RESEARCH PAPER

Quantification of cocaine and metabolites in exhaled breathby liquid chromatography-high-resolution mass spectrometryfollowing controlled administration of intravenous cocaine

Kayla N. Ellefsen & Marta Concheiro & Olof Beck &

David A. Gorelick & Sandrine Pirard & Marilyn A. Huestis

Received: 10 June 2014 /Revised: 11 July 2014 /Accepted: 17 July 2014 /Published online: 17 August 2014# Springer-Verlag Berlin Heidelberg (outside the USA) 2014

Abstract Breath has been investigated as an alternative ma-trix for detecting recent cocaine intake; however, there are nocontrolled cocaine administration studies that investigated thedrug’s disposition into breath. Breath was collected from 10healthy adult cocaine users by asking them to breathe into aSensAbues device for 3 min before and up to 22 h following25 mg intravenous (IV) cocaine dosing on days 1, 5, and 10,and assayed with a validated liquid chromatography-high-resolutionmass spectrometry (LC-HRMS)method to quantifybreath cocaine, benzoylecgonine (BE), ecgonine methyl ester(EME), and norcocaine. The assay was linear from 25 to1,000 pg/filter, extraction efficiencies were 83.6–126 %,intra- and inter-assay imprecision was <10.6 %, and biaswas between −8.5 and 16.8 %. No endogenous or exogenousinterferences were observed for more than 75 tested. Analyteswere generally stable under short-term storage conditions. Ionsuppression was less than 46 %. Of breath specimens collect-ed after controlled cocaine administration, 2.6 %were positivefor cocaine (26.1–66 pg/filter, 1–9.5 h), 0.72 % BE (83.3–151 pg/filter, 6.5–12.5 h), and 0.72% EME (50–69.1 pg/filter,

6.5–12.5 h); norcocaine was not detected. Methanolic extrac-tion of the devices themselves, after filters were removed,yielded 19.2 % positive cocaine tests (25.2–36.4 pg/device,10 min–22 h) and 4.3 % positive BE tests (26.4–93.7 pg/device, 10 min–22 h), explaining differences between the twoextraction techniques. These results suggest that the devicereflects the drug in oral fluid as well as lung microparticles,while the filter reflects only drug-laden microparticles. Asensitive and specific method for cocaine, BE, EME, andnorcocaine quantification in breath was developed and vali-dated. Cocaine in breath identifies recent cocaine ingestion,but its absence does not preclude recent use.

Keywords Cocaine . Exhaled breath . LC-HRMS .

Metabolites . SensAbues

Introduction

Cocaine is the second most frequently identified illicit drug inthe USA, after cannabis, and is the most widely used illicitstimulant in Europe [1, 2]. In 2012, there were approximatelytwomillion current cocaine users in the USA and Europe [2, 3].

Cocaine is rapidly metabolized, primarily in the liver, byhydrolysis of the two ester linkages [4]. Its two major inactivemetabolites are benzoylecgonine (BE) and ecgonine methylester (EME). Additionally, cocaine is metabolized via N-de-methylation by cytochrome P450 enzymes to produce a minoractive metabolite, norcocaine [5, 6].

There is increased interest in utilizing alternative matrices,other than blood and urine, for identifying drug use in drivingunder the influence of drugs (DUID), workplace, and clinicaland anti-doping programs [7–9]. Studies focused on oral fluid asa viable alternative matrix due to its non-invasiveness, observedcollection, reduced potential for adulteration, and general iden-tification of recent use [10–13]. However, oral fluid has several

Electronic supplementary material The online version of this article(doi:10.1007/s00216-014-8051-x) contains supplementary material,which is available to authorized users.

K. N. Ellefsen :M. Concheiro : S. Pirard :M. A. Huestis (*)Chemistry and Drug Metabolism, National Institute on Drug Abuse,NIH, Baltimore, MD 21224, USAe-mail: mhuestis@intra.nida.nih.gov

K. N. EllefsenProgram in Toxicology, University of Maryland, Baltimore,Baltimore, MD 21201, USA

O. BeckKarolinska Institute, 17177 Stockholm, Sweden

D. A. GorelickDepartment of Psychiatry, School of Medicine, University ofMaryland, Baltimore, MD 21224, USA

Anal Bioanal Chem (2014) 406:6213–6223DOI 10.1007/s00216-014-8051-x

disadvantages. The sample volume is small, especially afterintake of sympathomimetic drugs, and interpretation of resultscan be difficult due to oral mucosa contamination, yielding apoor correlation of oral fluid/plasma concentrations early in thetime course. Lately, several studies investigated the potential ofexhaled breath as an alternative matrix for detecting recentconsumption of drugs of abuse [14–23], including cocaine [20,22, 23]. This concept is based on the hypothesis that exogenoussubstances contaminate the airway lining fluid creating aerosolparticles (microparticles) that can be detected in exhaled breath[22]. Exhaled breath contains microparticles that can carry non-volatile compounds, including airway lining fluid, out of thelungs. Formation of these microparticles is believed to occurwhen the airway opens during inhalation [24]. Therefore, if thelungs are contaminated with exogenous substances, they canbecome trapped in the airway lining fluid and, thus, distributeinto the breath upon exhalation.

Three previous studies documented cocaine in exhaled breath[20, 22, 23]. Cocaine was detected (80 pg/min) in one patientrecovering from acute self-reported cocaine intoxication [20].Another study detected cocaine in 10 patients following acuteintoxication, approximately 24 h after drug intake [22]. Cocainebreath concentrations ranged from 29 to 13,000 pg/device (me-dian 60 pg/device); BEwas quantified in four of these cases withconcentrations of 18–560 pg/device (median 43 pg/device). BEwas only found when cocaine was detected. Two of the positivebreath cases did not have detectable plasma or urine concentra-tions, or self-reported cocaine use. More recently, a study exam-ined exhaled breath in parallel with urine specimens from prisoninmates following arrival at correctional facilities [23]. Only 22of 247 specimens were positive for drugs of abuse in urine andbreath. Cocaine (12 pg/device) and BE (3 pg/device) weredetected in breath in only 1 of 247 cases where breath and urinespecimens were available; however, individual data were notprovided in the article, so it cannot be determined how manyurine samples were positive for BE.

Currently, there are no controlled cocaine administrationstudies examining cocaine and its metabolites in exhaled breath.This investigation evaluated pharmacodynamic and pharmaco-kinetic interactions between controlled intravenous (IV) cocaineadministration and two potential medication adherence markers,acetazolamide and quinine. The SensAbues breath device col-lected cocaine andmetabolites in breath, and their concentrationswere quantified in a new validated liquid chromatography-high-resolution mass spectrometry (LC-HRMS) method.

Materials and methods

Participants

Ten healthy volunteers aged 18–50 years, who smoked orused IV cocaine for at least 6 months and at least three times

per month during the 3 months prior to screening, wereeligible. Participants were screened prior to enrollment in thestudy by recruitment staff. Volunteers receive extensive psy-chological and psychosocial testing including a StructuredClinical Interview for DSM-IV (SCID), drug use survey, andAddiction Severity Index (ASI). In addition, participants alsocompleted comprehensive medical history and physical ex-aminations to determine their eligibility for the study based onour inclusion and exclusion criteria. Exclusion criteria includ-ed pregnant or nursing women; current physical dependenceon any drug other than cocaine, caffeine, or nicotine; currentclinically significant medical or psychiatric disorder; currenthypertension or blood pressure readings consistently above140 mmHg systolic or 90 mmHg diastolic while at rest; heartrate consistently above 90 bpm or below 50 bpm while at rest;abnormal ECG; history of clinically significant adverse reac-tion to cocaine; or interest in drug abuse treatment, within3 months of study screening. Subjects provided written in-formed consent to participate in this NIDA Institutional Re-view Board-approved study. Participants resided on a secureresearch unit for 13 days and 12 nights.

Cocaine administration and breath collection

Participants were administered a single 25-mg IV cocaine dosethrough a peripheral venous catheter on three separate daysduring the study (days 1, 5, and 10). Cocaine was administeredalone on day 1, with oral acetazolamide on day 5 and with oralquinine on day 10. Breath specimens were collected over 3 minwith the SensAbues (SensAbues AB, Huddinge, Sweden) de-vice an hour prior to IV cocaine and 10min and 0.5, 1, 1.5, 2, 3,4, 6.5, 9.5, 12.5, and 21 h (day 1 and 10) or 22 h (day 5) post-administration. The SensAbues device consists of a mouth-piece, collection chamber, and polymeric filter [20]. Partici-pants were asked to breathe normally during sampling, inhalingthrough their nose and exhaling through the mouthpiece. After3 min, the mouthpiece was discarded as per manufacturer’sinstructions and both ends of the collection chamber weresealed and stored at −20 °C within 2 h. New devices were usedfor each time point. No oral intake or smoking was allowed10 min prior to collection. Devices are designed to protectagainst oral fluid contamination with barrier ledges inside themouthpiece [20].

Chemicals and materials

Cocaine, ecgonine methyl ester, and norcocaine HCl stan-dards in acetonitrile and benzoylecgonine standard in metha-nol (1 mg/mL) were purchased from Cerilliant (Round Rock,TX). Internal standards, cocaine-d3 (1 mg/mL), ecgoninemethyl ester-d3 (100 μg/mL), norcocaine-d3 HCl (100 μg/mL) in acetonitrile, and benzoylecgonine-d8 (1 mg/mL) inmethanol also were obtained from Cerilliant. Solid phase

6214 K.N. Ellefsen et al.

extraction (SPE) was performed with UCT Clean Screencartridges (United Chemical Technologies, Inc., PA). Formicacid, methanol, and water were from Fisher Scientific (FairLawn, NJ). Ammonium formate and ammonium hydroxide28–30 % were from JT Baker (Phillipsburg, NJ). All solventsemployed in the extraction were high-performance liquidchromatography (HPLC) grade and, in the chromatographicsystem, LC-MS grade. Water was purified in-house by anELGA PURELAB Ultra Analytic purifier (Siemens WaterTechnologies, Lowell, MA).

Instrumentation

LC-high-resolution MS was performed on a Thermo Scientif-ic UltiMate 3000 RSLCnano system coupled to a ThermoScientific Q Exactive Mass Spectrometer (Thermo Scientific,Fremont, CA). The UltiMate 3000 RSLCnano systemconsisted of a degasser, a tertiary loading pump, a binaryeluting pump, a column oven, and an RS Autosampler. Filterextraction was performed on a horizontal shaker fromEberbach© (Ann Arbor, MI) and SPE using a negative pres-sure manifold. Evaporation under compressed air was com-pleted using a TurboVap LV® evaporator from Zymark(Hopkinton, MA).

Preparation of standard solutions

Methanolic working solutions at 0.5–20 μg/L were prepared,yielding 25, 50, 100, 250, 500, and 1,000 pg/filter calibratorconcentrations when blank filters were fortified with 50 μL ofworking solution. Quality control (QC) working solutionswere prepared from different ampoules, to produce 75, 375,and 750 pg/filter QC samples. An internal standard solutioncontaining 2 μg/L of cocaine-d3, ecgonine methyl ester-d3,benzoylecgonine-d8, and norcocaine-d3 was prepared inmethanol.

Specimen procedure

Sample preparation was modeled after a previously describedmethod [21] with modifications. Blank filters were fortifiedwith their respective calibrator and QC concentrations. Inauthentic specimens, the filter was removed from the collec-tion chamber. Calibrator, QC, and specimen filters were trans-ferred with sterile tweezers to 20-mL amber glass screw-topvials and fortified with 50 μL internal standard solution. Sixmilliliters of 1 % formic acid in water was added to each vialprior to shaking on a horizontal shaker at 260 oscillations/min.After shaking, vials were allowed to sit for approximately5 min before transferring 3 mL to a 10-mL plastic SARS-TEDT tube. An additional 3 mL of 1 % formic acid in waterwas added to the filters, the vials were shaken for 5 min, and3 mL was removed to plastic tubes. This procedure was

repeated a total of three times (9 mL total). Other solventswere evaluated during method development, including meth-anol and 0.2 M sodium acetate buffer; however, water with1 % formic acid yielded the best recoveries from the filter,with less interference/background noise.

Supernatants were loaded onto UCT Clean Screen car-tridges preconditioned with methanol (3 mL) and water(3 mL). Columns were washed with 2 % formic acid in water(3 mL) and 2 % formic acid in methanol (3 mL) before dryingcolumns under nitrogen using negative pressure at 10 psi for20 min. Elution was performed with 3 mL methanol/ammonium hydroxide (95:5, v/v). Eluates were evaporatedto dryness under compressed air at 40 °C. Samples werereconstituted in 200 μL mobile phase A (1 mM ammoniumformate with 0.01 % formic acid in water), vortexed briefly,and centrifuged at 4,000×g for 3 min.

Additionally, a methanolic extraction of devices (withoutfilters) was performed, in the same manner as Beck et al. [22].The breath device was reassembled following removal of thefilter and 5 mL methanol directly added to the device. Thedevice was shaken to ensure methanol extraction of anyremaining drug from the device. The methanol was elutedinto a plastic SARSTEDT tube, and 50 μL internal standard(2 μg/L) and 20 μL 10% formic acid in methanol were added.Eluates were evaporated to dryness under compressed air at40 °C, reconstituted in 200 μL mobile phase A, vortexedbriefly, and centrifuged at 4,000×g for 3 min.

Liquid chromatography-tandem mass spectrometry

Chromatographic separation was achieved at 30 °C with aPhenomenex Synergi Polar-RP 100×2 mm 2.5 μm columnand identically packed defender guard cartridges (10×20 mm2.5 μm). Gradient elution with 1 mM ammonium formate and0.01 % formic acid (pH ~3.24) (solvent A) and methanol(solvent B) at 0.3 mL/min flow rate was performed. Initialcomposition (5 % B) was maintained for 1.5 min, increasedfrom 5 to 95 % over 5.5 min, held for 1.5 min, returned toinitial conditions over 30 s, and held for 1.5 min for a total runtime of 10.5 min. Injection volume was 10 μL.

Targeted-MS/MS scans were acquired at 35,000 resolution(full width at half maximum at m/z 200) with a heatedelectrospray ionization source (HESI-II) and operated in pos-itive ionization mode. Automatic gain control (AGC) targetwas set at 5×105 and maximum injection time at 100 ms.Precursor ions were selected in the quadrupole with a 3 m/zwindow and subsequently fragmented in the higher-energycollision dissociation (HCD) cell. A full scan of fragmentedions was performed, and two specific product ions utilized fordata analysis with a mass tolerance of 5 ppm (Table 1). Thespray voltage was 3 kV, capillary temperature 350 °C, heatertemperature 425 °C, S-lens RF level 50, sheath gas flow rate50, auxiliary gas flow rate 13, and sweep gas 3

Quantification of cocaine and metabolites in exhaled breath 6215

(manufacturer’s units). Nitrogen was used for spray stabiliza-tion, for collision-induced dissociation experiments in theHCD cell, and as the damping gas in the C-trap. The instru-ment was calibrated in the positive and negative mode every25 h. Thermo TraceFinder Clinical Research 3.1 software wasused for data collection and processing.

The following criteria identified compounds: retention time(RT) within ±0.2 min of average calibrator RT, presence oftwo product ions, and ion ratio of the quantifying ion/qualifierion within ±20 % of the average of all calibrators.

Method validation

Method validation was performed on the breath filter extrac-tion procedure, as the methanolic extraction of the breathdevices (without the filters) was solely to determine oralfluid drug contamination of the device. Validation parame-ters included linearity, limits of detection (LOD), limits ofquantification (LOQ), bias and imprecision, ionizationsuppression/enhancement, extraction efficiency, process effi-ciency, interference studies, carryover, and autosampler andshort-term stability studies. Linearity (R2) was evaluated byleast squares regression with ≥6 nonzero calibrators on5 days. Acceptable linearity was achieved when R2≥0.99and calibrators quantified within ±20 %. LOD and LOQwere determined with decreasing concentrations of drug-fortified filters; LOD was the lowest concentration withacceptable chromatography, signal/noise ratio ≥3, withanalytes identified according to previously described criteria.LOQ was the lowest concentrations that met LOD criteriaand a signal/noise ratio of at least 10, and bias and impre-cision within ±20 %. LOD and LOQ were evaluated intriplicate on three different days (n=9).

Assay bias and imprecision were determined at three con-centrations (low, medium, and high QCs) in triplicate over5 days (n=15). Bias was evaluated for each concentration as100×[(group mean observed concentration−known concen-tration)/known concentration]. Acceptable bias was ±20 % oftarget. Precision was expressed as coefficient of variation(%CV) and determined by the one-way analysis of variation

(ANOVA) approach to calculate combined within- andbetween-run imprecision, as described in the SWGTOXmeth-od validation guidelines [25]. Acceptable precision was<20 %CV.

Ion suppression/enhancement, extraction efficiency, andprocess efficiency for each analyte were measured at lowand high QC concentrations as described by Matuszewskiet al. [26]. Ion suppression/enhancement was assessed bycomparing analyte peak areas of neat samples (n=6) (set 1)to peak areas of 10 different blank samples fortified withanalyte and internal standard after extraction (set 2). Extrac-tion efficiencywas examined by comparing analyte peak areasof five different samples fortified at low and high concentra-tions with internal standard before extraction (set 3), to peakareas of set 2. Process efficiency examined the overall effect ofextraction efficiency and ion suppression/enhancement onquantification of analytes.

Endogenous and exogenous interferences were evaluated.Interferences from endogenous matrix components were in-vestigated by analyzing breath samples from 10 individualswithout the addition of internal standard. Exogenous interfer-ences including opiates, stimulants, benzodiazepines, canna-binoids, and synthetic cathinones (see Electronic Supplemen-tary Material Table S1) were analyzed by fortifying neatsamples with interferences equivalent to 1,000 pg/filter. Inter-ferences were considered insignificant if analytes of interestwere <LOD. Lack of carryover was demonstrated by injectingtriplicate internal standard-fortified blank samples after a sam-ple fortified at 2,500 pg/filter. Carryover was considered neg-ligible if the measured concentration was <LOD.

Autosampler stability was investigated by reinjecting lowand high QC samples stored for 48 h at 4 °C on theautosampler (n=3) and calculating results against the originalcalibration curve. In addition, short-term stability was evalu-ated with breath filters fortified at low and high QC concen-trations and stored in 20-mL amber glass vials for 24 h at roomtemperature (n=3) and 72 h at 4 °C (n=3) and −20 °C (n=3),and after three freeze-thaw cycles (n=3). Internal standardwas added to each sample just prior to extraction and proc-essed as described. Stability was considered acceptable if QC

Table 1 Liquid chromatography-high-resolution mass spectrome-try parameters and retention times(RT) for cocaine,benzoylecgonine (BE), ecgoninemethyl ester (EME), andnorcocaine. Quantitative ions areitalicized

NCE normalized collision energy

Analyte Precursor ion (m/z) Product ions (m/z) NCE (%) RT (min)

EME 200.1 182.1174, 82.0656 60 1.6

EME-d3 203.1 185.1363, 85.0844 60 1.6

BE-d8 298.1 171.1205, 110.0651 50 6.4

BE 290.1 168.1017, 105.0338 45 6.4

Cocaine 304.1 182.1174, 82.0656 50 7.7

Cocaine-d3 307.2 185.1363, 85.0844 50 7.7

Norcocaine-d3 293.1 171.1205, 136.0756 20 7.9

Norcocaine 290.1 136.0756, 168.1017 45 7.9

6216 K.N. Ellefsen et al.

samples were quantified within ±20 % of freshly prepared QCsamples (n=3).

Results

Method performance

A linear range from 25 to 1,000 pg/filter with 1/xweighting was achieved for all four analytes. 1/xweighting was selected for this method as correlationcoefficients were acceptable, and the data demonstratedhomoscedasticity, or random distribution of the individualresiduals around the zero line, suggesting that this linearmodel was optimal in comparison to other models. Cali-bration curves from five separate days yielded determina-tion coefficients (R2) above 0.99±0.003 with residualswithin ±20 %. LOD was 15 pg/filter for all analytes,except EME (25 pg/filter), and LOQ was 25 pg/filter for

all compounds. Although EME demonstrated a signal-to-noise ratio greater than 3, the ion ratios and peak shapeswere not acceptable at 15 pg/filter; however, at the LOQ(25 pg/filter), the LOD acceptance criteria were achieved,in addition to the acceptance criteria for establishing theLOQ. Figure 1 illustrates a chromatogram of a breathsample fortified at the LOQ for all analytes.

Bias and imprecision results are presented in Table 2.All analytes were <10.6 % for within- and between-runimprecision (%CV), and bias ranged from −8.5 to 16.8 %.Extraction efficiencies were 83.6–125.5 %, and processefficiencies 52.7–78.8 % (Table 2). Ion suppression wasless than 25 %, except for EME and cocaine (highconcentration), which exhibited suppression up to46 %. Despite observed ion suppression, variationamong 10 different breath sources was <15 % for EME;however, %RSD for the high cocaine concentration was21 % (Table 2).

There were no endogenous interferences in 10 differentbreath samples. Addition of 1,000 pg/filter potentially

Fig. 1 Liquid chromatography-tandem mass spectrometry chromatogram of a breath sample fortified at the limits of quantification (25 pg/filter) forcocaine, benzoylecgonine (BE), ecgonine methyl ester (EME), and norcocaine

Quantification of cocaine and metabolites in exhaled breath 6217

interfering drugs and metabolites to neat samples did notproduce any interfering peaks meeting the identificationcriteria. No carryover was observed for BE and EME ininternal standard-fortified blank samples after injection ofa 2,500 pg/filter (2.5× upper LOQ (ULOQ)) sample. Forcocaine and norcocaine, carryover was observed at2,500 pg/filter but not at the ULOQ (1,000 pg/filter). Inaddition, as an additional precaution, two mobile phaseinjections were performed after the highest calibrator andQC.

Analytes were stable in the autosampler for 48 h at low andhigh QC concentrations (n=3), with percent differences be-tween −7.4 and −0.11%.When stored at room temperature for24 h, at 4 °C for 72 h, and at −20 °C for 72 h, all compoundswere stable (−19.3 to 15.0 %), except for cocaine at roomtemperature (−20.9 %). Cocaine and metabolites also werestable after three freeze-thaw cycles (−10.2 to 9.4 %). Data areshown in Table 3.

Breath cocaine and metabolites

Breath specimens from 10 participants aged 35–50 years(Table 4) were analyzed (A–J) (Table 5). Two participantswere medically discharged prior to the second cocainedose on day 5 and one prior to the third cocaine dose(day 10). Only 6/228 specimens from three participants

were positive for cocaine, with concentrations from 26.1to 66 pg/filter (median 34.0 pg/filter). No specimens werepositive for any cocaine marker in the first breath speci-mens at 10 and 30 min after IV cocaine administrations.The majority of positive cocaine specimens were collected1–2 h post-dosing; however, two specimens were positiveat 4 and 9.5 h. Figure 2a illustrates an LC-MS/MS chro-matogram of a positive cocaine breath specimen (partici-pant D). Only two specimens from one participant (J)were positive for BE and EME 6.5 and 12.5 h post-dosing with concentrations of 83.3 and 151 pg/filter and50 and 69.1 pg/filter, respectively. Norcocaine was notdetected in any specimen.

Based on the low prevalence of positive specimensobserved in breath, methanolic extraction of the devicewas performed after the breath filter was removed, utiliz-ing the same extraction technique as Beck et al. [22].Methanolic extraction of the devices yielded 32/182 pos-itive cocaine tests (10 min to 22 h post-dosing), withsemi-quantitative concentrations from 25.2 to 364 pg/device (median 53.4 pg/device). Unlike for the filters,BE was detected in 10/230 devices, with semi-quantitative concentrations between 26.4 and 93.7 pg/de-vice (median 39 pg/device). In 9 of 10 BE-positive de-vices, cocaine also was present at higher concentrations.Figure 2b illustrates the LC-MS/MS chromatogram of a

Table 2 Within-run imprecision, between-run imprecision, and %biasfor cocaine, benzoylecgonine (BE), ecgonine methyl ester (EME), andnorcocaine in breath at low (75 pg/filter), medium (375 pg/filter), and

high (750 pg/filter) concentrations and efficiency, process efficiency, andion suppression/enhancement for cocaine, BE, EME, and norcocaine inbreath at low and high concentrations

Analyte Within-runimprecision(n=15, %CV)

Between-runimprecision(n=15, %CV)

%Bias(n=15)

Extraction efficiency(%) (n=5)

Process efficiency(%) (n=5)

Ion suppression/enhancement(%) (n=10) (%RSD)

Low Med High Low Med High Low Med High Low High Low High Low High

Cocaine 8.3 7.7 9.9 7.9 10.6 10.3 −7.4 −8.5 −2.9 89.0 90.0 69.5 63.2 −21.9 (9.6) −29.8 (21)BE 3.1 4.7 6.4 2.9 5.8 7.7 16.8 −0.02 3.8 80.4 82.5 65.0 69.9 −19.1 (3.4) −15.3 (5.8)EME 4.6 8.0 6.5 7.6 8.7 9.1 −3.0 −0.03 −0.40 125.5 97.8 78.8 52.7 −37.2 (5.5) −46.1 (13)Norcocaine 6.7 7.0 7.6 6.9 8.8 9.3 1.5 −3.8 −1.2 83.6 85.7 63.2 64.3 −24.3 (9.3) −25.0 (19)

Table 3 Short-term stability data(%difference) of cocaine,benzoylecgonine (BE), ecgoninemethyl ester (EME), andnorcocaine in breath at low(75 pg/filter) and high (750 pg/filter) concentrations after storageat 4 °C for 48 h in theautosampler, at room temperaturefor 24 h, at 4 °C for 72 h, at−20 °C for 72 h, and after threefreeze-thaw cycles (over 72 h)

Stability conditions Cocaine BE EME Norcocaine

Low High Low High Low High Low High

48 h 4 °C autosampler (n=3) −3.4 −4.3 −0.11 −7.1 −0.90 −7.4 −2.8 −0.2324 h room temperature (n=3) −19.3 −20.9 −4.3 −17.1 −12.1 −13.5 −14.2 −13.572 h 4 °C (n=3) −6.1 −6.7 15 −0.67 −10.4 −4.2 0.25 −0.2772 h −20 °C (n=3) −8.5 −12.2 −0.10 −4.1 −15.4 −8.0 −1.8 −3.23 freeze-thaw cycles (n=3) 4.8 −8.2 4.2 −0.92 −6.5 −10.2 0.12 9.4

6218 K.N. Ellefsen et al.

device positive for both cocaine and BE (participant D).EME and norcocaine were not detected in any device.

Discussion

A fully validated analytical method for the quantification ofcocaine, BE, EME, and norcocaine in exhaled breath wasdeveloped. There are no previous cocaine breath quantifica-tion methods including EME and norcocaine. Two previousstudies reported validation results for cocaine and BE only[20, 22]. LOQs for cocaine and BE in these investigationswere 0.6 and 0.1 pg on column [20] and 6 pg/filter [22],estimated from diluted calibrator extracts with criteria ofsignal-to-noise ≥10 and concentrations within ±20% of target.In the present method, with an empirical determination oflower and lower analyte concentrations required to meet iden-tification and quantitative criteria, the LOQs for all fouranalytes were 25 pg/filter. Based on concentrations reportedby Beck et al., 25 pg/filter should be adequate to detectcocaine and BE in authentic cases [22]; however, in previousstudies, the extraction was of the entire device and not just thefilter. This sensitivity following our controlled cocaine admin-istration study identified cocaine in breath in only a fewsubjects’ samples. Additional research is necessary to studycorrelations between cocaine dose, route of administration,and breath concentrations.

Sample preparation for this method required two steps:first, extraction of analytes from the breath filter, followedby solid phase extraction. Filters were removed from theSensAbues breath collection devices prior to extracting theanalytes with water and 1 % formic acid. Following extractionfrom the filters, a simple solid phase extraction procedure was

performed. The combination of analyte extraction from thefilters and solid phase extraction reduced matrix effects com-pared to filter extraction alone. Recently, published breathmethods extracted analytes from the filter by directly addingmethanol to the intact SensAbues breath collection devices,with no subsequent solid phase extraction [22, 23]. This likelyresulted in higher analyte concentrations due to contributionfrom the drug in oral fluid and also greater ion suppression.

The QExactive mass spectrometer was utilized for thismethod for its high resolution and high mass accuracy forthe two product ions employed for identification. This in-creased specificity and sensitivity by improving the signal-to-noise ratio. Typical breath cocaine concentrations are in thepicogram per filter range, which is why increased sensitivitywas an important feature of this method, enabling cocaine andmetabolite quantification in exhaled breath. No endogenous orexogenous interferences of structurally similar compoundswere observed. However, BE and norcocaine have identicalmolecular weights and common fragmentation. As such, high-resolution mass spectrometric identification was not sufficientto differentiate these two compounds; chromatographic sepa-ration was necessary for accurate identification.

Validation results for cocaine and its metabolites wereacceptable. Extraction efficiencies were 80–126 % for all fouranalytes, and ion suppression was less than 25 % except EME(−37 and −46 %) and the high cocaine concentration (−30 %).Although ion suppression was high for EME, its %RSD wasacceptable (<15 %), demonstrating little variation among the10 different matrix sources. Despite the high %RSD (21 %)for the high cocaine concentration, cocaine QC concentrationswere quantified within acceptable %bias and precision. Addi-tionally, inclusion of matched deuterated internal standardscompensated for these observed matrix effects. Previous ana-lytical breath cocaine methods did not quantify matrix effects

Table 4 Demographic and self-reported cocaine histories for 10cocaine users

Ppt participant, SM smoked, IVintravenous, AA African-Ameri-can, W White, M male

Ppt Race Sex Age (years) Weight (lbs) Days used inpast 14(SM or IV)

Mean use(SM or IV)

Age first use(SM or IV)(years)

A AA M 44.0 176 4 SM 2–3/week 38

B AA M 38.1 202 11 SM 5/week 32

C More than 1 M 42.4 162 5 IV 3/week 32

0 SM 3–4/month 21

D W M 35.2 165 3 SM 1/week 25

E AA M 45.3 206 0 IV N/A 34

5 SM 4/week 18

F W M 44.0 197 5 SM Daily 18

G W M 42.5 180 6 IV 3/week 28

7 SM 2/week 31

H AA M 43.7 150 14 SM Daily 21

I AA M 49.9 175 10 SM 4–5/week 25

J AA M 46.5 176 1 SM 3/week 15

Quantification of cocaine and metabolites in exhaled breath 6219

in breath [20, 22, 23]; however, THC and its metabolitesproduced breath matrix effects ranging from 12 to −35 %[21], similar to what was observed in this study. Processefficiencies were low (53–79 %) for all four analytes; thiscould be explained by loss of analyte during the extractionprocedure and the ion suppression observed in this method.Despite these lower process efficiencies and ion suppression, asensitive LOQ of 25 pg/filter was achieved.

To minimize possible carryover, additional mobile phaseswere analyzed after the highest calibrator (1,000 pg/filter) andhigh QC (750 pg/filter). However, in most reported cases,

cocaine breath concentrations are below 80 pg/device [20,22, 23]; therefore, cocaine and norcocaine carryover shouldnot be an issue. Additionally, breath specimens were stableunder all tested conditions, with the exception of cocaine after24 h at room temperature. Despite this observation, this wasnot problematic as collected breath specimens were stored at−20 °C within 2 hours of collection. Based on this investiga-tion, breath specimens should be stored refrigerated or frozenimmediately following collection in order to limit analyte loss.

Exhaled breath is commonly utilized and accepted for thedetection of alcohol. Its ease and accessibility in collection

Table 5 Breath cocaine, benzoylecgonine (BE), and ecgoninemethyl ester (EME) concentrations (pg/filter) in filters and cocaine and BE concentrations(pg/device) in breath collection devices (without the filters) from individuals following administration of 25 mg IV cocaine

b,e

−

− − − − − − − −−

− − − − − − −

−

−

−

−

− −

− −

− −

− − − − −

− − − −

− −

−

−

−

−

−

−

− −

− −

− −

−

−

− −

− − − − −

− −

−

− − − − − − −

− −

− −

− − − −

− − − −

−

− −

−

− − −

−

−

− − −

− −

− −

− −

− −

− −

−

−

− − −

−

−

− −

− − −

− − −

− − −

− − −

− − −

− − − − −

− −

−

−

−

−−

−

− − − − − − − −

− − − − − − − −

− − − − − −

− − − − − − −

− −

− −

− −

– <LOQaConcentrations from all breath filters collected an hour prior to IV cocaine administration were less than the limit of quantitation (LOQ)bOnly results from participants with positive results are presented; specimens from 10 participants were analyzedc All specimens were collected within ±15 min of scheduled collection timesd Breath specimens collected at 21 h on days 1 and 10 and at 22 h on day 5e Subscript 1 represents participant results from day 1, subscript 2 represents participant results from day 5, and subscript 3 represents participant resultsfrom day 10

6220 K.N. Ellefsen et al.

make it a more desirable matrix for roadside collection thanblood. Other matrices, like oral fluid, offer the same ease incollection as breath; however, dry mouth, especially afterstimulant abuse, can produce low-volume oral fluid collec-tions. Exhaled breath contains microparticles that carry non-volatile substances out of the lungs including components ofthe airway lining fluid and proteins of the surfactant phase [27,28]. Recent studies hypothesized that exogenous substances,including drugs of abuse, can contaminate the airway liningand, therefore, become detectable in exhaled breath [14–23].

Previous studies demonstrated that cocaine is detected inexhaled breath; however, this study is the first to investigatecocaine in exhaled breath after controlled cocaine administra-tion. We found few specimens (2.6 %) positive for cocaine inbreath following 25 mg IV cocaine, with even fewer (0.7 %)containing BE and EME. Cocaine was typically observed inbreath for up to 2 h, with some up to 9.5 h. BE and EMEwereobserved 6.5 and 12.5 h post-dosing, with concentrationsranging from 83.3 to 151 pg/filter and 50 to 69.1 pg/filter,respectively. This is the first report of EME in exhaled breath.Cocaine concentrations in our study ranged from 26.1–66 pg/filter, much lower than those previously reported. A probableexplanation for this difference could be that prior studies

extracted analytes from the filter by directly adding methanolto the intact SensAbues breath collection device, collectingthe drug in oral fluid on the device rather than solely that onthe filter [22, 23]. Our study removed the filters from thedevices prior to extraction, thereby, reducing the possibilityof contamination from the device itself.

To further investigate this hypothesis, we performed meth-anolic extractions in the same manner as Beck et al. [22] onthe breath collection devices, after the filters were removed,and identified the presence of cocaine and its metabolites inthe devices themselves. Cocaine was detected in the devicesfor a longer duration (up to 22 h), more frequently (19.2 %), athigher concentrations (25.2–363 pg/device), and with a me-dian concentration (53.4 pg/device) that more closely resem-bled those of previous studies [22]. BE also was detected in4.3 % of devices with concentrations from 26.4 to 93.7 pg/device. The increased presence of cocaine and BE in devicesas compared to filters suggests that the devices reflect oralfluid contamination as well as lung microparticles, while thefilters, which are designed to protect against oral fluid con-tamination, reflect only drug-laden microparticles.

The median concentrations found in the devices afterperforming a methanolic extraction without the filter were

Fig. 2 Liquid chromatography-tandem mass spectrometry chromatogram of a positive cocaine breath specimen (participant D) with 33.2 pg cocaine/filter and b cocaine 177.8 pg/device and benzoylecgonine (BE) 50.7 pg/device, 9.5 h after 25 mg IV cocaine

Quantification of cocaine and metabolites in exhaled breath 6221

comparable to what were observed in previous studies (co-caine median 60 pg/device; BE median 43 pg/device) [22].The higher prevalence and concentrations observed in thedevices as compared to the filters (Table 5) highlight the factthat the extraction procedure is critical in determining concen-trations of cocaine and its metabolites in exhaled breath andnot contaminated oral fluid.

Although the detection rates in the breath filters were low,this study quantified cocaine and its metabolites (BE andEME) in breath following controlled IV cocaine administra-tion. The route of administration and low dose (25 mg) im-pacted the presented results. Smoked or intranasal cocaineadministration could yield higher results in exhaled breaththan after IV cocaine administration, and methods of cocaineextraction also affect measureable cocaine markers. Addition-ally, a higher dosage could also increase the detection rates.Previous studies examining cocaine in exhaled breath werenot controlled; thus, route of administration and dosage infor-mation were unknown [20, 22, 23], making direct compari-sons with this current study difficult. Additional controlledadministration studies examining different routes of adminis-tration and dosages utilizing the presented extraction tech-nique are needed to determine if exhaled breath can be utilizedas a reliable alternative sampling matrix for identifying recentcocaine intake.

Conclusion

We developed and validated a sensitive and specific methodfor cocaine, BE, EME, and norcocaine quantification in ex-haled breath. Only 2.6 % of filters were positive for cocainefrom 1 to 9.5 h; BE and EME were detected in two specimens(0.72 %) 6.5 and 12.5 h post-dosing, and no specimens werepositive for norcocaine, following controlled administrationsof 25mg IV cocaine. This was the first report of EME detectedin exhaled breath. When extracting the device itself, withoutthe filters, 19.2 % were positive for cocaine and 4.3 % for BEup to 22 h after dosing, suggesting that the device reflects oralfluid as well as lung microparticles, while the filter, based onthe design of the device, reflects only drug-contaminatedmicroparticles. Cocaine in breath after controlled IVadminis-tration identifies recent cocaine ingestion, but its absence doesnot preclude recent use.

Acknowledgments This work was supported by the Intramural Re-search Program (IRP), National Institute on Drug Abuse (NIDA), Na-tional Institutes of Health (NIH). The authors acknowledge the contribu-tions of the clinical staff of the NIDA-IRP and the Clinical Research Unit,Johns Hopkins Bayview Medical Center, as well as the University ofMaryland, Baltimore, amember of the Graduate Partnership Program, NIH.

Conflict of interest O. Beck has part ownership in SensAbues produc-ing the sampling devices (O. Beck, US patent number 13/394,609).

References

1. U.S. Drug Encorcement Administration (DEA) (2013) NationalForensic Laboratory Information System: 2012 annual report.Springfield, Virginia

2. European Monitoring Centre for Drugs and Drug Addiction(EMCDDA) (2012) Annual report 2012: the state of the drugsproblem in Europe. Lisbon, Portugal

3. Substance Abuse and Mental Health Services Administration(SAMHSA) (2013) Results from the 2012 National Survey on DrugUse and Health: summary of national findings. Rockville, Maryland

4. Jatlow P (1988) Cocaine: analysis, pharmacokinetics, and metabolicdisposition. Yale J Biol Med 61:105–113

5. Misra AL, Nayak PK, Bloch R, Mule SJ (1975) Estimationand disposition of [3H]benzoylecgonine and pharmacologicalactivity of some cocaine metabolites. J Pharm Pharmacol27(10):784–786

6. Hawks RL, Kopin IJ, Colburn RW, Thoa NB (1974) Norcocaine: apharmacologically active metabolite of cocaine found in brain. LifeSci 15:2189–2195

7. Gallardo E, Queiroz JA (2008) The role of alternative specimens intoxicological analysis. Biomed Chromatogr 22(8):795–821

8. Frederick DL (2012) Toxicology testing in alternative specimenmatrices. Clin Lab Med 32(3):467–492

9. Anizan S, Huestis MA (2014) The potential role of oral fluid inantidoping testing. Clin Chem 60(2):307–322

10. Drummer OH (2008) Introduction and review of collection tech-niques and applications of drug testing of oral fluid. Ther DrugMonit 30(2):203–206

11. Drummer OH (2006) Drug testing in oral fluid. Clin Biochem Rev27(3):147–159

12. BoskerWM, HuestisMA (2009)Oral fluid testing for drugs of abuse.Clin Chem 55(11):1910–1931

13. Scheidweiler KB, Kolbrich Spargo EA, Kelly TL, Cone EJ, BarnesAJ, Huestis MA (2010) Pharmacokinetics of cocaine and metabolitesin human oral fluid and correlation with plasma concentrations aftercontrolled administration. Ther Drug Monit 32(5):628–637

14. Beck O, Leine K, Palmskog G, Franck J (2010) Amphetaminesdetected in exhaled breath from drug addicts: a new possible methodfor drugs-of-abuse testing. J Anal Toxicol 34(5):233–237

15. Beck O, Sandqvist S, Eriksen P, Franck J, Palmskog G (2010)Method for determination of methadone in exhaled breath collectedfrom subjects undergoing methadone maintenance treatment. JChromatogr B 878(24):2255–2259

16. Beck O, Sandqvist S, Eriksen P, Franck J, Palmskog G (2011)Determination of methadone in exhaled breath condensate by liquidchromatography-tandem mass spectrometry. J Anal Toxicol 35(3):129–133

17. Beck O, Sandqvist S, Bottcher M, Eriksen P, Franck J, Palmskog G(2011) Study on the sampling of methadone from exhaled breath. JAnal Toxicol 35(5):257–263

18. Beck O, Sandqvist S, Franck J (2011) Demonstration that methadoneis being present in the exhaled breath aerosol fraction. J PharmBiomed Anal 56(5):1024–1028

19. Beck O, Sandqvist S, Dubbelboer I, Franck J (2011) Detection ofdelta9-tetrahydrocannabinol in exhaled breath collected from canna-bis users. J Anal Toxicol 35(8):541–544

20. Beck O, Stephanson N, Sandqvist S, Franck J (2012) Detection ofdrugs of abuse in exhaled breath from users following recovery fromintoxication. J Anal Toxicol 36(9):638–646

21. Himes SK, Scheidweiler KB, Beck O, Gorelick DA, Desrosiers NA,HuestisMA (2013) Cannabinoids in exhaled breath following controlledadministration of smoked cannabis. Clin Chem 59(12):1780–1789

22. Beck O, Stephanson N, Sandqvist S, Franck J (2013) Detection ofdrugs of abuse in exhaled breath using a device for rapid collection:

6222 K.N. Ellefsen et al.

comparison with plasma, urine and self-reporting in 47 drug users. JBreath Res 7(2):1–11

23. Beck O (2014) Exhaled breath for drugs of abuse testing—evaluationin criminal justice settings. Sci Justice 54(1):57–60

24. Almstrand AC, Bake B, Ljungstrom E, Larsson P, Bredberg A,Mirgorodskaya E, Olin AC (2010) Effect of airway opening onproduction of exhaled particles. J Appl Physiol 108(3):584–588

25. Scientific Working Group for Forensic Toxicology (SWGTOX)(2013) SWGTOX standard practices for method validation in foren-sic toxicology. J Anal Toxicol 37(7):452–474

26. Matuszewski BK, Constanzer ML, Chavez-Eng CM (2003)Strategies for the assessment of matrix effect in quantitativebioanalytical methods based on HPLC-MS/MS. Anal Chem75(13):3019–3030

27. Papineni RS, Rosenthal FS (1997) The size distribution of droplets inthe exhaled breath of healthy human subjects. J Aerosol Med 10(2):105–116

28. ;Bredberg A, Gobom J, Almstrand AC, Larsson P, Blennow K, OlinAC, Mirgorodskaya E (2012) Exhaled endogenous particles containlung proteins. Clin Chem 58(2):431–440

Quantification of cocaine and metabolites in exhaled breath 6223