an evaluation of rapid point-of-collection oral fluid … · devices were evaluated (ora[lab |...

Journal of Analytical Toxicology, Vol. 27, October 2003

An Evaluation of Rapid Point-of-Collection Oral Fluid Drug-Testing Devices

J. Michael Walsh 1,*, Ron Flegel 1, Dennis J. Crouch 2, Leo Cangianelli 1, and Jakub Baudys 2 I The Walsh Group P..A., 6701 Democracy Blvd. Suite 300, Bethesda, Maryland 20817 and 2Center for Human Toxicology, University of Utah, 20 South 2030 East, Salt Lake City, Utah 84112

[Abstract ] New technology is currently being marketed to rapidly test oral fluids for drugs of abuse at the point of collection (PC)C). There are no nationally accepted standards or cutoff concentrations for detecting drugs in oral fluids and for most analytes there are significant differences in cutoff concentrations across devices. Four devices were evaluated (Ora[Lab | RapiScan, Drugwipe, and SalivaScreen | for their ability to meet manufacturers claims, and proposed federal standards for criminal justice and workplace programs. Human oral fluid fortified with known quantities of drug [drug(s) or metabolite(s)] was used to test these devices. Overall, the performance of these rapid POC oral fluid drug-testing devices was quite variable. Some devices performed well for the analysis of some drug classes but poorly for others. In general, most devices performed well for the detection of methamphetamine and opiates, but all performed poorly for the detection of cannabinoids. The ability to accurately and reliably detect cocaine and amphetamine was dependent on the individual device.

Introduction

A variety of specimens can be assayed for drugs (e.g., urine, blood, sweat, saliva, and hair). Each specimen is unique, and each provides different information about drug use. For example, drugs or their metabolites may be detected for many months after use in hair, while the detection window for urine and sweat can be days to weeks and that of blood and oral fluid, hours to days (1). Urine remains the most common specimen collected and tested for drugs of abuse because of the ease of analysis, relatively high drug and metabolite concentrations, and the variety of testing methods available.

In the current environment of increased testing for drugs of abuse in the workplace, in treatment programs, and in schools, there are inherent advantages to testing oral fluid. A number

�9 Author to whom correspondence should be addressed: J. Michael Walsh, Ph.D., 6701 Democracy Blvd., Suite 300, Bethesda, MD 208]7. E-mail: [email protected] g.

of studies and review articles have examined saliva/oral fluid as a diagnostic matrix for detecting drugs of abuse (2-6). Mixed saliva, or oral fluid, is perhaps the most accessible matrix used for the detection of drugs. Oral fluid consists primarily of se- cretions from the submaxillary (65%), parotid (23%), and sub- lingual (4%) glands (1,5). Detection times for drugs in oral fluids are roughly similar to that in blood, approximately 1-24 h [An extensive discussion on detection times in oral fluid by drug is provided by Huestis and Cone (1)]. Oral fluid normally contains parent drug rather than drug metabolites, which are most commonly detected in urine. Collection of oral fluid is generally considered less invasive than either blood or urine, and oral fluid could be an excellent matrix to relate drug use with behavioral impairment (1,2,5-7). However, testing oral fluid for drugs of abuse is relatively new, and the identification and interpretation of drugs in this medium can depend on the saliva/plasma ratio, which is often less than 1:1. Drug secretion in oral fluids is also pH dependent. Therefore, if the correlation between oral fluid and plasma concentration is needed for interpretation, then pH should be measured (8).

In the past, the analysis of oral fluid has normally been con- ducted in a laboratory. However, a number of new rapid im- munoassay testing devices have recently become available that permit immediate testing of the oral fluid specimen at the point of collection (POC) (e.g., a police station or, in some cases, even at the roadside). These devices use methods that appear to be similar to the rapid POC testing devices that have been shown to be useful for urine drug testing. Some of the newly developed oral fluid devices are modifications of urine test kits, and reports comparing POC oral fluid devices with other matrices have found them user friendly, but generally not as accurate as the POC urine testing devices (9,10). Studies ex- amining the effectiveness of specific oral fluid devices to detect drugs and comparing results with lab-based assays have yielded varying results (11-14). A few of the newly available POC oral fluids testing devices have potential for use in roadside testing, and others utilize desktop instruments and would need to be used at police stations in a manner similar to evidentiary breathalyzer devices.

Reproduction (photocopying) of editorial content of this journal is prohibited without publisher's permission. 429

The purpose of the research reported here was to compare a number of POC oral fluid testing devices to determine their accuracy and reliability when testing oral fluid samples fortified with known concentrations of amphetamine, methamphetamine, cocaine, opiates (morphine), Ag-tetrahydro- cannabinol (THC), 11-carboxy-THC (THC-COOH), and benzodiazepines (temazepam).

Methods

The evaluation was performed jointly by The Walsh Group (TWG, Bethesda, MD) and the Center for Human Toxicology (CHT) at the University of Utah (Salt Lake City, UT). We re- viewed the literature, researched the Internet, and contacted a variety of sources to identify the currently available POC oral fluid devices that were on the market or in development. The identified manufacturers were contacted, the evaluation ex- plained, and their participation was solicited. Each manufacturer was requested to provide the following information about their device: drugs tested, cutoff concentrations (cutoffs), an- tibody target analytes, cross-reactivity, whether the device was available in a single drug test or panel, device cost, training re- quirements, and the status of product development. In order to obtain this information, a cross-reactivity data sheet and questionnaire was sent to each manufacturer identified.

Device selection Based on the manufacturer's responses and device avail-

ability, the following products were made available for the evaluation: OralLab (Ansys Technologies, Inc., Lake Foster, CA), RapiScan (Cozart Bioscience Ltd., Abingdon, Oxfordshire, U.K.), Drugwipe (Securetec, Ottobrunn, Germany), and SalivaScreen (Ulti-Med, Ahrensberg, Germany). These devices represent four of the seven POC oral fluid testing products commercially available in the world at the time of this study. All of these devices employ an immunochromatographic analysis similar to that which is widely used in rapid urine test devices.

Training Each of the manufacturers was asked to provide technical

training about the use of their device. Device training allowed our analysts to ask questions regarding the procedures and allowed the manufacturer to explain the recommended procedures for their device. Training was scheduled onsite at Bethesda, MD or Salt Lake City, UT and was completed prior to the evaluation.

Quality control Using the manufacturers' product information and The Sub-

stance Abuse and Mental Health Services Administration (SAMHSA)'s draft guidelines (15) for the analysis of oral fluids, target drug concentrations for the evaluation protocols were established. The protocol was designed such that each device was challenged with a low, medium, and high concentration of the target drug as well as with a drug-free (negative) control. Each device was challenged 10 times at each fortified drug

430


concentration and 5 times with the negative control (i.e., a total of 35 tests per drug per device). As a general rule, the low concentration was one-half of the proposed SAMHSA cutoff, the medium concentration was twice the proposed cutoff, and the high concentration was 10 times the proposed cutoff. The cannabinoid concentrations were the only exception. It was clear from the product information provided that none of the devices could test for THC in the range suggested by SAMHSA (i.e., 4 ng/mL). Therefore, for THC, the low concentration was 1.25 times the proposed SAMHSA cutoff and the medium and high concentrations were 5 times and 25 times, respectively, the SAMHSA concentration. Although not planned in the orig- inal experimental design, additional evaluations were also con- ducted with low, medium, and high concentrations of marijuana metabolite (THC-COOH) to contrast with detection capabilities of THC. Based on the examination of the manufacturers' claims, the low and high concentrations for each drug were consistent with the lower and upper ends of the detection windows for most of the devices. (See Table I for the specific solutions and drug concentrations).

Quality control preparation Human oral fluid was purchased (Biochemed Pharmacolog-

icals, Inc., Winchester, VA) or pooled from drug-free subjects, frozen, thawed, centrifuged, and the supernatant collected. This process (freezing, thawing, centrifuging) was repeated three times to ensure consistency and clarity before the oral fluid was fortified with the target drug. Drug reference for amphetamine, methamphetamine, morphine, cocaine, THC, THC- COOH, and temazepam was purchased from Cerilliant (Round Rock, TX) and diluted for preparation of the control solutions. The low, medium, and high control solutions were prepared by adding known amounts of drug to a predetermined volume of the drug-free oral fluid. The fortified solutions were assayed by gas chromatography-mass spectrometry (GC-MS) or liquid chromatography-tandem mass spectrometry (LC-MS-MS) for quantitative concentrations of fortified drugs. An Agilent I I00 series HPLC system (Palo Alto, CA) interfaced with a Thermo/Finnigan Triple Stage Quadrupole (TSQ-7000) MS employing electrospray ionization, Excalibur software, and selected reaction monitoring was used for the quantitation of all of the drugs except the cannabinoids (16). An Agilent 5890 series II GC system interfaced with a Thermo/Finnigan Single Stage Quadrupole (SSQ-7000) MS employing negative chemical ionization and Excalibur software was used for the THC and THC-COOH analysis. For each analysis, a multipoint calibration curve was prepared in drug-free oral fluids. The peak-area ratio of the analyte to deuterium-labeled internal standard was used to determine the drug concentration in each control from the regression equation generated from the calibrators. Each control solution for amphetamine, methamphetamine, cocaine, opiates, and benzodiazepine was analyzed in duplicate and the mean concentration and the percent of target concentration cal- culated. Once the target concentration was verified (• 10%) the control solutions were transferred into smaller containers of no more than 25-50 mL and stored frozen at -20~ until the device evaluation was performed. The THC and THC-COOH solutions were prepared just prior to the device evaluation.


Aliquots of each control solution were removed, stored frozen at -20~ and the concentrations were verified after the device evaluation.

Evaluation procedure On each day of the evaluation the following procedure was

followed. A control solution was selected and thawed. Ten devices from each manufacturer were unpackaged and labeled. The analyses were performed, and the test results were read and recorded by the principal analyst. For devices in which the results were visually interpreted, a second analyst also read and recorded his/her results. Discrepant results between the primary and secondary analyst were noted. (Only two discrepan- cies were encountered in the entire evaluation, and these were interpreted in favor of the device). The procedure was repeated until all of the devices were evaluated with the selected control solutions.

The Ansys test kits are individually packaged and include a foam collection pad and a six-panel multi-test cassette. The test cassette consists of two membrane strips each containing three drug tests and an internal control. The cassette simultane- ously tests for amphetamine, methamphetamine, cocaine, opiates, cannabinoids, and phencyclidine (PCP) (PCP was not evaluated in this study). The cassette has built-in control lines on each membrane that indicate when the test is completed and valid (i.e., that the specimen has completed migration across all the test windows). This product requires three steps: (1) The oral fluid sample is collected from the donor using a foam collector. (2) To initiate a test, the collector is placed

into the sample well of the device and pushed down slowly to transfer the oral fluid from the collector into the test cassette. Oral fluid that is not absorbed by the test cassette-strips over- flows into a reservoir that can be sealed and sent to a laboratory for alternate testing (e.g., confirmation testing). (3) Test validity and drug-test results were read visually between 10 and 15 rain after initiating the test.

The Cozart system consists of an oral fluid collection swab, a disposable test cartridge, a handheld instrument that inter- prets and digitally displays the results, and an optional printer for a permanent record. The test kits are individually packaged and include a collection pad, transport tube, separator filter, and a test cassette that is inserted into the instrumented reader. This product requires eight steps: (1) The collection pad is placed in the donor's mouth for specimen collection until a blue indicator appears. (2) Transfer the collection pad into the transport tube. The transport tube contains a buffered solution that facilitates dissolution of the sample pad contents. (3) Cap the transport tube and tap tube to separate the collection pad from the stem. (4) Remove cap and stem from transport tube. (5) Insert the separator filter (similar to a serum separator) fully into the transport tube. (6) Transfer six drops of the fil- tered and buffered saliva into the cassette sample-well. (7) Wait approximately 2-30 s until the reagent appears on the test membrane. (8) Insert the cassette into the instrument and automatically read results after 12 rain. There is an internal control that indicates whether the test has been properly run. (Cozart provides an "oral fluid calibrator" cassette and recom- mends calibration right before test samples are run) Results are

Table I. QC Challenges and Drug Concentrations*

Drug Methamphetamine Opiates Benzodiazepines Cocaine Amphetamine THC THCA

Fortified Drug

Target Concentration

Solution Number

I 2

3

4 5

6

7

8 9

10

11

12

13

Methamphetamine Morph ine Temazepam

50 ng/mL 40 nglmL 5 nglmL

(ng/mL) (ng/mL) (ng/mL)

Neg (0) Neg (0) Neg (0)

Low (25) Low (20) Low (25)

Med (100) Med (80) Med (10)

High (500) High (400) High (50)

* Neg, negative conlrol; low, I/2x target conc.; medium, 2x larger conc.; and high, IOx target conc.

Cocaine Amphetamine THC THC COOH

20 ng/mL 50 ng/mL 4 ng/mL 4 ng/mL

(nglmL) (ng/mL) (nglmL) (nglmL)

Neg (0) Neg (0) Neg (0) Neg (0)

Low (10) Low (25)

Med (40) Med (100)

High (200) High (500)

Low (5)

Med (20)

High (100)

Low (10)

Med (100)

High (200)

431

displayed on an LCD screen and can be printed out on an in- tegrated printer. Two Cozart cassettes were tested: a five-panel multi-test cassette that includes amphetamine, cocaine, opiates, cannabinoids, and benzodiazepines and a single test cassette for methamphetamine.

Securetec's devices are individually packaged and each device contains a collection pad and a single test strip membrane. The removable cover contains a collection pad that is used to cob lect the specimen. The body of the device [detection element] contains a membrane with an attached absorbent fleece. The membrane contains the drug test and an internal control that indicates a valid test. This product requires four steps: (1) Specimen collection--in the standard protocol users are in- structed to wipe the test pad gently over the tongue. Securetec provided a modified protocol for testing with spiked saliva specimens specifying that 5 pL of the spiked specimen be pipetted directly onto the test pad area. This minimized the amount of spiked specimen required to conduct the test. (2) After collecting the specimen the collection pad cover is re- assembled onto the body of the device coming into direct con- tact with the drug-test membrane strip. (3) The absorbent fleece on the end of the membrane strip is dipped in tap water for 10 s which initiates flow across the membrane. (The absorbent fleece was touched to the side of a beaker to remove any excess water, and the testing device was placed horizontally on the laboratory bench while the analysis proceeded.) (4) Drug test and validity results were visually read after 10 rain (although results could often be read in 2 rain or less). Three single-test devices were evaluated: amphetamine, methamphetamine, and opiates.

The UltiMed test kit includes a foam collection pad, a saliva extraction tube and test cassette. The test cassette contains one membrane strip with five drug tests and an internal control.


This device requires three steps: (1) The foam collector is placed in the donors mouth for collection of the specimen. (2) The foam collector is then pushed into the extraction tube that is tapered at one end and is used to expel the saliva onto the test cassette. To initiate the test, three to four drops of the donor's saliva are applied to the sample well of the test cassette. (3) Read test results. Run time for the test is approximately 12 rain (2 rain for dissoMng reagents and 10 rain for chromatography). Test and test valid results are read visually and must be read within 10 to 20 rain of initiation of the test. A five- panel multi-test device was evaluated for methamphetamine, cocaine, opiates, and cannabinoids. The methadone test was not evaluated in this study.

Results

There are no nationally accepted standards or cutoff values for detecting drugs in oral fluids (either workplace or criminal justice) and, for most drugs, there are significant differences in cutoff values across devices (i.e., sensitivity to detect drug). Therefore, comparisons across devices became complicated. There were several ways to evaluate these devices: (1) evaluate performance against manufacturer's stated cutoff values; (2) evaluate performance against the proposed SAMHSA standards; or (3) evaluate the devices simply for practical use to detect recent drug use at the lowest concentration. These different ap- proaches to evaluating the devices produce very different re- suits. Therefore, we have attempted to evaluate the data in all three ways in order to provide the most comprehensive look at the overall performance characteristics of these devices.

Table II shows the low, medium, and high control concen-

Table II. Quality Conlrol Analysis by GC-MS

Drug Class Methamphetamine Opiates Cocaine Amphetamine Cannabinoids Cannabinoids Benzodiazepine Target Drug Methamphetamine Morphine Cocaine Amphetamine THC THC-COOH Temazepam

Units ng/mL ng/mL ng/mL ng/ml, ng/mL n~mL ng/mL

SAMHSA Target 50 40 20 50 4 4 NA*

Protocol Target 50 40 20 50 4 4 5

Negative 0 0 0 0 0 0 0

Assayed Negative Negative Negative Negative Negative Negative Negative

Low 25 20 10 25 5 10 2.5

Assayed 24.8 19.6 10.0 25.5 No Device 10.0 2.9

% Target 99,2% 98,0% 100,0% 102,0% Challenges 100.5% 116.0%*

Medium 1 O0 80 40 1 O0 20 1 O0 10

Assayed 102.1 83.1 42.0 102.4 19.3 89.8 10.9

% Target 102.1% 103.9 105.0% 102.4% 96.5% 89.8% t 109.0%

High 500 400 200 500 I00 200 50 Assayed 494,0 425,2 t 98,9 510.5 72.7 191,2 45.9

% Target 98,8% 106.3% 99.4% 102.1% 72.7% ~ 95.6% 91.8%

' Not applicable to SAMHSA guidelines. t Exceeded I0%.

432


trations and the target analytes selected for each drug class. The table also shows the assayed concentration of each control and its % of target by GC-MS analysis. The concentration of the low benzodiazepine control exceeded 10% difference from target. This control was accepted because the absolute error was 0.4 ng/mL and the device tested as expected. The concentration of the high THC control assayed at 72.7% of the target value, which clearly exceeded 10% from the target concentration and was taken into account when the devices were evaluated at this concentration. The concentration of the medium THC-COOH assayed at 89.8% of the target concentration, but given the cutoffs of the devices tested and the kit performance, this did not appear to affect the results.

Device performance The following tables and figures present the results for each

device by drug. Tables III-IX show the ability of the devices to

perform at the manufacturer's stated cutoffs. These tables show "expected response" (e.g., Positive or Negative) which is the an- ticipated response based on the manufacturer's cutoff value for their product. Results failing to conform to the manufacturer's stated cutoff value (either false-positive or false-negative re- suits) are highlighted. Figures 1-6 illustrate performance of the devices compared to the proposed SAMHSA cutoffs. The SAMSHA cutoffs are shown as a line in the figures, and the re- suits of each device above and below the cutoff are presented.

Amphetamine Three manufacturers' devices were evaluated for am-

phetamine performance. The manufacturer's cutoff concentrations for the devices were Ansys at 160 ng/mL, Cozart at 50 ng/mL, and Securetec at 100 ng/mL. Table III shows that, given the 160-ng/mL cutoff, the Ansys devices performed as expected at all control concentrations. Cozart's devices were able

Table III. Amphetamine Device Results

Ansys Cozart Securetec (160 nglmL) (50 ng/mL) (100 ng/mL)

Amphetamine Conc. Exp. Exp. Exp. Challenges ng/mL Resp TP TN FP FN Resp TP TN FP FN Resp TP TN FP FN

Negative 0 Neg 0 5 0 0 Neg 0 5 0 0 Neg 0 5 0 0 Low 2,5 Neg 0 10 0 0 Neg 0 5 �9 0 Neg 0 10 0 0 Medium 100 Neg 0 10 0 0 Pos 10 0 0 0 Pos 0 0 0 �9 High 500 Pos 10 0 0 0 Pos 10 0 0 0 Pos 10 0 0 0

Table IV. Methamphetamine Device Results

Ansys Cozart Securetec UltiMed (160 ng/mL) (50 ng/mC) (100 ng/mC) (50 ng/mt)

Methamphetamine Conc. Exp. Exp. Exp. Exp. Challenges n~/mL Resp TP TN FP FN Resp TP TN FP FN Resp TP TN FP FN Resp TP TN FP FN

Negative Low Medium High

0 Neg 0 5 0 0 Neg 0 5 0 0 Neg 0 5 0 0 Neg 0 5 0 0 25 Neg 0 I0 0 0 Neg 0 i0 0 i Neg 0 IO 0 i Neg 0 8 0 ~ 0

100 Neg 0 10 0 0 Pos 1 0 0 Pos 8 0 Pos 10 0 0 500 Pos 10 0 0 0 Pos 1 0 0 Pos 9 0 0 �9 Pos 10 0 0 0

Table V. Cocaine Device Results

Ansys Cozart UltiMed (20 rig/mr) (10 ng/mL) (30 ng/mt)

Cocaine Conc. Exp. Exp. Exp. Challenges ng/mL Resp TP TN FP FN Resp TP TN FP FN Resp TP TN FP FN

Negative Low Medium High

0 Neg 0 5 0 0 Neg 0 5 0 0 Neg 0 5 0 0 10 Neg 0 0 �9 0 Pos 0 0 0 �9 Neg 0 10 0 0 40 P o s t 0 0 0 0 Pos 0 0 0 1 P o s 3 0 0 ~0

200 Pos I0 0 0 0 Pos I 0 0 Pos I0 0 0

433


Table VI. THC Device Results

Ansys Cozart Cannabinoids (100 ng/mL) (200 ng/mL)

Ag.THC Conc. Exp. Exp. Challenges ng/mL Resp TP TN FP FN Resp TP TN FP

UltiMed (> 100 ng/mt)

Exp. FN Resp TP TN FP FN

Negative 0 Neg 0 5 0 0 Neg 0 5 0 Low 5 Neg -" -" -* -* Neg -* -* -* Medium 20 Neg 0 I0 0 0 Neg 0 I0 0 High I00 Pos 4 0 0 �9 Neg 0 I0 0

* Did not challenge device at low concentration.

0 Neg 0 5 0 0 -* Neg -* -* -* -* 0 Neg 0 I 0 0 0 0 Neg 0 I 0 0 0

Table VII. THC-COOH Device Results

Ansys Cozart UltiMed (6 ng/mL) (10 ng/mL) (15 ng/mL)

THC-COOH Conc. Exp. Exp. Challenges n~mL Resp TP TN FP FN Resp TP TN FP FN

Exp. Resp TP TN FP FN

Negative 0 Neg 0 5 0 0 Neg 0 5 0 0 Neg Low 10 Pos 10 0 0 0 Pos 5 0 0 �9 Neg Medium 100 Pos I0 0 0 0 Pos 10 0 0 0 Pos High 200 Pos I0 0 0 0 Pos I 0 0 0 0 Pos

0 5 0 0 0 1 9 0

10 0 0 0 10 0 0 0

Table VIII. Opiates Device Results

Ansys Cozart Securetec UltiMed (40 ng/mL) (10 ng/mr) (20 ng/mr) (30 ng/mC)

Opiate Conc. Exp. Exp. Exp. Exp. Challenges ngJmt Resp TP TN FP FN Resp TP TN FP FN Resp TP TN FP FN Resp TP TN FP FN

Negative 0 Neg 0 5 0 0 Neg 0 5 0 0 Neg 0 5 0 0 Neg 0 5 0 0 Low 20 Neg 0 10 0 0 Pos 0 0 0 �9 Pos 4 0 0 [ ] Neg 0 0 [ ] 0 Medium 80 Pos 8 0 0 ~ Pos 10 0 0 0 Pos 10 0 0 0 Pos 10 0 0 0 High 400 Pos 10 0 0 0 Pos 10 0 0 0 Pos 10 0 0 0 Pos 10 0 0 0

to discriminate positive from negative challenges. However, the devices had a 50% false-positive rate for the low control because its concentration of 25 ng/mL was below the Cozart cutoff. Securetec devices produced a false-negative rate of 100% at its cutoff of 100 ng/mL, but performed well at the high concentration challenge.

Figure I illustrates that all manufacturer's devices were able to discriminate between negative challenges and the high concentration, but were less effective around the cutoff.

Methamphetamine Four manufacturers' devices were evaluated for their ability

to detect methamphetamine. In general, the devices were able to correctly detect methamphetamine in the control challenges. However, because there were substantial differences in the manufacturers' cutoffs (Ansys at 160 ng/mL, Cozart at 50 ng/mL, Securetec at 100 ng/rnL, and UltiMed at 50 ng/mL),

434

Table IX. Benzodiazepine Device Results

Benzodiazepine Conc. Exp. Challenges (ng/mt) Resp TP

Cozart (5 ng/mL)

TN FP FN

Negative 0 Neg 0 10 0 0 Low 2.5 Neg 0 10 0 0 Medium | 0 Pos 0 0 0 10 High 50 Pos 10 0 0 0

the number of positive and negative results/control varied. Table IV shows that the Ansys, Securetec, and UltiMed devices performed well at detecting methamphetamine given their respective cutoffs. The errors highlighted in the table demonstrate that the Cozart methamphetamine devices failed


to detect 90% of challenges that contained drug at 2 and at 10 times their cutoff. The Securetec devices had a false-negative error rate of 10% for challenges greater than their cutoff. Ul- tiMed's devices had a 20% false-positive rate for challenges that contained drug at one-half the cutoff concentration.

Figure 2 compares performance of the devices around the proposed SAMHSA cutoff and illustrates that most of the devices did not produce false-positive results for the low (25 ng/mL) or negative controls. The Securetec and UltiMed devices successfully differentiated negative from 100 ng/mL and 500 ng/mL challenges. However, the Ansys device did not discriminate negative from positive challenges until the methamphetamine concentration was 500 ng/mL. The Cozart device produced false-negative results even on the A,~o~ IXmWv~ medium and high controls.

Cocaine Three manufacturers' devices were

evaluated for their ability to detect cocaine. Their cutoffs were as follows: Ansys at 20 ng/mL, Cozart at 10 ng/mL, and UltiMed at 30 ng/mL. Be- cause the cutoffs varied, performance on the control challenges also varied. The Ansys device performed error-free above its cutoff. However, all of the low challenges (10 ng/mL) tested positive, which is below their stated cutoff. The Cozart device performed poorly. It had a 100% false-negative rate at both the low (10 ng/mL) and medium (40 ng/mL) concentrations and a 90% false-negative rate at the high concentration. The UltiMed devices had a 70% false-negative rate with the medium control, but performed without errors at the high concentration (see Table V)

Figure 3 illustrates that the Ansys device was the most sensitive and produced the only consistently positive results near the SAMHSA cutoff.

THC Three manufacturers' devices were

evaluated for their ability to detect THC: Ansys at 100 ng/mL, Cozart at 200 ng/mL, and UltiMed at > 100 ng/mL. Table VI indicated that none of the devices was able to detect the positive challenges at 20 ng/mL. Ansys's devices detected 40% of the positive challenges at 100 ng/mL, but neither the Cozart devices nor the UltiMed devices detected any of these challenges. Athough the Cozart and UltiMed

devices performed in accordance with their specifications (i.e., all challenges were below their cutoff and all tested negative), they failed to detect any of the positive challenges. (Note: For THC, the devices were tested prior to GC-MS analysis being performed. The high concentration control analyzed at 72.7% of the target 100 ng/mL value.)

Figure 4 illustrates device performance compared to the SAMHSA proposed cutoff. Clearly, none of the devices have the sensitivity to detect THC at the proposed cutoff.

/ / , Controls

u~off- 100 ng/mL)

5O r~r~.) Oevk:, 0 n~n~.)

Figure I. Amphetamine result summary. (* Test device not available.)

Recorded positives

M # // ~, o

(Cutoff = ,50 ng/mL ) 'Cutoff = 100 ng/mL) r = 50 ng/mL) I)evice

16o ng/mL)

Figure 2. Methamphetamine result summary.

435


THC-COOH Devices from three manufacturers were evaluated for the

ability to detect THC-COOH in the control challenges. The device cutoffs were Ansys, 6 ng/mL; Cozart, 10 ng/mL; and UltiMed, 15 ng/mL. All three devices were able to differentiate positive from negative challenges at the low, medium, and high concentrations. As indicated in Table VII, the Ansys de-

o

Controls

Ix~81U'vu

Figure 3. Cocaine result summary. (* Test device not provided by Securetec.)

Recorded positives

/

10-

9-

8-

7.

6

S

4

3

vices performed error-free at all concentrations. The Cozart devices also performed as predicted given that one would expect that 50% of the samples would test positive at the cutoff. The UltiMed device performed as predicted for the medium and high challenges, and demonstrated the ability to discriminate between negative and the low concentration at 10 ng/mL. However, they tested positive for 90% of the challenges at 10

ng/mL, which is below their 15-ng/mL cutoff.

Figure 5 illustrates performance of the devices relative to the proposed SAMHSA cutoff for cannabinoids. Clearly, all devices demonstrate the ability to discriminate between negative and low concentrations of THC- COOH.

UlU I~1 (Cu~, , SO n ~ l . ) Icum~ �9 ~ (Cu~, , 10 r~/mL) Cu~, , 20 nO/mL)

C u ~ - lOO nO/ml,.)

Povlco �9 200 n,g/mL) r~mL)

Figure 4. Ag-THC (parent drug) result summary. (* Test device not provided by Securetec.)

Opiates All manufacturers were evaluated for

the ability to detect opiates. Although there were differences in cutoffs (Ansys at 40 ng/mL, Cozart at 10 ng/mL, Se- curetec at 20 ng/mL, and UltiMed at 30 ng/mL), they were not as pro- nounced as with some drugs. In general, all devices performed well at concentrations above the manufacturers' cutoff and most errant results were with challenges at, or near, the cutoff (see Table VIII). The Ansys devices had a 20% false-negative rate at the medium challenge concentration. The table shows that the Cozart, Securetec and UltiMed devices performed error- free at the medium and high concentrations. However, Cozart's devices had a 100% false-negative rate (i.e., missed all challenges) at the low concentration. The Securetec devices tested es- sentially as one would predict at their cutoff (40% positive/60% negative). The UltiMed devices demonstrated good sensitivity at low opiate concentrations, but failed to discriminate between their cutoff of 30 ng/mL and the low challenges at 20 ng/mL.

Figure 6 illustrates that all of the devices were able to accurately discriminate between negative and positive opiate challenges at both the medium and high concentrations and that both the UltiMed and Securetec devices performed well around the proposed 40- ng/mL SAMHSA cutoff value.

Benzodiazepines Cozart was the only manufacturer

that offered a benzodiazepine assay.

436


The Cozart cutoff is 5 ng/mL. The device was able to discriminate true negatives from positives only at the high concentration which was 10 times the stated cutoff value. Table IX shows a false-negative rate of 100% at twice the cutoff value. These results are not illustrated in a figure as SAMHSA has not stated a proposed cutoff for benzodiazepines.

nicians, they indicated that the "real cutoff value" was actually three times the stated value because the specimen is diluted with buffer during the testing process. In the case of the Cozart device, the claimed cutoff value apparently refers to the sensitivity of the assay to the amount of drug in the buffered, diluted specimen, not in the neat oral fluid. In the case of the Se-

Performance summary In order to provide some cross-

device comparisons, a summary of device and performance characteristics for the following five evaluation criteria is provided in Table X: (1) the lowest concentration at which the device was able to detect a positive challenge; (2) the various device configu- rations commercially available; (3) the number of technical steps required to complete a test; (4) whether the test is interpreted visually or by an instrument reader; and (5) an estimated cost per analyte.

Discussion

Overall, the performance of these rapid POC oral fluid drug-testing devices was quite variable. Some devices

Recorded posltlvu

curetec device, the undiluted oral fluid specimen is placed

L

Med (C.ton - 15 n ~ ) mmr

D.toll- lO ng/rnL) I)e~lce o11- e ng/mL)

performed well for the detection of some drugs, but poorly for the detection of others. No single device consistently out performed the other devices. In general, most of the devices performed well for the detection of methamphetamine and opiates, but poorly for the detection of THC. The ability to accurately and reliably detect cocaine, THC-COOH, and amphetamine varied with device.

A major difficulty in comparing the devices is that no widely accepted cutoff concentrations exist for identi- fying drugs of abuse in oral fluid, and there is a lack of consistency in cutoff concentrations across devices. We used the questionnaire data provided by the manufacturer to assess cutoff concentrations, target analytes, and cross-reactivity for each of the devices. In some cases it was difficult to assess the device manufacturer's cutoffs. For example, the package inserts for the Cozart devices indicate a specific cutoff value, but when the manufacturers' representatives went to the Center for Human Toxicology to train the tech-

Figure 5. THC-COOH (metabolite) result summary. (* Test device not provided by Securetec.)

Reeorr

Figure 6. Opiate result summary.

i Cooeols

u

/

r- SO nghnlJ -20 r,O~C)

437

directly on the membrane, and the tap water carrier used to initiate migration across the membrane does not appear to have a dilution effect.

The detection of marijuana use in oral fluid appears to be es- pecially difficult for these POC devices. Marijuana is the most widely abused and detected drug in the United States and for most testing applications (e.g., workplace, criminal justice, drugs/driving, etc.) the ability to detect recent marijuana use is critical. This evaluation clearly shows that these POC oral fluid testing devices are not capable of detecting ag-THC at concentrations that are meaningful. Based on available literature, the capability of detecting ag-THC at 100 ng/mL or greater would provide an extremely short window of detection (e.g., less than 1 h) (4). In the marketing materials for these devices, all of the manufacturers claim they can detect marijuana at low concentrations (e.g., 6-15 ng/mL). However, these claims appear to refer to the ability to detect THC-acid (THC-COOH), which is the principal metabolite of ag-THC. Our evaluation has demonstrated that, indeed, these devices can detect the THC-acid, which is an important factor in detecting marijuana use in urine, but this is particularly misleading as the literature indicates that the THC-acid metabolite is rarely ever detected in the oral fluid of a marijuana smoker (1,4). We believe the devices evaluated were ineffective in detecting THC at concentrations expected following recreational marijuana use for two reasons: the device manufacturers appear to be targeting the wrong an- aIyte and the cutoff concentrations are too high considering the low concentrations of THC generally present in oral fluid.

Overall, the performance data demonstrates that the rapid POC oral fluid devices evaluated in this study are probably not suitable for general workplace, criminal justice, drugged driver, or any testing programs where marijuana is the primary drug of interest. The devices did perform well, however, in detecting opiates and methamphetamine, and, for specialized programs


where one or both of these classes of drugs are the primary interest, rapid POC testing may be suitable. For programs where cocaine and amphetamine are of interest, some devices may be suitable and others are not.

Conclusions

The state of the art in oral fluid testing is evolving rapidly. There have been significant improvements over the last five years, and new methods and devices are currently in development. Three of the manufacturers we contacted indicated that within the next 3--6 months they will have new products on the market with increased sensitivity and specificity. We have also become aware of other major diagnostic manufacturers that will enter the oral fluid market within the next year. The search for a marijuana assay that can provide a reasonable window of detection appears to be the major hurdle for all of the device manufacturers. In our opinion, until a more sensitive marijuana assay is developed, testing oral fluids for illegal drugs will have limited use. The results of this evaluation strongly suggest that, at the current state-of-the-art, the manufacturers have overstated the capabilities of their POCT products to detect drugs in oral fluids. However, with the current focus of technology development in oral fluid testing, we believe there is every reason to be optimistic about the future for drug testing using the oral fluid matrix.

Acknowledgments

This research was sponsored by the Office of National Drug Control Policy, Counterdrug Technology Assessment Center, through a contract with the U.S. Army Electronic Proving

Table X. Summary of Device Characteristics

Manufacturer Ansys Cozarl Securetec UltiMed

Rapiscan Device

Evaluation Criteria

Lowest Detected Concentration

Testing Device Panel

Number of steps required for testing

Visual or Inst. Interpretation

Cost/Analyte

Oral lab

Meth Opi Coc Amp THC THCA Melh

soo no ~o soo ~oo m ~oo

6 Panel I. Meth/Opi/Coc/Amp/THC/ 1.

PCP 2.

3.

4.

5.

Opi Coc Amp THC THCA BDP

80 200 25 ND' 10 50

1, 2 and 5 Panel Methamphetamine Methadone/Opi Opi/Coc Opi/Coc/Amp/THC/BDP Opi/Coc/Amp/BDP/ Methadone

Drug Wipe

Melh Opi Coc Amp THC

loo 20 NP soo NP

1 and 2 Panel 1. All Drugs Single 2. Opi/Coc 3. Amp/THC

SalivaScreen

Melh Opi Coc Amp THC THCA

25 20 40 NT ND 10

5 Panel 1. Meth/Opi/Coc/lHC/

Methadone

3 Steps

Visual

$ 4.00/Analyte

8 Steps

Instrumental

$ 4.00/Analyte

3 Steps

Visual

$ 7.00/Analyte

3 Steps

Visual/Instrumental

$ 2.50/Analyte

* Abbreviations: ND, none detected; NP, test available but not provided by manufacturer; and NT, no test available.

438


Ground, Contract number N66001-01-C-6028. The views and conclusions are the authors' and not those of the government. The authors would also like to acknowledge the technical con- tribution of Dr. David Kidwell of Naval Research Laboratory, Washington, D.C.

References

1. M.A. Huestis and E.J. Cone. Alternative testing matrices. In Drug Abuse Handbook, S. Karch, Ed. CRC Press, Boca Raton, FL, 1998, pp 799-857.

2. E.J. Cone. Saliva testing for drugs of abuse. Ann. N.Y. Acad. Sci. 694:91-127 (1993).

3. E.J. Cone, J.M. Oyler, and W.D. Darwin. Cocaine disposition in saliva following intravenous, intranasal, and smoked administration. J. Anal. Toxicol. 21:465-475 (1997).

4. D. Malamud and L. Tabak, Eds. Saliva as a diagnostic fluid. Con- ference proceedings. Ann. N.Y. Acad. 5ci. 694:1-348 (1993).

5. R Kintz. The role of saliva and sweat in detecting cases of driving under influence. In Road Traffic and Drugs. Council of Europe Publishing, Strasbourg, France, 1999, pp 239-245.

6. N. Samyn, A. Verstraete, C. Van Haeren, and P. Kintz. Analysis of drugs of abuse in saliva. Forensic Sci. Rev. 11: 1-19 (1999).

7. R.S. Niedbala, K.W. Kardos, F.F. Dean, S. Kardos, F. Tiffany, J. Waga, J. Robb, and E.J. Cone. Detection of marijuana use by oral fluid and urine analysis following single-dose administration of

smoked and oral marijuana. J. Anal. ToxicoL 25:289-303 (2001). 8. C.L O'Neal, D.J. Crouch, D.E. Rollins, and A.A. Fatah. The effects

of collection methods on oral fluids codeine concentrations. J. Anal. ToxicoL 24:536--542 (2000).

9. A. Verstraete. ROSITA: Roadside Testing and Assessment. ROSITA Contract No. DGVII PL98-3032. Ghent University, Ghent, Bel- gium, 2000, pp 1-397.

10. N. Samyn, G. DeBoeck, and A.G. Verstraete. The use of oral fluid and sweat wipes for the detection of drugs of abuse in drivers. J. Forensic Sci. 47(6): 1-8 (2002).

11. P. Kintz, V. Cirimele, and B. Ludes. Codeine testing in sweat and saliva with the Drugwipe. Int. J. Legal Meal. 111:82-84 (1998).

12. N. Samyn and C. Van Haeren. On-site testing of saliva and sweat with Drugwipe and determination of concentrations of drugs of abuse in saliva, plasma and urine of suspected users. Int. J. Legal Med. 113:150-154 (2000).

13. C. Barrett, C. Good, and C. Moore. Comparison of point-of-collection screening of drugs of abuse in oral fluid with a laboratory based urine screen. Forensic Sci. Int. 122:163-166 (2001).

14. V. Spiehler, D. Baldwin, and C. Hand. Analysis of drugs in saliva. In On-site Drug Testing, A. Jenkins, Ed. Humana Press, Totowa, N J, 2002, pp 95-108.

1S. Substance Abuse and Mental Health Services Administration, Proposed Guidelines For Mandatory Drug Testing. http://www. drugfreeworkplace.gov/frames/frame_drugtest.htm. Accessed March 2001.

16. D. Crouch. Evaluation of saliva as an alternative drug-testing specimen. Final Report. National Institute of Standards and Tech- nology. Grant #70NANB7H003. Gaithersburg, MD, May 2000, pp 1-44.

439

an evaluation of rapid point-of-collection oral fluid … · devices were evaluated (ora[lab |...

Documents