CLIN.CHEM.

38/5, 71 7-719 (1992)

Marijuana Testing in Urine: Use of a Hexadeuterated Internal Standard for Extended Linearity, and Ion Trap vs Mass Selective Detector Gas Chromatograph/Mass Spectrometer Systems
W. A Joern The use of a hexadeuterated internal standard for the assay of the marijuana carboxy metabolite in urine resulted in two significant improvements. First, the linearity of the procedure was increased considerably because of the minimal chromatographic peak overlap of the internal standard and carboxy metabolitederivatives.Second, again because of minimal peak overlap, the same extract could be analyzed with similar results by both the ion trap detector and mass selective detector gas chromatograph/ mass spectrometer ,systems. AddftlonalKeyphrases:drug assay
abused drugs intermethod comparison

nor-9-carboxy-8-tetrahydrocannabinol-2H6 (D6-TA), is commercially available (4). Here, we used D6-TA in place of D3-TA in the procedure of Paul et al. (2). Materials Reagents TA was from the Research Triangle Institute (Research Triangle, NC). Working standard solutions were prepared by appropriate dilutions with ethanol. 1)6-TA was from ElSohly Labs (Meridian, MS). The internal standard solution was D6-TA, 4.00 mg/L in ethanol. Tetramethylammonium hydroxide (250 g/L in methanol) and iodomethane were from Aldrich Chemical Co. (Milwaukee, Wi). All other chemicals were ACS grade. Solvents were grade from Burdick & Jackson (Muskegon, Ml). The urine matrix used for standards and controls was prepared by adding 5 mL of 10 mol/L sodium hydroxide solution to 500 mL of drug-free urine, stirring the mixture for 5 mm, centrifuging for 15 mm at 1000 x g, and separating the supernate from the precipitate. We mixed appropriate volumes of stock standards, controls, and treated urine to prepare the urine standards and controls, which were stored at -80 #{176}C use. Polyethylene caps until (Tamer Tops; Fisher Chemical Co., St. Louis, MO) facilitated the use of disposable12 x 75mm and 16 x 100 mm borosilicate test tubes for all steps of the procedure. Apparatus The MSD CC/MS system was a Hewlett-Packard Co. (Palo Alto, CA) Model 5970A MSD/5790B GC. The capillary column was a J&W Scientific (Folsom, CA) no. DB-5 (30 m, 0.25 mm i.d., and 0.25 urn coating) coupled to the MSD through an open-split interface. Helium flows were 0.9 mL/min (32 cm/s linear velocity) through the column and 0.5 mLlmin through the interface. We programmed the temperature to increase from 180 to 270 #{176}C #{176}C/min, a 7-ruin hold, and the injector at 15 with and detector temperatures were 260 and 280 #{176}C, respectively. We used MSD in the electron-impact multipleion-monitoring mode, programmed to detect the ion peaks (m/z) at 313.3, 357.4, and 372.4 for TA and 322.3 and 378.4 for D6-TA. We set the electron multiplier at 400 V above autotune (with perfluorotributylamine). The peak window was 0.30, dwell time was 50 ms, and the wide peak setting was used. We turned on the filament 0.5 mm after injecting the sample. The lTD system was a Perkin-Elmer Corp. (Norwalk,
CLINICAL CHEMISTRY, Vol. 38, No. 5, 1992 717

To detect recurrent marijuana use in substance-abuse patients, one must accurately determine the concentration of the marijuana carboxy metabolite, 11-nor-9carboxy-#{176}-tetrahydrocannabino1 (TA) in urine (1). on admission, many patients have very high TA concentrations; with recurrent marijuana use, TA concentration can be very high. Therefore, to eliminate the need for specimen dilution and reanalysis, it is advantageous to have an assay method with a wide linearity range. The procedure of Paul et al. (2) produces very stable methyl derivatives of TA and its trideuterated analog, 11-nor-9-carboxy-#{176}-tetrahydrocannabinol-2H3 (D3-TA), and works quite well with the mass selective detector (MSD) gas chromatograph/mass spectrometer (GCIMS) system. However, because the most abundant ion of D3-TA (mlz 316) is also found in TA and because TA coelutes with D3-TA, the linearity range is limited. The data can be recalculated to account for the peak overlap, but most instrument data-reduction programs do not allow this. Attempts in this laboratory to adapt this procedure to the ion trap detector (IT!)) GC/MS system resulted in a very severely limited linearity range, again because of the coelution of TA and D3-TA. McCurdy et al. (3) studied the determination of TA with the IT!), but they did not use an internal standard. A hexadeuterated marijuana internal standard, 11Laboratory, St. Anthonys Medical Center, St. Louis, MO 63128. Nonstandard abbreviations: TA, 11-nor-9-carboxy-A9-tetrahydrocannabinol; D3-TA, 11-nor-9-carboxy-t9-tetrahydrocannabinol-2H3; D6-TA, 11-nor-9-carboxy-i5-tetrahydrocannabinol-2H6; MSD, mass selective detector, GC/MS, gas chromatograph/mass spectrometer, and ITD, ion trap detector. Received October 30, 1991; accepted February 26, 1992.

CT) Model 700 ITD/8500 CC, with a software upgrade the lTD equivalent to a Model 800. The column and flow rates were the, same as above except that a direct interface was used. We programmed the temperature to increase from 180 to 270 #{176}C #{176}C/min, at 30 with an 8-ruin hold. The injector and detector temperatures were 260 and 270 #{176}C, respectively. We used the lTD in the electron-impact mode, monitoring the scan range, m/z 305-385, and operated the electron multiplier at the autotune (with perfluorotributylamine) voltage. We turned the filament on at 8 nun and off at 12 ruin. We set the sensitivity (B value) at 15000 (autotune sensitivity usually is in the 3000-5000 range), the mass defect at 100, and the background mass at 48. Postrun data processing included integration of the ion peaks at m/z 313, 357, and 372 for TA and 322 and 378 for D6-TA.
making

S...

1..

4*1

72*

I.4I

4*1

3.5

#{149} S41

Cr42

1551

153143

5rU

irLl

rh

ll

1141

Ir

411

721

*...ISI

Ill

44242

52

41535

A
t?&
411
II?1

11?t1

Procedure The derivatization procedure was essentially that of Paul et al. (2). To a disposable 16 x 100 mm glass tube add 1.0 mL of 3.0 mol/L potassium hydroxide. Add 3.0 mL of urine and 25 MLof the internal standard solution. Mix, and incubate at 45#{176}C 15 mm. Cool at 4#{176}C5 nun. for for Add 3.5 mL of ethyl acetate:hexane mixture (1:19 by vol). Add 2.0 mL of 3.0 mol/L hydrochloric acid. Cap the tubes, and rock for 15 mm. Centrifuge for 10 mm at 1000 x g. Transfer most of the top layer to a 12 x 75 mm tube, and evaporate at 45#{176}C. a separate tube, add 50 iL of In tetrainethylammonium hydroxide to 1.0 mL of dimethyl sulfoxide, mix, and add 100 pL of the resulting solution to the (evaporated) extract in the 12 x 75 mm tube. Vortexmix for lOs. Add 20 zL of iodomethane (in the hood!), and vortex-mix for 30s. Add 500 pL of 0.10 mol/L hydrochloric acid solution and 2.5 mL of hexane. Cap the tube and vortex-mix for 30s. Let stand for 60s, and then transfer most of the top layer to another 12 x 75 mm tube. Evaporate at 45#{176}C. Reconstitute with 25 1cL of n-butyl acetate for MSD analysis or with 10 zL for IT!) analysis. Inject 1.0 pL into the appropriate CC/MS system. (Figure 1 shows the lTD data from a 20 pgfL urine standard.) Calculations We analyzed the data from five replicate injections f o an unextracted standard by using all possible combinations of the integrated ion areas. We obtained greater precision, especially for the ITD, if the sums of two or three ion areas were used for quantitation rather than the area of only the most abundant ion. Accordingly, we used the following formulas to calculate the concentrations (x = unknown, std = standard, A = area): For MSD, concn concn A313 + A7
322 +

l#

Ill

$41

Ill

It

41

leSt

3*542

-,

A
. iti#{149}

I
t1ti
1t#{149}

O3*t

Fig. 1. Total ion chromatogram and lndMdualIon chromatograms (ITD)from Injectionof a 20 ig/L urine standard extract
Scan range mlz 305-385

were calculated as A3SA3 and A372/A313 for TA.

Results Test Characteristics

for D6-TA and as A357/A313

A378

For ITD, concn

concn-

A313 + A7 + A372 A322 + A378

The ion ratios for the data from both CC/MS systems
718

Linearity. We assayed urine standards at concentrations of 0, 50, 200, 500, 2000, and 5000 pg/L with both the MSD and IT!) CC/MS systems. Acceptable linearity was defined as a set of concentration values with a linear-regression correlation coefficient (r) of 0.990 and ion-ratio values within 20% (MSD) or 30% (IT!)) of that of the 50 p.g/L standard. For the MSD, the correlation for the set of values from 0 to 2000 pg/L was acceptable (r = 0.9996). When the 5000 pg/L standard was diluted fivefold with n-butyl acetate and injected and these data were used in place of the undiluted data, the set of values from 0 to 5000 g/L was acceptable (r = 0.9999). For the IT!), the set of values from 0 to 500 pg/L was acceptable (r = 0.9998). When the 2000 pg/L standard was diluted 10-fold with n-butyl acetate and injected and these data were used in place of the undiluted data, the set of values from 0 to 2000 g/L was acceptable (r = 0.9999). (When the 5000 ig/L standard was diluted 10-fold for IT!) and injected, the internal standard peaks gave unacceptable signal/noise ratios.) Reproducibility. We assayed five urine controls at a concentration of 25 ugfL with the MSD CC/MS system. The coefficient of variation (CV) of the area ratios was 3.0%. The ion-ratio ranges were 73-74% for m/z 322/ 378, 72-74% for m/z 357/313, and 45-46% for miz 372/313. Nine urine controls at a concentration of 20 ig/L were assayed with the ITI) CC/MS system. The CV of the area ratios was 4.2%. The ion-ratio ranges were

CLINICAL CHEMISTRY, Vol. 38, No. 5, 1992

72-88% for m/z 322/378, 70-80% for mlz 357/313, and 37-41% for m/z 372/313. We assayed nine sets of urine controls and standards at concentrations of 20 and 50 ig/L, respectively, with both the MSD and IT!) CC/MS systems over a 20-day period. The CVs determined were 3.7% for MSD and 7.0% for lTD. The average differences of the ion-ratio values between the controls and the standards were 1.3% for MSD and 6.9% for lTD for m/z 322/378,2.3% for MSD and 5.3% for lTD for mlz 357/313, and 3.3% for MSD and 8.5% for IT!) for m/z 372/313. Sensitivity. We assayed five negative urine specimens of various creatinine concentrations (120-1860 g/L) by both the MSD and the IT!) CC/MS systems. The apparent concentration was calculated by using the area of the highest peaks near the expected retention time of the TA peaks. The sensitivity (detection limit) was calculated in two ways: the 3 x baseline method (5) and the limit of detection/limit of quantitation (LOD/ LOQ) method (6). For 3 x baseline, the sensitivity was 0.3 g/L for MSD and 1.4 g/L for ITD; for LOD (mean + 3 SDs), it was 0.5 g/L for MSD and 0.3 g/L for IT!); and for LOQ (mean + 10 SDs), it was 1.3 gfL for MSD and 1.2 g/L for lTD. A series-dilution sensitivity determination also was carried out with urine standards 1,2,3,4,5,6,7, and 25 g/L, assayed with both the MSD and IT!) CC/MS systems. The limit of quantitation by this procedure was defined as the lowest standard for which the ion ratios were all within 20% (MSD) or 30% (ITD) of the 25 ,ug/L standard and for which the signal/noise ratios were 5.O. Under these conditions, the limit of quantitation of the MSD was 2 tg/L and that of the lTD was 4 .tgfL (for the 3 g/L standard the ion ratios were acceptable, but one of the signal/noise ratios was 3.2). Cariyover. We injected an extractof a negative urine control immediately after injecting a 5000 gfL standard and then calculatedthe apparent concentration (in tg/L) of the negative control. Percent carryover, defined as (apparent concentration/5000)#{149} 00, was 0.09% for 1 the MSD and 0.06% for the lTD. Absolute recovery. We compared the MSD area ratios for a 25 ig/L urine standard with those of an unextracted standard equivalent to a 25 ugfL urine standard. The absolute recovery (assuming 100% completion of the derivatization step) of TA was 80% and that of D6-TA was 75%. Urine Specimens from Patients Urine specimens (n = 26) positive for marijuana were assayed with both the MSD and IT!) CC/MS systems over a 15-day period. For the correlation of MSD (x) to lTD (y), slope = 0.918, intercept = 1.8 pg/L, and r = 0.918. The average differences of the ion-ratio values of the controls from those of the standards were 2.1% for MSD and 5.7% for lTD for mlz 322/378, 3.2% for MSD and 12.6% for IT!) for miz 357/313, and 5.4% for MSD and 16.2% for lTD for m/z 372/313. For the MSD, the ion ratios of all but one specimen were within 20% of those of their respective standards. The ion ratios for the lTD

were less precise; however, the ion ratios for all but two specimens were within 30% of their respective standard ion ratios. The TA concentration of one of the outlier specimens was quite low (7 g/L), and the ion ratios of this specimen were out of the acceptable range for both CC/MS systems. Upon reassay in duplicate, one of the duplicates met the ion-ratio criteria for both CC/MS systems, whereas the other failed on both systems. Apparently there was spurious interference with this urine specimen. The other specimen that was outside the ion-ratio limits on the IT!) was reassayed, and the repeat ion ratio was acceptable.
DiscussIon

The retention times of 1)6-TA and TA differ by about 0.2

mmn. The use of !)6-TA as the internal standard definitely increases the linearity of this procedure and should facil-

itate the use of TA concentration data in determining recurrent marijuana use. This study and the work of McCurdy et al. (3) show that the ion ratios obtained from an IT!) CC/MS system are not as precise as those from an MSD system. However, the data obtained here indicate that the IT!) is an acceptable CC/MS system for marijuana analysis as long as the criterion for ion-ratio acceptability is set at 30% (there is nothing magic about the 20% criterion; it is just one characteristic of one type of CC/MS instrumentation). The signal/noise ratio of the lTD peaks are very good, and the test characteristics indicate that the lTD is capable of accurate quantitative determinations. In fact, the newer ion-trap instrumentation (e.g., Model ITS4O) apparently demonstrates much improved ion-ratio precision and full-scan sensitivity over those of Models 700 and 800. Although the IT!) can process data automatically, occasionally the integration program will split a peak, requiring the peak to be integrated manually. Also, it was discoveredquite late in this study that sensitivity settings(B values) of 25000-50000 produce greater signal/noise ratios than does 15 000, so it is possible that the limit of quantitation on the IT!) is lowered somewhat by the use of a higher B value.
We gratefully

acknowledge the toxicology technologists of St.

Anthonys Medical Center, who prepared and irected most of the

urine extracts.
References 1. Joern WA. Detection of past and recurrent marijuana Detection and quantitation of urinary

use by a modified GC/MS procedure. J Anal Toxicol 1987;1l:49-52. 2. Paul BD, Mell LI) Jr, Mitehell JM, McKinley RM, Irving J.
11-nor-delta-9-tetrahydro-

cannabinol-9-carboxylic acid, a met.abolite of tetrahydrocannabinol, by capillary gas chromatography and electron impact mass fragmentography. J Anal Toxicol 1987;11:1-5. 3. McCurdy HH, Lewellen U, Callahan LS, Childs PS. Evaluation of the ion trap detector for the detection of 1l-nor-delta-9THC-9-carboxylic acid in urine after extraction by bonded-phase adsorption. J Anal Toxicol 1986;10:175-7. 4. ElSohly MA, Stanford DF, Little TL Jr. 2H6-1l-Nor-delta-8THC-9-COOH: a new internal standard for the analysis of THC metabolite in biological fluids. J Anal Toxicol 1988;12:54-5. 5. Long GL, Winefordner JD. Limits of detection. Anal Chem
1983;55:712A-9A. 6. Needleman for some drugs

SB, Romberg RW. Limits of linearity and detection of abuse. J Anal Toxicol 1990;14:34-8. CLINICAL HEMISTRY, C Vol.38, No. 5, 1992 719