m/e 314, 299) from purified plasma extract.

Standard curves were prepared by adding

known amounts of 6-THC (0.5-100 ng/ml)to blank plasma samples and carrying outthe described procedure (Fig. 4). Thecorrectness of the standard curves werechecked during the day.

Figure 4

Fig. 4. Standard curves for 6 -THC ( 8 -THC) inplasma. 0-1- ng/ml (left), 10-100 ng/ml (right).

RESULTS AND DISCUSSION

Principles

The described method for the determinationo f 6-THC (cf. Agurell et al., 1974) inblood is basically identical to the one

published for l-THC (Agurell et al.,1973). The method is based upon the addi-

tion of the proper deuterated internalstandard ( 6-THC-d4) due to the blood plasmasample. After extraction with light petro-

leum, the 6-THC containing fraction ispurified by liquid chromatography on aSephadex LH-20 column. The pertinentfraction containing 6-THC (absorbed bysmoking a spiked Cannabis sample) and

6-THC-d4 (added as internal standard)is collected. The relative amounts of thetwo compounds are determined by mass frag-mentography monitoring the molecular peaks(m/e 314 and 318, respectively; Fig. 3.).

54

As subsequently discussed, this method canalso be used for the analysis of other canna-

binoids, such as THC:s of different mole-cular weight, CBD, and CBN. However, atpresent less information is available withregards to these latter compounds.

1"- and 2"-positions of the pentyl side

chain (Fig. 3).

There are generally only ng-amounts ofcannabinoids present in blood plasma and

although the deuterated internal standards

serve as carriers, the method requires

scrupulously clean, silanized glass wareand redistilled solvents.

Other deuterated cannabinoids. A synthesis

o f 1-THC-d4 was reported earlier(Agurell et al., 1973). In the presentpaper the syntheses of olivetol-d7 ,

olivetol-d3, and olivetol-d2 - mainly

based upon the procedures of Pitt et al.(1975) - are outlined. From these inter-

mediates, correspondingly labelled 1-THC,CBD, and CBN can be prepared as exemplifiedin Fig. 1.

Reference cannabinoids

In all methods pure cannabinoids are ne-cessary for identification and calibra-

tion purposes. Since cannabinoids areoften rather unstable and, with few ex-ceptions, not crystalline - to obtainand maintain the purity of cannabinoidsis not an entirely easy task. However,

purification on pre-washed TLC plates as

described previously (Agurell et al.,1974) followed by careful drying andstorage in ethanol solution at 4° in thedark provides cannabinoid standard solu-tions of satisfactory stability.

The syntheses of the deuherium labelledcompounds were accomplished by condensation

of suitable deuterated olivetols and (+)-

trans-p -mentha-2,8-dien-1-ol according tostandard methods (Mechoulam et al., 1976)- Fig. 1.

6-THC-d4. The synthesis of this internal The mass spectra of 1-THC-d7 and non-standard was described previously (Agurell labelled 1-THC are shown in Fig. 5et al., 1974). The label is located in the together with 1-THC-d 3 .

The sensitivity in the final mass frag-

mentographic assay may be limited by theamount of non-labelled (d0) compoundpresent in the deuterated internal standard.This interferes with the non-labelled canna-binoid present in the plasma. Thus, we havetried to minimize this interference by pre-venting exchange reactions, increasing thenumber of hydrogens substituted with deu-

terium, and by limiting the amount of in-

ternal standard in samples containing lowamounts of cannabinoids (Fig. 5). Thus,the present limit of sensitivity (ca. 0.3

ng THC/ml plasma) is partly due to theamount of THC-d0 in the internal standard

and not due to chromatographic or massspectrometric problems per se . As expected,d7-containing 1-THC, CBD, and CBN showed,

together with 1-THC-d3, the least conta-

mination with d0-analogues (2%). This is

in the same range as found for 6-THC-d4

(1.5%; Agurell et al., 1974).

55

Figure 5

Fig. 5. Mass spectra of 1-THC, 1-THC- d3, and 1THC-d7 .56

Extraction and purification bolites before mass fragmentography. Theelution volume for e.g. 6-THC is stable

for months if the column is operated at

low const. temp. (12°C) and, when not in

use, washed with fresh solvent regularly.Our present set-up contains five columns

which can be handled by a technician(10 samples per day). However, such sys-tems can be automated (Sippell et al.,

1975). Over 90% of the 6-THC peak iseluted in a 5 ml volume (Fig. 2) but a7-8 ml fraction is collected. As a pre-caution ca. 3-4 ml on each side of the

peak is collected and stored.

6-THC and its 1-THC isomer appear to be

stable in human plasma for months ifstored at -20°C i silanized glass tubes.

The extraction and purification proceduresfor 1-THC, CBD, or CBN are analogous.

The extraction procedure for 6-THC, asrevealed by experiments with 6-THC-3H,is quite efficient and the recovery afterboth extraction and liquid chromatographyis usually over 80%. Also. early studies

on 1-THC showed recoveries of 70 ± 6%(s.d.) after the column purification

(Agurell et al., 1973).

The Sephadex LH-20 separation is essential

in removing interfering lipids and meta-

We have investigated the elution patternsof different cannabinoids and their meta-

bolites on the Sephadex LH-20 column

(Fonseka et al., 1976). This is shown in

Fig. 6.

Figure 6

Fig. 6. Elution patterns of l -THC, 6 -THC, CBN, and CBD and their mono-oxygenatedderivatives on Sephadex LH-20. Abbreviations as exemplified by the l -series 6-0 =

6-oxo- l -THC, 3"-OH= 3"-hydroxy- l -THC, 6 , 7-dihydroxy- l - THC, etc.

57

The complexity of cannabinoid metabolism inman and animal is indeed great and the re-quirements for specificity when assaying a

certain cannabinoid are consequently con-siderable. To give an impression of the

complexity, we have in a review (Agurell

et al., 1976 b) listed 35 metabolites of1-THC, 6-THC, CBN, and CBD, which have

been isolated in our and other laboratoriesand are substituted in the pentyl side

chain. To make a complete list an equalnumber of metabolites oxygenated in theterpene nucleus (cf. Mechoulam et al.,1976) should be added.

Mass fragmentography

The sensitivity achieved in the quantifica-

tion of 6-THC and other cannabinoids is

partly due to the liquid chromatographyclean-up of blood plasma extract. Mainly,however, the sensitivity and specificity is

dependent upon the mass fragmentographicanalysis.

We have used the deuterium labelled analogue6-THC-d4 for the analysis of 6-THC in blood

plasma from four male, casual Cannabis smokerswho had smoked 8 mg of 6-THC. Such a mass

fragmentogram is shown in Fig. 3.

Two standard curves were prepared (0-10 and10-100 ng/ml) by adding known amounts of

6-THC to blank plasma samples and carrying

out the described procedure. Peak heightso f 6-THC-d4 (m/e 318) were plotted againstknown amounts of 6-THC (Fig. 4). Sincethe same ratios were obtained simply bymixing known amounts of 6-THC and

6-THC-d4, usually such standard curveswere used. The correctness of the stan-dard curves should be checked occasionallyduring the day of analysis.

There are advantages as well as disadvan-

tages in the use of deuterium labelledanalogues as internal standards. They are

carriers for the minute amounts of canna-binoids present in biological samples

and, being almost identical to the ana-

lyzed compound they can be added to theoriginal plasma sample and will then com-pensate for variations in extraction andpurification recoveries. A disadvantageis that deuterium labelled standards

have to be synthesized. Hopefully thesemight be provided by NIDA.

The present method for 6-THC can be useddown to 0.3 ng/ml. As pointed out (see

discussion on "Reference cannabinoids")the sensitivity is partly limited by thesmall amount of 6-THC-d0 present in the

deuterated internal standard. Thus, wehave tried to eliminate do-contamination

in the internal standards for 6-THC,l-THC, CBD, and CBN - with best results

in the d7 -analogues.

The d3-compounds can also be used asstandards but were mainly synthesized to

study single dose pharmacokinetics of1-THC, CBD, and CBN in heavy hashish

users. These users presumably have a high"steady-state" blood level of cannabinoids

and their metabolites. By the use of a

d3-labelled single dose of e.g. 1-THC,the kinetics of a particular 1-THC dose

can be followed in spite of high background

levels of non-labelled 1-THC (Fig. 7a). Thed7-labelled 1-THC is then used as internalstandard (Fig. 7b).

l-THC and CBD can both be determined in the

same mass fragmentogram using the samechannels - m/e 314 and e.g. 321 for inter-nal standard - whereas CBN is simultaneously

assayed using m/e 310 and 313 (Fig. 7c).

58

Figure 7a Figure 7b

Fig. 7 a. Mass fragmentation of purifiedplasma extract containing 1 -THC (0.9 ng/mlplasma, m/e 314) and 1 -THC-d3 as internalstandard (m/e 317). Column temp. 210°C.

Mass fragmentogram ofFig. 7 b.l -THC-d3, and

l -THC,1 -THC-d7 (ca. 5 ng/ml plasma

of each). Column temp. 210°C.

So far we have not developed assays forany of the possibly important l -THC meta-

bolites, e.g. 7-hydroxy- l-THC. However,

it is likely that if quantitative metabolicStudies in man warrant, this metabolitecould also be quantitated after elution andderivatization using mass fragmentography.

It is also possible that cannabinoids mono-

hydroxylated in the side chain can be assayed

as their trimethylsilyl ethers using speci-

fic fragments (Binder et al., 1974). Sidechain hydroxylation seems to be a general

metabolic route and of importance in atleast some species (Agurell et al., 1976 a,Harvey & Paton, 1976). Hydroxylation inthe 3"- and 4"-positions are favored -reactions which also seem to confer highpsychotomimetic activity in the rhesus

monkey (Agurell et al., 1976 a).59

Figure 7c

Fig. 7 c. Mass fragmentograms of CBD (m/e314) and internal standard CBD-d2 (m/e 316,retention time 3.0); l-THC (m/e 314) andinternal standard l-THC-d3 (m/e 317, re-tention time 4.0); and CBN (m/e 310) andinternal standard CBN-d3 (m/e 313, retentiontime 5.0). Column temp. 220° C.

Plasma levels

The plasma levels of 6-THC in man aftersmoking 8 mg 6-THC - of which about halfis absorbed in the lungs - are shown in Fig. 8.Immediately after smoking high values

(>100 ng/ml) of 6-THC are recorded but

drop rapidly to 10-20 ng/ml at 0.5 hourand are about 1 ng/ml at 4 hours. Thepresent sensitivity of 0.3 ng/ml, thusallows THC levels to be followed for12-24 hours. This is more than enough to

study any correlations of 6-THC levels

with physiological and psychological

effects (Agurell et al., 1974) but notenough to establish a true eliminationphase. Lemberger and co-workers (1971)

have estimated elimination phase half-lifes in man of 1-2 days.

Garrett and Hunt (1976 b) have shownthat the terminal half-life of l-THCin the dog is reached only slowly. They

also found that the return of l-THC fromthe tissues is the rate determining stepof the drug elimination process after the

initial distribution and metabolism.

Similar plasma levels of l-THC as for6-THC (Fig.8 ) have earlier been found

in man by us (Agurell et al., 1973) and

Rosenfeld et al. (1974) using mass frag-mentography, and by e.g. Galanter et al.(1972) using 1-THC-14C. Wall and co-

workers (1974) have published plasmalevels of both l-THC, CBD, and CBN andcertain metabolites as well as subjective

psychological effects after i.v. admini-stration of the labelled drug.

Hence, the sensitivity requirements forthe determination of l-THC are indicatedin Fig. 8. Plasma levels will obviouslybe modified by the amount of THC absorbedand by the rate of absorption but iflevels are to be measured later than 4hours after administration a sensitivityof 1 ng/ml is required. Such a sensitivi-ty with sufficient specificity is perhapslimited to the mass fragmentographic tech-

60

Figure 8

Fig. 8. Plasma levels of 6 -THC after smoking 8.3 mg 6-THC.

niques. We have so far encountered little Radioimmuno assay procedures, where also

interference in the mass fragmentographic certain metabolites cross-react, would

determination of l-THC provided redistilled clearly be applicable to the qualitative

solvents, particularly ethanol, and all identification of Cannabis users. However,

silanized glass ware are used. the potential cross-reactivity to e.g.

61

steroids and other drugs also has to be

ascertained.purification. With the non-automated systemnow in use, a technician can process aboutten plasma samples per day. This might be

The capacity of the mass fragmentographic improved by automation, by high pressuretechnique is more limited than the radio- liquid chromatography or by using the doubleimmune assay procedure, the main time re- extraction technique of Rosenfeld et al.quiring step being the liquid chromatography (1974).

ACKNOWLEDGMENTS

The support of the Swedish Medical Research

Council is appreciated. The mass spectrometric

work was supported by grants to the Department

of Toxicology.

REFERENCES

Agurell, S., Gustafsson, B., Holmstedt, B.,Leander, K., Lindgren, J.-E., Nilsson, I.,

Sandberg, F. & Åsberg, M. (1973). J. Pharm.Pharmac. 25, 554-558.

Agurell, S., Levander, S., Binder, M.,

Bader-Bartfai, A., Gustafsson, B.,Leander, K., Lindgren, J.-E., Ohlsson,A. & Tobisson B. (1974). Pharmacokineticso f 8-THC in man after smoking - relations

to physiological and psychological effects.Presented at Conference on the pharmacologyof Cannabis, Savannah, Dec. 3-6, 1974. Inpress in Pharmacology of Cannabis, Eds.S. Szara & M. Braude, Raven Press, NewYork, USA.

Agurell, S., Binder, M., Fonseka, K.,Gustafsson, B., Lindgren, J.-E., Leander,K., Martin, B., Nilsson, I.M., Nordqvist,

M., Ohlsson, A. & Widman, M. (1976 a).

Cannabinoids: Metabolites hydroxylated in

the side chain. In press.

Agurell, S., Martin, B. & Widman, M.(1976 b). To be published.

Binder, M., Agurell, S., Leander, K.& Lindgren, J.-E. (1974). Helv. Chim. Acta57, 1626-1641.

Buchta, E. & Andree, F. (1959). Ber. 92,3111.

62

Buchta, E. & Andree, F. (1960). Ber. 93,

1349.

Cais, M., Dani, S., Josephy, Y., Modiano,A., Gershon, H. & Mechoulam, R. (1975)

FBBS 55, 257-260.

Elkin, K., Pierrou, L., Ahlborg, U.G.,Holmstedt, B. & Lindgren, J.-E. (1973)

J. Chromatogr. 81', 47-55.

Fenimore, D.C., Freeman, R.R. & Loy,

P.R. (1973). Anal. Chem. 46, 2232-2234.

Fonseka, K., Widman, M.& Agurell, S.

(1976). J. Chromatogr. In press.

Galanter, M., Wyatt, R.J., Lemberger, L.,

Weingartner, H., Waughan, T.B. & Roth,

W.T. (1972). Science 176, 934-935.

Garrett, E.R. & Hunt, C.A. (1973). J.Pharm. Sci. 62, 1211-1214.

Garrett, E.R. & Hunt, C.A. (1976 a).Separation and sensitive analysis oftetrahydrocannabinol in biological fluidsby HPLC and GLC. To be published.

Garrett, E.R. & Hunt, C.A. (1976 b).Pharmacokinetics of 9 -THC in the dog.To be published.

Grli , L. (1974). Acta Pharm. Jugoslav.24, 63-72.

Gross, S.J., Soares, J.R., Wong, S.L.R.

& Schuster, R.E. (1974). Nature 252,581-582.

Harvey, D.J. & Paton, W.D.M (1976). Exami-nation of the metabolites of 1 -tetrahydro-cannabinol in mouse liver, heart and lung

by combined gas chromatography and massspectrometry. In press.

Lemberger, L., Axelrod, J. & Kopin, I.J.(1971). N.Y. Acad. Sci. 191, 142-152.

Loeffler, K.O., Green, D.E., Chao, F.C.& Forrest, I.S. (1975). Proc. West.

Pharmacol. Soc. 18, 363-368.

McCallum, N.K. (1973). J. Chromatogr.

11, 509-511.

Mechoulam, R., Yagnitinsky, B. & Gaoni, Y.

(1968). J. Am. Chem. Soc. 90, 2418.

Mechoulam, R., McCallum, N.K. & Burstein,

S. (1976). Chem. Rev. In press.

Petrzilka, T., Haefliger, W. & Sikemeier, C.

(1969). Helv. Chim. Acta 52, 1102-1118.

Pitt, C.G., Hobbs, D.T., Schran, H., TwineJr., C.E. & Williams, D.L. (1975). J. Label.

Comp. 11, 551-575

Rosenfeld, J.J., Bowins, B., Roberts, J.,Perkins, J. & McPhersson, A.S. (1974).

Anal. Chem. 46, 2232-2234.

Sippell, W.G., Lehman, P. & Hellman, G. (1975)

J. Chromatogr. 108, 305-312.

Teale, J.D., Forman, E.J., King, L.J., Piall,E.M. & Marks, V. (1975). J. Pharm. Pharmac.

27, 465-472.

Wall, M.E., Brine, D.E. & Perez-Reyes, M.(1974). Metabolism of cannabinoids in man.Presented at Conference on the pharmaco-logy of Cannabis, Savannah, Dec. 3-6,1974. In press in Pharmacology of Canna-bis, Eds. S. Szara & M. Braude, RavenPress, New York, USA.

63

A METHOD FOR THE IDENTIFICATION OF ACIDMETABOLITES OF TETRAHYDROCANNABINOL (THC)BY MASS FRAGMENTOGRAPHY

Marianne Nordqvist1, Jan-Erik Lindgren2,3, Stig Agurell1,3

1Faculty of Pharmacy, University of Uppsala, Uppsala

2Department of Toxicology, Karolinska Institutet, Stockholm

3Astra Läkemedel AB, Södertälje, Sweden

INTRODUCTION

Plans to identify Cannabis users by thedetection of 1-THC in urine had to be

abandoned after the reports by Lembergeret al. (1970, 1971). Their experimentswith humans given 14C- 1-THC i.v. and

later work by Wall et al. (1974) showedthat the drug is almost completely meta-bolized and the metabolites excretedmainly via faeces. Most of the urinaryradioactivity derived from acidic meta-bolites. The contents of unchanged l-THCwere less than 0.02% of the administereddose. Only small amounts of 7-hydroxy- l -THC, the major metabolite of 1-THC in

vitro, were indicated in the urine by TLC(Lemberger et al., 1970). In contrast,

the faecal contents of 7-hydroxy- 1-THC

are high, about 20% (Lemberger et al.,

1971; Wall et al., 1974). The low amounts

o f l-THC and also 7-hydroxy- l-THC ex-

creted in the urine seem to be a resultof further metabolism yielding more polarcompounds with the 7-methyl group beingfurther oxidized to a carboxyl group viathe aldehyde (Ben-Zvi & Burstein, 1974)with or without introduction of additionalhydroxyl group(s). The l-THC-7-oic acidhas been identified as a major metabolite

in urine, faeces, and plasma from humans

given l-THC or 7-hydroxy- l-THC i.v.(Wall et al., 1974).

64

A method for detection of l-THC or 7-hydroxy- l-THC in urine or in plasma

samples taken more than 12 h after theinhalation of a normal occasional dose of

l-THC would have to be extremely sensi-tive. The plasma level of unchanged drugis then less than 0.3 ng/ml, which isthe limit of our method for detection of

THC. 7-Hydroxy- l-THC, 6 -hydroxy- 1-THC,

-hydroxy- l-THC and 6,7-dihydroxy- 1-

THC are all minor metabolites in plasma

with concentrations much lower than thatof the parent compound at all times, accor-ding to Wall et al. (1974). On the other

hand, the amount of l-THC-7-oic acid inplasma is almost equal to that of 1-THC

within 50 min and then remains slightlyhigher for more than 24 h.

The possible occurrence of other major acidmetabolites must not be overlooked, as

only l-THC-7-oic acid is identified inman so far. The major ones isolated from

rabbit urine - 1"- and 2"-hydroxy- l-THC-7-oic acid (Burstein et al., 1972) and

4",5"-bisnor- 1-THC-7,3"-dioic acid

Nordqvist et al., 1974) could also be ex-

pected to be formed in man by furtherhydroxylation and oxidation of 7-hydroxy-

l-THC. The unidentified part of metabolitesin human plasma, urine, and faeces classifiedas more polar acids by Wall et al. (1974)are probably similar to the hydroxylated

THC-oic acids identified by Harvey andPaton (1976) in mice.

Our proposed method for the identification

of Cannabis intoxication is based upon

the finding that l-THC-7-oic acid is amajor metabolite. For forensic purposes,

when the identification but not accuratequantitation is required, it should be apractical method, if gas chromatography -

mass spectrometry (GC-MS) equipment is

available. It is related to our method for

detection of THC in blood plasma (Agurellet al., 1973) and can as described here be

used for qualitative identification andsemiquantitative assay of a major acidicmetabolite in urine and plasma fromCannabis users. This method is probably

with slight modifications applicable also

to related acid metabolites.

METHODS

Analysis of THC-7-oic acid in blood plasma

The procedure used was a modification of

that described for 1-THC (Agurell et al.,1973). All glass ware was carefully washed

and silanized before use. The solvents were

of analytical grade and those constituting

the eluent mixture were distilled twice.

Extraction. 50 ng 6-THC-7-oic acid

(Mechoulam et al., 1973) dissolved in 25µl ethanol was pipetted into a glass

stoppered centrifuge tube containing 2.5

ml human plasma. It was equilibrated and2.5 ml 1 M citrate-HCl buffer pH 4.1 was

added 5 min before extraction with 10 mldiethyl ether. After centrifugation for

10 min at 4000 rpm the ether layer wasdrawn off, transferred into a conic tube

and evaporated to dryness under a streamof nitrogen at 40°.

Methyl ester formation. The residue was

dissolved in 100 µl methanol, an etheralsolution of diazomethane was added inexcess and the solution allowed to reactfor 10 min before evaporation of reagentand solvent under nitrogen.

65

C!olumn chromatography. The resulting

residue was transferred to a jacketedSephadex LH-20 column (1 x 55 cm, V0 =17 ml) using three consecutive 200 µlportions of the elution solvent lightpetroleum-chloroform-ethanol (10:10:1).

At the elution rate 0.2 ml/min and thetemperature 12° the methyl ester of 6-THC-7-oic acid was eluted between 37 and

47 ml. This fraction was collected in a10 ml conic tube. The solvent was evapo-rated under nitrogen, the residue dissolvedin ethanol and transferred to a' 1.0 ml

conic vial, where it was stored at 4' in

the dark until analysis.

Silylation. On the day of analysis the

methylated extract was dried under nitrogen,dissolved in 25 µl dry acetonitrile, mixedwith 10 µl silylating agent [N,0-bis-(tri-methylsilyl)acetamide] and kept at 50-60°for 10 min. It was then dried again undernitrogen and redissolved in 25 µl acetoni-trile.

Mass fragmentography. The silyl ether of6-THC-7-oic acid methyl ester was subjected

to mass fragmentography (3% SE-30 Gas Chrom

Q. 100/120 mesh, 220°). The mass spectrometer(LKB Model 9000) was adjusted to record theintensity of m/e 430 (M+), 415, and 374 at70 eV on three different channels. For

further details, see previous paper in thisvolume (Fig. 1).

Analysis of THC-7-oic acid in urine

The procedure described above for analysisof human plasma can also be applied tourine. In a preliminary study we equilibrated100 ng 3H- 6-THC-7-oic acid for 20 min with5.0 ml human urine and added 2.5 ml 1 Mcitrate-HCl buffer to maintain pH 4.1 during

the extraction with equal volume of diethyl

ether (7.5 ml x 1).

Figure 1

Fig. 1. Mass fragmentograms of the silylethers of the isomeric THC-7-oic acid methylesters from purified plasma extracts.

RESULTS AND DISCUSSION

The partition of 3H- 6-THC-7-oic acid (pre-

pared by acid catalyzed exchange) between

buffer solutions of different pH and anorganic phase (diethyl ether, light petro-leum with 1.5% isopentanol, and toluene,respectively) was studied to determine the

combination of pH and extraction solventgiving the best recovery. Diethyl etherprovedtobe the most efficient solvent(log D = 2.2, pH 4.1). The recovery from

66

plasma of added 3H- 6-THC-7-oic acid (in

the range of 50-200 ng) was over 85%after extraction with a double volume of

ether. The recovery from urine was over

90% using equal volumes. The major invivo metabolite l-THC-7-oic acid is ex-

pected to behave almost identically.

The diazomethane was effective in esteri-fying the 6-THC-7-oic acid but thecorresponding acid metabolite of l-THC

was more difficult to esterify. However,

the proceduredescribed using a 10 minincubation with excess of reagent wasapplicable to both acids. The elutionvolume on Sephadex LH-20 of the methylester of 6-THC-7-oic acid was determinedby calibration with ng amounts of tritia-ted compound. From earlier experiments

l-THC is known to have an elution volumeslightly higher than that of 1-THC. In

our study the delayed elution of the

ester of l-THC-7-oic acid, in comparison

to i t s 6-isomer, had to be compensated

for by a wide fraction of collection.

Total recovery of 3H- 6-THC-7-oic acidin plasma after extraction, methylation,

and column purification was around 70%.The silylated methyl ester of 6-THC-

7-oic acid is not fully separable from thel-isomer on the conventional GLC columns

tested (SE-30, SE-52, OV-17, JXR, XE-60).When not silylated, the methyl esters

were resolved but that of the 1-isomerpartly decomposed on the column. Thus,

the silylated methyl ester of 6-THC-7-oicacid was used as an external standard for

t h e l-isomer since both compounds havevery similar retention times but differentintensities in the mass fragmentograms(Fig. 2, 3). The 6-isomer can also serveas a reference for semiquantitative estima-

Figure 2

Fig. 2. Mass spectra of the silyl ethers of 6- (upper panel) and l -THC-7-oic acidmethyl ester (lower panel).

67

Figure 3a Figure 3b

Fig. 3. a. T h e s i l y l e t h e r o f l -THC-7-oicacid methyl ester (m/e 430, 374, 415).

tions of l-THC-7-oic acid in plasma or urine.

The possible concomitant occurrence of 6 -THC-7-oic acid in plasma or urine samplesfrom smokers should be of little significancesince 6-THC at most occurs in 2% of theamount of 1-THC in Cannabis (Ohlsson et al.,1977). There is only a limited need foran accurate quantitation method for l -THC-7-oic acid since pharmacological testwith Rhesus monkeys (Edery et al., 1971)have shown that the isomeric 6-THC-7-oic

Fig. 3. b. The silyl ether of 6 -THC-7-oicacid methyl ester.

acid is inactive in doses up to 10 mg/kg

i.v. If necessary, the deuterium labelledl-THC-7-oic acid can be synthesized as

an internal standard according to methodsdescribed by Pitt and Wall (1974) and Pittet al., (1975).

There are no studies on the urinary excretionrates of l-THC-7-oic acid in humans afterCannabis smoking. However, the studies on

l-THC metabolism in humans by Wall et al.(1974) suggest that identification of thisacid may be a practical method to identifyCannabis users. The sensitivity of themethod has not been thoroughly tested yetbut is at least in the range of a few ng/ml.Further studies are in progress.

68

ACKNOWLEDGMENTS

The support of the Medical Research

Council was appreciated. Mass spectro-

metry was supported by grants to the

Department of Toxicology. Reference

acids were kindly provided by Drs. R.

Mechoulam and M. Wall.

Lemberger, L., Silberstein, S.D., Axelrod,

J. & Kopin, I.J. (1970). Science, 170,

1320-1322.

Lemberger, L., Axelrod, J. & Kopin, I.J.(1971). N.Y. Acad. Sci., 191, 142-152.

Mechoulam, R., Ben-Zvi, Z., Agurell, S.,

Nilsson, I.M., Nilsson, J.L.G., Edery, H.& Grunfeld, Y. (1973). Experientia, 29,1193-1195.

Nordqvist, M., Agurell, S., Binder, M.

& Nilsson, I.M. (1974). J. Pharm. Pharmac.

6, 471-473.

REFERENCES

Agurell, S., Gustafsson, B., Holmstedt, B.,Leander, K., Lindgren, J.-E., Nilsson, I.,

Sandberg, F. & Åsberg, M. (1973). J. Pharm.Pharmac. 25, 554-558.

Ben-Zvi, Z. & Burstein, S. (1974). Res.

Commun. Chem. Pathol. & Pharmacol.

8, 223-229.

Burstein, S., Rosenfeld, J. & Wittstruck, T.

(1972). Science N.Y. 176, 422-423.

Edery, H., Grunfeld, Y., Ben-Zvi, Z. &

Mechoulam, R. (1971). Ann. N.Y. Acad. Sci.

191, 40-53.

Harvey, D.J. & Paton, W.D.M. (1976).

Examination of the metabolites of l-

tetrahydrocannabinol in mouse liver,

heart and lung by combined gas chromato-graphy and mass spectrometry. In press.

Ohlsson, A., Abou-Chaar, C.I., Agurell,

S., Nilsson, I.M., Olofsson, K. &Sandberg, F. (1971). Bulletin on Nar-

cotics, 23, 29-32.

Pitt, C.G., Fowler, M.S., Sathe, S.,Srivastava, S.C. & Williams, D.L. (1975)J. Am. Chem. Soc., 97, 3798-3802.

Pitt, C.G.& Wall, M.E. (1974). Synthesisof radiolabeled cannabinoids and meta-

bolites. Annual report from ResearchTriangle Institute, North Carolina, USAto National Institute on Drug Abuse.

Wall, M.E., Brine, D.E. & Perez-Reyes,M. (1974). Metabolism of cannabinoidsin man. Presented at Conference on the

pharmacology of Cannabis, Savannah, Dec.3-6, 1974. In press in Pharmacology of

Cannabis, Eds. S. Szara & M. Braude,Raven Press, New York, USA.

69

QUANTITATION OF CANNABINOIDS IN BIOLOGICALSPECIMENS USING PROBABILITY BASEDMATCHING GC/MS

Donald E. Green, Ph.D.Veterans Administration HospitalPalo Alto, California

INTRODUCTION

The major effort in our laboratory in thedevelopment of THC analytical methodologyhas been on evaluating the efficacy of extrac-tion procedures, on the TLC fractionation ofindividual in vivo metabolites, and on thedevelopment of an automated GC/MS quantita-tive procedure. For the first two aspects,we have relied heavily upon 14C-studies. InCalifornia it is almost impossible to admini-ster 14C-labeled drugs to humans for researchpurposes, so our model animals were usuallythe higher primates, Rhesus and baboon (PPA).We found that the excretion rates of totalradioactivity in both urine and feces forRhesus monkey and for man (Lemberger et al,1971) were very similar (Figures 1 and 2), somost of our early development efforts em-ployed Rhesus specimens.

EXPERIMENTAL

Baboon feces

Extractions: Fecal specimens, althoughof no direct practical importance, were use-ful for orientation and for methods develop-

Figure 1

Figure 1. Comparative urinary excretion of9-THC and its metabolites in man and three

primates.

70

Figure 2 Table 1

Figure 2. Comparative cumulative fecalexcretion of parent drug and metabolitesafter oral administration of 9-THC in manand three primates. (Human data fromLemberger et al, 1971)

ment because they contained copious quanti-ties of the parent drug and its less exten-sively metabolized forms. (In this lattersense, they can serve as a general model forTHC in blood.) In order to minimize the unpleasantness of working with feces, a generaltechnique, based upon continuous extractions,was developed. In typical experiments inwhich 14C- 9-THC was administered to chroni-cally THC-dosed baboons, 24-hour fecal col-lections were homogenized in methanol usinga Waring Blender and then transferred into asoxhlet thimble where overnight extractionwith methanol removed more than 99% of thetotal counts.

Methanolic extracts from individual specimenswere evaporated to dryness and the residueswere suspended in distilled water and ex-tracted for 48 hours with petroleum etherin a liquid/liquid extractor (Fraction I).The aqueous phases were adjusted to pH 10 andextracted with diethyl ether for 18 hours(Fraction II). After adjusting the aqueousphases to pH 2.5, final ether extractions werecontinued for another 18 hours (Fraction III).These three extractions removed nearly 90% ofthe original counts as shown in Table 1 (Greenet al, 1975).

Fraction I, which contained 4/5 of the totalradioactivity, was separated by TLC into threemain groups of compounds: A non-polar band

Continuous Liquid/Liquid Extractionsof Soxhlet-Extracted Residues

Pet ro l eum Ether (neut ra l p I I ) 80 .2%

Diethyl Ether (pII 10) 3 . 2

Diethyl Ether (pII 2.5) 6 . 4

Total Recovery 8 9 . 8 %

Figure 3

Figure 3. Thin layer radiochromatogram ofpetroleum ether extract of baboon feces.(Silica gel, petroleum ether/diethyl ether1:1, 3 passes.)

(42%) near the Rf of 9-THC; a moderatelypolar diffuse region (9%) in the approximateposition of mono-hydroxy- 9-THC's; and somehighly polar compounds near the origin (16%)(Figure 3). A corresponding non-radioactivepetroleum ether extract - obtained from fecescollected on the day prior to radio-label dos-ing - was separated by preparative TLC (onsilica gel with 3 passes in chloroform) into5 Fast Blue B-reacting bands: Band I-A, Rf0.88; Band I-B, Rf 0.68; Band I-C, Rf 0.14;Band I-D, Rf 0.12; and Band I-E, Rf 0.08.

High pressure liquid chromatography: Analiquot of Band I-A was dansylated and a smallportion was analyzed using a gradient, highpressure liquid chromatograph (Varian Aero-Graph Model 8500) equipped with a 25 cm NH2-bonded 10 µ silica column and an experimentalfilter fluorometer detector (Figure 4)

71

Figure 4 Figure 5

Figure 4. High pressure liquid chromatog-raphy of Dansylated Fraction I-A from petro-leum ether extract of baboon feces (top) andmixture of Dansylated standards (bottom).

(Loeffler et al, 1975). From this chromato-gram, it can be seen that I-A had a major com-ponent with the same retention time as 9-THC.Minor components at the approximate retentiontimes of CBN and CBD were also present.

Gas chromatography/mass spectrometry: Asimilar aliquot of Band I-A was silylated withBSTFA and analyzed by GC/MS (Figure 5). Theautomated GC/MS report (the inset in Figure 5)shows that 9-THC and CBN were abundantlypresent (2.02 µg and 544 ng, respectively),but that little, if any, of the common mono-and di-hydroxylated or carboxy metaboliteswere present in this TLC band.

On our instrument, the automated GC/MS anal-yses can also be generated and displayed in amore complete and accurate mode, if desired.The actual mass fragmentographic data fromwhich the analytical results are derived canbe printed for any single individual compoundat that point on the GC peak at which the bestmass spectral match between the experimentalsample and the library spectrum occurred.ThiS type of data for the two compounds whichwere judged to be present in Band I-A areshown in Figures 6 and 7 which show that

Figure 5. FID Chromatogram of silylatedFraction I-A of petroleum ether extract ofbaboon feces. (Column: 6' X 2 mm ID pyrexwith 1.5% OV-17 on 100/120 mesh HP ChromasorbG; 30 ml nitrogen/min.; 275°C.) Inset: Auto-mated GC/MS analytical data. Shaded areasindicate retention time intervals during whichpositively identified components were scanned.

eight characteristic ions from the spectrumo f 9-THC were present in the proper inten-sity ratios (and at the correct retentiontime) and that nine such ions were recordedfor CBN. As an illustration, the five high-est mass ions from the 9-THC contracted spec-trum were monitored as a function of time(Figure 8) and it can be clearly seen thatthe GC peak is entirely homogeneous with res-pect to 9-THC. The ion currents reveal thatthe typical pattern was already recognizableat about 10 ng.

The quantities of 9-THC and CBN in these in-dividual compound assays represent a totaldaily output of 1.48 mg and 390 µg, respec-tively. The first of these figures indicatesthat about 8.5% of the total 9-THC which wasadministered was excreted as unmetabolizeddrug at "steady state."

The value obtained from CBN helps to settlethe question as to whether CBN, whose presencehas been reported in plasma (Wilman et al,1975; McCallum et al, 1975), is truly a meta-

72

Figure 6 Figure 7

O L F A X / G C A N A L Y S I S

T S - 2 0 7 T I - 2 9 5> H E X A N E E X T R A C T O F P P A # 3 0 F E C E S : 0 . 2 3 % O F F R A C T I O N I - A .

D E L T A - 9 - T H C +POS Q T Y 3 . 0 8 E + 3 C O N F 1 0 2 RT 81

MASS I N T E N S I T Y

6 7 C 2 . 2 2 E - 79 3 C 6 . 1 6 E - 8

1 8 1 5 . 6 8 E - 91 9 6 3 . 5 2 E - 92 1 9 8 . 3 8 E - 92 3 1 6 . 3 2 E - 92 4 6 1 . 2 0 E - 83 3 0 1 . 4 9 E - 83 4 3 2 . 7 8 E - 83 7 1 8 . 0 3 E - 8

P L O T Y O R N ? Y# OF DECADES=1

F S = 2 . 2 2 E - 7

0 1 2 3 4 5 6 1 8 9 1 0+ - - - - + - - - - + - - - - + - - - - + - - - - + - - - - + - - - - + - - - - + - - - - + - - - - +

++

6 7 C + > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > >9 3 C + > > > > > > > > > > > > > >

1 8 1 + >1 9 6 + > >2 1 9 + > >2 3 1 + >2 4 6 + > > >3 3 1 + > > >3 4 3 + > > > > > >3 7 1 + > > > > > > > > > > > > > > > > > >

+++

O L F A X / G C A N A L Y S I S

T S = 2 0 9 T I = 2 9 8> H E X A N E E X T R A C T O F P P A # 3 0 F E C E S : 0 . 2 3 % O F F R A C T I O N I - A .

C A N N A B l N O L +POS Q T Y 8 . 0 8 E - 1 C O N F 1 3 6 RT 1 0 9

MASS I N T E N S I T Y

1 6 5 C 7 . 6 8 E - 91 7 8 6 . 8 4 E - 92 3 8 9 . 3 6 E - 92 9 5 1 . 3 4 E - 33 1 0 2 . 0 5 E - 33 2 3 7 . 6 8 E - 93 6 7 2 . 2 7 E - 73 6 8 8 . 2 0 E - 83 6 9 2 . 2 4 E : 33 8 2 1 . 0 4 E - 6

P L O T Y O R N ? Y# OF DECADES=1

F S = 2 . 2 7 E - 7

0 1 2 3 4 5 6 7 8 9 10+ - - - - + - - - - + - - - - + - - - - + - - - - + - - - - + - - - - + - - - - + - - - - + - - - - +

++

1 6 5 C + > >1 7 8 + > >2 3 3 + > >2 9 5 + > > >3 1 3 + > > > > >

3 3 3 + > >3 6 7 + > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > > >3 6 8 + > > > > > > > > > > > > > > > > > >3 6 9 + > > > >

3 8 2 + > >+++

Figure 6. Single compound GC/MS analysis of Figure 7. Single compound GC/MS analysisFraction I-A of petroleum ether extract of (Confirmation mode) of Fraction I-A of petro-baboon feces indicating 3.08 ng of 9-THC with leum ether extract of baboon feces indicatinga Confidence Index of 102 at a retention time 0.808 µg of CBN with a Confidence Index ofof 81 secs. Note that only 2 ions (m/e 67 136 at a retention time of 109 secs. (onlyand 93) out of 10 are contaminated. m/e 165 is not present in the required rela-

tive abundance for pure CBN.)

bolic product or arises as the excretion ofa CBN impurity in the administered THC. Ourwork with standards indicates that there isusually a maximum of about 3% CBN in the

9-THC obtained from NIDA. If this were alsotrue of the lot which was used to dose thisbaboon, a maximum quantity of about 0.5 mgcould have been administered. The finding ofnearly 0.4 mg in this specimen indicates arecovery of nearly 80% of the maximum amountof impurity possibly present. Leo Hollister'sgroup, with which we have collaborated closelyfor several years in their intensive studiesof the urinary excretion of various metabo-l i t e s o f 9-THC, CBN, CBD, and 11-HO- 9-THC,has shown that orally administered CBN isvery extensively metabolized (Kanter, 1975).

Thus it seems unlikely that more than half ofthe "administered" CBN should be recoverablefrom the feces in a single fraction espe-cially after ail the precessing this samplereceived - and therefore it would appear thatCBN is a genuine metabolite of 9-THC. Theremaining four bands derived from the petro-leum ether extraction contained none of theeight compounds in our test panel.

The continuous ether extraction at pH 10(Fraction II) contained 54 µg 11-HO- 9-THC(total daily output) and 210 µg cannabinol-11-oic acid (providing further evidence for themetabolic origin of CBN), but the presence of

9-THC-11-oic acid could not be unequivocallydemonstrated. The ether extract obtained at

73

Figure 8

Figure 8. Top: Single ion mass chromatogram est masses used in the automatic assay pro-(m/e 386) for 1 µg 9-THC. Center: 5-Mass gram). Bottom: Same as center, but withselected ion record for 1 µg 9-THC (5 high. 3-decade logarithmic presentation.

pH 2.5 (Fraction III) contained 93 µg of 8 ,11-dihydroxy- 9-THC, but no other metabolite inour panel could be conclusively identified.

The foregoing GC/MS results were obtained usinga computer-managed Olfax II spectrometer sys-tem (Universal Monitor Corp., Pasadena, CA)which utilizes Processor McLafferty's "Proba-bility Based Matching" (PBM) algorithm(McLafferty et al, 1974). In the PBM pro-cedure, reverse sear& logic (Abramson, 1975)is used to compare a many-line mass spectralpattern derived from an experimental samplewith a corresponding spectrum stored in alimited library. In the present study, thelibrary was specially tailored to the Problemof differentiating between various cannabinoidsand their metabolites.

Probability based matching

McLafferty's PBM technique was specificallydesigned to monitor individual compounds inmixtures by "resolving" their merged mass spec-tra and therefrom to automatically generate twokinds of information: 1) quantitation and2) an estimate of the degree of specificity ofthe determination. Both of these analyticalvalues are derived from an ability to automa-tically ascertain whether or not any particularmass spectral line (fragment ionj is contami-nated by the simultaneous presence of an impu-rity. Such judgments are possible because ofone particular characteristic physical propertyof mass spectra; when two or more differentcompounds contribute an identical molecularweight fragment ion to a merged spectrum. the

74

combined ion currents are always additive.This undirectional variation can be used by thecomputer to detect an artificially enhancedabundance of any individual ion in the massspectral pattern of a pure substance whicharises as result of a contribution from asecond substance.

Since Probability Based Matching concerns themanner of specific ion recording and the pro-cessing of these data, a brief description ofthe PBM algorithm may help in understandingthe type of results which it produces. In thereverse search process, it is not necessary toscan an entire mass spectrum, but only to mea-sure a set of ions which are particularlycharacteristic of the target compound. A prin-cipal advantage of this concept is that a com-plete search for a limited, or “contracted,”spectrum can be accomplished in less than asecond and therefore the contracted spectraof several different compounds can be ac-curately recorded repetitively during theelution of even very sharp GC peaks.

In the Olfax II GC/MS system a microprocessordirects the spectrometer to record the ioncurrents for each of the pre-selected ionsin the contracted spectrum and subtracts theinstrument background from these values(Hertel et al, 1975). It then automaticallycomputes the relative intensities and com-pares the resulting ratios with the storedlibrary spectrum in order to detect any con-taminated ions; it simultaneously comparesthe absolute intensities of each ion with theintensities of the corresponding lines in thebackground and disregards any ions which areso weak that they are approaching backgroundlevels. All contaminated lines and thosewhich are near background intensity are ig-nored in subsequent calculations.

Confidence Indices: From the remaining un-contaminated lines, a statistical probabilityis automatically computed which reflects thequality of match between the experimentalspectrum and the reference spectrum of thepure compound. McLafferty calls this prob-ability a “Confidence Index” (K-score). Itsmagnitude depends mainly upon the absolutenumber of uncontaminated lines, the relativeimportance of each specific uncontaminatedline, and the “dilution” of the sample.

Of these parameters, the concept of relativeimportance of a particular ion - “linevalue” - is unique to the PBM algorithm. Theline value (or V-score) is a combination ofthe uniqueness of that particular mass number(frequency of occurrence in McLafferty’slibrary of 18,800+ compounds) and the rela-tive abundance of that particular ion in theentire mass spectrum of that specific com-

pound (Pesyna et al, 1975). Unequivocalidentification of a spectrum is obviouslyeasier when using unusual mass peaks thanwhen one is forced to rely on common peaksand when using strong ions rather than weakones. Actually, it is often better to havea weak unique ion than a strong common ion.V-scores provide a convenient way to auto-matically make such judgments without beingexperienced in the interpretation of massspectra.

Quantitation: All of these criteria (andmore, which time does not permit mentioning)are incorporated into the K-score which iscalculated each time the contracted spectrumis scanned (i.e., several times a second).Simultaneously, the absolute intensity ofeach uncontaminated ion is compared with theabsolute intensity of each correspondingion - resulting from a given quantity of thepure material and stored as a part of thecomputer library - to produce an independentmass fragmentographic quantitation for eachuncontaminated ion. All of these scaled ioncurrents are then averaged to provide anestimate of the quantity, Q, of substancepresent at that point in time at which thecontracted spectrum was scanned. A signi-ficant aspect of PBM quantitation is thatthe computer automatically selects the un-contaminated ion which represents the lowestquantity of the target compound (based uponthe required relative abundances) and there-fore is a self-adaptive system which reportsthe maximum quantity of that compound whichcould possibly have been present in thesample. Since the K-score reflects the massspectral specificity of the uncontaminatedions upon which quantitation is based, itsmagnitude is helpful in establishing theconfidence one can place in the quantitativeresults.

A highly unusual feature of the Olfax systemis that it is entirely digital; the onlytruly analog output it provides is a conven-tional FID record from the GC. The manner inwhich data are recorded is shown in Figure 9.Each repetitive set of 5 ions constitutes asingle cycle ( i .e . , one scan of the con-tracted spectrum) which is compared againstthe library values and for each cycle a newK and Q are calculated. The computer reportsonly the very best fit observed during anyseries of scans (highest K-score) and themaximum ion currents generated (highest Q),e.g., cycle 5 in this illustration.

To show how these digital data relate to con-ventional specific ion recording, a set ofdata like those of Figure 9 have been plottedin Figure 10. (In normal operation, at leastthree to five cycles are scanned between each

75

Figure 9 Figure 10

2 / 8 / 7 6 O P E R A T O R # 2 3P R O G R A M D 9 - T H C T Y P E S C N 2RUN # 2 TS=197 TI=298> 0 . 7 5 U G D E L T A - 9 - T H C / T M S

M A S S I N T E N S I T Y

2 4 6 3 . 7 6 E - 1 23 3 0 8 . 2 8 E - 1 23 4 3 2 . 2 6 E - 1 23 7 1 1 . 1 1 E - 1 13 8 6 1 . 8 1 E - 1 1

2 4 6 3 . 6 6 E - 1 13 3 3 3 . 5 2 E - 1 13 4 3 1 . 9 4 E - 1 03 7 1 3 . 8 4 E - 1 03 8 6 3 . 0 4 E - 1 0

2 4 6 5 . 4 4 E - 1 03 3 0 5 . 5 2 E - 1 03 4 3 1 . 5 8 E - 93 7 1 3 . 8 4 E - 93 8 6 2 . 4 0 E - 9

2 4 6 2 . 4 3 E - 93 3 0 3 . 4 0 E - 93 4 3 7 . 0 0 E - 93 7 1 1 . 6 5 E - 83 8 6 7 . 8 8 E - 9

2 4 6 3 . 2 8 E - 93 3 0 4 . 2 6 E - 93 4 3 9 . 9 2 E - 93 7 1 2 . 3 4 E - 33 8 6 1 . 0 9 E - 8

MaximumCycle

2 4 6 2 . 7 2 E - 93 3 0 3 . 4 8 E - 93 4 3 6 . 2 4 E - 93 7 1 1 . 7 8 E - 8

Figure 9. Representative digital datarecorded during elution of 0.75 µg 9-THC GCpeak. (Intensities are in amperes, e.g. m/e371 in maximum cycle is 2.34 X 10-8A.)

of the points depicted in Figure 10 - and atypical cycle contains 10 to 15 ions, ratherthan 5 as shown here.) Figure 11 shows threemodes in which an Olfax contracted spectrummay be displayed. At the top is a single ionmass chromatogram; in the middle is a fairlyconventional specific ion record (note therecurring “pattern” of relative ion currents);and at the bottom is a logarithmic presenta-tion of the same data as in the middle (notehow much more easily the relative abundancescan be compared from cycle to cycle and alsothat the correct pattern can be readily dis-cerned at a very low level of signal - i.e.,at less than 10 ng).

Selection of specific fragment ions

The PEN reverse search technique is such apowerful pattern matching procedure that

Figure 10. Plot of typical digital data(like that of Fig. 9) for 0.9 µg 9-THC.

almost any set of ions selected from thespectrum of a compound will give acceptablequalitative and quantitative results. How-ever, for sets of compounds having the simi-larity of the cannabinoids, especially wherehigh sensitivity is desired, selection of theoptimum mass lists requires careful considera-tion of many factors.

The main types of differential criteria whichcan be incorporated into a PBM assay programare: 1) positive attributes of the targetcompound (namely, high V-scores); 2) negativeattributes of challenges (compounds which arelikely to be encountered at similar retentiontimes); and 3) absolute intensities of theselected ions relative to their backgroundintensities (which affect the minimum de-tectable quantity at a given mass number).Thus one must attempt to maximize the K-scorefor the target compound while simultaneouslyminimizing the K-score for challenges (bydeliberately selecting lines which result incontaminated or weak ions for the challenges)and all the while keep in mind the backgroundion currents for each mass being considered.

76

Fugure 11

Figure 11. Top: Mass chromatogram for 9-THCin Fraction I-A of petroleum ether extract of

tion). Bottom: Same as center, but with

baboon feces. Center: 5-Ion selected ion3-decade logarithmic presentation. (5-Highest

record for the same sample (linear presenta-masses used in automatic assay program.)

The closer the challenge comes to the reten-tion time of the target, the more critical itis to select contaminated or weak ions. Thusfor 11-hydroxy- and 8 , 11-dihydroxy- 9-THC/TMS, the retention times differed by less than1% and it became more important to screenagainst the challenging compound than toscreen for the target. Because their spectraare quite different this could be accomplished,but only by sacrificing some sensitivity. Thepair 8 , 11-dihydroxy- and 8 , 11-dihydroxy-

9-THC/TMS had similar spectra but differedmarkedly in retention time so that they couldbe easily differentiated by GC alone. Un-fortunately, a different situation was encoun-tered in the pair 8 -hydroxy- and 8 -hydroxy-

9-THC/TMS which have nearly identical spectraand nearly identical retention times; there-

fore even PBM was unable to produce uniqueassays for the individual compounds. Accord-ingly, a single program was produced whichwas labeled “8 -hydroxy- 9-THC,” but whichwas in reality an 8 - and/or 8 -hydroxy- 9 -THC assay.

Responses of the PBM algorithm

The performance of the PBM algorithm can bestbe understood by looking at the manner in whichit responds to specific single compounds. Ifthe Confidence Index is monitored as a func-tion of sample size, a decreasing value isobserved with decreasing quantity of sample(Figure 12). This is due primarily to twoeffects. At the upper end of the plot, itdecreases due to a “dilution” correction be-

77

Figure 12

Figure 12. Effect produced on PBM ConfidenceIndex by varying the quantity of pure 9-THCinjected.

cause it is assumed that low ion currents arefrequently the result of dilution with impur-ities and, if this is truly the case, suchreadings are statistically less valid. At thebottom end of the curve, in addition to thedilution correction, the weaker lines ap-proach background levels, and are dropped fromconsideration because they are too weak tomeasure reliably.

Numerically, K is a binary logarithmic repre-sentation of the probability of an accidental(random) match of the observed quality be-tween the unknown spectrum and the libraryspectrum. A rough approximation of the signi-

ficance of the K-scores is that 2Kth spectra,chosen at random, would have to be comparedwith the library spectrum before one encoun-tered a spectrum which would match as well asdoes the experimental spectrum. Thus a K of10 means roughly that 210 (approximately 1,000)spectra would have to be examined in order tofind as good a fit.

Obviously, the numbers grow very rapidly -K=20 is approximately one million - so thatonly a qualitative meaning can be given tovery large K-scores. As the result of twoyears of experience with the method, we cansay that a K of 30 to 40 is somewhat equivo-cal, but a Confidence Index of 60 to 70 isvirtual certainty of proper identification,within the limits of low resolution massspectrometry (even without retention timeconsiderations). Differential K-scores areof great significance as is easily seen inFigure 12 where 8-THC, when used to chal-lenge the pattern for the 9-THC program,gives a K-score of 11, more than 100 pointslower than does 9-THC (the target compound).Differentials in K-score are likewise used toestablish the "thresholds for reliable iden-tification" for each program. Somewhat arbi-trarily, this threshold is set at 10 unitsgreater than the maximum K-score generated byany known challenging compound or control spe-cimen ( i .e . a "safety" factor of 1 ,000) . Inthe case of 9-THC shown in Figure 12, theConfidence Index threshold for reliable iden-tification was set at 21, indicating that onecan be reasonably certain that it is 9-THCwhich is actually being monitored in anyanalysis which results in a K of 22 orgreater.

The ability of these PBM assay programs (allof which contained ten fragment ions) to dis-tinguish between closely related compounds canbe seen in the matrix shown in Figure 13. TheConfidence Index (the upper number in eachbox) for each target compound is compared withthat of the THC metabolites eluting immediate-ly before it and after it. In most cases, thedifference (in K-scores) between the correctcompound and its challenge is greater than100. Besides this Confidence Index screening,many adjacent pairs can be differentiated bytheir GC retention times (the lower number ineach box).

As one gains experience with these ConfidenceIndices, the analytical significance of thisstatistical parameter becomes "abundantlyclear". Thus, for each and every quantitation,the likelihood that the data are valid is im-mediately apparent by inspection.

Along with the computation of the ConfidenceIndex, the PBM algorithm provides a quantita-tive response which is linear for well overtwo decades (as may be seen in Figure 14).Olfax II quantitation usually has less than a10% error and its precision has typically a2-5% coefficient of variation.

For the test panel used in the present study,the 10 best lines which were found for each ofthe compounds, along with some of the other es-sential parameters for the Olfax algorithm

78

Figure 13 Figure 14

Figure 13. Confidence Indices and retentiontimes for nine cannabinoids. Reported as

K for 1 µg samples. Column tempera-RT(sec)ture 275°C.; nitrogen carrier flow rate30 ml/min.)

Table 2OLFAX II ASSAY PROGRAMS

Figure 14. Quantitation reported by the massspectrometer (PBM algorithm) as a function ofthe quantity of 9-THC actually injected intothe GC.

arator temperatures (TS), and line-values (V)

Table 2. Confidence Index “thresholds for for the ions (m/e) used in the 8 cannabinoidreliable identification” (KT), molecular sep- assay programs.

79

Figure 15 Figure 16

Figure 15. Extraction scheme used for-glucuronidase / aryl sulfatase hydrolyzed9-THC urines indicating derivation of the

H, E-I and E-II Fractions.

(V-scores for the individual lines, ConfidenceIndex “thresholds”, and optimum temperaturesfor the molecular separator), are listed inTable 2.

Human urines

Although working with feces is, relativelyspeaking, easier than with most other biologi-cal specimens, no great practical value can bederived from their analysis. Therefore, wehave concentrated our efforts upon investiga-ting urine, which would appear to be the mostpractical biological specimen, provided thatsome metabolite could be identified and moni-tored which follows the time course of physi-cal impairment due to THC. (We have some pre-liminary evidence that suggests that we haveseen such a metabolite.)

Extractions: Four years ago, we developed arather elaborate procedure for fractionatingTHC metabolites according to their polaritiesand acidities using simple changes of solventsand/or pH (Figure 15) (Forrest et al, 1972;Melikian et al, 1973). This scheme separatedthe total radioactivity in enzyme-hydrolyzedRhesus urine into six approximately equal frac-tions: Hexane extractable, H (non-polar neu-t ra l s ) ; ether (pH 12) extractable, E-I (weaklypolar neutrals); ether (pH 2.5) extractable,E-II and E-III (weakly polar acids); ethyl ace-tate (pH 2.5) extractable, EA (moderately polaracids) ; and tetrahydrofuran (pH 2.5) extract-able, THF (highly polar acids). Figure 16shows the relationships between the sizes ofthese fractions in hydrolyzed, as well as un-hydrolyzed, urines. The actual values aregiven in Table 3. Subsequently, this procedure

Figure 16. Distribution according to polarity(extractability) of 9-THC metabolites in vivo(hydrolyzed and unhydrolyzed Rhesus urine).

Non-polar neutrals = extractable with hexaneat neutral pH

Weakly polar neutrals = extractable with etherat pH 12

Weakly polar acids = extractable with ether atpH 2

Moderately polar acids = extractable withethyl acetate at pH 2

Highly polar acids = extractable with tetra-hydrofuran at pH 2.

Table 3

Extraction Efficiencies of Various Solvents (Percentof total counts extractable from Rhesus monkey urine)

Solvent and pII Unhydrolyzed Urine Hydrolyzed Urine

Hexane 0.1% 6.7%(natural pII)

Ether 0.9% 11.0%(pII 12)

Ether 47.8% 48.3%(pII 2)

Ethyl acetate 26.8% 15.2%(pII 2)

Tetrahydrofuran 23.9% 14.6%(pII 2)

Residue 1.4% 1.3%

Total 100.9% 37.1%

80

was refined by resolving the weakly polar acids(the largest single fraction) into strong andweak acids according to their solubility inNaHCO3 (E-III and E-II, respectively)(Kanteret al, 1974).

Most of our efforts have been concentrated onthe three least polar fractions, H, E-I, andE-II, which contain the parent cannafioidsand the mono- and di-hydroxys, as well as the“11-oic” acids. Standards do not distributecleanly between these categories; 9-THC and11-hydroxy- 9-THC, for instance, distribute inthe ratio of about 4:1 between the hexane frac-tion and the ether at pH 12 fraction (E-I).8,11-Dihydroxy- 9-THC, on the other hand, isfound predominantly in the E-I fraction (80%)with only about 5% each in fractions H, E-IIand E-III.

Hollister’s group has applied this fractiona-tion scheme very extensively to urines obtain-ed from human volunteers after oral administra-tion of pure standards of 9-THC, CBN, CBD,11-HO- 9-THC, singly and in combinations(Hollister, 1973; Kanter, 1975). They haveobserved many interesting phenomena which canbest be summarized briefly as: 1) the exis-tence of a multitude of drug-related metabo-lites; 2) very few matches with availablestandards; and 3) the extreme persistance (10-15 days) of highly polar metabolites after asingle administration of drug.

High dose specimens: As part of our own workusing these same procedures, our group wasvery-fortunate to be able to obtain urine spe-cimens from subjects used in some unusualexperiments last spring at the University ofCalifornia Langley Porter Clinic in San Fran-cisco (Jones et al, 1975). These subjectsreceived the extremely high oral doses of 210milligrams of 9-THC per day.

The urine was hydrolyzed with -glucuronidase/aryl sulfatase and fractionated according tothe foregoing extraction scheme. Aliquots ofthe dried extracts were derivatized with BSTFAand screened by GC/MS. All the “positives”which were detected, and any determinationswhich showed high K-scores (although “nega-tive”), were re-run in the more comprehensivesingle compound mode.

The hexane fraction showed only CBN on initialscreening (Figure 17), but the confirmationmode re-run showed a marginal positive for9-THC at 10 ng/ml (Figure 18). Since thispositive identification was achieved with onlytwo ions. a more conventional mass chromato-gram was run (Figure 19). This showed clearlythat there indeed was a GC component with thehighly unique m/e 371 which eluted at theretention time of 9-THC.

In addition to the 9-THC, a confirmation modere-run showed a fairly clear-cut presence ofCBN (K=28) at a level of 12 ng/ml (Figure 20).

The weakly polar neutral fraction (E-I) showeda very marginal positive for CBN at 0.8 ng/ml,which probably was due to a slight carry-overfrom the hexane fraction, and a fairly strongpositive for 8 ,11-dihydroxy- 9-THC at 24 ng/mlin the screening run (Figure 21). When frac-tion E-I was re-run in the single compoundmode, it no longer showed a definite positivealthough the 8 ,11-dihydroxy- 9-THC was a near-miss with three weak ions in the correct ratiosand corresponding to a maximum concentration of22 ng/ml. The screening run for the weaklypolar, weak acid fraction (E-II) indicated pos-i t i v e s f o r 9-THC (K=24), 9-THC-11-oic acid(K=94) and CBN-11-oic acid (K=37) (Figure 22))but only the THC acid could be unequivocallyconfirmed (Figure 23). The 9-THC-11-oic acidconcentration corresponded to 206 ng/ml in ahighly specific determination. That theseintensity ratios were in fact derived from asingle substance which eluted at the properretention time can be clearly seen in Figure 24(top), as can the fact that the observed inten-sity ratios do, in fact, correspond to those ofauthentic 9-THC-11-oic acid (bottom).

One aspect of PBM quantitation should be stres-sed in connection with the concentration of

9-THC-11-oic acid which was measured. The PBMalgorithm provides an upper limit to the pos-sible quantity present. From this we must con-clude that a maximum quantity of about 0.4 mgof this acid could have been excreted in the24 hours during which the intake was 210 mg of

9-THC. Since the urinary portion of the totalexcretion is about 20% (Lemberger et al, 1971),there would probably have been about 40 mg ofmetabolites in all forms. Therefore the 9-THC-11-oic acid represents something of the orderof 0.5 to 1% of the total metabolite pool.Allowing for dose-dependent variations of up to10-fold, this particular acid could amount tono more than 10%, at best, of the total drugexcreted in urine. This finding is entirelyconsistent with the TLC data which have beenobtained in our laboratory and in Hollister’slaboratory (Kanter, 1975), as well as inJones’ laboratory (Ellman, 1975). There arecopious quantities of many acidic metabolites,but the 9-THC-11-oic acid per se is definite-ly a minor component.

Moderate dose specimen: An examination of aslightly more typical urine specimen was car-ried out on some extracts provided to us bySaul Kanter of Hollister’s group. These frac-tions, H, E-I, and E-II, were extracted fromthe urines of an individual before and afterhe received a single oral dose of 30 mg of

9-THC; specimens were collected 12 hours priorto drug intake and during the interval of 12 to24 hours post-drug.

81

Figure 17 Figure 18

too high (c. f. pattern in Figs. 8, 9 and 11).

Figure 18. Single compound GC/MS analysis(Confirmation mode) of Fraction H from hydro-lyzed human urine (same as Fig. 17) indicating45.4 ng of 9-THC in 5 ml urine with K=24 at aretention time of 82 secs. Note that the com-

non-polar neutrals (H) fraction of human urinebecause it knew that they would be too close

(enzyme hydrolyzed).- GC conditions same asto background levels and that the computer

Fig. 5. Inset: Automated GC/MS analyticaljudged 3 of the remaining ions to be contam-inated, i.e. their relative abundances were

Figure 17.puter elected not to scan the lowest 5 masses

FID Chromatogram of silylated

data (Autoscan mode).

Figure 19

Figure 19. Single ion (m/e 371) mass chro-matogram for same sample as Figs. 17 and 18.

Note that this ion becomes cleanly measurableat about 5 ng.

82

Figure 20 Figure 22

Figure 20. Confirmation mode GC/MS analysisof Fraction H from hydrolyzed human urine(same as Fig. 17) indicating 52.0 ng of CBN in5 ml urine with K=28 at a retention time of110 secs. Nine out of the 10 masses werescanned, but 7 were contaminated.

Figure 21

Figure 21. FID Chromatogram of silylatedweakly polar neutrals (E-I) fraction of humanurine (enzyme hydrolyzed). GC conditionssame as Fig. 5. Inset: Automated GC/MS anal-ytical data (Autoscan mode).

Figure 22. Chromatogram of silylated weaklypolar, weak acids (E-II) fraction of humanurine (enzyme hydrolyzed). GC conditions sameas Fig. 5. Inset: Autoscan mode GC/MS data.

Figure 23

Figure 23. 9-THC-11-oic acid/TMS analysisof Fraction E-II (same as Fig. 22) showing1.39 µg/5 ml hydrolyzed urine with K=52 at aretention time of 227 secs. All 10 lineswere scanned, but 6 were contaminated.

83

Figure 24 Figure 25

Figure 24. Top: Selected ion record of 9-THC-11-oic acid GC peak from Fraction E-II(same sample and quantity as Figs. 22 and 23). Figure 25. Effect of pH on extractability ofBottom: Same kind of selected ion record of 9-THC in vivo metabolites from unhydrolyzed0.75 µg of authentic 9-THC-11-oic acid. Rhesus urine.

We found a maximum quantity of about 50 ng of9-THC per 10 mg creatinine in the hexane

fraction, but the Confidence Index was too lowto identify the compound with certainy. TheCBN, however, was clearly evident (K=59) at alevel of about 100 ng/10 mg creatinine;

The E-I fraction (weakly polar neutrals) gavea very marginal confirmation for 8 -HO- 9-THCat a level of about 60 ng/10 mg creatinine anda suspiciously high reading of nearly 0.6 µgof 8 , 11-dihydroxy- 9-THC per 10 mg creatin-ine with a K-score of 35. The E-II fraction(weakly polar, weak acids) from this specimen,surprisingly, failed to give a positive for

9-THC-11-oic acid.

Unhydrolyzed urines: Four years ago, we re-ported that an appreciable proportion of thetotal counts in Rhesus urine could be extractedwith ether, even without prior hydrolysis(Forrest et al, 1972; Melikian et al, 1973).This extraction was highly pH-dependent withthe curve's inflection point being near pH 4

(Figure 25) indicating that it was probablycarboxylic acids which were being removed bythe ether.

As one part of an experiment on urines obtainedfrom the Langley Porter Clinic Study, we ex-tracted and another specimen with ether before andafter treatment with enzyme at pH 5.5. Unfor-tunately this pH was far from the optimum pHfor removal by ether, but the results areinteresting, nevertheless. A highly signifi-cant Confidence Index (K=68) for 9-THC-11-oicacid was obtained along with a quantitation of36 ng/ml (Figure 26). Had the extraction beenmade at pH 2.5, a quantity of about 90 ng/mlwould probably have been extracted, if thecompound follows a normal ionization curve forcarboxylic acids. (This level suggests thatabout half of the 9-THC-11-oic acid which ispresent in urine can be extracted directly,without hydrolyzing the specimen.)

Single specific compound analysis confirmedthe screening results as did a selected ionrecord, as shown at the bottom of Figure 27.

84

Figure 26 Figure 28

Figure 26. Silylated ether extract (pH 5.5)of human urine prior to enzyme hydrolysis.Inset: Autoscan analysis.

Figure 28. Silylated ether extract (pH 5.5)of human urine after enzyme hydrolysis.Inset: Autoscan analysis.

Figure 27

Figure 27. Top: Selected ion record of urine at RT of 9-THC-11-oic acid. Bottom:0.75 µg of authentic 9-THC-11-oic acid.Center: E-II Fraction of hydrolyzed human

Ether extract of unhydrolyzed human urine atRT of 9-THC-11-oic acid.

85

Figure 29

Figure 29. 8α,11-Dihydroxy- 9 - T H C / T M Sanalysis of ether extract of human urineafter enzyme hydrolysis showing 1.20 µg/5 mlwith K=35 at a retention time of 150 secs.Three ions are uncontaminated.

This selected ion record also shows that oneion (m/e 355) becomes contaminated shortlyafter the peak has eluted and is the type ofsituation which would have been rejected bythe computer if it had occurred a few secondsearlier. The center record was obtained fromthe E-II (weak acids) fraction of the previ-ously discussed enzyme hydrolyzed urine, andthe top record is authentic 9-THC-11-oic acid.

Figure 28 shows the results from the post-hydrolysis ether extraction. Only 8 ,11-di-hydroxy- 9-THC appeared to be present, but itsidentification was very marginal (K=32)although it was slightly better in the confir-mation mode (K=35) (Figure 29).

SUMMARY

The results from these urine studies and a fewother isolated extracts we have run may begeneralized as follows: The hexane fractionsshowed low levels of 9-THC (ca. 10 ng/ml)with marginal Confidence Indices (25-30), butboth the quantities and the K-scores were wellabove the values seen in extracts of controlurines; the hexane fractions showed similarlevels of CBN (ca. 12-25 ng/ml) with signifi-cant K-scores (60-80); the ether-extractableneutral fraction contained 8 ,11-dihydroxy-

9-THC (25-150 ng/ml) with low confidence andsometimes a suggestion of 8 -hydroxy- 9-THC(15 ng/ml) with a marginal identification; andthe E-II fraction (ether-extractable, weakacids usually contained a strong indication(CONF 60-95) of 9-THC-11-oic acid (100-200ng/ml).

CONCLUSIONS

Obviously, the limited scope of these datawhich I have presented cannot be considered tobe much more than a feasibility study, but theresults are quite promising. Three metabo-l i t e s o f 9-THC are consistently observed and

9-THC per se is seen and quartitated with adegree of certainty which requires only slightimprovement. Since the 9-THC determinationswere not entirely satisfactory, suggestionsfor future efforts aimed at improving thisparticular assay might include partially shift-ing the burden of identification from the massspectrometer to the GC by improving the GCseparation using such means as operating at alower column temperature or increasing thelength of the column. The effectiveness ofthe PBM assay program might be improved byprocessing a larger sample to decrease theeffect of weak ions in the algorithm or byimproving the sensitivity of the mass spectro-meter itself. Also, since 11-hydroxy- 9-THCwas not encountered in any urine samples, the8α,11-dihydroxy- 9-THC program should be modi-fied so as to make it better able to withstandgeneral challenges even though in the processit becomes less specific with respect to11-hydroxy- 9-THC.

In all of the work described here, astandard test panel of TMS derivatives of

9-THC and seven of its metabolites wasemployed. Assay programs for additional meta-bolites, such as the side-chain hydroxylatedcompounds described by Professor Agurell,could easily be added to the panel if appro-priate reference samples were made availableto us.

The Olfax method, in which GC retention timescreening is combined with the PBM mass spec-trum matching technique, appears to be a very

86

apparatus which are available in any labora-tory, the labor and analysis time have beenminimized, and the final assay uses a com-mercially available, completely automatedinstrument which any skilled technician canoperate and which does not require priorexperience with mass spectrometers or anability to interpret mass spectra.

promising procedure for the highly specificquantitative analysis of cannabinoids be-cause it automates and makes practical thebest features of mass fragmentography. Itcannot do anything which would not be pos-sible using conventional GC/MS systems, butits degree of automation provides an advan-tage in speed and convenience while retainingthe inherent specificity of mass fragmento-graphy. The extractions employ solvents and

ACKNOWLEDGMENTS

Parts of this work were supported by ContractHSM-42-71-100 and by USPHS Grants DA 00424-01and DA 00748-02.

The baboon specimens were very kindly suppliedby Eva E. K. Killam of the University of Cali-fornia (Davis) whose cooperation and interestis greatly appreciated. Special thanks areexpressed to George L. Ellman of the Universityof California (San Francisco) for providing theunique urine specimens and also to Saul L.Kanter of this Hospital for providing specialurine extracts. I wish to thank RichardSchneider of Syva Corporation for the sample ofCBN-11-oic acid and Robert H. Hertel of Uni-versal Monitor Corporation for valuable discus-sions during the course of this work. The veryable technical assistance of Fu-Chuan Chao andKay O. Loeffler is gratefully acknowledged.

Loeffler, K.O., Green, D.E., Chao, F-C., andForrest, I.S. (1975) Proc. West. Pharmacol.Soc. 18, 363-368

McCallum, N.K., Yagen, B., Levy, S. andMechoulam, R. (1975) Experientia 31, 520-521

McLafferty, F.W., Hertel, R.H. and Villwock,R.D. (1974) Org. Mass Spectrom. 9, 690-702

Melikian, A.P., Green, D.E., Skinner, J.L.and Forrest, I.S. (1973) Proc. West.Pkarmacol. Soc. 16, 234-239

Pesyna, G.M., McLafferty, F.W., Venkataragha-van, R., and Dayringer, H.E. (1975)

Finally, most special thanks are due to Duane P. Anal. Chem. 47, 1161-1164Littlejohn and the Universal Monitor Corpora-tion for the loan of a prototype Olfax II Widman, M., Nordqvist, M., Dollery, C.T. andGC/MS instrument system, without which this Briant, R.H. (1975) J. Pharm. Pharmacol.work would not have been possible. 27, 842-848

REFERENCES

Abramson, F.P. (1975) Anal. Chem. 47, 45-49

Ellman, G.L. (1975) Personal communication

Forrest, I.S., Green, D.E., and Würsch, M.S.(1972) Proc. Fifth Intern. Cong. onPharmacol.

Green, D.E., Killam E.E.K., Loeffler, K.O.,Chao, F-C. and Forrest, I.S. (1975)Pharmacologist 17, 268

Hertel, R.H., Green, D.E. and Strauss, P.A.(1975) Proc. 26th Pittsburgh Conf. on Anal.Ckem. and Appl. Spectr.

Hollister, L.E. (1973) Experimentia 29, 825-826

Jones, R., Ellman, G.L. and Lysko, H. (1975)Unpublished data

Kanter, S.L., Hollister, L.E., Moore, F. andGreen, D.E. (1974) Res. Comm. Chem. Path.Pharmacol. 9, 205-213

Kanter, S.L. (1975) Personal communication

Lemberger, L., Tamarkin, N.R., Axelrod, J.and Kopin, J. (1971) Science 173, 72-74

87

QUANTITATION OF -TETRAHYDROCANNABINOLIN BODY FLUIDS BY GAS CHROMATOGRAPHY/CHEMICAL IONIZATION -MASS SPECTROMETRY

Ruthanne Detrick and Roger FoltzBattelle, Columbus LaboratoriesColumbus, Ohio

As part of a program for the develop-ment of methodology for quantitationof cannabinoids in body fluids, wehave validated an extraction and GC/CI-MS analysis for the determinationo f 9-THC in plasma. The procedurehas proven suitable for the analysisof relatively large numbers of sam-ples, yet has sufficient sensitivityand selectivity for analysis of9-THC concentrations as low as 0.5

ng/ml in plasma. Using this proce-dure, we are currently analyzing avariety of body fluid samples fromdifferent sources submitted by out-side researchers.

Previous work at Battelle and otherlaboratories indicated that substan-tial clean-up of the body fluid ex-tract was necessary prior to GC/MSanalysis in order to achieve adequatesensitivity. Although several clean-up procedures have been used success-fully, many of these are cumbersomeand not well suited for analysis of

large numbers of samples. The pro-cedure we are currently using is amodification of the RTI solvent ex-traction procedure (Rosenthal, 1975).The analysis scheme is essentially anorganic solvent extraction and clean-up, concentration and derivatizationof the 9-THC, followed by quantita-tion by GC/CI-MS analysis.

Specifically, the plasma sample istransferred with a calibrated pipetto a culture tube. To this is addedpH 7.0 buffer solution and a knownamount of 9-THC-d3 internal stan-dard. The mixtures is vortexed toassure complete mixing, then extract-ed twice with hexane. The combinedhexane extract is successively wash-ed with 0.1 N NaOH and 0.1 N HCl.The hexane extract is transferred toan evaporation tube and evaporatedto dryness under a stream of nitrogenusing a tube heater maintained at be-low 50 c. The tube is rinsed with

88

pentane and the pentane solutiontransferred quantitatively to aReacti-Vial(R). After removal of thepentane, BSTFA [N,O-bis-(trimethyl-silyl)-trifluoroacetamide] with 1%TMCS is added, and the vial is heat-ed for a minimum of 30 minutes at75 C.

The concentration of 9-THC is deter-mined using the technique of select-ed ion monitoring (SIM) with a GC-MS.A 6 ft x 2-mm glass column packedwith 3% OV-17 on 100/120 mesh GasChrom Q is temperature programmedfrom 180 to 280 C at 10°/minute.Methane is used as the carrier gaswhile a small amount of ammonia isbled into the ion scource through amake-up gas inlet. The masses moni-tored are m/e 387, the protonatedmolecule ion of the trimethylsilylether ( 9-THC-TMS) and m/e 390, theprotonated molecule ion of the TMS

ether of the trideuterated internalstandard ( 9-THC-d3-TMS).

Previously we used methane as thereagent gas for chemical ionizationof the GC effluent. We have nowfound that addition of ammonia to theion source improved the signalstrength for a specific quantity of9-THC-TMS by a factor of about 3.Furthermore, ammonia being a verymild reagent gas is more selectivethan methane so that there tends tobe less interference from other com-ponents of body fluid extracts. Also,the relative intensity of the M-2 peakis significantly lower when ammoniais used rather than methane. There-fore, the 9-THC-d3-TMS will makeless of a contribution to the MH+ ioncurrent of 9-THC-d0-TMS. Figure 1shows the methane and methane-ammoniaCI mass spectra of 9-THC-TMS, while

Figure 1

Figure 1. Comparison of mass spectra of 9-THC-TMS89

Table 1

IonizationMethod m/e Monitored

% of Sample Relative Response Per

Ion Current Unit Weight of 9-THC

EI 386 (M+) 7 50

CI (CH4) 387 (MH+) 24 2 0

CI (CH4+NH3)

CI (N2+NH3)

387 (MH+)

387 (MH+)

67

67

100

80

TABLE 1. RELATIVE RESPONSES FOR PROMINENTIONS IN THE EI AND CI MASSSPECTRA OF 9-THC-TMS

Table 1 compares the response perunit weight of 9-THC for differentmodes of ionization.

A disadvantage of using ammonia asthe reagent gas is the lack of asecond prominent ion which can bemonitored as a check on possible con-tributions to the MH+ ion currents byother components of the body fluidextracts. Chemical ionization withmethane gives several prominent frag-ment ions in addition to the MH+.However, methane is a less selectivereagent gas than ammonia, and in ourexperience intefering peaks are moreof a problem with methane than withammonia. Of course, it is also im-portant to carefully select the GCcolumn and operating conditions tominimize interference from endogenouscomponents of the body fluids. Wehave found it necessary to use temp-erature programming of the GC column.

Quantitation of the 9-THC is achievedby monitoring ion masses associatedwith the 9-THC-TMS and the deuterat-ed internal standard, and measuringthe peak height ratios of the ion9currents. The actual amount of 9 -THC is determined by comparison ofthe ratio to a standard curve estab-lished at the time of each set ofanalyses. The standard curve is pre-pared by simply mixing aliquots ofstandard solutions of 9-THC and

9-THC-d3, removal of the solvent byevaporation, derivatization, and SIManalysis. Figure 2 shows a typicalcalibration curve prepared in theabove manner. The slope is 1.03 andthe zero intercept corresponds to apeak height ratio of 0.0074 or a

9-THC concentration of 0.3 ng. Thecurve has a correlation coefficientof 1.000. In almost all cases thestandard curve has been linear overthe concentrations measured, and the

90

Figure 2

Figure 2. Calibration curve for analysis of 9 -THC-TMS

Table 2

9-THCAdded

Sample Ht./Standard Ht. RelativeExperimental Standard

Theoretical (Average) DeviationNumber of

Determinations

96.8 ng 2.20 2.28 1.23 3

48.4 1.10 1.11 0.32 3

9.7 0.22 0.23 6.08 3

4.8 0.11 0.13 6.65 3

1.0 0.02 0.03 14.00 2

0.5 0.01 0.02 1.80 2

0.0 0.00 0.02 1

TABLE 2. RELATIVE STANDARD DEVIATIONS FOR POINTS ON CALIBRATION CURVE(44.1 NG of 9-THC-d3 ADDED TO EACH)

relative standard deviation at anygiven peak height ratio has typicallybeen less than 20, even for the lowerconcentrations. The relative stan-dard deviations for the points onthis particular calibration curve arelisted in Table 2. Changes in theslope and intercept of the curve havebeen observed, but the greatest

differences in the 9-THC valuesassigned to a sample by using differ-ent calibration curves have been lessthan 10%. To allow for changes dueto instrument settings or to a changein the concentration of the internalstandard solution, we continue toestablish a calibration curve for eachset of samples.

91

Table 3 summarizes the precision andaccuracy data for a series of analyses.Known quantities of 9-THC and 9-THC-d3 were added to 1-ml portions ofplasma. The final volume of eachderivatized extract was about 25 µl.Injections into the GC were between 2and 4 µl. The actual quantity of

9-THC injected on column is there-fore approximately one-tenth of theamount of 9-THC added to each ml ofplasma. Figure 3 is a typical com-puter plot of an extract of 1 ml ofplasma containing 1 ng of 9-THC. Inthis example the peak in the m/e 387ion current curve corresponds to about100 pg of 9-THC-TMS injected on col-umn. The corresponding peak in them/e 390 ion current curve correspondst o 9-THC-d3 concentration of 44.1 ng/ml. The m/e 390 ion current signal isattenuated by a factor of 10 relativeto the m/e 387 ion current signal.

Figure 3

Figure 3. Computer plot of ion currents form/e 387 and 390 corresponding to 1 NG/ML of

9 -THC and 44 NG/ML of 9 -THC-d3. The m/e390 ion current curve is attenuated by afactor of 10.

Table 39-THC 9-THC

Added Measured(ng/ml) (Average)

RelativeStandardDeviation

Number ofDeterminations

PercentError

9.7 7.7 1.7% 10 21%

4.8 3.8 2.5% 10 21.%

1.0 0.65 6.4 9 35%

0.5 0.28 19.0% 10 44%

0 -0.13(a) 15.0% 8 -

(a) The average of the measured weights corresponded to a value below thezero intercept of the calibration curve.TABLE 3. PRECISION AND ACCURACY DATA FOR MEASUREMENT OF 9-THC ADDED

TO PLASMA (44 NG OF 9-THC-d3 ADDED PER ML OF PLASMA ASINTERNAL STANDARD)

92

Figure 4

Figure 4. Plot of data points from SIM analysis of extracts of 1-MLplasma samples containing 0 to 9.7 NG of 9 -THC and 44.1 NG of 9 -THC-d3.

Finally, Figure 4 plots data pointscorresponding to 50 separate SIM deter-minations on 2 sets of extracts fromplasmas to which known amounts of9-THC and 9-THC-d3 were added. The9-THC concentrations ranged from 0 to

9.7 ng/ml of plasma.

Using this procedure, we are currentlyanalyzing a variety of plasma samplessubmitted through NIDA by other re-searchers. These have included rat,

monkey, and human plasma samples withvolumes in the range of 0.25 to 2.0 ml.Levels of 9-THC in the samples havebeen typically in the range of 5-50 ng/ml of plasma, although samples contain-ing concentrations of 9-THC as highas 400 ng/ml of plasma have been ana-lyzed. The actual procedure for ana-lyzing the unknowns is as previouslydescribed. To increase our confidencein the integrity of the results forany given set of samples, spiked plasma

93

samples containing a known amount of9-THC are analyzed at the same time.

These have been useful in pointing outvarious problems with interferences,and help to establish some confidencelimits for the measured values, par-ticularly at the lower levels. In addi-tion, a GC column resolution check anda sensitivity check of the GC-MS aremade at the beginning of each day topinpoint any instrumental problems.

Although further modifications in theanalysis scheme are likely, the pro-cedure presently being used hasproven suitable for the analysis of9-THC in plasma to concentrations as

low as 0.5 ng/ml and can be conve-niently used for relatively largenumbers of samples.

Experimental Section

The plasma sample (usually 1 ml) istransferred to a 16-ml culture tubeusing a calibrated pipet and the vol-ume is recorded. To this is added 1.0ml of pH 7.0 buffer solution and 100µl of 441 ng/ml solution of 9-THC-d3in ethanol. The mixture is vortexedfor 30 seconds. The plasma is thenextracted with 5 ml of Distilled-in-Glass(R) hexane for 30 minutes on arotator. The mixture is centrifugedand the hexane extract transferred toa second tube. The hexane extractionis repeated and the extracts combined.The combined hexane extract is suc-cessively washed with 2.5 ml 0.1 NNaOH, and 2.5 ml 0.1 N HCl. The -hexane extract is transferred to anevaporation tube and evaporated todryness under a stream of nitrogenusing a Kontes tube heater maintainedat below 50 C. The tube is rinsedwith 1 ml of Distilled-in-Glass(R)

pentane and the pentane solutiontransferred quantitatively to a 1.0-ml Reacti-Vial(R). The pentane isremoved under a stream of nitrogen atroom temperature. The sides of thevial are washed down with 100 µl ofpentane, and the pentane is againevaporated. To the vial is added 25µl of BSTFA with 1% TMCS. The vialis capped using a Teflon linerand heated for 2 hours at 75 C.

The concentration of 9-THC is deter-mined using the technique of selectedion monitoring with a Finnigan 3200GC-MS. A 6 ft x 2-mm glass columnpacked with 3% OV-17 on 100/120 meshGas Chrom Q is temperature programmed

from 180 to 280 C at 10o/minute.Methane is used as the carrier gaswhile a small amount of ammonia isbled into the ion source through amake-up gas inlet. The partial pres-sure of ammonia in the ion source is100 µ. The flow rate of methane is20 ml/min giving a partial pressure inthe ion source of about 700 µ. Othermass spectrometer parameters are:electron energy, 210 v; ion energy,15 v; lens voltage, %30 v; electron

multiplier voltage, 1.4 to 1.6 KV;and filament emission, 0.8 to 1.0 ma.

Cross-contamination from the syringeused for injecting the samples intothe gas chromatography is avoided byrinsing 10-20 times with pentane be-tween injections.

For every 5 unknowns, a spiked samplecontaining 9.7 mg 9-THC/ml of plasmais prepared by adding 100 µl of a 97ng/ml solution of 9-THC in ethanolto 1 ml of human plasma. This isextracted in the same manner and atthe same time as the unknowns.

A standard curve is established atthe time of each set of analyses byadding 100 µl of the 441 ng/ml 9-THC-d3 solution in ethanol and 100 µl ofone of various concentrations of 9 -THC in ethanol directly to a Reacti-

A seriesVial(R).standards with d0/d3

of four suchratios ranging

from 0-1 for 50 ng of 9-THC-d3 areprepared. The ethanol is removed atbelow 40 C, and the THC derivatizedin the same manner as the unknowns:25 µl of BSTFA with 1% TMCS is added,and the vial is heated at 75 C for 2hours. These standard samples areanalyzed by monitoring the m/e 387 and390 peaks of the mass spectrum. Sincethere should be no interferences inthese standards, the gas chromato-graphic column is used isothermally at205 C. The values for the ratio ofd0/d3 are used to prepare a standardcurve.

All reusable glassware used in theanalysis is rinsed with methanol,cleaned with detergent and water,rinsed with water, soaked in hot sul-furic and nitric acid solution, rins-ed successively with tap water, dis-tilled water and methanol, and finallydried in an oven at >100 C.

Aliquots of standard 9-THC and 9-THC-d3 solutions are measured out

94

using a 100-µl Eppendorf pipet withdisposable tips. Syringes used for The concentration of the 9-THC-d3measuring out BSTFA or injecting stock solution in ethanol was deter-samples into the GC-MS are cleaned in mined by comparison with a primarymethanol. The hexane and pentane used standard of 9-THC in ethanol usingare high-purity, Distilled-in-Glass(R) SIM analysis.solvents.

REFERENCE

D. Rosenthal, Second NIDA Tech-nical Review on Mass Spectro-scopy, Oak Brook, Illinois,October 22, 1975.

95

HPLC-MS DETERMINATION OF-TETRAHYDROCANNABINOL

IN HUMAN BODY SAMPLES

Jimmie L. Valentine, Ph.D., Paul J. Bryant, Ph. D., Paul L. Gutshall, M.S.,

Owen H.M. Gan, B.S., Everett O. Thompson, B.S., Hsien Chi Niu, Ph. D.

University of Missouri, Kansas City, Missouri

INTRODUCTION

Social and chronic abuse of cannabisis believed to occur throughout theUnited States. However, most infor-mation on the societal use of mari-juana, in particular, comes fromquestion-response type surveys. Pre-cise quantitative data obtainable viabody specimen analysis has been un-available due mainly to the lack ofan accurate biological assay tech-nology for the chemical constituentsfound in cannabis.

The marijuana commonly smoked in thiscountry contains four principal con-stituents, viz., 9-tetrahydrocanna-binol ( 9-THC), cannabidiol (CBD),cannabinol (CBN) , and cannabichromene(CBC) (Doorenbos, et al., 1971). Oneof these constituents, 9-THC is be-lieved to be responsible for the psy-chomimetic properties of marijuana(Edery, et al., 1971). Likewise someof the physiological responses in manhave been shown to change followingsmoking of cigarettes containing 9-THC and these correlate with plasmalevels of 9-THC and its metabolites

(Galanter, et al., 1972). However,such studies were accomplishable onlyby administering radiolabeled 9-THCto the study subjects and followingits deposition in accessible bodyfluids.

More recently there have been a num-ber of methods proposed for determi-nation of 9-THC in blood plasma. Aglc procedure which utilized an elec-tron capture detector was detailedand reported to be capable of detect-ins 0.5 nq per 1 ml of blood plasma(Fenimore; et al., 1973). This me-thod which required precolumn deriva-tization and a dual oven apparatuswas not used to assay the blood plas-ma of actual cannabis smokers. An-other reported method employed glc-mass fragmentography with d2-

9-THCas the internal standard (Agurell, etal., 1973). This later method wasemployed to analyze 5 ml samples ofblood plasma from three volunteerseach of whom smoked a cigarette con-taining 10 mg of 9-THC. A chromato-graphic elution of the blood plasma

96

on a Sephadex column was requiredprior to glc-ms analysis. Blood sam-ples were taken from the volunteersat 0, 0.2, 0.5, 1 and 2 hours follow-ing smoking. Peak levels of 9-THCwere found to be 19-26 ng/ml at 10minutes following smoking. The lev-els declined to 5 ng/ml or less at 2hours. Another method which has beenreported to be capable of detecting9-THC is based upon radioimmunoassay

procedures (Teal, et al., 1974 andGross, et al., 1974).

The method reported herein makes useof a high pressure liquid chromatog-raphy-mass spectrometry (hplc-ms)technique for determining ng/ml quan-tities of 9-THC in human'body speci-mens. Problems associated with di-rect coupling of the hplc to the mshave been circumvented by collectingfractions of the mobile phase eminat-ing from the hplc and subsequentlyanalyzing the fraction via the directinsertion probe of the ms. Inherentin this method is the use of d3-

9-THC for controlling extraction effi-ciency, as a marker for collection ofthe hplc effluent and as a conven-ient internal standard for ms quan-tification. Thus by this method thebody specimen has enough d3-

9-THCadded to allow its detection by au.v. spectrophotometer connected tothe output of the hplc. Once thed3-

9-THC is detected, fractions ofthe mobile phase eluant are collect-ed and introduced to the ms via thedirect insertion probe for quantifi-cation. This method was then vali-dated in various human body samplesover the concentration range of 1-100 ng/ml or ng/g depending upon thespecimen of interest.

The new assay technology was thenused to determine the blood plasmalevels of 9-THC in eleven male vol-unteers during a 24 hour period fol-lowing smoking of a marijuana cigar-ette. Also the new procedure hasbeen used to determine the presenceo f 9-THC in exhaled breath and sa-liva of volunteers following mari-juana smoking. Similarly, a corre-lation has been made, using the me-thod, between the blood plasma, bileand brain levels of 9-THC in humanpost-mortem specimens. The methodhas also been used to analyze humanblood samples from a 9-THC aerosolinhalation study on a blind basis.

EXPERIMENTAL

High pressure liquid chromatography

All hplc analyses were conducted ona Varian 8520 gradient elution liq-uid chromatograph utilizing a Varian635M spectrophotometer set at 273.7mµ as the detector. The column wasa 10 µ silica gel (Varian Si-10) , 25cm X 2 mm (i.d.). A gradient elu-tion program was developed using hep-tane and methylene chloride. For asatisfactory separation of the canna-binoids as well as to assure accuratems quantitation, it was necessary toroutinely record the u.v. spectrum ofeach lot of heptane and methylenechloride prior to its use in the gra-dient program. Only those lots ofheptane and methylene chloride whichgave minimum absorbance in the‘regionof 260-280 mµ were used.

The gradient elution program, devel-oped for this application, began at95:5%, heptane:methylene chlorideand proceeded to 95:5%, methylenechloride:heptane over 9 minutes.The program was reversed, i.e., from95:5% methylene chloride:heptane tothe initial 95:5% heptane:methylenechloride mixture, thereby regenerat-ing the column. A solvent flow rateof 120 ml/hr was used for all deter-minations. By employing these con-ditions 9-THC as well as d3-

9-THCwere found to have a retention timeof 4.7 minutes, i.e., they appearat a gradient elution mixture of 52:48 methylene chloride:heptane. Theother major constituents of mari-juana, CBD, CBN and CBC were foundto have retention times of 4.4, 4.6and 5.6 minutes, respectively. Theamount of d3-

9-THC added to thebody sample was sufficient to allowu.v. detection of 9-THC (labeledplus unlabeled) as it eminated fromthe column. A 10 cm "zero dead vol-ume" stainless steel tube was at-tached to the flow cell of the spec-trophotometer to facilitate collec-tion of the effluent droplets almostinstantaneously after passing throughthe flow cell.

Mass spectrometer quantification

All ms analyses were accomplishedusing a Varian MAT SMI-B high reso-lution, double-focus mass spectrome-ter. A new ion-counting technique

97

was developed in conjunction withthe peak matching accessory whichprovided for a rapid comparison be-tween data from the internal stan-dard (d3-

9-THC, mass 317) and theassayed compound ( 9-THC, mass 314).Each sample from the hplc was intro-duced into the ms via the direct in-sertion probe. The instrument wasinitially focused exactly on the 317(d3-

9-THC) mass signal, then throughthe action of the peak matching unit,and with the high resolution capabil-ity alternately focused to the 314( 9-THC) mass signal, cf., Figure 1.

Figure 1Liquid Sample (5 mcl)

D i rec t Inser t ion Probe

Mass Spectrometer

Peak Matching Acces.

3 1 7 3 1 4

Ion Coun te r

Data Processor - P lo t te r

Standard Data

R e g i s t e r A Reg is te r B

As this alternation from one signalto another occurs the exact numberof ion counts for each compound isrecorded and stored in two channelsof a Princeton Applied Research ModelSSR1110 dedicated computer. Thisunit performs a summation of the num-ber of ion events occurring in bothmass peaks (314 and 317) and storesthese in two registers. Thus a run-ning total of ions detected from the317 internal standard and the un-known amount of the 314 mass are

Figure 2Ion Countinq Cycle

Counting Period Counting Period

t RA = 0.067 sec. t RB = 0.067 sec.

M/e = 317 AMU M / e = 3 1 4 A M U

Ten Data Points

Fifty Cycles

5 0 5 0

Ion Counts (317) Ion Counts (314)

t R A = 1 tRB = 1

T = 5 0 tR A + 5 0 tR B

=6.7 sec.

One Data Point

1 0 x T = 6 7 s e c .

500 Counting Cycles

1,000 Counting Periods

stored separately by the counter asshown schematically in Figure 2. Thepeak matcher accessory is set todwell for 67 milliseconds on eachmass signal before it alternates tothe other mass signal. Repeated ex-perimentation has shown that a totalcounting time of 67 seconds combinedwith a probe temperature of 65° gaveoptimum results. Thus in 67 secondsthe alternating cycle is repeated500 times yielding 1000 bits of datafor comparison and quantification.With time the sample is depleted butthe ratio based on the internal stan-dard remains linear and provides fordependable quantification as shownin Figure 3. The curve shown in thisfigure was obtained from a Hewlett-Packard 9820 computer-plotter byusing the data from register B forthe ordinate and register A for theabscissa and determining the leastsquare best straight line. Theslope of the data line gives the 314/317 ratio. The amount of 9-THC(mass 314) is determined by multiply-ing this ratio times the knownamount of internal standard added.In actual practice the d3-

9-THC em-ployed contained a small amount ofundeutrated 9-THC (mass 314). A msdetermination of this amount of mass314 was made by analyzing 10 samplesof 1.6 µg d3-

9-THC which had beenadded to 1 ml of blood plasma by the

98

Figure 3

Table 1Table 1. Precision and Accuracy in Recovery of

9-THC Added to Human Plasma.

Added Found,a

ng/ml ng/ml0 11.3 (9.35-13.6)b

1 1.2 (0.6-1.5)5 4.5 (4.0-4.7)

10 10.5 (9.5-12.4)20 20.0 (19.1-21.1)40 41.2 (39.6-42.5)60 62.1 (61.3-62.8)80 82.8 (74.9-90.7)

100 103.7 (99.3-106.0)

n

10331033333

Average percent recovered 103.8

±RSD RE

11.15 --43.3 20.09.0 -11.12

10.0 5.05.14 0.03.58 3.01.26 3.59.55 3.53.65 3.7

b This value represents the amount of unlabeled9-THC in 1.6 µg of d3-

9-THC. Other foundvalues have been corrected for this amount.

a Average (range) of n determinations.

method described above, cf. Table I.This background value for 9-THC from the total amount of 9-THC re-(mass 314), inherently present in corded to give the actual foundthe internal standard, is subtracted value.

99

Figure 4

Methods of blood plasma analysis

All glassware used was silinized us-ing a reported method (Garrett andHunt, 1974). Whole blood was cen-trifuged at 2600 rpm for 20 minutesto obtain blood plasma. To 1 ml ofblood plasma was added 1.6 µg of d3-9-THC followed by three repetitiveextractions with 2 ml of petroleumether for each extraction. The ex-tracts were combined and evaporatedto dryness under nitrogen at roomtemperature. The resultant residuewas reconstituted in 300 µl of hep-tane and the entire solution injectedinto the hplc. A 100 µl wash of hep-tane was used on the vessel whichcontained the extracts and it was al-so injected into the hplc and thegradient elution program begun asdescribed above. When the peak for9-THC was noted on the recorder, the

eluant was collected in a silinizedscrew cap vial. Collection of theeluant was synchronized with the re-corder tracing such that equal amountswere obtained from either side ofthe symmetrical peak. In general,approximately 1 ml of eluant was col-lected for each 9-THC peak. Samplesthus collected from the hplc werestored at -5° until ms analysis.Prior to ms analysis each sample wasevaporated under nitrogen at roomtemperature to approximately a 10 µl

volume. A microsyringe was used totransfer this solution in two por-tions to a 5 µl gold cup. The solu-tion was allowed to air evaporateand the gold cup was introduced intothe ms by attaching it to the directinsertion probe. Quantification wasaccomplished as described above.

Validation of the assay method inblood plasma

Blood from 10 laboratory workersknown to be non-users of marijuanawas drawn and analyzed as detailedabove. These blood plasmas consti-tuted the control samples. Each sam-ple contained some background amountof 9-THC (mass 314) since the 1.6µg of d3-

9-THC, added to each plasmasample, contained an average of 11.3ng of 9-THC (mass 314), cf. TableI. To demonstrate the reproduci-bility and accuracy of the devel-oped method, pooled human plasmasamples were analyzed to which knownamounts of 9-THC had been added.Table I is a summary of this study.Figure 4 is a plot of the data givenin Table I. As outlined above, theintercept in Figure 4 does not passthrough the origin or zero valuesince some 9-THC (mass 314) ispresent in the d3-

9-THC used asthe internal standard.

100

Methods of breath analysis

For human breath collection a stan-dard face mask (Welch model 7500-30G) was modified by placing a 2.5cm diameter port directly in frontof the mouth. On the interior ofthe mask was placed a 4.5 cm (o.d.)by 2.0 cm thick rubber ring of suf-ficent flexibility to firmly holda specially molded polyethylene fil-ter. The polyethylene was formedinto a small wafer size, i.e., adisc 3.0 cm in diameter and 0.25 cmthick. This polyethylene wafer wasthus held directly in front of thesubject's mouth and approximately1.5 cm away from the lips. Figure5 gives a cross-sectional view of

this modified mask. For actualbreath collections each subject wasasked to breathe for one minute withthe mask positioned over the noseand mouth using deep inhalations andexhalations through the mouth. Thepolyethylene wafers were placed inand taken from the mask using dis-posable examination gloves. Afterbreath collection, each sample wasplaced in 10 ml of methanol in asilinized beaker and then ultra-sonicated for 30 minutes. To eachextract was added 1.6 µg d3-

9-THCand the solution was evaporated todryness at room temperature, undernitrogen. The residue was recon-stituted in 300 µl of heptane andanalyzed via hplc-ms as describedabove.

Figure 5

101

Method of saliva analysis

Saliva was collected from each vol-unteer in silinized collection tubesand immediately frozen until analy-sis. To 100 µl of saliva was added1.6 µg of d3-

9-THC followed by ex-traction 3 times with 2 ml of petro-leum ether for each extraction. Theextracts were combined and evapor-ated to dryness under nitrogen atroom temperature. The residue wasreconstituted in 300 µl of heptaneand analyzed by hplc-ms as discussedabove.

Method of bile analysis

Bile taken at post-mortem examina-tion was frozen until analysis. To100 µl of the bile was added 1.6 µgof d3-

9-THC and 1 ml of pH 7.0 buf-fer. The mixture was extracted 3times with 2 ml of petroleum etherfor each extraction and the extractscombined and evaporated to drynessunder nitrogen at room temperature.The residue was reconstituted in 300µl of heptane and analyzed by hplc-ms as discussed above.

Method of brain analysis

Brain samples from the cerebrum ta-ken at post-mortem examination werefrozen until analysis. To 5 g ofbrain sample was added 20 ml of pH7.0 buffer and 1.6 µg of d3-

9-THC.The entire mixture was homogenizedthen extracted 3 times with 10 ml ofpetroleum ether each time and theextracts combined and evaporated todryness under nitrogen at room tem-perature.

Marijuana smoking studies

Eleven healthy male volunteers be-tween the ages of 21 and 26 wereused in the marijuana smoking studies.Each subject was within 10% of idealbody weight and received both medi-cal and psychological exams prior toadmission. Values for the followingtests were determined prior to thestudy and each subject was requiredto be within the normal range:electrocardiogram, chest x-ray,creatinine, BUN, LDH, SGOT, alka-line phosphatase, blood sugar, cal-cium, phosphorus, bilirubin, totalprotein, albumin, cholesterol, uricacid, hematocrit, hemoglobin,platelet count and prothrombin time.

All subjects were moderate marijua-na smokers. Each was requested notto smoke marijuana for 2 days priorto reporting for the study. Thesubjects were brought into the hos-pital ward 12 hrs. prior to smokingand they were not allowed any foodor drink after 12:00 am of thestudy day. At approximately 8:00am of the test day, each subjecthad a heparin-lock placed in a fore-arm vein and 5 ml of blood was with-drawn and placed in a silinized hep-arin vacutainer tube. Each bloodsample was handled so that the blooddid not come in contact with therubber stopper. All blood sampleswere immediately centrifuged at 2600rpm and the plasma removed andplaced in another silinized tube andfrozen for later analysis. Thisinitial 5 ml of blood drawn priorto smoking constituted the 0 hr sam-ple. Each subject was then allowedto smoke one marijuana ci arettewhich contained 10.8 mq 9-THC.2.16 mg CBN, 0.9 mg CBC and 0.63 mgCBD. Upon completion of smoking,timing was begun. Blood samples (5ml) were withdrawn at 0.25, 0.5, 1,2, 3, 4, 12 and 24 hours. Eachblood sample was handled as des-cribed above. At the same time in-tervals saliva and breath sampleswere taken from each subject.

9-THC aerosol and oral administra-tion studies

Two normal male volunteers, who metthe same physical criteria listedfor the marijuana smoking volunteers,were used. Each subject was admin-istered an aerosol spray which con-tained 10 mg of 9-THC. Blood sam-ples were taken from each subject at0, 0.25, 0.5, 1, 1.5, 2.0, 3.0, 4.0,5.0 and 6.0 hours. Samples were sup-plied to our laboratory with a codenumber. In addition to these samples,some additional blood plasma samplesfrom subjects in an oral cannabanoidstudy were also coded and submittedfor analysis. All samples were ana-lyzed as discussed above.

RESULTS AND DISCUSSION

Analysis of trace substances in abiological system is often limitedby the sensitivity of an analyticalmethod. This was precisely the casewith the analysis of 9-THC sinceprior to the present work there were

102

not many analytical methods whichoffered the selectivity and sensi-tivity needed. The newly developedmethod reported herein was shown tobe reproducible, accurate and linearin the range of 1-100 ng/ml of humanblood plasma. Use of hplc rior toms quantification permits 9-THC tobe selectively separated from theother cannabinoids present in mari-juana preventing an erroneous anal-ysis with the other mass 314 canna-binoids, viz., CBD and CBC. Alsowhen the ms analysis is performedthe high resolution capability ofthe ms prevents any possible confu-sion of compounds, i.e., only 9-THCwill have the correct mass number.

Use of the stable isotopic' form of9-THC in the developed method allows

for a control of extraction efficien-cy as well as a marker for peak col-lection and quantification. Thus thed 3-

9-THC was added to the plasmaprior to extraction and gave a con-trol for any losses which might oc-cur during extraction or chromato-graphic procedures. In the describedmethod the plasma was extracted threetimes. Ordinarily three extractionswould be redundant when an internalstandard has been added. However,the main reason for using the multi-ple extractions is to introduce tothe ms a sufficient amount of thestable isotope for accurate quanti-fication while not using excessive

amounts of the isotope for the inter-nal standard. The other importantuse of the d3-

9-THC was to permitthe total amount of 9-THC (both la-bled and unlabled) to be detectableby the u.v. spectrophotometer at-tached to the output of the hplcsince 9-THC has a low extinctioncoefficient.

A human marijuana smoking study wasconducted and blood samples taken atappropriate time intervals for 24hours. Each sample was analyzed byhplc-ms and the results are givenin Table II. As observed from themean values in this table, the peakblood plasma concentration of 9-THCoccurs in all individuals, exceptL.G., at 0.25 hours. Also worthy ofnote in Table II is the fact thatmost subjects showed a level of 9 -THC in their control samples. Thisis readily explainable since all sub-jects were prior users of marijuanaand most likely had smoked sometimebefore reporting to the study.

The breath samples from each subjectin the controlled smoking study werealso analyzed by the new method.Since the volume of exhaled air wasnot measured, no meaningful quantifi-cation data could be obtained. Thatis, the subjects were asked tobreathe through the breath apparatusfor a period of one minute and inthis time the rate of exhalation in

Table 2Table 2. Amount of 9-THC (ng/ml) found in human

blood plasma following marijuana smoking.

SUBJECTS 0 . 2 5

23.039.034.166.738.457.939.743.715.334.620.6

.5 4

D.L. 3.6B.N. 1.2B.B. 1.6D.J. 0F.R. 4.2B.W. 0.4W.Z. 1.2R.B. 5.7L.G. 7.6T.V. 11.3K.C. 4.7

10.111.723.319.720.124.819.718.939.916.522.4

TIME (hrs.)

1

12.23.8

16.811.86.8

14.86.8

27.68.1

10.114.5

7.10.91.43.08.4

10.75.36.72.15.18.4

2

12.1 5.2MEAN 3.8 37.5 20.7 3.9 3.6 2.5 1.9

3

0.901.13.82.82.44.94.4

11.14.27.7

3.50.70

4.36.36.22.02.57.52.34.7

12 24

2.7 00 1.7

2.0 0.90 3.1

0.6 0.95.8 2.50 00.1 3.49.7 2.51.6 2.85.0 2.5

103

Table 3

SUBJECT 1

SUBJECT 2

ORALCANNABANOIDSUBJECTSd

Table 3.Blood Plasma Levels of 9-THC

Code No.aTime (min)

Following 9-THC ng of 9-THC

11A 0 7.4

10A 5 30.3

7A 15 9.6

16A 30 4.3

22A 60 14.6

12A 90 15.2

13A 120 14.95A 180 lostb

15A 240 24.5

4A 300 20.3

2A 360 4.918A 0 2.2

19A 5 10.4

14A 15 13.3

20A 30 11.8

9A 60 no samplec

3A 90 13.6

17A 120 no samplec

8A 180 7.1

21A 240 7.61A 300 1.4

6A 360 lostb

3 0e 0.1

4 300f 0.0

5 300g lostb

6 300e 0.0

7 300e 0.0

8 Of 0.49 0g 0.0

10 Of 4.0

4 days prior to receiving the oral formulation.e Receiving oral CBN.f Receiving oral CBD.g Receiving oral 9-THC.

104

a Code number of sample as supplied by Dr. Tashkin.b Laboratory accident resulting in sample loss.c Sample broken in transit.d All subjects were confined in the hospital ward for

each subject may have varied. Moreimportant, however, was the fact thata positive and statistically meaning-ful level of 9-THC could be deter-mined in the breath of these knownmarijuana smokers during the initial60 minutes following smoking.

Saliva samples from marijuana smokershave to be quantitatively treatedvery much like the breath samples.That is, in the present study no at-tempt was made to accurately controlthe amount of saliva exudated. Thusany attempt to quantitatively assignlevels of 9-THC per unit volumewould be meaningless. However, itwas observed by analyzing the salivasamples that a positive, statistic-ally significant level of 9-THCcould be detected in the saliva ofknown marijuana smokers during theinitial 60 minutes following smoking.

Analysis of the blood samples whichcame from a 9-THC aerosol studyprovided an opportunity to analyzesamples on a blind basis. The dataobtained from this study is shownin Table III. Positive levels of9-THC for the control samples (0

hour) is readily explainable sinceeach subject was a marijuana userand was not confined prior to thestudy. In contrast, however, theoral cannabanoid subjects were con-fined for 4 days prior to the col-lection of their control samples.Also as shown in Table III, othercannabanoids were being administeredwhich could have conceivably inter-fered with the 9-THC determination.However, the hplc program success-fully separated the 9-THC fractionas was verified by correlation fol-lowing the blind sample assays. For

both studies there was a good corre-lation between the physiologicalstates recorded following administra-tion and the 9-THC levels measuredlater by means of the quantificationtechnology reported in this paper(Tashkin, 1976).

CONCLUSIONS

A method was developed for quantita-tively determining 9-THC in humanbody specimens. The method was shownto be accurate and precise over aconcentration range of 1-100 ng/mlin human blood plasma. This methodwasused to assay blood plasma, breath andsaliva of human subjects followingsmoking of a marijuana cigarette.Results from these studies demonstra-ted that during the early time inter-vals following marijuana smoking thelevels of 9-THC measured are of Suf-ficient magnitude to be clearly dis-cernable from the amount of 9-THC(i.e., 11.3 ng) added in the internalstandard.

The newly developed method has alsobeen used to study the relationshipbetween 9-THC levels in blood plas-ma, bile and brain of post-mortemsamples. Such data has revealedthat 9-THC levels are higher inbrain and bile than in blood plasma.

Data was also accumulated on bloodplasma samples which came from sub-jects receiving either an aerosolformulation of 9-THC or oral formu-lations of CBD, CBN and 9-THC.Analysis of these samples on a blindbasis provided further validation ofthe new technology and gave resultswhich were consistant with the ob-served physiological responses.

105

ACKNOWLEDGMENTS

The authors wish to thank Mr. RobertNowlan, Ms. Phyllis Lessin and Dr.Donald Tashkin of the University ofCalifornia - Los Angeles for theircooperation in the human studies per-formed there.

This work was performed under con-tract DOT HS 4 00968 from the Depart-ment of Transportation, NationalHighway Traffic Safety Administration.The technical advise given by Dr.Fred B. Benjamin of this agency andDr. Robert Willette of the NationalInstitute on Drug Abuse was greatlyappreciated.

REFERENCES

Agurell, S., Gustafsson, B., Holm-stedt, B., Leander, K., Lindgren,J., Nilsson, I., Sandberq, F., andAsberg, M.: J. Pharm. Pharmacol.,25,554 (1973).

Doorenbos, N.J., Fetterman, P.S.,Quimby, M.W. and Turner, C.E.: N.Y.Acad. Sci., 191,3 (1971).

Edery, H., Grunfeld, Y., Ben-Zvi, Z.,and Mechoulam, R.: N.Y. Acad. Sc.,191,40 (1971).

Fenimore, D.C., Freeman, R.R., andLoy, P.R.: Anal. Chem., 45,2331(1973).

Galanter, M., Wyatt, R.J., Lemberger,L., Weingartner, H., Vaughan, T.B.and Roth, W.T.: Sci., 176,934(1972).

Garrett, E.R. and Hunt, C.A.: J.Pharm. Sci., 63,1056 (1974).

Gross, S.J., Soares, J.R., Wong, S-L.R. and Schuster, R.E.; Nature, 252,581 (1974).

Tashkin, D.P., Private communication.

Teal, J.D., King, L. J., Forman, E.J. and Marks, V.: Lancet, 1974, 553.

106

ANALYTICAL METHODS FOR THE DETERMINATIONOF CANNABINOIDS IN BIOLOGICAL MEDIA

M. E. Wall, Ph.D., T.M. Harvey, Ph.D., J.T. Bursey, Ph.D.,

D.R. Brine, B.S., D. Rosenthal, Ph.D.

Research Triangle Institute, North Carolina

ABSTRACT

A pharmacokinetic tudy of the blood plasmalevels in man of 9-tetrahydrocannabinol,11-hydroxy- 9-tetrahydrocannabinol and canna-binol has been carried out by means of com-bined gas chromatographic-mass spectralanalysis. In some cases comparison of thedata was obtained on the same sample usingthin layer chromatography of radiolabeledsamples and electron capture gas-liquid chro-matography. For the mass spectral studiesappropriately deuterium labeled analogs ofthe previously named compounds were usedboth as internal standards and as a carrierfor the relatively small amounts of non-labeled drug present in plasma. Bloodsamples were obtained at periodic intervalsup to 24 hours from volunteers receiving 4-5m g 9-THC intravenously. After extractionand "clean-up" by Sephadex chromatography,the extracts were concentrated and subjectedto glc-ms in the electron impact (ei) modeor alternatively with a chemical ionization(ci) source, in which case preliminary chro-matography could be omitted. In all casescalibration curves were obtained from repli-cate analyses of spiked plasma containing

the internal standard and various quantitiesof the cannabinoid under analysis. A typicalbiphasic elimination of the drug was observedwith rapid elimination of 9-THC from theblood over a period of 40 min followed by amuch slower elimination up to 24 hours. Theexperimental data show that 11-hydroxy- 9-THC is found in the plasma in quantitiesonly of about one-twentieth to one-twenty-fifth the values found for 9-THC. Cannabinolwas not found in significant quantities. Goodagreement was obtained between the massspectral analyses and the thin layer chroma-tography or electron capture gas-liquidchromatographic procedures.

I N T R O D U C T I O N

In recent years there has been a great in-crease in interest in the pharmacology,metabolism and biodisposition of the cannabi-noids [for recent reviews cf. Mechoulam, Patonand Crown, Wall (1975) and Wall, et al.(1975)]. Until recently quantitation of thevarious cannabinoids in blood, urine, fecesand other biological tissues could be carried

107

out only by means of the use of appropriatelyradiolabeled analogs of the cannabinoids understudy (Wall, et al. (1975) and Lemberger) .Because of the widespread and increasingopposition to the use of radiolabeled isotopesin studies involving man and because many ofthe studies currently being conducted withvarious cannabinoids involve the use of largescale experiments in which radiolabeled canna-binoids are not used, the need for the devel-opment of non-radiolabeled quantitativemethodology for certain key cannabinoids hasbecome increasingly apparent. In addition,the use of radiolabeled thin layer chromatog-raphy techniques, while useful in initialstudies, lacks sufficient accuracy in thesense that when-biological extracts arestudied, separation of 9-THC from cannabinoland 11-hydroxy- 9-THC from other monohydroxy-lated analogs is poor. If such interferingsubstances are present in considerable quan-tity, one would obtain erroneously highvalues, This will increasingly be the casewhen one is analyzing biological materialsobtained from marihuana smokers which contain

9-THC, cannabinol, cannabidiol, and 11-hydroxylated analogs of these compounds.

Quantitative gas liquid chromatography com-bined with mass spectrometry (glc-ms) hasbeen used with excellent results for thequantitative analysis of drugs in biologicalmaterials, combining as it does the separa-tive powers of glc and the inherent sensi-tivity of ms detection. Pioneer studies byHammar, Holmstedt and Ryhage introduced theconcept of mass fragmentography [now alsocalled multiple ion detection (MID)] andgreatly increased the sensitivity of msmethodology so that it could be applied tothe nanogram and picogram levels. The con-cept has been applied to many drugs and re-cently by Agurell, Holmstedt and co-workersto the determination of 9-THC in bloodplasma. We wish to present in this papermethods for the quantitative determination bycombined glc-ms of 9-tetrahydrocannabinol( 9-THC), 11-hydroxy- 9-THC and cannabinol(CBN) in human plasma and results on the samesamples obtained by the tlc-radiolabeled orelectron capture glc procedures.

M E T H O D S

Clinical protocol

Human, male volunteers who were experiencedmarihuana users were administered 4.0-5.0 mgof 9-THC by the intravenous infusion methodof Perez-Reyes et al. The volunteers werekept under medical supervision for 24 hoursin the Clinical Research Unit of the Univer-sity of North Carolina, School of Medicine.Blood samples (10 ml approximately) were

collected at periodic intervals over a periodof 24 hours. Plasma was obtained by centri-fugation, immediately frozen and stored infrozen condition until analyzed.

Internal standards

A key feature of our quantitative procedureswas the use of appropriate deuterated analogsof the cannabinoids under study as both car-riers for the small quantity of cannabinoidsexpected to be present in many cases and asinternal standards for quantitation by massspectrometry. The structures of the cannabi-noids and their deuterated analogs used inthese studies are shown in Figure 1. All of

Figure 1

Figure 1.Structure of Cannabinoids and Internal Standards

the compounds used were synthetic and weremade available by the National Institute onDrug Abuse Synthesis Program.1 Syntheticmethods for the various deuterated cannabi-noids utilized in these studies have beenpresented by Pitt, et al.

Analytical procedures prior to glc-ms

General Precautions.--Close attention must bepaid to the precedural details presented belowin order to obtain reproducible and quan-titative data. In general in working withcannabinoids exposure of samples or extracts

108

to light or air should be minimized. Allsolvent evaporations should be conducted invacuo or under nitrogen at low temperature.Cannabinoids in nanogram levels are subject toadsorption on the surface of glassware. Inorder to minimize this problem all glassware,including chromatography columns, were coatedby silylation using 5% DMCS in toluene.

Extraction and Purification Prior to Analysisby GLC-MS in Ei Mode.--When the mass spectro-meters were operated in this mode the molecu-lar ions or charged fragments utilized for thequantitative analysis of underivatized canna-binoids were in a range of m/e of 320 orlower. Preliminary studies-with plasmaextracts indicated that interference fromendogenous plasma constituents would beencountered. This could be avoided by carry-ing out a preliminary cleanup by SephadexLH-20 chromatography prior to the glc-msstep. The methods which are pretsented are forthe combined determination of 9-THC (1a),11-hydroxy- 9-THC (2a), and cannabinol (3a).The methods, of course, are equally utilizablefor the determination of individual consti-tuents. Deuterated internal standards (cf.,Figure 1) were added to a sample of 3.0 mlof cold (not frozen) plasma as follows: 1b,150.0 ng; 2b, 15.0 ng; and 3b, 15.0 ng. Eachinternal standard was added in 15-30 µlethanol. Following addition of each internalstandard the plasma sample was stirred for3-5 seconds in a vortex agitator and thensubjected to sonication (Cole-Parmer ultra-sonic cleaner) for the same time. Theplasma samples (contained in a screw cappedcentrifuge tube) were then extracted 4 timeswith 6.0 ml petroleum ether (bp 30-60°,Nanogram Grade or Burdick and Jackson)containing 1.5% isoamyl ether. The tubeswere agitated 15 minutes each time in avortex agitator and the layers separated bycentrifugation after each extraction. Thepetroleum ether extracts were combined, evapo-rated in vacuo at room temperature and freezedried overnight to remove water and isoamylalcohol. The dried residue was dissolved ina minimal volume of petroleum ether/chloro-form/ethanol (10:10:1) and chromatographedon 1 x 40 an water jacketed Sephadex LH-20columns at 26°C in the same solvent mixture.Twenty-seven ml of column eluant were col-lected and discarded. Seven ml of eluantwere then collected as the fraction contain-i n g 9-THC. The next 8 ml of eluant werecollected as the CBN-containing fraction.Thirty-eight ml of column effluent were thencollected and discarded. Finally, 17 ml ofeluant were collected as the fraction con-taining 11-hydroxy 9-THC. The 9-THC andCBN fractions were evaporated to dryness anddissolved in 35 µl hexane. The 11-hydroxy-

9-THC fraction was evaporated to dryness

under vacuum and heated with 75 µl of Regisil(BSTFA + 1% TCMS) in a closed vial at 110°for 3 hours. The reagent was removed invacuum and the residue dissolved in 20 µlhexane.

Extraction Prior to Analysis by GC-MS in CIMode.--In this mode of operation 9-THC andCBN were determined as their pentafluoropro-pionate ester derivatives, Fig. 1, 1c and3c, respectively, using the PFP ester ofhexahydrocannahinol, 4b as the internalstandard for both analyses. It was foundthat endogenous plasma constituents did notinterfere at the molecular ion positions util-ized which were in the range of m/e 455-465.Hence preliminary purificatin of the extractswas not required. To a sample of 2.0-3.0ml of plasma were added the internal standardhexahydrocannabinol 4b, 15 ng/ml plasma,under conditions described in the sectionabove. The samples were extracted fourtimes with hexane, 2.0 ml/ml plasma asdescribed in the above section. The combinedhexane extracts were washed sequentiallywith 2.0 ml 0.1 N sodium hydroxide, 2.0 ml0.1 N hydrochloric acid, and 4.0 ml distilledwater. The washed hexane extracts wereevaporated and dried overnight in vacuo. Tothe dried extracts were added 0.2 ml hexane,50 µl pyridine/benzene (5:95), and 50 µlpentafluoropropionic acid anhydride. Afterheating at 40°C for one hour, the reactionmixture was washed with 0.2 ml of saturatedsodium bicarbonate solution, which was thenbackwashed with an additional 0.2 ml hexane.The hexane extracts were combined and evapo-rated under nitrogen. The derivatizedextracts were dissolved in 20 µl hexane.

Gas chromatography conditions

LKB-GLC-MS.--With this instrument we utilized2% OV-17 on Chromosorb W-HP, 80-100 mesh in3' or 6' x 1/4" glass columns, temp. 220°C,the former being used for 11-hydroxy- 9 - T H C -bis-TMS ether; the latter for both 9-THC temp.220°C; and CBN, temp. 240°C. Helium was usedas the gas phase at a rate of 35 ml/min.Under the above stipulated conditions reten-tion times of 4-6 minutes were observed foreach compound.

Finnigan-GLC-MS.--Only 9-THC and CBN wereanalyzed on this instrument in the ei mode.We used 6' glass columns containing 1% SE-30on 100/120 mesh Chromatosorb W-HP, columntemperatures 200-230°; He flow 30-35 ml/min.In the ci mode, which was used only withpentafluoropropionate derivatives, the samecolumn was used at 180°.

109

Mass spectrometry

Only basic operating details are presented inregard to thems-computer systems describedbelow. For a much more detailed report cf.Rosenthal et al.

LKB 9000 GLC-MS .--This instrument is of themagnetic sector type and was operated atev in the ei mode. The instrument was fittedwith a peak matching accessory, modified franan instrument described by Klein, which per-mits detection of very law levels of compoundsby the technique of acceleratig voltagealternation (AVA) (Hammar, 1968 and Agurell1973). For 9-THC the mass spectrum was setto focus alternately on the ions m/e 314 and317 which correspond to the molecular ion ofthe compound and its trideuterio analog. ForCBN the molecular ions were m/e 310 and 313.For analysis of 11-hydroxy- 9-THC as the bis-tms-ether, the strong M-103 ion (Wall, etal., 1970, Wall, et al., 1976) at m/e 371and 374 respectively was selected. The AVAaccessory used with the mass spectrometerscans through each peak in turn; a hardwarecounter determines the area under the peakseach time they are scanned. The ratio of theselected response (either area or peakheight) for the two ions in question is the

value used to determine the quantity of unknowncannabinoid in the plasma. Appropriate cali-bration curves (Fig. 2, 3) are constructed byadding variable amounts of cannabinoid to afixed quantity of corresponding trideuterioanalog in human plasma. Extraction, prelimi-nary purifcation, and glc are carried out asdescribed above, Analysis by GLC-MS-EI.

Finnigan 3000-GLC-MS.--This instrument is aquadrupole mass spectrometer and was operatedat 70 ev in the ei mode. Its scan is controlledby a dedicated PDP-12 camputer which alsoacquires data from the instrument. Althoughthe computer operates on a different principle,the ratios of peak area or peak heights aredetermined as described above using the samemolecular or fragment ions described in theprevious section.

In the ci mode isobutane was used as the car-rier gas at a pressure of 400 µ; 9-THC andthe internal standard HHC were determined asthe corresponding PFP ester as the MH+ ionat m/e 461 and 463, respectively. Similarly,for the case of CBN the MH+ ions were m/e 457and 463, respectively. Plasma calibrationcurves were obtained in the manner describedpreviously.

Figure 2 Figure 3

Figure 3.Figure 2. Finnigan 3300-ei Plasma Calibration CurveLKB 9000-ei Plasma Calibration Curve for 9-THC for 9-THC

110

TLC-radiolabel procedure

The volunteer subjects (described in Clinicalprotocol) all received 100 µCi of tritiumlabeled 9-THC, along with the standard 4.0-5.0 mg IV dose. Two to three ml aliquots ofplasma were analyzed by the procedure describedby Wall, et al. (1975).

Electron capture-glc procedure

In this procedure 9-THC was determined as thepentafluropropionate ester derivative usinghexahydrocannabinol as internal standard.Extraction and derivatization were carriedout as described under the procedure andextraction for gc-ms in ci mode. The instru-mentation utilized was constructed at RTI byDr. Edo Pellizzari as a modification of thedesign reported by Fenimore, et al.

A packed precolumn is utilized for initialcleanup of the sample followed by a packedcolumn or a capillary column to furnish therequisite resolution. By interposing avalving system and cold trap between the twocolumns, a small portion of the effluent fromthe first column can be introduced to thesecond column with minimum loss of efficiency.

Furthermore, this instrumental design providesthe capability for temperature programmingwhich is seldom used in ec systems because of

excessive base-line drift due to the detector’ssensitivity to column bleed. In the reporteddual column-dual oven system, the precolumncan be temperature programmed while the highefficiency column is maintained at isothermaltemperature. Thus, in effect, rapid andefficient separations are achieved on the pre-column without disturbing the base-line re-sponse of the ec detector.

The instrument which we have assembled in ourlaboratory consists of two separately con-trolled ovens (Varian Model 1440, Walnut Creek,California). As shown in Figure 4 , one ovenhouses the packed columns and possesses thecapability of linear temperature progamming;the second oven contains the capillary columnor a second packed column and is operatedisothermally. The effluent from the firstpacked column leads to an eight-port, hightemperature, low dead volume switching valve(CF-8-HTa, Valco Instruments, Houston, Texas).The gas flow from the first column when in thenormal operating position passes to the valveand through a capillary-loop trap and finallyto a flame ionization detector. The trap con-sists of about 1 ft of 1/8 in stainless steeltubing to which is silver soldered a 2 meterlength of 1/16 in O.D. x 0.20 i.d. Ni-200capillary tubing. A water supply at 4°C isprovided through the 1/2 in tubing for coolingpurposes.

Figure 4

Figure 4. Dual Gas Chromatographic System

111

At the exact time interval (retention time) atwhich trapping of the constituents in theeffluent from the packed column is to occur,the valve is rotated to the trapping position,and the coolant is passed through the 1/2 intubing to cool the capillary loop. In thisposition, a second carrier gas flows to thehigh resolution capillary or second packedcolumn; thus, the base-line is undisturbed.The trapped constituents are swept into thesecond oven and column by returning the valveto the original position and shutting off theambient water supply. The heat in the firstoven is sufficient to rapidly raise the tem-perature on the capillary trap for samplevaporation. A heated gold-plated Ni transferline connects the packed column, valve andcapillary or second column. The transfer lineis also coated with the same stationary phaseas in the final column.

Our system allows the utilization of flameionization, alkali flame, or electron capture(ec) in the second oven for measuring drugs.The second oven will also accept conventionalpacked columns, metal (Ni) or glass capillar-ies. A linear ec amplifier (Hewlett Packard,Avondale, Pennsylvania) and a miniature ecddeveloped in our laboratories (Pellizzari,1974) are employed in this dual gas chromato-graphed system.

Because of the low nanogram levels of cannabi-noids found in human plasma, it was decided touse the electron capture detector to allowmaximum sensitivity in the cannabinoid analy-sis. To make the analytical method morewidely applicable by other researchers, it wasdecided to use packed columns in both ovens.

Glc conditions.--Plasma impurities were sepa-rated on a ten-foot column of 2% SE30 on Supel-coport (80/l00 mesh) at 205°C with a carriergas (N2) flow rate of 15 ml/min. Columneffluent was collected at the retention timesof HHC and 9-THC and transferred for quantita-tive analysis to a six-foot column of 2% OV225on Supelcoport (100/120 mesh) at 178°C witha carrier gas (5% methane - 95% argon) flowrate of 9.5 ml/min.

Plasma calibration curves for 9-THC obtainedwith the LKB and Finnigan mass spectrometersin the ei mode are shown in Figures 2 and 3;the calibration curve obtained with theFinnigan instrument in the ci mode is shownin Figure 5. In each case a linear calibra-tion curve was obtained in range 1-100 nano-grams g-THC/ml plasma. Detection of 9 - T H Cbelow the lower limit could be obtained (de-tection limits 0.1). 0.5 Ng/ml of plasma isregarded as the minimal concentration at

Figure 5

Figure 5.Finnigan 3300-ci Plasma Calibration Curvefor 9-THC

which reliable data could be obtained. Thestandard error of estimation for LKB dataas obtained by linear regression analysiswas 0.114, and the correlation coefficientwas 0.9970. For the Finnigan in ei mode, thecomparable values were 0.034 and 0.9994, re-spectively. In the ci mode of the Finniganmass spectrometer the corresponding valueswere 0.20 and 0.9977. Plasma calibrationcurves for CBN are shown in Figures 6 and 7,which respectively present the data obtainedon the LKB-ei and Finnigan-ei. Linear curveson both instruments were obtained between0.2-10.0 ng/ml plasma with detection limitsabout 0.1 ng/ml. The standard error of esti-mation for the LKB and Finnigan instrumentswas respectively 0.05 and 0.078; the corre-lation coefficients in each case were 0.998.Preliminary data for CBN determined in the cimode as the PFP ester indicate that similarresults are to be expected. The plasmacalibration curve for 11-hydroxy- 9-THC-bis-tms-ethrer obtained with the LKB-mass spectro-meter is shown in Figure 8. The curve waslinear in the range of 0.2-10.0 ng of 11-hydroxy- 9-THC/ml plasma. As in previouscases the accuracy of the determination de-creases below concentrations of 1.0 ng/ml.The standard error of the estimation was0.027 and the correlation coefficient 0.9975.

RESULTS

112

Figure 6 Figure 8

Figure 6.LKB 9000-ei Plasma Calibration Curve for CBN

Figure 7

Figure 7.Finnigan 3300-ei Plasma Calibration Curvefor CBN

Figure 8.Figure 8.LKB 9000-ei Plasma Calibration Curve forLKB 9000-ei Plasma Calibration Curve for11-Hydroxy- 9-THC

Figure 9

Figure 9.

Plasma Levels of 9-THC, 11-Hydroxy- 9-THCand CBN Found Over a 24-Hour Period in HumanPlasma from Volunteers Receiving 9-THC byIV Administration

113

Figure 9 presents the average values withstandard error obtained for 9-THC, ll-hydroxy-

9-THC and cannabinol from plasma of malevolunteers receiving 9-THC by intravenousinfusion. The measurements covered a 24 hourperiod.

9-THC values obtained with the LKB-9000 ei source were in close agreement withthe data obtained on the Finnigan 3300-cisource. Other data not presented here alsoshow close agrerment between the LKB andFinnigan instruments when both were in the eimode.

9-THC values increased rapidly duringthe first 10-20 minutes, the peak values in therange of 50-60 ng/ml coinciding With the maxi-mal psychomimetric activity. A typical biphasic

elimination pattern was noted; the 9-THCplasma levels decreased rapidly between 15-40minutes and then fell at a much slower rate.With a particular group of volunteers (3 sub-jects) levels after 24 hours were between 3-5ng/ml. Spot checks at lower levels utilizingthe Finnigan MID program confirmed that thesubstance being evaluated was indeed 9-THCand not instrument "noise".

Figure l compares the results obtained fromthe average of four subjects, analyzed by ms-ei tlc radiolabel and electron capture glc.Correlation coefficients are calculated inFigure 11. The results are in reasonableagreement, and in particular the glc-ms andelectron capture glc procedures gave goodagreement for most points over the wholecurve. In the case of 11-hydroxy- 9-THCmuch lower levels were found; peak valuesin the neighborhood of 2.0 ng/ml were notedbetween 30-40 minutes. The maximal valuesdeclined in a more gradual manner than wasthe case for 9-THC, falling to a level of1.0 ng/ml in 60-90 minutes and 0.5 ng/mlafter 24 hours. The values for CBN are con-sidered unreliable as obtained by the eitechnique since for the most part the levelswere well below 1.0 ng/ml and show no consis-tent pharmacokinetic pattern.

DISCUSSION

The basic objective of this investigation wasto establish sensitive methodology, whichwould not depend on radiolabeling for thequantitative estimation of 9-THC, its primarymetabolite (Wall, 1970), 11-hydroxy- 9-THC andcannabinol, which has been reported to be ametabolite of 9-THC in the rat (McCallum,1975). This objective has been realized,utilizing glc-ms with a variety of techniquesand instruments. In addition, a specializeddouble glc procedure which utilizes one columnfor clean up and one for final estimation hasbeen developed. Several aspects of our resultsmerit detailed discussion.

Figure 10

Figure 10.9-THC Found in the Plasma Following Intra-

venous Administration of 5 mg 9 - T H C ,Average of 4 Subjects

Figure 11

Figure 11.Least-Squares Best Lines Comparing All DataObtained by Each Two Methods of Analysis

114

Choice of instrument

Two completely different types of mass spec-trometers coupled with different means forquantitation of data were utilized. One was arelatively old (1968) magnetic sector ms, theLKB-9000 which was coupled with an alternationvoltage acceleration device (Rosenthal, inpress: Klein, 1972) which permitted measurementof the ratio of the peak area of the unknown ascompared with that of the internal standard.The other was a relatively new (1974) quadru-pole ms, the Finnigan 3300 which was interfacedto a PDP-12 computer. The Finnigan ms has bothei and ci sources. As shown in Figures 2, 3, 6and 7 and in the text of the Results section,

both instruments in the ei mode gave virtuallyidentical plasma calibration curves with iden-tical linear range and quite similar standerror of estimation. Figure 9 gives pharmaco-kinetic data in man obtained on the LKB in theei mode and the Finnigan in the ci mode. Theresults are quite similar. It is thus evidentthat a wide variety of mass spectrometers canbe used with comparable results provided appro-priate internal carriers and standards areadded. Before concluding this discussion oneword of caution should be given: the nature ofthe separators is most important; the LKB withthe Ryhage separator and the Finnigan with asilylated glass jet separator gave appropriatesensitivity. On the other hand, another massspectrometer which utilized a double glass coilseparator showed poor sensitivity and could notbe utilized for cannabinoid studies.

General analytical considerations

Internal standards.--As indicated previously,the final mass spectrometric measurements canbe conducted with great accuracy. The key tosuccess in the various analytical studies wasthe utilization of appropriate compounds whichcould be used as both carriers and internalstandards. For this purpose deuterium labeledcannabinoids identical to the parent compoundexcept for the label are ideal and were uti-lized for all of the ei studies. It is possi-ble to use with equal success an internalcarrier which is not isotopically labeled aslong as its properties are very similar to thatof the cannabinoid being studied but permitseparation by glc. Hexahydrocannabinol wasexcellent fostudies of

this purpose and was used in ci9-THC and cannabinol.

Ei vs ci source.--In recent years quantitativeglc-ms using ci source has become more andmore popular for estimation of drugs inbiological materials. This is due to thefact that in many cases greater sensitivitycan be obtained in the ci mode since thereis little molecular fragmentation as comparedto the situation when operating in the ei

mode. In our studies we have found no par-ticular advantage in terms of sensitivity asfar as the two methods are concerned, theuseable lower limits in both cases beingabout 1 ng/ml plasma and detection limits aslow as 0.2 ng/ml. However there would seemto be a major advantage for utilizing the cisource if a choice is available. As notedearlier (cf., Methods section) a generalizedbackground interference is found in humanplasma in the region for determination ofunderivatized 9-THC and cannabinol in theregion of m/e 310-320. This endogenous inter-ference cannot be removed by glc techniquesalone but requires a preliminary "clean up",utilizing Sephadex LH-20 chromatography. Thismethod which was described by Agurell et al.for 9-THC and extended by us to mixtureso f 9-THC, 11-hydroxy- 9-THC and CBN, workswell for the former two compounds but istime consuming. In addition, there is dangerof some conversion of 9-THC to CBN because ofthe longer exposure period. All of theseproblems are avoided when the cannabinoids areconverted to the corresponding pentafluoropro-pionate esters. The molecular ions are thenin the region of m/e 455-465 and no interfer-ence was found after extraction of plasma withhexane, derivatization and glc-ms determina-tion. The method at present has been appliedt o 9-THC and CBN which can be extracted withnon-polar solvents. Whether more polar canna-binoids which require more polar solvents canbe successfully analyzed without prior purifi-cation by the ci technique is an open questionat this time.

Metabolic and pharmacokinetic data

The development of sensitive and accurate glc-ms methodology permitted a preliminary studyin man utilizing these techniques for theprecise determination of 9-THC, 11-hydroxy-

9-THC and CBN in plasma. Previously we havemade an extensive studof the metabolism of

(Wall, et al., 1975)9-THC in man using

radiolabeled tracers and thin layer chromatog-raphy. The procedures utilized (in additionto the undesirability of a radiolabeled tracerin man) suffer from two potential sources oferror. The method utilized would not permitseparation of 9-THC from CBN, and in thecase of 11-hydroxy- 9-THC, would not permitseparation from other monohydroxy-metaboliteswhich might be present (Wall, et al., 1973;Wall, et al. 1975). The data in Figure 9for 9-THC are quite comparable to pharmaco-kinetic data obtained in earlier studies(Wall, et al., 1975). In both instances abiphasic elimination curve is noted with asharp decline after the initial maximum levelfollowed by a much more gradual decrease.Maximal values in the current studies were50-60 ng/ml. After 24 hours, 3-5 ng/ml of9-THC were still found in the plasma. Our

115

results for 11-hydroxy- 9-THC are probablythe most accurate data yet reported in man.The concentration of this active metabolite(cf. Figure 9) was only 2-3 ng/ml at peaklevels declining at a slower rate than 9-THC to 0.5 ng/ml after 24 hours. Although9-THC is readily converted to 11-hydroxy- 9-

THC in the liver (Wall, et al., 1975) onlysmall quantities find their way into the blood.

Our interest in CBN was aroused by reportsfrom McCallum (1975) and McCallum, et al.(1975) which indicate that CBN might be atransitory metabolite found at very earlytime periods after administration of 9-THC.

As shown in Figure 9, the level of CBN wasbelow our reliability limits in the ei mode.Other studies we have carried out by electroncapture glc or glc-ms in the ci mode indicatethe virtual absence of this substance at alltime periods. Since we have found that CBNhas the same general pharmacokinetic pattemas 9-THC in man (Wall, 1975), we must con-

ACKNOWLEDGMENTS

These studies were conducted with thesupport of the National Institute on DrugAbuse under contracts HSM-41-71-95 and ADM-45-74-109. We wish to thank Mario Perez-Reyes, M.D. for clinical material used insome of these studies and express to Mrs.Valerie H. Schindler and Mr. M. Taylorour appreciation for their technical assis-tance.

clude that CBN can be disregarded in termsof its importance as a metabolite in man.

Comparison of gc-ms with other procedures

As shown in Figures 10 and 11, the gc-ms pro-cedures show reasonable agreement in the caseof 9-THC with data obtained by two indepen-dent procedures, involving respectively thinlayer chromatography of radiolabeled cannabi-noids, and a double glc-electron captureprocedure. In preliminary studies on canna-binol levels of subjects who received 9-THC,good agreement was found between the massspectrometric glc-ci method and electron cap-ture glc. In each case no cannabinol couldbe found. Studies on 11-hydroxy- 9-THClevels by the glc-electron capture methodare in progress. Initial attempts to usehigh pressure liquid chromatography asmother method indicate that the sensitivityby this procedure with current detectors isof the order 10 ng/rnl. Hence at presenthplc techniques do not have requisitesensitivity.

FOOTNOTE1Research Triangle Institute Contract HSM-42-71-95. Appropriately qualified investi-gators may obtain a variety of labeled andunlabeled cannabinoids by application toDr. Robert Willette, Acting Chief, TheResearch Technology Branch, Division ofResearch, National Institute on Drug Abuse,Rockwall Building, 11400 Rockville Pike,Rockville, Maryland 20852.

REFERENCES

Agurell, S., Gustafsson, B., Holmstedt, B., Klein, P. D., Hauman, J. R. and Eisler, W. J.Leander, K., Lindgren, J.-E., Nilsson, I., Gas chromatograph-mass spectrometer-accel-Sandberg, F. and Asberg, M. Quantitationo f 9-tetrahydrocannabinol in plasma from

erating voltage alternator system for themeasurement of stable isotope ratios in

cannabis smokers. J. Pharm. Pharmac., organic molecules. Anal. Chem., 1972, 44:1973, 25: 554-558. 410-493.

Fenimore, D. C., Free R. R. and Loy, P. R. Lemberger, L. The metabolism of the tetra-Determination of 9-tetrahydrocannabinol in hydrocannabinoids. In S. Garratini, F.blood by electron capture gas chromatog- Hawking, A. Golden and I. Kopin (Eds.),raphy. Anal. Chem., 1973, 45: 2331-2335. Advances in Pharmacology and Chemotherapy.

Academic Press, New York, NY. 1972: 221-Hammar, C.-G. and Holmstedt, B. Mass fragmen- 251.

tography identification of chlorpromazineand its metabolites in human blood by a new McCallum, N. K. The effect of cannabinol onmethod. Anal. Biochem., 1968, 22: 532- 1-tetrahydrocannabinol clearance from the548. blood. Experimentia, 1975, 31: 957-958.

116

McCallum, N. K., Yagen, B., Levy, S. andMechoulam, R. Cannabinol: a rapidlyformed metabolite of 1- and 6-tetra-hydrocannabinol. Experimentia, 1975, 31:520-521.

Mechoulam, R. (Ed.), Marihuana. AcademicPress, New York, NY, 1973: 1-409.

Paton, W. D. and Crown, J. Cannabis and itsderivatives. Oxford University Press,London, England. 1972: 1-198.

Pellizzari, E. D. High resolution electroncapture gas-liquid chromatography. J.Chromatog., 1974, 92: 299-308.

Perez-Reyes, M., Timnons, M., Lipton, M.,Davis, K. and Wall, M. Intravenous injec-tion in man of 9-tetrahydrocannabinol and11-OH- 9-tetrahydrocannabinol. Science,1972. 177: 633-635.

Pitt, C. G., Hobbs, D. T., Schran, H., Twine,C. E. and Williams, D. L. The synthesis ofdeuterium, carbon-14 and carrier-freetritium-labeled cannabinoids. J. Label.Comp., 1975, 11: 551-575.

Rosenthal, D., Harvey, M. T., Bursey, J. T.,Brine, D. R. and Wall, M. E. Comparison Ofgas chromatography-mass spectrometrymethods for the determination of 9-tetra-hydrocannabinol in plasma. Biomed. MassSpec., in preparation.

Wall, M. E. Recent advances in the chemistryand metabolism of the cannabinoids. InV. C. Runeckles (Ed.), Recent Advances inPhychochemistry. Plenum Publishing Co., NewYork, NY, 1975: 29-61.

Wall, M. E. and Brine, D. R. Application ofmass spectrometry to structure of metabo-lites of 9-tetrahydrocannabinol. Sum-maries, International Symposium on MassSpectrometry in Biochemistry and Medicine.Milan, Italy, 1973, p. 52.

Wall, M. E., Brine, D. R. and Perez-Reyes, M.Identification of cannabinoids and metab-olites in biological materials by combinedgas-liquid chromatography-mass spectrom-etry. In G. G. Nahas (Ed.), Marihuana:Chemistry, Biochemistry, and CellularEffects. Springer-Verlag, New York, NY,1976: 51-62.

Wall, M. E., Brine, D. R. and Perez-Reyes, M.Metabolism of cannabinoids in man. In M.Brade and S. Szara (Eds.), Pharmacology ofMarihuana. Raven Press, New York, NY,1975.

Wall, M. E., Brine, D. R., Brine, G. A., Pitt,C. G., Freudenthal, R. I. and Christensen,H. D. Isolation, structure and biologicalactivity of several metabolites of 9-tetrahydrocannabinol. J. Am. Chem. Soc.,1970, 92; 3466-3468.

117

LIST OF CONTRIBUTORS

Quantitation of Cannabinoids in Biological Fluids by RadioimmunoassayArleen R. Chase, Paul R. Kelley, Alison Tunton-Rigby

Collaborative Research, Inc.1365 Main StreetWaltham, Mass. 02154

Reese T. JonesLangley Porter Neuropsychiatric InstituteUniversity of CaliforniaSan Francisco, Cal. 94143

Theresa HarwoodDrug Enforcement Administration1405 Eye Street, N.W.Washington, D.C. 20537

Separate Radioimmune Measurements of Body Fluid 9 -THC and 11-nor-9-Carboxy- 9-THC.Stanley J. Gross, M.D. and James R. Soares, Ph.D.

Department of Anatomy, School of MedicineUniversity of California, Los AngelesLos Angeles, Cal. 90024

Radioimmunoassay of 9-TetrahydrocannabinolClarence E. Cook, Ph.D., Mary L. Hawes, B.A., Ellen W. Amerson, B.A.,Colin G. Pitt, Ph. D., and David Williams, B. A.

Research Triangle InstituteP.O. Box 12194Research Triangle Park, North Carolina 27709

Determination of THC and its Metabolites by EMIT® Homogeneous Enzyme Immunoassay:A Summary Report

G.L. Rowley, Ph.D., T.A. Armstrong, C.P. Crowl, W.M. Eimstad, W.M. Hu, Ph.D.,J. K. Kam, R. Rodgers, Ph.D., R.C. Ronald, Ph.D., K.E. Rubenstein, Ph.D.,B.G. Sheldon. and E.F. Ullman. Ph.D.

Syva Research Institute3181 Porter DrivePalo Alto, Cal. 94304

Separation and Sensitive Analysis of Tetrahydrocannabinol in Biological Fluids by HPLC and GLCEdward R. Garrett and C. Anthony Hunt

College of Pharmacy, J. Hillis Miller Health CenterBox J-4University of FloridaGainesville, Fla. 32601

Determination of 9 -Tetrahydrocannabinol in Human Blood Serum by Electron CaptureGas Chromatography

David C. Fenimore, Ph.D., Chester M. Davis, Ph.D., and Alec H. HornInstrumental Analysis SectionTexas Research Institute of Mental Sciences1300 Moursund StreetHouston, Texas 77025

118

Detection and Quantification of Tetrahydrocannabinol in Blood PlasmaAgneta Ohlssonl, Jan-Erik Lindgren2 3, Kurt Leander2, Stig Agurell1 3

1Faculty of Pharmacy, University of UppsalaBMC, Box 579, S-751 23 Uppsala

2Department of ToxicologyKarolinska Institutet,S-104 01 Stockholm 60

3Astra Läkemedel AB,S-151-85Södertälje, Sweden

A Method for the Identification of Acid Metabolites of Tetrahydrocannabinol (THC)by Mass Fragmentography

Marianne Nordqvist1, Jan-Erik Lindgren2 3, Stig Agurell1 3

Faculty of Pharmacy, University of UppsalaBMC, Box 579, S-171 23 Uppsala

2Department of ToxicologyKarolinska Institutet,S-104 01 Stockholm 60

3Astra Läkemedel AB,S-151-85Södertälje, Sweden

Quantitation of Cannabinoids in Biological Specimens Using Probability Based MatchingGas Chromatography/Mass Spectrometry

Donald E. Green, Ph.D.Biochemistry Research Lab 151-FVeterans Administration HospitalPalo Alto, Cal. 94304

Quantitation of 9-Tetrahydrocannabinol in Body Fluids by Gas Chromatography/Chemical Ionization-Mass Spectrometry

Ruthanne Detrick and Rodger L. FoltzBattelle Memorial Institute505 King AvenueColumbus, Ohio 43201

HPLC-MS Determination of 9-Tetrahydrocannubinol in Human Body SamplesJimmie L. Valentine, Ph.D., Paul J. Bryant, Ph.D., Paul L. Gutshall, M.S.,Owen H. M. Gan, B.S., Everett D. Thompson, B.S., Hsien Chi Niu, Ph.D.

University of Missouri, Kansas City5100 Rock Hill RoadKansas City, Missouri 64110

Analytical Methods for the Determination of Cannabinoids in Biological MaterialsM.E. Wall, Ph.D., T.M. Harvey, Ph.D., J.T. Bursey, Ph.D., D.R. Brine, B.S.,and D. Rosenthal, Ph.D.

Research Triangle InstituteP.O. Box 12194Research Triangle Park, North Carolina 27709

119

monograph series

Each monograph in the NIDA Research Monographseries is for sale from the following sources:

1Findings of Drug Abuse ResearchAn annotated bibliography of NIMH and NIDA-supported extramural grant research 1967-74Volume 1: Stock #017-024-0467; pp. 384; $7.00Volume 2: Stock #017-024-0466-9; pp. 377; $5.05Order from: Superintendent of Documents

U.S. Government Printing OfficeWashington, D.C. 20402

2Operational Definitions in Socio-behavioralDrug Use Research 1975Editors: Jack Elinson, Ph.D. and David Nurco, Ph.D.Task Force articles proposing consensual defin-itions of concepts and terms used in psycho-social research to achieve operational compara-bility.Order by PB nwnber: PB #246 338Papercopy: $6.75; Microfiche: $2.25.Order from:NTIS (National Technical Information Service)U.S. Department of CommerceSpringfield, Virginia 22161

3Aminergic Hypotheses of Behavior:Reality or Cliche?Editor: Bruce J. Bernard, Ph.D.Articles examining the relation of the brainmonoamines to a range of animal and humanbehaviors.Order by PB number: PB #246 687Papercopy: $6.25; Microfiche: $2.25Order from:NTIS (National Technical Information Service)U.S. Department of CommerceSpringfield, Virginia 22161

4Narcotic Antagonists: The Search for Long-Acting PreparationsEditor: Robert Willette, Ph.D.Articles reporting current alternative insertedsustained-release or long-acting drug devices.Order by PD number: PB #247 696Papercopy: $4.00; Microfiche: $2.25Order from:NTIS (National Technical Information Service)U.S. Department of CommerceSpringfield, Virginia 22161

orStock #017-024-0048-0 @ $1.10Order from: Superintendent of Documents

U.S. Government Printing OfficeWashington, D.C. 20402

5Young Men and Drugs: A Nationwide SurveyAuthor: John A. O’Donnell, Ph.D. et al.Report of a national survey of drug use by men20-30 years in 1974-5.Order by PB number: PB #247 446Paper copy: $6.75; Microfiche: $2.25Order from:NTIS (National Technical Information Service)U.S. Department of CommerceSpringfield, Virginia 22161

6Effects of Labeling the “Drug-Abuser”- An InquiryAuthor: Jay R. Williams, Ph.D.Analysis and review of the literature examiningeffects of drug use apprehension or arrest onthe adolescent.Orderby PB number: PB #249 092Paper copy: $4.00; Microfiche: $2.25Order from:NTIS (National Technical Information Service)U.S. Department of CommerceSpringfield, Virginia 22161

120