Copyright © 2005 John Wiley & Sons, Ltd. Biomed. Chromatogr. 19: 677–688 (2005)

Cocaine and pyrolytic cocaine metabolites in human blood and urine 677ORIGINAL RESEARCHORIGINAL RESEARCH

Published online 19 April 2005

Copyright © 2005 John Wiley & Sons, Ltd.

BIOMEDICAL CHROMATOGRAPHYBiomed. Chromatogr. 19: 677–688 (2005)Published online in Wiley InterScience (www.interscience.wiley.com).DOI: 10.1002/bmc.495

Concentration profiles of cocaine, pyrolytic methylecgonidine and thirteen metabolites in human bloodand urine: determination by gas chromatography–massspectrometry†

Buddha D. Paul,1* Shairose Lalani,1 Thomas Bosy,1 Aaron J. Jacobs1 and Marilyn A. Huestis2

1Office of the Armed Forces Medical Examiner, Division of Forensic Toxicology, Armed Forces Institute of Pathology, 1413 Research Boulevard,Rockville, MD 20850, USA

2Chemistry and Drug Metabolism, Intramural Research Program, National Institute of Drug Abuse, Baltimore, MD 21224, USA

Received 31 July 2004; revised 4 January 2005; accepted 4 January 2005

ABSTRACT: When cocaine is smoked, a pyrolytic product, methyl ecgonidine (anhydroecgonine methyl ester), is also consumedwith the cocaine. The amount of methyl ecgonidine formed depends on the pyrolytic conditions and composition of the illicitcocaine. This procedure describes detection of cocaine and 10 metabolites—cocaethylene, nor-cocaine, nor-cocaethylene, methylecgonine, ethyl ecgonine, benzoylecgonine, nor-benzoylecgonine, m-hydroxybenzoylecgonine, p-hydroxybenzoylecgonine andecgonine—in blood and urine. In addition, the detection of pyrolytic methyl ecgonidine and three metabolites—ecgonidine(anhydroecgonine), ethyl ecgonidine (anhydroecgonine ethyl ester) and nor-ecgonidine (nor-anhydroecgonine)—are included. Thenewly described metabolites, ethyl ecgonidine and nor-ecgonidine, were synthesized and characterized by gas chromatography–mass spectrometry (GC-MS). All 15 compounds were extracted from 3 mL of blood or urine by solid-phase extraction and identi-fied by a GC-MS method. The overall recoveries were 49% for methyl ecgonine, 35% for ethyl ecgonine, 29% for ecgonine andmore than 83% for all other drugs. The limits of detection were between 0.5 and 4.0 ng/mL except for ecgonine, which was 16 ng/mL. Linearity for each analyte was established and in all cases correlation coefficients were 0.9985–1.0000. The procedure wasapplied to examine the concentration profiles of analytes of interest in post-mortem (PM) blood and urine, and in urine collectedfrom living individuals (LV). These specimens previously were shown to be positive for the cocaine metabolite, benzoylecgonine.Ecgonidine, the major metabolite of methyl ecgonidine, was present in 77% of PM and 88% of the LV specimens, indicatingsmoking as the major route of cocaine administration. The new pyrolytic metabolites, ethyl ecgonidine and nor-ecgonidine, werepresent in smaller amounts. The urine concentrations of nor-ecgonidine were 0–163 ng/mL in LV and 0–75 ng/mL in PMspecimens. Ethyl ecgonidine was found only in PM urine at concentrations 0–39 ng/mL. Ethanol-related cocaine metabolites,ethyl ecgonine or cocaethylene, were present in 69% of PM and 53% of cocaine-positive LV specimens, implying alcohol con-sumption with cocaine use. The four major metabolites of cocaine—benzoylecgonine, ecgonine, nor-benzoylecgonine and methylecgonine—constituted approximately 88 and 97% of all metabolites in PM and LV specimens, respectively. The concentrationsof nor-cocaine and nor-cocaethylene were consistently the lowest of all cocaine metabolites. At benzoylecgonine concentrationsbelow 100 ng/mL, ecgonine was present at the highest concentrations. In 20 urine specimens, benzoylecgonine and ecgoninemedian concentrations (range) were 54 (0–47) and 418 ng/mL (95–684), respectively. Therefore, detection of ecgonine is advant-ageous when benzoylecgonine concentrations are below 100 ng/mL. Copyright © 2005 John Wiley & Sons, Ltd.

KEYWORDS: use of smoked cocaine and alcohol; metabolite concentrations; human blood and urine; gas chromatography-massspectroscopy

*Correspondence to: B. D. Paul, Office of the Armed Forces MedicalExaminer, Division of Forensic Toxicology, Armed Forces Instituteof Pathology, 1413 Research Boulevard, Rockville, MD 20850, USA.E-mail: Buddha.Paul@afip.osd.mil†The opinions expressed herein are those of the authors and are notto be construed as official or as reflecting the views of the Depart-ment of the Army or the Department of Defense.

Abbreviations used: Bz, benzoylecgonine; CE, cocaethylene; COC,cocaine; DMF, dimethylformamide; DMF-DEA, DMF-diethylacetal;DMF-DPA, DMF-di-n-propyl acetal; EC, ecgonine; ED, ecgonidine;EEC, ecgonine ethyl ester; LV, living individuals; MEC, ecgoninemethyl ester; MED, methyl ecgonidine; m-HBz, meta-hydroxy-benzoylecgonine; NBz, nor-Bz; NC, nor-cocaine; NCE, nor-cocaethylene; NED, nor-ecgonidine; PFPL, pentafluoropropanol;PFPA, pentafluoropropionic anhydride; PM, post-mortem.

INTRODUCTION

The major routes of illicit cocaine (COC) adminis-tration are intranasal, intravenous and smoking. Overthe last two decades (Schnoll et al., 1985; Brower et al.,1986; Wesson and Washburn, 1990; Paul et al., 1999)smoking has emerged as the preferred route due toits rapid delivery of drug to the brain and resultantpleasurable effects (Perez-Reyes et al., 1982). Smok-ing also causes an intense craving for additionaldrug. Generally, free-base COC (‘crack’) is smoked by

Copyright © 2005 John Wiley & Sons, Ltd. Biomed. Chromatogr. 19: 677–688 (2005)

678 B. D. Paul et al.ORIGINAL RESEARCH

heating the compound in a smoking device and inhalingthe vapor in two to three puffs.

When COC is smoked, methyl ecgonidine (MED)is also consumed as a pyrolytic product. The amountsof COC and MED in the smoke vary considerablydepending upon the composition of the ‘street’ sampleand the temperature used to volatilize COC. At tem-peratures of 255–420°C, the amount of COC convertedto MED was 50–80% and at 650°C the conversioncould be as high as 89% (Martin et al., 1989; Nakaharaand Ishigami, 1991). Owing to variations in the com-position of cocaine smoke, bioavailability of the com-pound and its pharmacology are difficult to predict(Jones, 1990).

COC and its metabolite nor-cocaine (NC)are pharmacologically active. Other metabolitesare either minimally active or inactive. Meta-hydroxybenzoylecgonine (m-HBz) appeared to be re-lated to increased jitteriness in infants exposed to COC(Singer et al., 2000). In a study of the pulmonary effectsof the pyrolytic product MED in guinea pigs, decreasedairway conductance and induced lung damage werefound (Chen et al., 1995; Willetts et al., 1995). Thecompound also exerted a negative inotropic effect onisolated ferret and human myocardium (Woolf et al.,1997), and acted on M2 cholinergic receptors in theheart to produce cardiac toxicity (Huang et al., 1997).In addition, large doses of MED caused structuraldamage to isolated myocytes in ferrets. Both MED anda hypothetical metabolite, nor-ecgonidine methyl ester,decreased blood pressure and heart rate while increas-ing respiratory rate in anesthetized rabbits (Erzoukiet al., 1995). Although MED appeared to be toxic inanimals, pharmacological and pharmacokinetic pro-perties of the compound in humans have not beenestablished. Information on the relative amounts ofcocaine, MED and their metabolites in blood andurine could be helpful in understanding drug actionsand metabolic profiles in human. In this investigation,we report the concentration profiles of cocaine, 10of its metabolites, pyrolytic MED and its metaboliteecgonidine (ED), and two new pyrolytic metabolites,nor-ecgonidine (NED) and ethyl ecgonidine (EED), inpost-mortem urine and blood and in urine collectedfrom living persons. From the presence of pyrolyticand alcohol related metabolites, it may be possible toestimate cocaine use by smoking and in combinationwith alcohol consumption.

EXPERIMENTAL

Reagents and supplies. COC, d3-COC, cocaethylene (CE),NC, d3-NC, nor-cocaethylene (NCE), benzoylecgonine (Bz),d3-Bz, nor-Bz (NBz), m-hydroxy-Bz (m-HBz), methylecgonine (ecgonine methyl ester, MEC), d3-MEC, ethyl

ecgonine (ecgonine ethyl ester, EEC), methyl ecgonidine(anhydroecgonine methyl ester, MED), ecgonidine (anhydro-ecgonine, ED), and ecgonine (EC) were purchased fromCerilliant Corporation (Austin, TX, USA). para-Hydroxy-Bz(p-HBz), d3-p-HBz and d3-m-HBz were purchased fromElSohly Laboratories (Oxford, MS, USA). d3-MED andd3-ED were prepared according to the procedures previouslydescribed (Paul et al., 1999; Shimomura et al., 2001). Solid-phase extraction (SPE) columns containing silica-based C8

and benzene sulfonic acid (sorbent 200 mg) were purchasedfrom United Chemical Technologies (Bristol, PA, USA,catalog number ZCDAU020). Dimethylformamide (DMF),DMF-diethylacetal (DMF-DEA), DMF-di-n-propyl acetal(DMF-DPA), pentafluoropropanol (PFPL) and pentafluoro-propionic anhydride (PFPA) were purchased from AldrichChemical Corporation (Milwaukee, WI, USA). Other rea-gents and solvents were analytical or HPLC grade. Reacti-vials were purchased from Pierce (Rockford, IL, USA).

Instrument. The Hewlett-Packard (HP) GC-MS system con-sisted of an HP 5890 Series II Plus gas chromatograph and aHP 5972 quadrupole mass-selective detector. An HP 18593Bautoinjector was used to inject the samples into the GC-MS.

Preparation of EED (anhydroecgonine ethyl ester). Amethanolic solution of 120 µg of ED in a 10 mL tube wasevaporated to dryness. DMF (50 µL) and DMF-DEA (50 µL)were added to the compound. The tube was capped, vortex-mixed, and heated at 80°C for 4 h. The solution was cooled toroom temperature and diluted to 25 mL with dry acetonitrile.When the solution was tested by GC-MS using the full-scanmode (50–300 amu) and the area under the curve (AUC)compared with that of a known amount of MED, the concen-tration of EED was found to be 5.26 µg/mL. The yield was94% and no unreacted ED was detected in the sample.

Preparation of propyl d3-ecgonidine (d3-PED). This com-pound was used as an internal standard for quantitation ofMED and EED. The preparation was similar to that of EED:d3-ED (125 µg) (Paul et al., 1999) was propylated with 50 µLDMF-DPA in 50 µL DMF as solvent. The product wasdiluted to 25 mL with dry acetonitrile. When the AUC wascompared with that of a known amount of MED using GC-MS full scan mode (50–300 amu), the concentration of d3-PED was found to be 5.22 µg/mL (yield 84%). No unreactedd3-ED was detected in the sample.

Preparation of NED (nor-anhydroecgonine). A methanolicsolution of NBz (225 µg) in a 5 mL Reacti-vial was evapor-ated to dryness. The compound was dissolved in hydrochloricacid (500 µL, 11.6 mol/L), capped with a PTFE-lined disk,and heated in a sand bath at 110°C for 18 h. The solutionwas cooled to room temperature and evaporated to dryness.The product, NED hydrochloride, was dissolved in 25 mL ofmethanol.

GC-MS analysis of NED. NED solution (200 µL) was mixedwith a solution of ED (50 µL, 40 µg/mL) and evaporated todryness. The compounds were derivatized to N,O-dipentylNED and O-pentyl ED by n-iodopentane according to theprocedure previously described for propylation of Bz (Paul

Copyright © 2005 John Wiley & Sons, Ltd. Biomed. Chromatogr. 19: 677–688 (2005)

Cocaine and pyrolytic cocaine metabolites in human blood and urine 679ORIGINAL RESEARCH

et al., 1993). The derivatives were dissolved in 60 µL of dryacetonitrile and tested by GC-MS. The GC oven tempera-ture was increased from 80°C (held for 0.5 min) to 140°C at20°C/min and then to 265°C (held for 2.0 min) at 50°C/min.The injector and detector temperatures were 140 and 260°C,respectively. The GC injector was operated in the splitlessmode and the back solvent was purged 0.5 min after sampleinjection. The mass spectral scan range was 50–300 amu.When the AUC of NED was compared with that of theknown amount of ED, the concentration of NED in the solu-tion was found to be 4.9 µg/mL (yield 98%). No unreactedNBz was detected in the solution.

Specimen collection. Eight post-mortem (PM) blood andurine specimens were collected from previously investigatedcases. Sodium fluoride (10 mg/mL) was used as a preservativein the blood. Five additional PM urine specimens also weretested but corresponding blood specimens were not available.After testing positive for Bz at an immunoassay cut-off con-centration of 100 ng/mL for blood and 150 ng/mL for urine,specimens were stored at −18°C for 6 months prior to use inthis study. Specimens also were positive for Bz at a GC-MScutoff concentration of 100 ng/mL.

Seventeen urine specimens collected from living individuals(LV) subject to the military’s random drug-testing programwere studied. Specimens were positive for Bz at immunoassayand GC-MS cut-off concentrations of 150 and 100 ng/mL,respectively, and were stored at −18°C for 12 months. Twentyadditional urine specimens (Lvl), positive by immunoassaybut negative by GC-MS, were also investigated. The averagestorage time for these specimens was less than 3 weeks.

Preparation of stock solutions. A stock solution of COC,NC, CE, NCE, MEC, EEC, MED and EED at concen-

trations of 2.5, 1.25, 1.25, 1.25, 2.5, 1.25, 1.25 and 1.25 µg/mL,respectively, was prepared in acetonitrile. Another stocksolution of Bz, NBz, m-HBz, p-HBz, EC, ED and NED atconcentrations of 5.0, 1.25, 1.25, 1.25, 5.0, 7.0 and 1.25 µg/mL,respectively, was prepared in methanol. The calibrator wasprepared by adding 60 µL of each of the stock solutionsinto 3 mL of certified negative urine or blood. The finalconcentrations of drugs in the calibrator were as follows:COC and MEC at 50 ng/mL; Bz and EC at 100 ng/mL; EDat 140 ng/mL; and all others at 25 ng/mL. The low and highcontrols were prepared from different stock solutions andat final concentrations of 0.5 and 2.0 times the concentrationsof the calibrator.

Urine and blood specimen preparation. Specimens wereextracted according to Scheme 1. A stock solution of internalstandards containing d3-COC, d3-MEC, d3-NC, d3-PED(60 µL) at concentrations of 2.5, 2.5, 1.25 and 1.25 µg/mL,respectively, in acetonitrile, was added to 3 mL of calibrator,low and high controls, blank and specimens. A second stocksolution of internal standards containing d3-Bz, d3-m-HBz, d3-p-HBz, d3-EC and d3-ED (60 µL) at concentrations of 5.0,1.25, 1.25, 5.0 and 5.0 µg/mL, respectively, in methanol, wasadded to each sample. Two milliliters of phosphate buffer(0.1 m, pH 6.0) were added to blood and 0.8 mL of phosphatebuffer (1.0 m, pH 5.5) was added to urine specimens. Solu-tions were mixed and centrifuged at approximately 3800 g for40 min. The clear supernatants were passed through a set ofSPE columns (C8 + benzene sulfonic acid, 200 mg) pre-washed with 3 mL each of methanol, deionized water, andphosphate buffer (0.1 m, pH 6.0). The aqueous eluents con-taining unabsorbed EC, ED and NED were collected andsaved. The column materials were washed with 3 mL each ofdeionized water, 0.1 m HCl, and 2-propanol and then dried for

Scheme 1. Extraction and GC-MS detection of 15 cocaine and pyrolyticanalytes. Compounds derived from cocaine metabolism: cocaine (COC),cocaethylene (CE), nor-cocaine (NC), nor-cocaethylene (NCE), benzo-ylecgonine (Bz), nor-benzoylecgonine (NBz), m-hydroxy-benzoylecgonine(m-HBz), p-hydroxy-benzoylecgonine (p-HBz), methyl ecgonine (MEC), ethylecgonine (EEC) and ecgonine (EC). Compounds derived from pyrolyticcocaine metabolism: methyl ecgonidine (MED), ethyl ecgonidine (EED),ecgonidine (ED) and nor-ecgonidine (NED). Compounds derived fromethanol-related metabolism: CE, NCE, EEC and EED.

Copyright © 2005 John Wiley & Sons, Ltd. Biomed. Chromatogr. 19: 677–688 (2005)

680 B. D. Paul et al.ORIGINAL RESEARCH

5 min under reduced pressure. Basic drugs were eluted fromthe columns with a mixture of methylene chloride–methanol–aqueous ammonia (14.8 m; 3 mL, 9:1:0.2 by volume).

Solutions were divided into two equal portions (Scheme 1)and evaporated just to dryness at room temperature under aslow stream of nitrogen. MED and EED are volatile, andtherefore, prolonged evaporation was avoided. One portionwas dissolved in 60 µL of acetone and tested for MED andEED and for COC and CE by two separate GC-MS proce-dures. The remaining solution of this portion was tested forBz, m-HBz and p-HBz as propyl derivatives. The second por-tion of the extract after derivatization with PFPL and PFPA,was tested for MEC, EEC and NBz, and for NC and NCEby two separate GC-MS procedures. The EC, ED and NEDin the aqueous extract was processed according to the pro-cedure described previously for ED (Paul et al., 1999). Inthe method, the solutions were treated with 1 mL of 2 mHCl. The pH of the solutions was between 2.0 and 3.0. Thesolutions were poured onto a new set of SPE columns (C8 +benzene sulfonic acid, 200 mg) prewashed with 3 mL each ofMeOH, 0.1 m HCl and deionized water. The solutions wereallowed to pass through the columns by gravity flow. Thecolumn materials were washed with 3 mL each of deionizedwater, 0.1 m HCl and MeOH. After drying the column mater-ials for 2 min using suction, the EC, ED and NED wereeluted with 3 mL of a mixture of MeOH-2-propanol-aqueousammonia (14.8 m; 4:6:0.25 by volume). The solutions wereevaporated to dryness at 50°C under nitrogen.

GC-MS analysis. Sample volumes for injection were 1–4 µL.GC analysis was performed using a DB-5MS capillarycolumn (5:95 phenyl-methylsiloxane, 15 m × 0.25 mm i.d.;J&W Scientific, Folsom, CA, USA) at 10 psi helium pressurein the splitless mode. The excess solvent at the injection portwas purged after 0.3–0.5 min of sample injection. The detec-tor temperature was 270°C. The MSD was operated in theelectron ionization (EI) mode at 70 eV with a source tem-perature of 200–250°C. The electron multiplier voltage of thedetector was set at 200–700 V above autotune and the dwelltime for each ion was 50 ms.

GC-MS procedures for MED and EED. The acetonesolutions of the extract (60 µL) were transferred into screw-capped injection vials. Approximately 2 µL were injectedonto the GC-MS. The GC-MS procedure was similar to thatdescribed previously by Shimomura et al. (2001). The oventemperature was increased from 80°C (held for 0.5 min)to 140°C at 20°C/min and then to 265°C at 50°C/min. Theinjector temperature was 140°C. Ions monitored were asfollows: m/z 181, 166 and 152 for MED; m/z 195, 166 and 138for EED; and m/z 212 and 183 for d3-PED. To preventartifact formation of MED from other co-extracted cocainecompounds, injector temperatures more than 140°C wereavoided. The retention times for MED, EED and d3-PEDwere 3.75, 4.08 and 4.47 min, respectively.

GC-MS procedures for COC and CE. After analysis of MEDand EED, the same solutions were tested for COC and CE.Approximately 2 µL of the solutions was injected onto GC-MS. The oven temperature was increased from 170°C (heldfor 0.5 min) to 240°C at 20°C/min and then to 270°C (held

for 2 min) at 30°C/min. The injector temperature was 190°C.The ions monitored were as follows: m/z 303, 272 and 182 forCOC; m/z 317, 272 and 196 for CE; and m/z 306 and 185 ford3-COC. The retention times for d3-COC, COC and CE were4.06, 4.06 and 4.30 min, respectively.

Derivatization and GC-MS procedures for Bz, m-HBz andp-HBz. After analysis of MED, EED, COC and CE, 50 µLof a mixture of DMF and DMF-DPA (2:1 by volume) wereadded to the samples in the screw-capped vials. On-columnderivatization was performed at an injection port temperatureof 280°C. The oven temperature was increased from 190°C(held for 0.5 min) to 240°C at 20°C/min and then to 270°C(held for 3 min) at 30°C/min. Ions monitored were as follows:m/z 331, 272 and 210 for propyl Bz; m/z 334 and 213 forpropyl d3-Bz; m/z 389, 330 and 210 for propyl m- and p-HBz-O-propyl esters; m/z 392 and 213 for propyl m- and p-d3-HBz-O-propyl esters. To maximize sensitivity, the Bz ionsand the HBz ions were monitored separately at electron mul-tiplier voltages of 200 and 400 V, respectively, above autotunevoltage. The retention times for d3-Bz,Bz, d3-m-HBz, m-HBz,d3-p-HBz and p-HBz derivatives were 3.57, 3.57, 4.97, 4.97,5.37 and 5.38 min, respectively.

Derivatization and GC-MS procedures for MEC, EEC,NBz, NC and NCE. The second portion of the dried extract(Scheme 1) was treated with 50 µL each of PFPL and PFPAand heated at 50°C for 15 min. Excess reagent was evapor-ated to dryness at room temperature under nitrogen. ThePFP derivatives were dissolved in 50 µL of dry acetonitrileand approximately 2–3 µL was injected into the GC-MS totest for MEC, EEC and NBz. The oven temperature wasincreased from 100°C (held for 0.5 min) to 160°C at 15°C/minand then to 260°C (held for 2.0 min) at 50°C/min. The injec-tor temperature was 180°C. Ions monitored were as follows:m/z 345, 314 and 182 for MEC-PFP; m/z 348 and 185 for d3-MEC-PFP; m/z 359, 314 and 196 for EEC-PFP; m/z 431, 312and 214 for NBz; and m/z 424 and 303 for d3-Bz-PFP. Thecorresponding retention times were 3.88, 3.87, 4.34, 6.71 and6.59 min, respectively. To maximize sensitivity, MEC/EECions and NBz/d3-Bz ions were monitored in two separategroups. The electron multiplier voltage was 300–400 V abovethe autotune voltage.

Generally, the amounts of NC and NCE in specimenswere considerably less than that of MEC, EEC and NBz.Therefore, after the MEC, EEC, and NBz analysis, thederivatized extract was re-injected with a larger injectionvolume (4 µL) and at a higher electron multiplier voltage of500 V above autotune. The GC conditions were the sameas those described for MEC, EEC and NBz analysis. Ionsmonitored were as follows: m/z 313, 194 and 166 for NC-PFP;m/z 316 and 197 for d3-NC-PFP; and m/z 327, 208 and 214 forNCE. The corresponding retention times were 4.18, 4.17 and4.35 min, respectively.

Derivatization and GC-MS procedures for EC, ED andNED. The dried extract (Scheme 1) was dissolved in a mix-ture of tetramethylammonium hydroxide (25% in methanol),trimethylphenylammonium hydroxide (0.2 m in methanol) anddimethylsulfoxide (10 µL–200 µL–2 mL). The sample wastreated with 20 µL of 1-iodopentane and heated at 50°C for

Copyright © 2005 John Wiley & Sons, Ltd. Biomed. Chromatogr. 19: 677–688 (2005)

Cocaine and pyrolytic cocaine metabolites in human blood and urine 681ORIGINAL RESEARCH

15 min in a closed tube. The solution was cooled to roomtemperature and treated with 2 mL each of 0.15 m sulfuricacid and ethyl acetate. After vortex-mixing and centrifuga-tion, the supernatant was discarded. The acidic solution wasmade basic with 1 mL of 1.5 m carbonate buffer (pH 9.5) and150 µL of 2 m sodium hydroxide, and drugs were extractedwith 2 mL of ethyl acetate. The solvent was evaporated todryness at room temperature under a slow stream of nitrogen.Prolonged evaporation was avoided to minimize the loss ofpentyl ED. The residue was dissolved in 60 µL of acetone andtested by GC-MS. The oven temperature was increased from80°C (held for 0.5 min) to 140°C at 20°C/min and then to265°C (held for 2.0 min) at 50°C/min. The injector tempera-ture was 140°C. Ions monitored were as follows: m/z 237, 208and 138 for pentyl ED; m/z 240 and 211 for pentyl d3-ED;m/z 293, 264 and 236 for dipentyl NED. The correspondingretention times were 4.52, 4.51 and 5.31 min, respectively.Specimens also were tested for EC under the same GC condi-tions. The monitored ions were m/z 255, 238 and 168 forpentyl EC, and m/z 258 and 171 for pentyl d3-EC. Retentiontimes for both compounds were 4.96 min.

RESULTS AND DISCUSSION

Drugs were extracted and quantified according toScheme 1. Initially, specimens were adjusted to pH 6.0and allowed to flow through the SPE columns. Themajority of the compounds were retained on thecolumn except the more polar EC, ED and NED.The unabsorbed compounds were collected, readjustedto pH 2–3 and passed through another set of SPEcolumns (C8 + benzene sulfonic acid, 200 mg). Under

these acidic conditions the carboxylic acids were un-dissociated and lipophilic and the solid-phase materialsabsorbed the compounds more efficiently. A polarsolvent mixture containing methanol, 2-propanol andaqueous ammonia (4:6:0.25 by volume) was necessaryto elute EC, ED and NED from the SPE columns. Acommonly used solvent mixture (methylene chloride–2-propanol–aqueous ammonia, 8:2:0.2 by volume) foreluting basic drugs gave poor recovery (<2%). Elevencocaine compounds and four pyrolytic cocaine com-pounds were extracted and detected by this method.A schematic diagram of cocaine metabolism is pre-sented (Fig. 1). De-esterification and oxidation arethe major pathways of cocaine metabolism. Ethanoltransesterification was also observed when alcohol andcocaine were consumed simultaneously. If smoking isthe route of administration, pyrolytic product methylecgonidine (MED) is formed and consumed. Likecocaine, MED undergoes de-esterification, oxidationand ethanol transesterification (Fig. 2). The new meta-bolites, EED and NED, are the products of MEDmetabolism. Like alcohol-specific cocaine metabolites(CE, EEC or NCE), EED is an ethanol transester-ification product of MED. The reference compoundwas prepared by esterification of ED by DMF-DEA.In GC-MS, EED showed an elution and mass frag-mentation pattern similar to that of MED (Fig. 3).The oxidative metabolite, NED, was prepared frompyrolysis of NBz under acidic conditions. For identifica-tion, the compound was derivatized to N,O-dipentyl-NED or di-PFP-NED, and tested by GC-MS. The

Figure 1. Cocaine metabolism after hydrolysis (HOH), oxidation (Ox), and ethanol transester-ification (ethanol) in human.

Copyright © 2005 John Wiley & Sons, Ltd. Biomed. Chromatogr. 19: 677–688 (2005)

682 B. D. Paul et al.ORIGINAL RESEARCH

Figure 2. Metabolism of pyrolytic (smoked) cocaine after hydrolysis (HOH), oxidation (Ox) andethanol transesterification (ethanol) in human.

Figure 3. Mass fragmentation of methyl ecgonidine, ethyl ecgonidine, propyl ecgonidine (d3-), ecgonidine (O-pentyl-) and nor-ecgonidine (N,O-dipentyl- and N,O-di-PFP-).

fragmentation pattern of the N,O-dipentyl-NED wassimilar to that of other alkylated ecgonidine com-pounds (Fig. 3). d3-PED was used as an internal stand-ard for the quantification of MED, as it offered a betterselection of ions than d3-MED. The compound wasprepared by esterification of d3-ED by DMF-DPA andidentified by GC-MS (Fig. 3). Non-deuterated PEDwas avoided as an internal standard for MED becausethe starting compound (ED), if left unreacted, couldinterfere with detection of ED in specimens. Internalstandards for each analyte are given in Table 1.

The drugs in the specimens were identified bycomparing retention times (±2%) and relative ion

abundances (±20%) with those of known referencecompounds. Overall recoveries were 49% for MEC,35% for EEC, 29% for EC, and more than 83% forall other drugs. Within the linear range, correlationcoefficients were 0.9901 (NBz) to 1.0000 (COC and CE;Table 1). The same identification and quantificationcriteria were used for all concentrations within the limitof linearity and, therefore, the limit of quantitation(LOQ) was considered as the limit of detection (LOD).At LOD, chromatographic backgrounds for urine andblood samples were essentially the same (peak:noise>4). For batch analysis, a calibrator (standard) and twoquality control samples of known drug concentrations

Copyright © 2005 John Wiley & Sons, Ltd. Biomed. Chromatogr. 19: 677–688 (2005)

Cocaine and pyrolytic cocaine metabolites in human blood and urine 683ORIGINAL RESEARCH

Table 1. Method validation results for fifteen cocaine analytes in blood and urine

Internal Urine BloodDrug standard Linearity, ng/mL r Linearity, ng/mL r

COC d3-COC 1–2000 1.0000 1–2000 0.9999CE d3-COC 1–2000 0.9999 1–50 1.0000NC d3-NC 2–100 0.9963 2–50 0.9987NCE d3-NC 2–50 0.9946 4–50 0.9913MEC d3-MEC 1–2000 0.9993 1–2000 0.9998EEC d3-MEC 0.5–2000 0.9935 2–2000 0.9995Bz d3-Bz 4–2000 0.9999 4–2000 0.9999NBz d3-Bz 2–1000 0.9901 2–1000 0.9901m-HBz d3-m-HBz 1–2000 0.9999 1–2000 0.9997p-HBz d3-p-HBz 1–2000 0.9995 1–2000 0.9999EC d3-EC 16–2500 0.9992 16–2500 0.9999MED d3-PED 1–1000 0.9985 1–50 0.9997EED d3-PED 2–50 0.9996 2–50 0.9988ED d3-ED 2–2500 0.9996 2–2500 0.9998NED d3-ED 1–250 0.9938 2–50 0.9985

were used. A drug-free control was also used in theanalysis. The calibrator and the controls were preparedfrom different stock solutions. In batch analysis, theaccuracy was monitored by control concentrationswithin ±20% of the expected values. In-run precisionswere evaluated at calibrator and control concentrations.Excepting NBz and NED, the coefficients of variationwere within ±10% (n = 3). For NBz and NED, thecoefficients of variation were within ±20%. Differencein chemical properties between the nor-compounds andthe internal standards may be the reason for highercoefficients of variation of NBz and NED. In GC-MSanalysis, all compounds showed good chromatographiccharacteristics (Fig. 4). In specimen analysis the back-ground noise may vary. Therefore, the ion-ratiorequirement within ±20% is important for drug iden-tification, because it controls background noise within20%.

The procedure was applied to test eight PM bloodand urine specimens. An additional five PM urinespecimens were also tested. Most victims died frominjury or multiple drug use, and therefore, the causeof death could not be directly related to cocaine use(Table 2). Seventeen urine specimens, collected fromliving individuals (LV), were also tested. The results ofthe PM and LV specimens are summarized in Tables 3and 4, respectively. EED, a marker for combined useof smoked cocaine and alcohol, was detected in onlyseven of 13 PM urine specimens. In three EED posi-tive specimens (PM-1, PM-2 and PM-8), alcohol wasdetected in blood, urine and vitreous fluids, indicatingthat EED was a true metabolite of smoked cocaineand ethanol rather than a post-mortem artifact. Theconcentrations of EED were relatively small (0–39 ng/mL) compared with those of the alcohol-related COCmetabolite, EEC. To determine the combined use ofsmoked COC and alcohol, detection of ED and EECappeared to be better than detection of EED alone.

Table 2. Drug use profile and cause of death for post-mortemspecimens

Specimen Incidence Drug positive

1 Hit by car Coc, Alc2 Found dead Coc, Op, Alc3 Overdose Coc, Op, Alc4 Overdose Coc, Op5 Drug addict Coc, Op6 Suicide Coc7 Gun-shot wound Coc8 Found dead Coc, Alc9 Gun-shot wound Coc, MDMA

10 Distended bladder Coc, Op11 Suicide Coc, Alc12 Gun-shot wound Coc, Alc13 Not known Coc

Coc, cocaine; Op, opiates; Alc, alcohol.

However, like COC and CE, EED may have pharma-cological properties similar to that of MED. Theoxidative metabolite, NED was detected in both PM(10/13) and LV (12/17) urine specimens and the con-centrations were 0–75 and 0–163 ng/mL, respectively.The pharmacological properties of NED are unknown.

The cocaine pyrolytic metabolite, ED was detectedin 10 of 13 PM specimens (77%) supporting our previ-ous observation that smoking is the major route ofcocaine administration (Paul et al., 1999). In three otherspecimens the ED concentrations may be below theLOD because the overall drug concentrations werelow. However, for these specimens, the other routes ofdrug administration cannot be ruled out. The concen-trations of alcohol-specific cocaine metabolites were inthe order of EEC > CE > NCE. Nine urine specimenshad detectable EEC, indicating concomitant use ofalcohol in 69% of these PM cases. In almost all EECpositive urine specimens (8/9), ethanol was detected inat least one body fluid. Cocaine oxidative metabolites,

Copyright © 2005 John Wiley & Sons, Ltd. Biomed. Chromatogr. 19: 677–688 (2005)

684 B. D. Paul et al.ORIGINAL RESEARCH

Figure 4. Total and selected ion chromatogram of 15 cocaine and smoked cocaine compounds at calibratorconcentration.

Copyright © 2005 John Wiley & Sons, Ltd. Biomed. Chromatogr. 19: 677–688 (2005)

Cocaine and pyrolytic cocaine metabolites in human blood and urine 685ORIGINAL RESEARCH

Tab

le 3

. Con

cent

ratio

n pr

ofile

s of

15

coca

ine-

rela

ted

com

poun

ds in

pos

t-m

orte

m b

lood

and

uri

ne s

peci

men

s

Pyr

olyt

ic c

ocai

neC

ocai

ne m

etab

olite

s, n

g/m

La

met

abol

ites,

ng/

mL

aE

than

olb

Spec

imen

CO

CC

EN

CN

CE

ME

CE

EC

Bz

NB

zm

-HB

zp-

HB

zE

CM

ED

EE

DE

DN

ED

mg/

dL

PM

-1, U

rine

2543

2576

960

8949

3170

1853

113

130

346

420

3331

811

335

Blo

od3

87

883

5123

227

(339

)P

M-2

, Uri

ne41

880

046

4137

5573

1334

322

524

473

310

1842

210

203

Blo

od26

2094

43

3085

199

(235

)P

M-3

, Uri

ne45

6413

634

348

3598

178

1141

116

814

177

553

1212

53

92B

lood

14

114

3753

011

1812

292

37N

Dc (

21)

PM

-4, U

rine

266

161

9113

050

667

333

225

1802

376

1385

2753

Blo

od7

174

641

011

646

733

8N

D (

ND

)P

M-5

, Uri

ne12

2731

6217

714

233

2244

286

186

3686

522

1545

19N

DB

lood

231

225

643

223

101

14 (

ND

)P

M-6

, Uri

ne32

521

6816

301

272

321

8723

3630

295

12N

DB

lood

352

543

153

585

29N

D (

ND

)P

M-7

, Uri

ne83

4013

266

5743

066

7711

525

1409

4146

204

1340

26N

DB

lood

414

7526

0817

518

5013

4236

138

ND

(N

D)

PM

-8, U

rine

4505

520

6115

2289

4462

315

271

552

2081

5062

399

3919

3127

40B

lood

147

558

5221

9717

431

8882

933

012

(37

)P

M-9

, Uri

ne12

93

ND

(N

D)

PM

-10,

Uri

ne87

370

619

195

39N

D (

ND

)P

M-1

1, U

rine

205

202

1888

166

1911

462

(11

1)P

M-1

2, U

rine

1159

3425

577

397

9037

427

213

811

2232

1185

436

76

2228

5211

(14

)P

M-1

3, U

rine

6523

3322

212

011

6430

6102

424

5654

0910

277

638

6303

75N

D (

NA

d )

a Con

cent

rati

ons

belo

w t

he L

OD

are

not

rep

orte

d.b E

than

ol c

once

ntra

tion

s in

vit

reou

s flu

id a

re i

n pa

rent

hese

s.c N

D,

not

dete

cted

(<1

0 m

g/dL

).d N

A, n

ot a

vaila

ble.

Copyright © 2005 John Wiley & Sons, Ltd. Biomed. Chromatogr. 19: 677–688 (2005)

686 B. D. Paul et al.ORIGINAL RESEARCH

Table 4. Concentration profiles of 15 cocaine-related compounds in urine collected from living individualsa

Pyrolytic cocaineCocaine metabolites, ng/mLb metabolites, ng/mLb

Specimen COC CE NC NCE MEC EEC Bz NBz m-HBz p-HBz EC MED EED ED NED

1 8 2 218 4 10 5 73 131 22 4 96 16 284 14 8 20 8613 1 20 1 297 14 4 11 2864 8 280 11 4 181 225 64 317 8 518 14 35 28 247 506 600 5 40 4 287 67 467 10 7 9 201 6 18 61 426 10 906 29 47 48 717 752 129 204 808 1114 109 49 182 6820 10 2148 38

10 6 155 1516 48 32 84 1926 78 411 5 206 7 2015 61 67 54 705 8 23 412 5 3506 98 34 74 1373 7 10 113 435 68 42 14 3107 132 9831 514 57 534 872 13 460 1514 26 1943 9567 414 46 258 1111 8 14 115 118 6207 21243 448 732 1224 2263 18 82 516 71 16 10151 27 132355 14880 1367 1358 24036 121 2937 4917 113 21 523731 72172 1525 6939 341187 224 2348 163

a Ethanol was not detected in any of the specimens.b Concentrations below the LOD are not reported.

specimens LV 1–3, 5, 8, 11, 13, 16 and 17) and thosewithout (specimens LV 4, 6, 7, 9, 10, 12, 14 and 15).Assuming unequal variances, no significant differencewas observed for either oxidative metabolite (NBz,tCrit,0.05(2),15 = 2.13 > tCal = 1.11; HBz tCrit,0.05(2),8 = 2.20 > tCal

= 0.21). The t-test was calculated by standard MicrosoftExcel software or from statistical formulas (Zar, 1999).In both cases the t-calculated values were less than thet-critical values (95% confidence), suggesting a lack ofinfluence of alcohol on metabolic oxidation of Bz inthese specimens. However, this observation is limitedto specimens containing no alcohol, but having at leastone of the alcohol-related COC metabolites. In PMcases, the mole percentage of NBz and HBz in spec-imens containing and devoid of alcohol, also showedno significant differences (NBz, tCrit,0.05(2),7 = 2.36 > tCal =0.23; HBz, tCrit,0.05(2),8 = 2.31 > tCal = 0.40), suggesting nomajor influence of alcohol on metabolic oxidation ofBz.

In six of 10 LV specimens the molar concentrationsof EC appeared to be greater than those of Bz whenBz concentrations were less than 2000 ng/mL. The con-centration profiles of Bz and EC in 20 Lvl urine spec-imens, positive in immunoassay (>150 ng/mL) butnegative by GC-MS (<100 ng/mL), were also studied.The concentrations of oxidative metabolites, m- and p-HBz, and the major pyrolytic metabolite, ED, also arereported in Table 5. All specimens had EC concen-trations higher than any other analyte. It appearedthat EC is the major metabolite at the end-phase ofBz excretion. ED was detected in 11 specimens. In sixspecimens the molar concentrations of ED were greaterthan Bz. Both m- and p-HBz were minor analytes. In

NC and NCE, were not detected in any blood spec-imen. A similar pattern of COC use by smoking andconcomitant use of COC and alcohol were observed inthe LV specimens (88 and 53%, respectively).

Molar concentrations of HBz (m- + p-) and NBzin all 13 PM urines had a linear relationship with thetotal concentration of cocaine metabolites (r = 0.9613and 0.9693, respectively). When compared only toBz concentrations, correlation coefficients were 0.9371and 0.9418, respectively. This indicates that Bz is theprimary compound for metabolic hydroxylation andoxidative N-demethylation. Similar results also wereobserved in 17 LV specimens. Correlation coefficientsfor HBz and NBz were 0.9751 and 0.9994, respectively,when compared with total metabolite concentrations,and 0.9838 and 0.9986, respectively when comparedonly with Bz concentrations. Four urinary meta-bolites, Bz, NBz, MEC and EC, represented almost88% (59–98%) of all PM cocaine metabolites and 97%(91–99%) of all LV cocaine metabolites. In PM urinespecimens, pyrolytic metabolites were 0–14.4% of non-pyrolytic metabolites, while in LV specimens the rangewas 0–63%. Three of the LV specimens had highermolar concentrations of ED than Bz (LV-1, LV-8 andLV-9). Without knowledge of kidney re-absorption andexcretion kinetics of these compounds, it is difficultto estimate the pyrolysis of cocaine based on the urineresult alone.

Alcohol-effect on oxidative metabolism of Bz wasinvestigated indirectly using LV urine specimens. Themolar concentrations of NBz and HBz as a percentageof Bz (mole percentage) were compared between thegroups with alcohol markers (CE, NCE and EEC in

Copyright © 2005 John Wiley & Sons, Ltd. Biomed. Chromatogr. 19: 677–688 (2005)

Cocaine and pyrolytic cocaine metabolites in human blood and urine 687ORIGINAL RESEARCH

Table 5. Concentration profile for selected cocaine com-pounds in 20 Lvl urine specimens that were positive inimmunoassay (Bz >>>>> 150 ng/mL) and negative in GC-MSconfirmation (Bz <<<<< 100 ng/mL)

Specimen Bz m-HBz p-HBz EC EDng/mLa

1 74 6.0 447 1172 54 23 10 953 5.1 5124 51 3.7 596 165 77 12 542 436 3997 34 9.3 253 228 76 3589 54 16 362 8510 25 56411 58 21712 43 13413 67 345 5214 67 494 3415 76 662 3116 437 5117 52 684 21718 79 22 203 2519 59 34520 445

a Concentrations below the LOD are not reported.

four specimens, Bz could not be detected (<LOD). Thepositive immunoassay response may be due to EC incombination with other COC metabolites. For forensicinvestigation of COC use, testing for EC is advant-ageous when Bz concentrations are below 100 ng/mL.In urine drug testing, sometimes the accused claims theBz positive results are due to COC contaminationof collection devices. This claim appears justified be-cause COC, when added to urine, may spontaneouslyhydrolyze to Bz. To refute the claim, the presence ofoxidative metabolite (NBz, m-HBz or p-HBz) substan-tiates COC use. The claim for unknowing ingestion infood or drink (Baselt and Chang, 1987) or unknowingexposure from dermal absorption (Baselt et al., 1990;ElSohly, 1991) may also be verified by testing for the

presence of pyrolytic compound, ED. The presence ofED would indicate ingestion by the smoked route.Free-base COC is generally administered by heatingthe compound in a device and inhaling the vapor intwo to three puffs. This limits the amount of cocaine inthe environment and availability for possible exposure.Therefore, a positive result with Bz more than 100 ng/mL from side-stream passive exposure is unlikely(Baselt et al., 1991).

The stabilities of COC, NC, MEC, Bz, NBz, m-HBz,p-HBz, MED, ED and NED in urine were evaluatedover a period of 6 months. The samples were preparedat calibrator and control concentrations (0.5 × and 2.0× of calibrator), stored at −18°C and tested at threedifferent times. The highest coefficient of variation wasobserved with NBz (34%). Differences in chemicalproperties between NBz and internal standard (d3-Bz)may be the reason for the high value. In all othercases the coefficients of variation were 3.8–14%. Thestabilities of ethyl esters (CE, NCE, EEC and EED)were not evaluated. But the compounds are expectedto be as stable as the methyl esters. Blood specimenscontaining COC, MEC, Bz, m-HBz, p-HBz, EC andED were stored frozen at −18°C and tested at differenttimes (Table 6). The stability of EC was evaluated onlyfor 2 months because the method was not availablebefore this period. The slope and correlation coefficientindicated that EC was stable (m = 0.98 and r = 0.9758).All other compounds were stable for at least 24months. During analysis all compounds were stable forat least 48 h at room temperature. No noticeable lossin GC-MS sensitivity was observed. When injectionswere delayed, the samples were stored at −18°C. Thecompounds were stable under this condition for at least7 days.

The concentration profiles of 15 COC-related com-pounds in blood and urine provided useful informationon metabolism of COC and pyrolytic MED. Based onmedian concentrations (µmol/L) the relative amountswere Bz > MEC > (EC, NBz) > COC > (EEC, m-HBz,p-HBz) > CE > (NC, NCE) for PM urine specimens

Table 6. Stability of drugs in blood specimens after stored for 2–36 months at −−−−−18°C

COC MEC Bz m-HBz p-HBz EC ED36 months 24 months 24/36 months 24 months 24 months 2 months 2/36 months

3/5 342/558 39/30/NT 13/15 2.8/2.8 485/585 29/27/294/12 1323/1475 38/32/NT 29/31 68/87 821/829 330/340/314

96/87/NT 28/18 77/50 1395/1342 138/90/143446/543/458 192/223 85/71/NT527/628/NT 103/51 101/92/NT

3304/2197/NT 600/467 23/19/NT1791/2071/NT 469/292 37/32/NT2271/2608/1831

The first result represents initial concentration, ng/mL.The second and third results represent concentrations (ng/mL) after storage.NT, not tested.

Copyright © 2005 John Wiley & Sons, Ltd. Biomed. Chromatogr. 19: 677–688 (2005)

688 B. D. Paul et al.ORIGINAL RESEARCH

(n = 13) and Bz > EC > (NBz, p-HBz, MEC, m-HBz) >COC > EEC > CE > (NC, NCE) for LV urine spec-imens (n = 17). Pyrolytic compounds were present inthe order ED > MED > NED > EED in both PM andLV specimens. Hydrolysis is the major metabolic path-way of COC metabolism. The average mole percent-ages (median, range) of parent compounds (COC, CE),oxidative metabolites (NC, NCE, NBz, m-HBz, p-HBz)and hydrolytic metabolites (Bz, NBz, MEC, EEC, m-HBz, p-HBz, EC) were 5.9 (4.1, 0–23.6), 9.2 (5.9, 1.7–27.7) and 94.0 (95.9, 76.4–100), respectively in PMurine. In LV specimens, the mole percentages of thesame metabolites were 0.7 (0.1, 0–4.0), 4.7 (4.3, 2.3–9.0)and 99.2 (99.9, 96.0–100), respectively. From COC andpyrolytic COC metabolites, it appears that smoking wasthe major route of cocaine administration (PM 77%and LV 88%). Combined use of COC and alcohol, asevident from the alcohol-related COC metabolites, wasalso substantial (PM 69% and LV 53%). EED, a newmetabolite that relates to concurrent use of smokedCOC and alcohol, was present only in PM specimensin small amounts. Therefore, ethanol-related meta-bolite, EEC, and pyrolytic metabolite, ED, togetherappeared to be better markers than EED for identify-ing combined use of smoked COC and alcohol. ECappeared to be the major metabolite when Bz concen-trations are below 100 ng/mL, and therefore, in forensicinvestigation of COC use, detection of EC is advant-ageous when Bz concentrations are less than 100 ng/mL. Combined use of smoked COC and alcohol inhuman is common and therefore, pharmacological andpharmacokinetic investigations of cocaine use shouldinclude measurement of appropriate metabolites.

REFERENCES

Baselt RC and Chang R. Urinary excretion of cocaine andbenzoylecgonine following oral ingestion in a single subject.Journal of Analytical Toxicology 1987; 11: 81.

Baselt RC, Chang JY and Yoshikawa DM. On the dermal absorptionof cocaine. Journal of Analytical Toxicology 1990; 14: 383.

Baselt RC, Yoshikawa DM and Chang JY. Passive inhalation ofcocaine. Clinical Chemistry 1991; 37: 2160.

Brower KJ, Hierholzer R and Maddahian E. Recent trends incocaine abuse in a VA psychiatric population. Hospital andCommunity Psychiatry 1986; 37: 1229.

Chen LC, Graefe JF, Shojaie J, Willetts J and Wood RW. Pulmonaryeffects of the cocaine pyrolysis product, methylecgonidine, inguinea pigs. Life Sciences 1995; 56: 7.

ElSohly MA. Urinalysis and casual handling of marijuana andcocaine. Journal of Analytical Toxicology 1991; 15: 46.

Erzouki HK, Allen AC, Newman AH, Goldberg SR andSchindler CW. Effects of cocaine, cocaine metabolites and cocainepyrolysis products on the hindbrain cardiac and respiratory centersof the rabbit. Life Sciences 1995; 57: 1861.

Huang L, Woolf JH, Ishiguro Y and Morgan JP. Effect of cocaineand methylecgonidine on intracellular Ca2+ and myocardial contrac-tion in cardiac myocytes. American Journal of Physiology 1997;273: 893.

Jones RT. The pharmacology of cocaine smoking in humans. NIDAResearch Monographs 1990; 99: 30.

Martin BR, Lue LP and Boni JP. Pyrolysis and volatilization ofcocaine. Journal of Analytical Toxicology 1989; 13: 158.

Nakahara Y and Ishigami A. Inhalation efficiency of free-basecocaine by pyrolysis of ‘crack’ and cocaine hydrochloride. Journalof Analytical Toxicology 1991; 15: 105.

Paul BD, McKinley RM, Walsh JK, Jamir TS and Past MR. Effect offreezing on the concentration of drugs of abuse in urine. Journal ofAnalytical Toxicology 1993; 17: 378.

Paul BD, McWhorter LK and Smith ML. Electron ionization massfragmentometric detection of urinary ecgonidine, a hydrolyticproduct of methylecgonidine, as an indicator of smoking cocaine.Journal Mass Spectrometry 1999; 34: 651.

Perez-Reyes M, Di Guiseppi S, Ondrisek G, Jeffcoat AR andCook CE. Free-base cocaine smoking. Clinical Pharmacology andTherapy 1982; 32: 459.

Schnoll SH, Karrigan J, Kitchen SB, Daghestani A and Hansen T.Characteristics of cocaine abusers presenting for treatment. NIDAResearch Monographs 1985; 61: 171.

Shimomura ET, Hodge GD and Paul BD. Examination of post-mortem fluids and tissues for the presence of methylecgonidine,ecgonidine, cocaine and benzoylecgonine using solid-phase extrac-tion and gas chromatography-mass spectrometry. Clinical Chem-istry 2001; 47: 1040.

Singer LT, Arendt R, Minnes S, Farkas K and Salvator A.Neurobehavioral outcomes of cocaine-exposed infants. Neuro-toxicology and Teratology 2000; 22: 653.

Wesson DR and Washburn P. Current patterns of drug abuse thatinvolved smoking. NIDA Research Monographs 1990; 99: 5.

Willetts J, Chen LC, Graefe JF and Wood RW. Effects ofmethylecgonidine on acetylcholine-induced bronchoconstrictionand indicators of lung injury in guinea pigs. Life Sciences 1995; 57:225.

Woolf JH, Huang L, Ishiguro Y and Morgan JP. Negative inotropiceffect of methylecgonidine, a major product of cocaine basepyrolysis, on ferret and human myocardium. Journal of Cardiovas-cular Pharmacology 1997; 30: 352.

Zar JH. Biostatistical Analysis, 4th edn. Prentice Hall: EngelwoodCliffs, NJ, 1999; 122.