Differentiating New Marijuana Use From Residual Drug Excretion in

Journal of Analytical Toxicology, Vol. 22, October 1998

Differentiating New Marijuana Use From Residual Drug Excretion in Occasional Marijuana Users*

Marilyn A. Huestis t and Edward J. Cone Intramural Research Program, National Institute on Drug Abuse, National Institutes of Health, Baltimore, Maryland 21224

Abstract 1

Increases in urine drug concentration that result from changes in urinary output may be mistakenly interpreted as new drug use rather than carryover from previous drug exposure. Normalization of drug excretion to urine creatinine concentration reduces the variability of drug measurement attributable to urine dilution. A specimen ratio of 1.5 or greater between two creatinine normalized positive urine cannabinoid tests was previously proposed as an indicator of new marijuana use. This approach has received wide attention for potential use in treatment and employee assistance programs associated with workplace drug testing. Unfortunately, there has been limited evaluation of the usefulness of this ratio under controlled-dosing conditions with marijuana smokers. A controlled clinical study was conducted to examine the excretion profile of creatinine and marijuana metaholites in a group of six marijuana users who smoked Iwo different doses of marijuana over a 4-week period. A relative operating characteristic curve was constructed from sensitivity and specificity data for 26 different specimen ratios ranging from 0.1 to 2.0. The most accurate specimen ratio (85.4%) for differentiating new use from residual excretion was 0.5. Use of this ratio provided a sensitivity of 80.1%, a specificity of 90.2%, and 5.6% false-pesitive and 7.4% false-negative predictions. To substantiate the validity of the 0.5 specimen ratio, urine cannabinoid and creatinine data from a controlled clinical trial specifically addressing water dilution as a means of specimen adulteration were evaluated. Sensitivity, specificity, accuracy, and percent false-positive and percent false-negative predictions were 71.9%, 91.6%, 83.9%, 5.4%, and 10.7%, respectively. These data compared favorably with the results from the first clinical study, with the exception of slightly lower sensitivity and higher false-negative percentages in the water dilution study. This would be expected because of the ingestion of large amounts of water and consequent dilution of urine drug concentration. These data indicated that selection of a specimen ratio to evaluate sequential creatinine normalized urine drug concentrations can improve the ability to distinguish residual excretion from new marijuana usage. The selection of an appropriate specimen ratio can be made based on the needs of a specific urine drug-testing program taking into account sensitivity, specificity, and accuracy data.

* A preliminary presentation of these data was made at the Society of Forensic Toxicologists annual meeting, Denver, CO, October 14-19, 1996.

t Author to whom correspondence should be addressed.

Introduction

Urine formation serves to regulate the composition of blood, eliminate endogenous waste products, and excrete toxic drugs and metabolites from the body. Creatinine, a waste product of muscle metabolism, is excreted in urine in relatively constant amounts of 1.0 to 2.5 g/day (1-6). Creatinine is freely filtered from the blood through glomeruli in the kidneys. The concentration of creatinine in the urine increases in the tubules as water is reabsorbed with little reabsorption or active secretion of creatinine during urine formation. Creatinine measurements are clinically useful for determining the completeness of 24 h urine collections, for evaluating renal function with creatinine clearance ratios, and for measuring the ratio of clearance of drug to creatinine. Normalization of the excretion of a drug to the creatinine concentration reduces the variability of analyte measurement attributable to urine dilution. Ideally, urine creatinine would have minimal intra- and interindividual variability, would be excreted in a constant daily amount, and would not be influenced by the quantity of urine excreted, dietary intake, or exercise. Creatinine meets many of these criteria, although 10% and 29% variation in daily creatinine concentration were reported within and between individuals, respectively, in radioisotope excretion studies (3). Other researchers have reported within- and between-subject creatinine concentrations ranging from 4 to 50% (1,4,5). Increased meat consumption (5) and exercise (6) may also temporarily elevate urinary creatinine. Daily creatinine concentration generally is not influenced by diuresis (6).

In contrast to creatinine, which is excreted in urine in approximately equal amounts each day, a drug dose is excreted in urine over a few days following drug administration and usually in rapidly decreasing concentration. The concentration of drug or rnetabolite in urine depends upon many factors including dose, route of administration, and rates of absorption, metabolism, and excretion. For illicit drugs such as marijuana, the excretion in urine of the metabolite, 11-nor-9-carboxy-Ag-tetrahydrocannabinol (THCCOOH), reaches a peak in 6 to 28 h and declines exponentially over a period of 1-2 weeks (7). Although the rate of excretion of THCCOOH is relatively constant, the concentration does not continually decrease and often follows a "sawtooth" pattern of increases and decreases dependent upon urine volume.

Reproduction (photocopying) of editorial content of this journal is prohibited without publisher's permission. 445

Dow

nloaded from https://academ

ic.oup.com/jat/article-abstract/22/6/445/781835 by guest on 10 April 2019


For example, Ellis et al. (8) reported fluctuations in immunoassay response with specimens obtained from heavy marijuana users during the terminal elimination phase. Specimens had alternating negative and positive responses around the 20-ng/mL cutoff concentration. A key factor in the explanation of these concentration svAngs is the amount of liquid ingested by the individual, and hence, the dilution of the urine specimen.

During the terminal elimination phase of THCCOOH, increases in concentration may be mistakenly interpreted as new drug use. Hawks (9) first suggested creatinine normalization of urine test results to correct for variations in urine volume. Whereas urine volume is highly variable because of changes in liquid, salt, and protein intake, exercise, and age, creatinine excretion is much more stable. Therefore, apparent variability in drug excretion because of urine volume changes is reduced when drug concentration is normalized to creatinine concentration and reported as nanogram of drug per milligram of creatinine. Following marijuana smoking, division of urinary THCCOOH concentration by creatinine ideally produces a metabolite/creatinine ratio that should decrease until a new episode of drug use occurs. MetaboliteYcreatinine ratios of specimens collected over time can be compared to determine if new marijuana use has occurred. Variations in the amount of fluid intake should not alter the metabolite/creatinine ratio. Con- sequently, an abrupt Lncre~e in the ratio in a subsequent specimen would indicate new drug use, Manno et al, {10) proposed a change in the metabolite/creatinine ratios of 1.5 in two positive specimens as an indicator of new use. This approach has received wide attention for potential use in treatment and employee-assistance programs associated with workplace drug testing. Unfortunately, there has been limited study of this technique under controlled-dosing conditions with marijuana smokers.

This controlled-drug-administration study examined the excretion profile of cannabinoids and creatinine in a group of six marijuana users who smoked two doses of marijuana over a 4-week period. THCCOOH/creatinine concentrations were calculated for all specimens. Specimen ratios (ratio of metabolite/ creatinine concentration in two cannabinoid-positive specimens from a single individual) were examined in an attempt to ascertain their usefulness in predicting new marijuana use versus residual excretion from previous usage. In addition, specimen ratios from another controlled-marijuana-administration study that included extensive urine dilution were evaluated against proposed criteria. Drug-testing a~d drug-treatment programs report many positive cannabinoid test results and frequently collect a series of urine specimens over days, weeks, or months to monitor an individual's drug usage or abstinence. A method to accurately differ- entiate new marijuana use from residual drug excretion would improve interpretation of urine cannabinoid results.

Experimental

Subjects Six healthy male subjects with a history of marijuana use

resided on the clinical ward of the Addiction Research Center,

National Institute on Drug Abuse, NIH while participating in a protocol designed to characterize the pharmacokinetics and pharmacodynamics of acute marijuana smoking. The protocol was approved by the Francis Scott Key Medical Center Insti- tutional Review Board and adhered to Federal guidelines for the conduct of ~es~arch on drugs of abuse in human subjects. All subjects provided written informed consent, were under contin- uous medical supervision, and were financially compensated for their participation. Subject characteristics and history of drug use have been previously published (11). In brieL the men ranged from 28 to 36 years of age, had an average weight of 77.6 kg, and a mean percent body fat content of 14.7%. Subjects began marijuana smoking between 13 and 16 years of age and smoked between 0.4 and 7.9 marijuana cigarettes per week (mean, 2.3 cigarettes). Subjects resided on the ward for approximately 4 weeks. Access to the clinical ward was restricted to prevent use of unauthorized licit or illicit drugs by the subjects. In addition, frequent urine drug tests were performed to deter additional drug exposures. Water, soft drinks, coffee, and tea were available to the subjects ad lib throughout the study.

Protocol Once a week for 3 consecutive weeks, subjects smoked a single

marijuana Cigarette (placebo, 1.75% or 3.55% Ag-tetrahydro - cannabinol ITI-tgl) according to a randomized Latin-square design. The low and high dose cigarettes contained approximately 15.8 mg and 33.8 mg THC, respectively. Subjects, medical staff, and investigators were blind to the order in which marijuana doses were administered. Marijuana cigarettes, assayed for THC content, were obtained from the Research Technology Branch, National Institute on Drug Abuse. Marijuana smoking was computer-paced, with the number of cigarette puffs, the length of time of each inhalation, the time between puffs, and the puff hold time controlled to reduce intersubject variability. Data collection, including collection of all urine specimens, continued for approximately 4 weeks.

Specimen collection and analysis All individual urine specimens were collected from six

research subjects over an approximately 4-week period. Specimens (N = 955) were collected in polypropylene containers and refrigerated immediately after urinatLon. The volume of each specimen was measured. Urine was stored in a 30-mL polypropyIer~e screw cap bottie at -30~ on the day of collection. After the 7-month protocol was complete, frozen specimens were assembled, coded, and randomized. This procedure ensured that all specimens were randomized for analysis within the large batch to eliminate any potential bias based on subject, dose, time of collection, and individual calibration. Each specimen was analyzed under blind conditions for THCCOOH by gas chromatography-mass spectrometry (GC-MS) with a one-point calibration at 15 ng/mL and a 0.5-ng/mL limit of detection according to a previously published procedure (12). Specimens were diluted and reanalyzed if results exceeded the linear portion of the curve. A modified Jaffe method on a Hitachi 704 automated clinical analyzer (Boehringer Mannheim Corp., Indianapolis, IN) was employed to obtain urine creatinine concentrations. Reagent

446

Dow




volumes and standard concentrations of the substrate-triggered rate method were adapted for creatinine measurement in urine rather than serum or plasma. Specimens were thawed just prior to analysis and subjected to only a single freeze-thaw cycle. Quality control samples were included within the randomized batch and served to assure within-and between-batch repro- ducibility. Twenty-four-hour creatinine excretion was determined each day by calculating the amount of creatinine excreted in each urine specimen (creatinine concentration in milligrams per deciliter times the volume of the specimen in deciliters) and summing the amounts of creatinine excreted in each 24-h period.

Statistics The mean creatinine concentrations for individual urine

specimens and 24 h urine collections were calculated for each subject and for the total group. In addition, the mean creatinine concentrations for all individual specimens collected during each 6-h interval (12:00 a.m. to 6:00 a.m., 6:00 a.m. to 12:00 p.m., 12:00 p.m. to 6:00 p.m, and 6 p.m. to 12:00 a.m.) were compared to determine if creatinine concentrations differed significantly during different times of the day. A repeated-measures analysis of variance with Huynh-Feldt correction and Tukey post-hoc analysis was performed to determine if significant differences in creatinine concentrations were present within subjects for 24-h collections and for different marijuana doses, between subjects for creatinine concentrations in individual and 24-h collections, and between creatinine concentrations in the four time intervals.

Normalization of urinary THCCOOH excretion to urine creatinine concentration (metabolite/creatinine ratio)

THCCOOH concentration (ng/mL) of all positive urine specimens (THCCOOH > 15 ng/mL) was divided by the specimens' urine creatinine concentration (mg/mL) to obtain the normalized THCCOOH/creatinine concentration (ng/mg) (rnetabolite/ creatinine ratio).

Calculation of the ratio for two urine cannabinoid test results (specimen ratio)

The metabolite/creatinine ratio of a positive specimen was divided by the metabolite/creatinine ratio of an earlier positive specimen to obtain a specimen ratio. Specimen ratios were determined for all positive urine specimen pairs collected over a 4-week period.

Calculation of true-positive, false-positive, true-negative, and false-negative predictions

All urine specimens were collected from subjects residing on a closed clinical ward. Each subject participated in one placebo and two active marijuana smoking sessions within the 4-week study period. Therefore, it was known when subjects had smoked marijuana. Twenty-six different specimen ratios from 0.1 to 2.0 were evaluated. If the specimen ratio was greater than or equal to the selected criterion, new drug use was predicted. If the specimen ratio was less than the criterion, residual drug

excretion from a previous drug exposure was predicted. The correctness of each prediction was determined. A true positive (TP) correctly predicted new marijuana use, a true negative (TN) correctly predicted residual excretion, a false positive (FP) predicted that the individual had smoked marijuana when residual excretion was the true reason for the positive test result, and a false negative (FN) predicted residual drug excretion when new marijuana use had occurred. Only specimens with a minimum 24 h separation between collections were compared.

Determination of sensitivity, specificity, and accuracy at each decision level (ratios of 0.1 to 2.0)

Sensitivity, specificity, and accuracy were calculated for each of the 26 specimen ratio criteria from the number ofTP, FP, TN, and FN according to standardized NCCLS definitions (13). Sensitivity = [TP/(TP + FN) ] * 100, specificity = [TN/(TN + FP)] �9 100, and accuracy = [(TP + TN)/(TP + TN + FP + FN)] * 100. The percentage of FP predictions and FN predictions were calculated by dividing each by the total number of comparisons and multiplying by 100.

Relative operating characteristics (ROC) curve An ROC curve illustrates the relationship between the true

positive rate or sensitivity of a procedure and its false positive rate or specificity (13-17). An ROC curve was constructed from sensitivity and specificity data for 26 specimen ratios. For each ratio (0.1 to 2.0), the true-positive rate was plotted on the ver- tical axis and the false-positive rate (1-specificity) was plotted on the horizontal axis. The individual points were connected to construct the ROC curve. The most accurate point on an ROC curve is the point closest to the upper left comer of the graph (coordinates 0.0, 1.0).

In vivo adulteration study The urine cannabinoid and creatinine concentrations of spec-

imens collected in an in vivo adulteration study with controlled administration of marijuana were evaluated for differentiation of residual drug excretion from new marijuana use (18). The goal of this study was to investigate the effects of excessive fluid ingestion on immunoassay test results following marijuana and cocaine administration. Several flushing conditions were part of the evaluation, including a commercial product, Naturally Klean Herbal Tea TM (Klean Tea), Golden Seal root capsules taken with excess water, and a commonly used diuretic medica- tion (hydrochlorothiazide) taken with excess water. Results were compared to control conditions in which only a small amount of water (12 oz) or excess water (1 gal) was ingested. Urine specimens were assayed for cannabinoids with the Abbott TDx | fluorescence polarization immunoassay on a TDx instrument (Abbott Laboratories, Abbott Park, IL) according to manufacturer's recommended procedures. Semiquantitative cannabinoid results ___ 15 ng/mL were normalized to the urine creatinine concentration. Sensitivity, specificity, and accuracy were calculated for predicting new marijuana use from normalized cannabinoid/creatinine concentrations at six different ratio criteria, 1.5, 1.0, 0.5, 0.25, 0.1, and 0.05.

447

Dow




Results

Urine creatinine concentrations in individual and 24-h specimen collections

A total of 955 individual urine specimens were collected over a 4oweek period from six healthy male subjects. Participants had free access to fluids throughout the study, and there were no positive or negative incentives for subjects to concentrate or dilute their urine specimens. The mean, standard deviation (SD), coefficient of variation (CV), and minimum and maximum creatinine concentrations in individual urine specimens for each subject and across the group are shown in Table I. The mean plus or minus SD creatinine concentration ranged from a low of 67.7 • 47.3 mg/dL for Subject G to a high of 188.8 • 63.3 mg/dL for Subject H. Creatinine concentration of individual specimens varied substantially within a single subject with coefficient of variations from 33.5 to 69.9%. Subjects had an 8- to 45-fold difference in their minimum and maximum creatinine concentrations. Four of six subjects produced at least one dilute specimen (<20 mg/dL) over the 4-week period (Table I). Twenty-six specimens or 2.7% of the total were found to be dilute, with 17 or >65% of these produced by a single subject. A statistically significant difference in mean individual urine creatinine concentration was found between subjects (F = 79.3,p < 0.001). The mean creatinine concentration for all individual specimens was 118.5 • 62.8 mg/dL with a range of 6-360 mg/dL.

Twenty-four-hour urine collections were obtained for six subjects on 21 different days (Table I). Mean 24 h creatinine excretion ranged from 1.03 • 0.34 to 2.34 • 0.25 g/day. There was less variability in 24 h creatinine excretion than individual urine cre-

400.

o 8 0

0 0 0 0

o o 0 0 0 0 0 0 0

o 0 o ~ 1 7 6

8oOo

0Oo O o ~

O O ~9 0 ~0 " / ) ~

(R)0 1200 1800 2400

100-

0 q

A .,J "O

E c .2 (11 Im

c o o Iz 0 0

Time (h)

Figure 1. Urine creatinine concentrations (mg/dL) in individual urine specimens (N = 955) collected from six healthy male marijuana smokers over a 3-week time period. Creatinine concentration data are displayed against the time of specimen collection. Dilute specimens (<20 mg/dL) are shown as closed circles.

atinine concentration both across subjects' and within subjects' 24-h collections. The coefficients of variation for 24 h creatinine excretion ranged from 10.8 to 33.2% across subjects with a 1.6- to 3.9-fold variation in 24 h excretion within a single subject. The overall mean 24 h creatinine excretion was 1.56 + 0.51 g/day with a CV of 32.6%.

Creatinine concentration in individual urine specimens at different times of the day

The mean creatinine concentration for individual urine specimens over four 6-h intervals (12:00 a.m. to 6:00 a.m., 6:00 a.m. to 12:00 p.m., 12:00 p.m. to 6:00 p.m., and 6:00 p.m. to 12:00 a.m.) ranged from 109.7 • 62.9 mg/dL to 122.6 • 62.0 mg/dL (Table II). Considerable variability, 50 to 57%, was noted across each interval. There were no statistically significant differences in creatinine concentration between the four time intervals. Although low creatinine concentrations were found in each interval, more dilute specimens were found between 6:00 a.m. and 12:00 p.m. (N = 8) and 6:00 p.m. and 12:00 a.m. (N = 10) than in the afternoon and late night intervals. Figure I illustrates creatinine concentrations in individual urine specimens (N = 955) collected at various times during a 24-h period.

Creatinine concentration in individual urine specimens following smoking of active and placebo marijuana cigarettes

A significant difference (F = 5.1, p < 0.006) was found in the creatinine concentration of individual urine specimens collected after smoking different potency marijuana cigarettes. The mean creatinine concentrations across subjects following smoking of a placebo, 1.75% THC, and 3.55% THC cigarette were 112.2 • 62.6

mg/dL, 120.4 • 52.8 mg/dL, and 124.7 • 70.8 mg/dL, respectively. Tukey's post-hoc analysis revealed that there was a significant difference in creatinine concentrations of individual urine specimens collected after smoking placebo marijuana cigarettes and high dose (3.55%) THC cigarettes (p < 0.01). There was no significant difference between placebo and low dose THC cigarettes or between the low and high-dose drug administrations.

Urinary THCCOOH excretion following marijuana smoking

Urine THCCOOH concentration by GC-MS for specimens collected from six subjects who smoked a high-dose (3.55% THC) marijuana cigarette are shown in Figure 2A. Data are shown for the first 48 h after drug administration, although THCCOOH was still detectable (0.5 ng/mL cutoff) in all subjects' urine specimens at the end of the 7-day collection period. Substantial intersubject variability was noted between subjects in patterns of THCCOOH excretion. In addition, substantial variability was noted in drug excretion for a single subject in sequential specimens depending upon urine dilution at the time of specimen collection. Figure 2B shows the cannabinoid concentrations normalized to the

448

Dow




Table I. Creatinine Concentrations in Individual and 24-h Urine Specimens from Six Healthy Male Subjects

Individual specimens 24-h specimens

Mean CV t Minimum Maximum #<20 Mean CV Minimum Maximum Subject N (mg/dL) SD* (%) (mg/dL) (mg/dL) (mg/dL) N (g/day) SD (%) (g/day) (g/day)

F G H

Mean (Total)

B 131 117.4 47.1 40.1 8 241 1 21 1.24 0.30 24.1 0.72 1.76 C 116 143.8 48.6 33.8 8 268 0 21 1.72 0.19 18.5 1.21 2.36 E 249 121.4 49.0 40.4 19 256 1 21 2.34 0.25 10.8 1.71 2.90

206 102.5 63.7 62.1 8 360 7 21 1.48 0.26 17.4 0.96 1.90 149 67.7 47.3 69.9 6 258 17 21 1.03 0.34 33.2 0.43 1.70 104 188.8 63.3 33.5 40 326 0 21 1.51 0.30 19.7 0.96 1.99

(955) 118.5 62.8 53.0 (26) (126) 1.56 0.51 32.6

* SD, standard deviation. t CV, coefficient of variation.

Table II. Creatinine Concentration in Individual Urine Specimens at Different Times of the Day

Subject Interval B C E F G H Mean/(Total)

Midnight-6 a.m. No. specimens 29 37 65 48 37 33 (249) Mean 131.7 117.5 116.0 95.8 75.2 191.2 118.1 Standard deviation 52.0 57.1 44.3 59.2 51.6 71.1 64.0 CV'(%) 39.5 48.6 38.2 61.8 68.5 37.2 54.2 Minimum 43 38 45 13 8 40 Maximum 241 231 213 208 184 326 No.<20 mg/dL 0 0 0 2 3 0 (5)

6 a.m.-noon No. specimens 28 19 52 43 36 15 (I 93) Mean 91.4 178.4 118.3 97.6 55.9 190.3 109.7 Standard deviation 50.8 45.0 49.3 51 .I 46.0 49.5 62.9 CV (%) 55.5 25.2 41.7 52.4 82.3 26.0 57.3 Minimum 8 71 32 26 6 127 Maximum 224 268 256 274 154 283 No.<20 mgldl_ I 0 0 0 7 0 (8)

Noon-6 p.m. No. specimens 31 31 74 60 39 35 (270) Mean 120.8 140.1 115.4 119.1 73.2 184.7 122.6 Standard deviation 42.7 29.2 50.2 76.2 33.6 63.9 62.0 CV (%) 35.3 20.9 43.5 64.0 45.9 34.6 50.6 Minimum 51 67 19 8 24 56 Maximum 224 210 229 360 149 324 No.<20 mg/dL 0 0 1 2 0 0 (3)

6 p.m.-midnight No. specimens 43 29 58 55 37 21 (243) Mean 122.1 158.7 138.0 94.2 65.7 190.8 Standard deviation 38.4 36.8 49.8 59.4 55.1 61.7 62.0 CV (%) 31.4 23.2 36.1 63.0 83.9 32.3 51.1 Minimum 43 110 37 14 10 45 Maximum 185 243 249 308 258 309 No.<20 mgldL 0 0 0 3 7 0 (I0)

* CV, coefficient of variation.

121.3

449

Dow




A

300.

I ,- loo.

0 - i i i (ll) 0 e 12 18 24 30 36 42 48

2so] Subject B 200 j'l ~ B J , F

/ Z % . c- -o -

(61 0 -6 1'2 1'8 2'4 3.8 36 ,~2 4~

T i m e (h )

Figure 2. A, Forty-eight-hour urinary excretion profiles of THCCOOH as measured by gas chromatography/mass spectrometry in six subjects following smoking of a single 3.55% tetrahydrocannabinol (THC) cigarette. B, Forty-eight-hour urine THCCOOH concentrations (ng/mL) normalized to urine creatinine concentrations (ng/mg).

150- 150

100,

t=

50.

0 0

--o-- THCCOOH (ng/mL) OO reati ne (ng/mg) 100

50

o 3'0 6'0 90 120 150 180

T i m e (h)

i r

Figure 3. Urinary excretion profile of THCCOOH as measured by GC-MS in Subject B following smoking of a single 3.55% THC cigarette (open circles). The horizontaI Iine at 15 ng/mL represents the current GC-MS cutoff used in most testing programs. Results above this line indicate positive urine cannabinoid tests; results below this line would be considered negative. Subject B's urine THCCOOH concentrations (nanograms per milliliter) normalized to urine creatinine concentrations (milligrams per milliliter) are illustrated with closed circle symbols.

creatinine concentrations for the six subjects. Figure 3 illustrates the urinary excretion of THC- COOH by GC-MS over 168 h in Subject B following smoking of a single 3.55% THC cigarette.

Sensitivity, specificity, and accuracy of prediction of new marijuana use with different specimen-ratio criteria

Sensitivity, specificity, accuracy, and percentages of false-positive and false-negative predictions at 26 specimen-ratio criteria are included in Table III. Specimen ratios from 0.01 to 2.0 were applied to predict if new drug use had occurred when another positive (> 15 ng/mL) result was obtained. Figure 4 is the ROC curve developed from true-positive and false-positive rates for the 26 ratio criteria evaluated in this controlled clinical study of marijuana administration. Positive specimen results for each individual were compared with all other positive specimen results collected between 24 h and up to 4 weeks after collection of the first positive specimen (N= 1837). Using a specimen ratio of 1.5, the sensitivity for detecting new drug use was 33.4%, specificity was 99.8%, and the overall accuracy of prediction was 74.2%. The highest accuracy was obtained with a specimen ratio of 0.5. Sensitivity, specificity, and accuracy for the 0.5 ratio criteria were 80.1%, 90.2%, and 85.4%, respectively. Lowering the ratio improved the sensitivity of detection of new drug use, but there was a con- comitant loss of specificity and higher false positive predictions. An increased specimen ratio improved specificity, that is, fewer false predictions of new use, but sensitivity in the detection of new use was decreased, that is, a higher false-negative rate.

Evaluation of specimen-ratio criteria to predict new marijuana use in a study of in vivo adulteration of urine cannabinoid concentrations

There were 531 cannabinoid positive (TDx > 15 ng/mL) urine specimens collected from seven healthy men in this controlled clinical study of smoked marijuana. These individuals drank vary- ing quantities of water and consumed commercial diuretic and masking agents over a 7-week period. A total of 26,112 comparisons between creatinine normalized cannabinoid data were evaluated against six specimen-ratio criteria (1.5, 1.0, 0.5, 0.25, 0.1, and 0.05). Sensitivity, specificity, accuracy, and false-positive and false-negative prediction rates are included in Table IV. Specificity, sensitivity, and accuracy with a 1.5 ratio were 28.9%, 98.8%, and 65.2%, respectively. The false-negative rate was greater than 33%. With a specimen ratio criterion of 0.5, sensitivity

450

Dow




improved to 71.9% and specificity decreased to 91.6%, for a higher accuracy of 83.9%. Accuracy and sensitivity increased as the specimen ratio was lowered to 0.25 and 0.1, but there was an increase in false-positive predictions from 5.4% (0.5 ratio) to 13.8% (0.1 ratio).

Discussion

Urine creatinine has been suggested as an appropriate analyte for normalization of cannabinoid concentration to reduce the variability in concentration that is attributable to changes in urine volume. Quantitative adjustment to creatinine concentration has been used for more than 50 years in the environmental and industrial toxicology fields (19). Normalization of urine drug concentration to creatinine concentration has been attempted for marijuana (10,20-24), amphetamine (25), cocaine (26), nico- tine (27), and buprenorphine (28) metabolite levels to compen- sate for dilution effects and to improve interpretation of test results. When a second cannabinoid-positive urine test is obtained, with or without intervening negative test results, it may be important to determine if new drug use has occurred. An evaluation of the specimen ratio for two positive specimens can aid in determining if the individual used marijuana between specimen collections.

For creatinine normalization to be useful in the interpretation of urine cannabinoid results, an understanding of the stability of creatinine excretion is necessary. The present study evaluated the creatinine concentration of 955 individual urine specimens collected over 4 weeks. The mean creatinine value, 118.5 + 62.8 mg/dL, was in agreement with other reported averages ranging from 93.8 to 183.1 mg/dL (4, 22, 29, 30). A wide variation was noted in mean individual creatinine concentrations across subjects and in individual creatinine concentrations for a single subject over the 3-week study period. The pres- ence of significant variation in creatinine excretion between individuals poses a problem for establishment of a normalized metaboliteJcreati- nine cutoff level for determining positive or negative test results for a single specimen. However, this variation does not affect the monitoring of metabolite/creatinine ratios within a single subject. Variation in creatinine concentration of individual urine specimens for a single subject reflects changes in urine dilution that similarly influence urine drug concentration. It is the stability of an individual's 24-h urine creatinine excretion that affects the usefulness of cannabinoid/creatinine normalization. No significant differences in mean 24-h creatinine excretion were found for individual subjects when 24-h excretion was monitored for 21 days. Mean 24-h creatinine excretion varied from 10 to 30% for an individual. Thus, the urinary excretion of

cannabinoid metabolites over time will ideally produce a smooth excretion curve with continually decreasing concentration in sequential spe-cimens, when normalized to the creatinine concentration.

Objective criteria are needed for the interpretation of an indi- viduars metabolite/creatinine ratios over time. The specimen ratio of a later specimen's metabolite/creatinine ratio to an earlier specimen's results can be interpreted against empirically determined specimen ratio criteria. A specimen ratio of 1.5 has been commonly used to predict new marijuana exposure, although there are few data to substantiate its use. Data from this controlled clinical study of marijuana smoking were used to develop objective specimen-ratio criteria for evaluating multiple positive test results. An ROC curve was constructed from sensitivity and specificity data at 26 specimen-ratio criteria for differentiation of new marijuana use from residual drug excretion in occasional marijuana smokers. As shown in Table III, the accuracy of differentiating new use from residual drug excretion with a ratio of 1.5 was less then 75%. New marijuana use was misiden- tiffed as residual excretion one-fourth of the time. On the other hand, new drug use was almost always correctly predicted. The most accurate differentiation occurred with a ratio of 0.5. Accurate differentiation of new marijuana episodes from residual excretion occurred in 85% of the cases. New use was missed 7.4% of the time, and new use was incorrectly predicted in 5.6% of cases.

In vivo adulteration of urine specimens through water loading and ingestion of diuretics and other masking agents complicate drug-test interpretation. Urine dilution decreases drug concentration and increases false-negative test results. Urine cannabi-

* T r u e - p o s i t i v e and f a l s e - p o s i t i v e ra tes for 26 s p e c i m e n ra t io c r i t e r i a (0.1 to 2 .0 )

1,oo

~ , . , 0.75"

' ;

~ .~ o.5o- O ~

' - 0.25; I -

O.1 = h ighes t sens i t i v i t y , lowest specificity

ccurate j combination of

sensitivity and specificity

2.0 : lowest sensitivity, highest specificity

0.00 0.00 0.25 0.50 0.75 1.00

Fa lse-pos i t ive rate (1 - Speci f ic i ty)

Figure 4. Relative operating characteristic (ROC) curve for the prediction of new marijuana use based on metabolite/creatinine ratios. ]he true-positive and false-positive rate at 26 different specimen ratio criteria (0.1 to 2.0) determines the ROC curve. As the specimen ratio decreased from 2.0 to 0.1 sensitivity increased, but specificity decreased, ]'he highest accuracy was achieved at a ratio of specimen ratio of 0.5.

451

Dow



noid and creatinine data from an in vivo adulteration study were evaluated against the empirically derived specimen-ratio criteria. Similar sensitivities, specificities, and accuracies were obtained in both studies at each of the six criteria tested (Tables III and IV). With a specimen ratio of 1.5 the sensitivity, specificity and accuracy of the marijuana smoking study and the in vivo dilution study were 33.4%, 99.8%, and 74.2%, and 28.9%, 98.8%, and

Table III. Relative Operating Characteristic (ROC) Curve for Selecting the Appropriate Specimen Ratio Criteria to Use when Comparing Two Metabolite/ Creatinine Ratios.

Specimen Mean Mean Mean % False % False ratio criteria sensitivity specificity% accuracy% positive negative

2.00 25.0 100 70.6 0.0 30.5 1.75 28.1 100 71.9 0.0 26.2 1.50 33.4 99.8 74.2 0.1 24.0 1.25 38.4 99.3 75.9 0.4 21.6 1.20 39.8 99.3 76.4 0.4 21.0 1.15 42.0 99.1 77.2 0.5 20.1 1.10 43.6 98.9 77.7 0.6 19.3 1.05 46.3 98.8 78.7 0.7 18.2 1.01 47.8 98.2 78.9 0.9 17.4 1.00 48.7 98.1 79.2 1.0 17.1 0.95 50.0 97.6 79.4 1.3 16.6 0.90 51.4 96.8 79.4 1.6 16.0 0.85 54.9 96.6 80.4 1.8 15.1 0.80 56.8 96.1 80.9 2.1 14.2 0.75 59.9 95.8 81.9 2.2 13.2 0.70 62.6 95.5 82.7 2.4 12.1 0.65 66.8 94.7 83.6 2.8 11.0 0.60 71.5 93.6 84.5 3.5 9.9 0.55 76.1 92.1 85.3 4.4 8.5 0.50 80.1 90.2 85.4 5.6 7.4 0.45 83.4 87.4 85.1 7.0 6.4 0.40 86.5 85.4 84.9 8.3 5.8 0.35 90.8 83.2 85.1 9.7 5.1 0.30 92.6 79.4 83.8 11.9 4.1 0.25 94.5 75.8 82.7 14.1 3.2 0.01 98.8 62.4 77.5 21.3 0.8

* The accuracy, sensitivity, specificity, % FP and % FN at each specimen ratio were determined in a controlled-drug-administration study.


65.2%, respectively. Sensitivity, specificity, and accuracy were 80.1%, 90.2%, and 85.4% in the marijuana smoking study and 71.9%, 91.6%, and 83.9% in the dilution study when a specimen ratio of 0.5 was used for prediction. Further reduction in the specimen ratio improved sensitivity for the detection of marijuana use in the dilution study, but accuracy remained constant because of a substantial increase in false-positive predictions.

Specimen-ratio criteria were derived from GC-MS concentrations of THCCOOH using a 15-ng/mL cutoff in a controlled clinical study of marijuana smoking. These ratio criteria were tested against urine cannabinoid data from a controlled clinical study of marijuana smoking that involved water dilution. Cannabinoid data from the latter study were obtained with a semi-quantitative immunoassay (Abbott | fluorescence polarization immunoassay on a TDx | instrument). For consistency, a 15-ng/mL cutoff concentration also was employed in the water-dilution study. Similar accuracy, sensitivity, and specificity data were noted, suggesting that these criteria may be applicable to quantitative GC-MS data and semiquantitative cannabinoid data from a variety of immunoassays.

An important issue in the use of specimen-ratio criteria for the evaluation of positive urine cannabinoid results is the amount of time elapsed between specimen collections. Previous research in our laboratory documented positive urine cannabinoid- screening results (> 50 ng/mL) in six individuals for only 0 to 78 h after the smoking of one low- or high-dose marijuana cigarette when analyzed with six different immunoassay tests (11). The average detection time for the last positive urine test was approximately 24 h after a low-dose and 48 h after a high-dose marijuana cigarette in occasional drug users. Consequently, an occasional user should not produce positive results beyond 2-4 days at the 50-ng/mL cutoff concentration and possibly 4-8 days at the 20-ng/mL threshold. In addition, specimen ratios for creatinine-normalized cannabinoid test results should diminish during drug excretion. An increase in the specimen ratio between two positive cannabinoid tests that are separated in time by more than 8 days is highly suggestive of new use in an occasional user. It is uncertain if the sensitivity, specificity, and accuracy of prediction of new drug use will be the same when the proposed specimen-ratio criteria are applied to frequent users' specimens. Clearly, these criteria should be tested in chronic users under controlled experimental conditions before applica-

Table IV. The Accuracy, Sensitivity, Specificity, % FP, and % FN at Representative Specimen Ratios Applied to Water Dilution Study of Marijuana Smoking'

Specimen Mean Mean Mean % False % False ratio criteria sensitivity % specificity % accuracy % positive negative

1.5 28.9 98.8 65.2 0.9 33.9

1.0 44.2 97.9 72.8 1.4 25.8

0.5 71.9 91.6 83.9 5.4 10.7

0.25 90.0 80.8 85.5 11.1 3.4

0.1 97.8 73.2 85.6 13.8 0.6 0.05 99.4 71.9 85.6 14.3 0.2

* See reference 18.

452

Dow




tion to heavy marijuana users. Currently, to our knowledge, there are no cannabinoid excre-

tion data from chronic marijuana users available in the literature to evaluate the proposed specimen-ratio criteria. These criteria were derived from a controlled clinical study of marijuana smoking in occasional users, and the validity of their application to chronic user excretion data cannot be verified. There are several reports in the literature of an extended cannabinoid excretion time following cessation of marijuana use in chronic users (8,24,31). The limited number of studies that have been performed either include no urine cannabinoid or creatinine data, or subjects were not confined to a controlled environment to preclude additional marijuana use. Johansson et al. (24) reported a mean urinary excretion half-life of 3.0 • 2.3 days (range 0.8 to 9.8 days) in 13 heavy marijuana users. Individual subjects' metabolite/creatinine data were not provided, although graph- ical representations of these data were furnished for three subjects. Visual inspection suggests that a specimen ratio of __. 1.5 might be useful for prediction of new drug use in chronic marijuana users. It is important to obtain an accurate drug history from individuals in treatment and employee-assistance programs, including age at first use, duration of use, and frequency and pattern of use, to aid in the interpretation of urine cannabinoid data. The longest cannabinoid detection time reported for a chronic long-term marijuana user was 46 days using a broadly cross-reacting immunoassay with a 20-ng/mL cutoff (8). Individuals frequently claim residual drug excretion more than 60 days after a positive urine drug test. These assertions do not appear to be realistic in light of controlled drug administration and urinary excretion case studies (24). Frequent urine drug tests can be performed during drug cessation to better interpret test results. The pattern of urine drug-test results may help to establish the chronicity of marijuana usage.

Two additional findings about urine creatinine excretion in this study are worthy of comment. Early morning urine specimens are generally considered to be highly concentrated and thus specimens of choice for drug detection. However, there were no significant differences in creatinine concentration of urine specimens collected over the four time intervals, 12:00 a.m to 6:00 a.m., 6:00 a.m. to 12:00 p.m, 12:00 p.m to 6:00 p.m., and 6:00 p.m. to 12:00 a.m. Furthermore, as shown in Figure 1, dilute urine specimens occurred throughout the day, and urine creatinine concentrations were fairly evenly distributed over a 24-h day. An additional finding was the significant difference in mean creatinine concentration in individual specimens following the smoking of placebo compared with high-dose marijuana cigarettes. This has not been previously reported to our knowledge. The mean creatinine concentration in specimens collected after placebo marijuana (112.2 • 62.6 mg/dL) was significantly lower than in specimens collected after the 3.55% THC cigarettes (124.7 • 70.8 mg/dL). Mean creatinine concentration in specimens collected after smoking low-dose marijuana cigarettes was intermediary in concentration (120.4 • 52.8 mg/dL); however, this result was not significantly different from the placebo or high-dose averages. Additional research is needed to determine if these results are artifacts of this subject popula- tion or study design. There is some evidence to support the validity of these findings. In a chronic oral THC administration

study (32), plasma volume increased 21% during drug administration. The authors hypothesized that the increased plasma volume was due to increased aldosterone secretion but added that other drug effects, including renal perfusion changes, prostaglandin inhibition, and modified pituitary-adrenal function, could contribute to increased plasma volume. Sodium and water retention were marked during chronic dosing but returned to normal within 48 h of cessation of drug administration. Perhaps increased reabsorption of water by the kidneys following marijuana administration could result in increased concentration of the urine and higher creatinine concentration in individual urine specimens.

Conclusion

Drug-testing programs have diverse strategies, procedures, and goals. Response to a second positive marijuana urine test varies by program and may result in vastly different consequences for participants. High sensitivity (low false-negative rate) may be more important than high specificity (low false-positive rate) to drug-treatment programs in which identification of new drug use generates critical discussions and enables staff to evaluate adherence to drug-abstinence contracts. High specificity may be more important than sensitivity in probation and child custody cases in which dire consequences may follow the identification of new drug use. This study provides sensitivity, specificity, accuracy, false-positive and false-negative rates at a variety of specimen ratios, which enables selection of the best criteria to fit a program's needs.

References

1. D.B. Morgan, S. Dillon, and R.B. Payne. The assessment of glomerular function: creatinine clearance or plasma creatinine? Postgrad. Med. J. 54:302-310 (1978).

2. K. Spencer. Analytical reviews in clinical biochemistry: the estima- tion of creatinine. Ann. Clin. Biochem. 23:1-25 (1986).

3. S. Narayanan and H.D. Appleton. Creatinine: a review. Clin. Chem. 26(8): 1119-1126 (1980).

4. M.D.S. Shephard, L.A. Penberthy, and C.G. Fraser. Short- and long- term biological variation in analytes in urine of apparently healthy individuals. Clin. Chem. 27(4): 569-573 (1981 ).

5. S.A. Bingham and J.H. Cummings. The use of creatinine output as a check on the completeness of 24-hour urine collections. Hum. Nutr.: C/in. Nutr. 39C: 343-355 (1985).

6. L. Alessio, A. Berlin, A. Dell'Orto, E Toffoletto, and I. Ghezzi. Reliability of urinary creatinine as a parameter used to adjust values of urinary biological indicators. Int. Arch. Occup. Environ. Health 55:99-106 (1985).

7. M.A. Huestis, J.M. Mitchell, and E.J. Cone. Urinary excretion profiles of 11-nor-9-carboxy-&%tetrahydrocannabinol in humans after single smoked doses of marijuana. J. AnaL Toxicol. 20:441-452 (1996).

8. G.M. Ellis, M.A. Mann, B.A. Judson, N.T. Schramm, and A. Tashchian. Excretion patterns of cannabinoid metabolites after last use in a group of chronic users. C/in. Pharmacol. Ther. 38:

453

Dow




572-578 (1985). 9. R.L. Hawks. Developments in cannabinoid analyses of body

fluids: implications for forensic applications. In The Cannabinoids: Chemical, Pharmacologic, and Therapeutic Aspects. S. Agurell, W. Dewey and R. Willette, Eds. Academic Press, Rockville, MD, 1983, pp I - I 2.

I0. J.E. Manno, K.E. Ferslew, and B.R. Manno. Urine excretion patterns of cannabinoids and the clinical appliation of the EMIT-d.a.u. cannabinoid urine assay for substance abuse treatment. In The Cannabinoids: Chemical, Pharmacologic, and Therapeutic Aspects. S. Agurell, W.L. Dewey, and R.E. Willette, Eds. Academic Press, Inc., New York, NY, 1984, pp 281-290.

11. M.A. Huestis, J.M. Mitchell, and E.J. Cone. Detection times of marijuana metabolites in urine by immunoassay and GC-MS. ]. Anal Toxicol. 19:443-449 (1995).

12. B.D. Paul, L.D. Mell, Jr., J.M. Mitchell, R.M. McKinley, and J. Irving. Detection and quantitation of urinary 11-nor-delta-9-tetrahydrocannabinol-9-carboxylic acid, a metabolite of tetrahydrocannabinol, by capillary gas chromatography and electron impact mass fragmentography. J. AnaL ToxicoL 11 : 1-5 (1987).

13. M.H. Zweig. Assessment of clinical sensitivity and specificity of laboratory tests. National Committee for Clinical Laboratory Standards, GP 10-P. 7(6): 143-171 (1987).

14. V. Spiehler, H. Wilson, S. El Shami, T. Sneath, and N. Jain. Relative operating characteristic (ROC) curves used to compare drug screening immunoassays for amphetamine and methamphetamine. AACC ZD003: Clin. Chem. 35(6): 175 (1989).

15. J.A. Hanley and 8.J. McNeil. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 143: 29-36 (I 982).

16. D.A. Noe. Selecting a diagnostic study cutoff value by using its receiver operating characteristic curve. Clin. Chem. 29:571-572 (1983).

17. J.R. Beck and E.K. Shultz. The use of relative operating characteristic (ROC) curves in test performance evaluation. Arch. Pathol. Lab. Meal. 110:13-20 (1986).

18. E.J. Cone, R. Lange, and W.D. Darwin. In vivo adulteration: excess fluid ingestion causes false-negative marijuana and cocaine urine test results. J. Anal. Toxicol. 22:460-473 (1998).

19. J.N. Wilkins. Quantitative urine levels of cocaine and other sub- stances of abuse. NIDA Res. Monograph 175:235-252 (1997).

20. P.C. Painter, J.H. Evans, J.D. Greenwood, and W.W. Fain. Urine cannabinoids monitoring. Diag. Clin. Testing 27:29-33 (1989).

21. R. Bell, E.H. Taylor, B. Ackerman, and A.A. Pappas. Interpretation of urine quantitative 11-nor-delta-9-tetrahydrocannabinol-9-carboxylic acid to determine abstinence from marijuana smoking. Clin. Toxicol. 27: 109-I 15 (1989).

22. P. Lafolie, O. Beck, G. Blennow, L. Boreus, S. Borg, C.E. Elwin, L Karlsson, G. Odelius, and P. Hjemdahl. Importance of creatinine analyses of urine when screening for abused drugs. Clin. Chem. 37: 1927-1931 (1991).

23. R. Rodgers, C.P. Crowl, W.M. Eimstad, M.W. Hu, J.K. Kam, R.C. Ronald, G.L. Rowley, and E.F. Ullman. Homogeneous enzyme immunoassay for cannabinoids in urine. Clin. Chem. 24:95-I00 (I 978).

24. E. Johansson and M.M. Halldin. Urinary excretion half-life of A1-tetrahydrocannabinol-7-oic acid in heavy marijuana users after smoking. J. Anal. Toxicol. 13:218-223 (I 989).

25. K.A. Haberny, S.L. Walsh, D.H. Ginn, J.N. Wilkins, J.E. Garner, D. Seloda, and G.E. Bigelow. Absence of acute cocaine interactions with the MAO-B inhibitor selegiline. Drug Alcohol Depend. 39: 55-62 (I 995).

26. R.D. Weiss and F.H. Gawin. Protracted elimination of cocaine metaboilites in long-term, high-dose cocaine abusers. Am. J. Med. 85:879-880 (1988).

27. S.G. Thompson, R.D. Barlow, N.J. Wald, and H.V. Vunakis. How should urinary cotinine concentrations be adjusted for urinary creatinine concentration? Clin. Chim. Acta 187:289-296 (1990).

28. I.D. Watson. Analysis of commonly abused drugs in urine at selected threshold (cutoff) concentrations. Clin. Chem. 38(2): 441 (1992).

29. E.M. Gowans and C.G. Fraser. Biological variation of serum and urine creatinine and creatinine clearance: ramifications for interpretation of results and patient care. Ann. Clin. Biochem. 25: 259-263 (1988).

30. S.B. Needleman, M. Porvaznik, and D. Ander. Creatinine analysis in single collection urine specimens. ]. Forensic Sci. 37:1125-1133 (1992).

31. R. Swatek. Marijuana use: persistence and urinary elimination. J. Subst. Abuse Treat. I : 265--270 (I 984).

32. R.T. Jones. Cardiovascular effects of cannabinoids. In Marihuana '84. S.W. Paton and G.G. Nahas, Eds. IRL Press Limited, Oxford, U.K., 1985, pp 325-334.

Manuscript received March 27, 1998; revision received May 22, 1998.

454

Dow



Documents

Differentiating New Marijuana Use From Residual Drug Excretion in