Development and Evaluation of Immunochromatographic Rapid Tests ...

22 downloads 533 Views 3MB Size Report
Rapid test development ..... RIA assay was shown to be due to a calibration bias of the RIA ... FPIA (y = 6.5 + 2.17x; r = 0.90; RIA versus FPIA with 19 urine.
Journal of Analytical ioxk:ology, Vol. 22, March/April 1998

Development and Evaluationof Immunochromatographic Rapid Testsfor Screening of Cannabinoids,Cocaine, and Opiates in Urine R. Wennig 1, M.R. Moeller 2, J.M. Haguenoer ~, A. Marocchi 4, E Zoppi 4, B.L. Smith s, R. de la Torre 6, C.A. Carstensen 7, A. Goerlach-Graw 7, J. Schaeffler 7, and R. Leinbergerh* I Laboratoire National de Sant6, Centre Universitaire, 162A, av. de la Fafencerie, L- 1511 Luxembourg; 21nstitutf[ir Rechtsmedizin, Universitaetsklinilg 66424 Homburg, Germany; ~Laboratoirede Toxicologie, Place de Verdun, 59045 Lille, France; 40spedale Niguarda, Piazza Ospedale Maggiore 3, 20162 Milano, Italic ~The Bethlem Royal Hospital, Beckenham, Kent BR3 3BX, Great Britain; ~'lnstitut Municipal d'lnvestigaciO M~dica, Dr. Aiguader 80, 08003 Barcelona, Spain; and ZBoehringerMannheim GmbH, 68298 Mannheim, Germany

Abstract The test principle and the optimization of the reactive ingredients are described for the one-step dip-and-read immunochromatographic ERONTLINE ~ rapid tests for drugs-of-abuse testing in urine samples. In a multicenter evaluation the rapid tests were compared with EPIA and EMIT immunoassays. Discrepant results were further analyzed by gas chromatography-mass spectrometry methods. In the comparison of the cannabinoids rapid tests versus both immunoassays using clinical and forensic urine samples (399 versus FPIA and 755 versus EMIT), sensitivities and specificities were 97% or better for both comparisons. For cocaine, a sensitivity of 100"/o versus both routine technologies was obtained, whereas the specificity was reduced somewhat to 91% because of some cross-reactivity with metabolltes of methadone and of clozapine. Specificity was very high for the cocaine rapid tests (98-100%) when applied to urine samples of persons not in a methadone maintenance program. Sensitivities and specificities for the opiates rapid tests were 99% or better at all sites when compared with the routine methods. In the screening of about 1200 clinical urine samples for cannabinoids, cocaine or opiates misuse only six samples would have stayed undetected by rapid test analyzes. These results show the FRONTLINE assays allow a reliable and immediate screening for drugs of abuse.

(5), particulary in the criminal justice system, the law enforcement, hospital emergency (12) and in monitoring of patient compliance in the doctors office. The test optimization of a reagent-carrier immunochromatographic technology (FRONTLINErapid tests) for the detection of drugs of abuse in urine is described, and the degree of agreement with FPIAand EMIT assays is investigated in a multicenter evaluation.

Material and Methods

Test principle The test principle of the FRONTLINE rapid tests (Figures 1 and 2) is identical for all drugs (cannabinoids, cocaine, opiates, and amphetamines are available from Boehringer Mannheim GmbH, Germany; a benzodiazepines test is in development) and is based on tile gold-labeled immunochromatography technology. The rapid tests consist of a carrier foil as support with a number of fleece compartments. Urine is drawn along the rapid test by capillary action. Analyte reacts with a cover foil

ii

immersion mark

color identification

carrier foil

I

/"

J"

i

Introduction It has become impossible to ignore the drug problems in the socio-political arena. Hence, it was not surprising that several companies used their experience in carrier-bound reagent technology (e.g., Reflotron| clinical chemistry system [1,2]), in dry reagent immunology test systems (e.g., cardiac marker troponin T [3}) and in dry chemistry coagulation testing (4) to provide simple tools for decentralized drug screening (5-11). Onsite testing for drugs of abuse has increased in popularity *AddressI(Ir com.t,p~,~derlce:Robert[einbergel,BoehrhlgerMannlwin~Gnlt)tl, Sandhofer Sil,tst,e I 16, [J-(,l129[IM,.1.hr (](!lf~lall~/',

148

/

absorbent conjugate fleece capture matrix detection field fleece antibody-gold-conjugate immobilized analyle color sample gold W W drug ~ ! drug drug ยง ~,- gold drug + T " gold

gold

drug ~

gold !

drug ~ _ l +

drug y gold

~ gold

4. gold drug . ~ |

Figure 1. Test principle of FRONTLINE rapid tests.

Reproduction (photocopying) of editorial content of this journal is prohibited wilhout publisher's i)ermission.

Journal of Analytical loxicology, Vol. 22, March/April 1998

conjugate comprised of anti-analyte antibody attatched to intensely red-colored gold particles. Excess conjugate is removed by passage through a specific immobilized binding agent ("polyhapten"). The amount of analyte in the original sample is therefore proportional to the intensity of the red-colored analyte-bound conjugate reaching the detection zone. Detection of the reaction color is performed by visual comparison to a color scale.

Components Each of the various components has different properties optimized to meet the analytical requirements of the rapid test. The polyester absorbant fleece rapidly absorbs a uniform volume of urine, providing a reservoir for the rest of the upstream rapid test components. It obviously must not bind to the test analyte. The conjugate fleece is made of polyester, which has a

3

. . . .

homogenous, loosely bound, coating of conjugate, which is completely eluted by the passage of the test urine. The conjugate, which is water soluble and intensely red, consists of gold particles onto which a number of anti-analyte antibody molecules are firmly bound. The bound antibody must react very quickly with the test analyte, reaching a sufficient degree of reaction during the passage of the urine sample (20--40 s). The capture zone (or capture matrix) consists of a cellulose matrix on which all the protein-binding sites are irreversibly occupied by streptavidin. A biotinylated carrier protein onto which an analogue of the test analyte is bound (polyhapten) is fixed to the streptavidin. The detection zone consists of a cellulose fleece that has high capillary activity for easy and complete filling. The reaction color must spread homogenously.

Test reaction sequence The test analyte reacts with the antibody moieties on the conjugate. If the analyte concentration in the urine sample is low, most of the antibody sites will not have reacted with analyte. These binding sites will thus be available to bind to the immobilized analogue of the analyte (the polyhapten) and thus be captured when the conjugate enters the capture zone, leaving little conjugate to enter the detection zone. If the analyte concentration is high, most of the antibody binding sites will be blocked by test analyte. This conjugate will not ~: bind to the polyhapten in the capture zone and 9 will pass through to the detection zone where it is detected visually by its red color. Test instruction

The rapid tests have to be dipped into the urine sample for 3-5 s; the liquid level is between the two marks on the cover foil. The rapid tests are placed on a horizontal surface. The test results can be read visually after 2 min. The reaction colors are estimated by comparison with a color scale on the vial that contains the rapid tests. The color results indicate a negative (white to apricot color), a single-positive (red), or a doublepositive (intense red) color block. Threshold limits for a single (double) positive test result were greater than or equal to 50 ng/mL (_>200 ng/mL) for cannabinoids, greater than or equal to 300 ng/mL (>_3000 ng/mL) for benzoylecgonine (BZE), and greater than or equal to 200 ng/mL (>_1000 ng/mL) for opiates (stated by the manufacturer). Intermediate reaction colors have to be allocated to the next lower color block. Reaction colors have been shown to be stable for at least 10 rain (internal data).

Figure 2. Top view of FRONTLINE rapid tests.

Rapid test development Optimization of test components. Different quantities of gold conjugate in the conjugate fleece, antibody on the gold conjugate, polyhapten 149

journal of Analytical Toxicology, Vol. 22, March/April 1998

carrier protein in the capture zone, or haptens coupled to the carrier protein were used to optimize performance with respect to true- and false-negative results, cutoff concentrations and true- and false-positiveresults. Different test variants were primarily tested with solutions of the drug in phosphate buffer (ptt 7.4, 10mM phosphate, Boehringer Mannheim ID No. 773921). Quantitation of the reaction color was achieved with a Minolta CR-321 ReflectanceChromameter. The intensity of the reaction color is directly related to the Chromameter reading. The color system used to quantitate absolute chromaticity was the L*C*H~system according to the Commission Internationale de l'Eclairage 1976 with C* = metric chroma or C-value. The higher the C-value, the more intense is the reaction color. By empirical results, a C-value above seven units can be clearly recognized as a red color by most individuals. The rapid tests were calibrated using 50-ng/mL ll-nor-A9-tetrahydrocannabinol-carboxylic acid (THCCOOH), 300-ng/mL BZE, or 200-ng/mL morphine. Cross-reactMties. Compounds that might be expected to give a positive reaction were added to drug-free urine samples. Minimum concentrations at which a positive reading was observed are given in Table I. Each spiked urine sample was tested twice by each of three individuals operating blind and independently. Effect of low urhw temperatures. The reaction colors of opiates and cocaine rapid tests were examined at urine temperatures of 20, 10, and 4~ The reaction colors were evaluated by 10 operators using drug concentrations around the cutoff. Multicenter evaluation The multicenter evaluation was performed at six European sites each following tile same protocol. Familiarization. In a familiarization period, all personnel (lab technicians) had the opportunity to become experienced with the allocation of the reaction colors to the comparison color scale in the different laboratories. For this purpose, 10 urine samples were examined in comparison with the routine method. One of the sites did not initially allocate values to the next lowest color but to the nearest color, which caused a relatively high incidence of incorrect results. Correct color comparison was re-demonstrated at this site, and the incidence of blind false results was reduced. The wording on the test instructions was duly amended. Method comparison. At each site, about 200 fresh urine specimens from the daily workload were determined with each of the three rapid tests and also by their standard procedures. If necessary, known drug-containing clinical urine specimens were additionally examined to achieve at least 50 positive specimens per evaluated parameter. The individuals who read the rapid tests were unaware of the routine method results. The comparison methods, enzyme-multiplied immunoassay technique (EMITd.a.u., Behringwerke Marburg, Germany) (three sites), fluorescence polarization immunoassay (FPIA) on TDx instruments (Abbott Laboratories, Abbott Park, IL) (two sites), and radioimmunoassay (DPC, Rueil Malmaison, France) (one site), were performed according to the instructions of the manufacturers.

150

Statistical evaluation Urine samples with discrepant results between rapid tests and routine FPIAor EMITassays were re-analyzed by gas chromatography-mass spectrometry (GC-MS) method (13), which was used as the reference method. Sensitivities and specificities were calculated in relation to FPIA and EMITassay for each individual site and for the compiled results. The comparison methods were assigned positive when greater than or equal to 50 ng/mL for cannabinoids, 300 ng/mL for BZE, or 200 ng/mL for opiates. Comparisons of rapid test results versus EMIT results were assessed in the form of 3 x 3 field contigency tables. The percentage agreement (concordance) between rapid tests and EMITwas calculated by summarization of the three fieldswith concordant results in relation to all results. The method 3O

l..~,J 25

/

20

I--

O O

I

I

I

200 400 600 800 Concentration of analytc (ng/ml.)

1000

Figure 3. Influence of conjugate on test performance of FRONTLINE Cannabis. Analyte: THCCOOH. Low (OD 5), , ; medium (OD 10), A; high (OD 14), 9 concentration of gold conjugate. Low (60 pg/mL), O; medium (90 pg/mL), ,~; high (120 pg/mL), [Z], amount of antibody on gold conjugate.

25 1

0

I

0

~---

I

I

I

1000 2000 3000 4000 Concentration of analyte (ng/mL)

5000

Figure 4. Influence of capture zone on test performance of FRONTLINE Cocaine. Analyte: BZE. Low (3 pg/cm 2) 0, medium (5 pg/cm2), A; high (7 pg/cm2), 9 concentration of polyhapten. Low (5:1), O and high (10:1 ), [] degree of coupling on polyhapten.

Journal of Analyticalloxicology, Vol. 22, March/April 1998

comparison results of rapid tests versus the quantitative FPIAassay were evaluated as double histogram figures.

Table I, Compounds Tested that Gave a Positive Result at the Concentration Shown* Test

Compound

FRONTLINE~

11-Nor-z~'LTHCCOOH 11-Nor-AILTHCCOOH 11-Hydroxy-k%THC A%THC &~LTHC 11-Nor-A%lHCCOOH-~,-D-Glucuronide Cannabinol Cannabidiol

Cannabis

FRONTLtNE|

Thresholdconcentration

Benzoyiecgonine Cocaine

Cocaine

50 ng/mL (Calibrant) 50 ng/mL 500 ng/mL 1000 ng/mL 7(]0 ng/mL 250 ng/mL 1000 ng/mL >100 ~lg/mL

300 ng/mL (Calibranl) 100 ng/mL

Ecgonine Ecgoninemethylester

100 pg/mL >100 bJg/mL

FRONTLINE|

Morphine

Opiates

Morphine-3-/3-p-glucuronide Morphine-6-~,-i>glucuronide Codeine Ethylrnorl)hine Heroin Dihydrocodeine Iqydromorphone

200 ng/mL (Calibranl) 300 ngJmL 300 ng/mL 200 ng/mL 200 ng/mL 300 ng/mL 200 ng/mL 500 n~mL

Hydrocodone Oxycodone

300 ng/mL > 1(),000 ng/mL

N-Norcodeine t.evorphanol Nalorphine Thel)aine

20,000 ng/mL 2500 ng/mL 20,000 ngh'nL 300 ng/mL

" Conlpounds in solution in blank urine,

Table II. Multicenter Evaluation of FRONT/INE Tests: Results of the Individual Sites Site 1

Parameter Cannabinoids

Comparison method

Numberof samples

Specificity

Sensitivity

(%)

(%)

Cocaine Opiates

EMIT DAU EMIT DAU EMIT DAU

374 385 323

97 94 100

98 100 99

Cannabinoids Cocaine Opiates

EMITST EMIT ST EMIT ST

120 97 102

1(]0 98 100

I(]0 100 100

Cannabinoids Cocaine

Opiates

RIA RIA RIA

2O3 23.5 231

99 100 92

92 100 100

Cannabinoids Cocaine Opiates

FPIA FPIA FP]A

284 289 286

98 99 99

98 1[)0 100

Cannabinoids Cocaine Opiates

EMIT DAU EMIT DAU EMIT DAU

261 264 253

96 85 99

100 100 I t')O

Cannabinoids Cocaine Opiates

FPIA FPIA FPIA

115 223 223

100 99 99

99 100 100

Results

Rapid tests optimization The examples in Figures 3 and 4 show the relationship between color intensity as measured by the chromameter and the analyte concentration. For cannabinoids (Figure 3), the cutoff concentration should be at 50-ng/tJLTHCCOOHand require a C value of at least seven units. For cocaine (Figure 4), this concentration should be at 300-ng/IJL BZE. Variables investigated were conjugate fleece, which is the quantity of gold conjugate particles per strip and the density of antibody (IJg/mL) coating these particles, and capture zone, which is the amount of protein polyhapten bound to the cellulose matrix (IJg polyhapten/cm 2) and the relative number of drug analogue haptens coupled to each protein molecule (e.g., degree of coupling 5:1 or 10:1). Conjugate fleece This is prepared by saturating the polyester fleece with a solution of gold conjugate, draining off the excess, and then drying the fleece. The quantity of gold conjugate in the fleece is dependent on the concentration of the saturating solution. This concentration can be estimated by its optical density (OD) at 520 nm (e.g., OD 5, 10, 14) representing increasing quantities of gold conjugate in the dried fleece. As shown in Figure 3, analyte sensitivity (color difference between cutoff and high analyte concentration) increases with increasing amounts of gold conjugate (OD) on the rapid test. For THCCOOH, they were 600 ng/1JLat OD 5, 100 ng/IJL at OD 10, and 50 ng/IJLat OD 14. Equally important for analyte sensitivity is the density of the coating of the antibody (l~g Ab/OD) on the gold particles. A minimum antibody density is necessary to prevent false negatives with C-values higher than 7. With increasing amounts of antibody, the dynamic range of the curves is optimized. An easily distinguished C-value of 7 at the desired cutoff concentration of 50 ng/lJL THCCOOH is best achieved with a gold conjugate OD of 12 and an antibody density of 90 IJg/mL. Capture zone

If the polyhapten concentration was zero, all the conjugate, both analyte bound and unbound,

151

Journal of Analytical loxicology, Vol. 22, March/April 1998

would reach the detection zone. A polyhapten concentration of 3 I~g/cm'~ is just sufficient for correct negative results; 5 or even 7 IJg/cm2 gives better reliability for distinguishing correct negatives but a decreased analyte sensitivity. The resulting cutoff concentrations for BZE for the first visible colors are 100,200, and 400 ng/IJL respectively. The degree of coupling has an even greater impact on the test performance. Five moles of hapten per mole of carrier protein gives a cutoff of 250 ng/IJL BZE, whereas a ratio of 10:1 gives a cutoff of 500 ng/l~L. The optimum combination for BZE was found to be 5.2 lJg/cm 2 and a ratio of 5:]. Each of these factors have to be optimized in relation to each other to give the desired test performance, that is, no false positives and an easily visible cutoff color. Table I shows the cutoff concentrations for the test analytes and for some related compounds using the optimized rapid test.

Multicenter evaluation

Method comparison: results of hTdividual sites. Table II summarizes results obtained in each evaluation center. For cannabinoid rapid tests, calculated sensitivity was greater than 98% at all sites but one. The lower sensitivity found versus the RIA assay was shown to be due to a calibration bias of the RIA method when compared with other immunoassays such as FPIA (y = 6.5 + 2.17x; r = 0.90; RIA versus FPIA with 19 urine samples). GC-MS for THCCOOH was not available at this site. Specificities were equal or greater than 96% at all sites. With cocaine rapid tests, a 100% sensitivity was obtained at all sites. The outcomes for rapid test specificities varied from site to site. At sites 2, 3, 4, and 6, specificities were between 98 and 100%, whereas at sites 1 and 5, 94 and 85%, respectively, were obtained. There was a very good reproducibility of results at all sites for the opiates determination. Sensitivities and specificities were equal or greater than 99%, with the exception of site 3 where

Table III. Summarized FRONTLINE Tests Results versus EMIT Results: 3 x 3 Field Contingency Tables

FRONTLINE Cannabis

0-49

*discrepant results

double positive >200 ng,'mL

2*

43

63

n = 755 urine samples

positive >50 ng/mL

12*

134

3

Sensitivity 99% Specificity 97%

negative

496

2*

0

Agreement 92%

0-299

BZE (EMIT) (ng/mL) 300-2999

>3000

12'

22

37

n = 746 urine sarnples

positive >300 ng/mL

44*

29

11

Sensitivity 100% Specificity 91%

negative

591

0

0

Agreement 88%

FRONTLINE Cocaine double positive >3000 ng/mL

FRONTLINE Opiates

0-199

Opiates(EMIT)(ng/mL) 200-1000 >1000

*discrepant results

*discrepant results

1"

45

121

n = 728 urine samples

positive >200 ng/mL

0

35

43

Sensitivity 99% Specificity 99%

negative

482

0

1

Agreement 88%

double positive >1000 ng/mL

152

THCCOOH (EMIT) (ng/mL) 50-199 >200

Journal of Analytical Toxicology, Vol. 22, March/April 1998

by FPIAmethod. Positive results with the rapid tests are shown a calibration bias of the RIA assay was shown (y = 16.8 + 0.45x; in the upper field; negative results are in the lower field. r = 0.98; RIA versus FPIA with 10 urine samples). For cannabinoids determination (395 specimens), 4 of the 10 Method comparison: results versus EMIT assay. The comoriginal discrepant results FPIAresults were shown to be incorpiled FRONTLINEtest results versus the EMIT assay are sumrect by GC-MS confirmation analysis. Specificity and sensimarized in Table Ill. tivity were 99 and 98%, respectively, for the remaining six Some of the original EMIT results were discrepant with the discrepant rapid test results rapid test results, and, after analysis by GC-MS, the EMIT For cocaine method comparison, only five discrepant results results where shown to be incorrect for 28 cannabinoids results out of 512 specimens were observed. All five specimens were (26 EMIT false-negative results for urine samples up to 450 confirmed as rapid test false positives by GC-MS analysis; one ng/mL according to GC-MS), 3 cocaine results (2 EMIT false specimen containing clozapine and four samples having BZE negatives), and 5 opiates results (4 EMIT false negatives). Three concentration, between 200 and 270 ng/mL by GC-MS were out of 755 urine samples showed a very dark color and were not therefore close to the 300-ng/mL threshold limit of the rapid suitable for the analysis by either rapid test or EMIT. tests. Four FPIA results were shown to be incorrect by GC-MS For a total number of 245 cannabinoid EMIT positive samples (50-ng/mL cutoff), 243 single- or doublepositive rapid test results and only two false FRONTUNE Cannabis negatives were found, which gave a sensitivity 108 j n = 399 urine samples threshold limit of 99%. Quantitation of the two false negatives (50 ng/mL) Sensitivity 98% by FPIAassay gave results of 69 and 75 ng/mL. Specificity was 97%. Of the 510 EMIT negaSpecificity 99% 2' 1M [][L ,~11. tives, only 14 samples were false positive with 1" 1" 1" the rapid tests. Agreement in the nine-field comparison table for cannabinoids was 92% with some urine samples assigned double pos"discrepant results itive instead of single positive by the opera87 140 25-49 50-74 ,'5-100 101-150 > 150 5-25