Forensic Science International 119 (2001) 2327

A survey of extraction techniques for drugs of abuse in urine

J.F. Wilson*, B.L. Smith, P.A. Toseland, I.D. Watson, J. Williams, A.H. Thomson, N.E. Capps, G. Sweeney, L.N. Sandle
Received 4 August 2000; received in revised form 4 September 2000; accepted 6 September 2000 Steering Committee for the United Kingdom National External Quality Assessment Scheme for Drug Assays

Abstract Sixty nine participants in the United Kingdom national external quality assessment scheme for drugs of abuse in urine reported details of their sample extraction technique by questionnaire. Laboratories were categorised by differences in technique and their analytical test results compared for samples containing D-amfetamine 0.4 (4) and 0.8 (3) mg/l, morphine 0.4 (4) and 0.8 (4) mg/l, and benzoylecgonine 0.15/0.2 (2) and 0.45/0.5 (4) mg/l. Values in parentheses are numbers of samples. For amfetamine, there was no signicant difference in the frequency of true positive results between liquidliquid or solid phase extraction and the Toxi-Lab A system at 0.8 mg/l. Toxi-Lab A gave signicantly fewer positives when operating below its specied threshold at 0.4 mg/l. Paradoxically, laboratories using >5 ml urine volume performed less well. Acidication of the extract before volume reduction gave signicantly more true positives. For extraction of morphine, solid phase systems signicantly outperformed both liquidliquid and the Toxi-Lab A system at both 0.8 and 0.4 mg/l. No signicant differences between extraction techniques were demonstrated for analysis of benzoylecgonine. # 2001 Elsevier Science Ireland Ltd. All rights reserved.
Keywords: Amfetamine; Morphine; Cocaine; Liquidliquid extraction; Solid phase extraction

1. Introduction The analytical strategy generally employed for drugs of abuse testing in urine is a two-stage process [1]. First, a screening test is used to differentiate between negative and positive samples. Screening tests may be immunoassays or one of a number of chromatographic procedures that include thin-layer (TLC), high-performance liquid (HPLC) or gas chromatography (GC). Results of positive screening tests are conrmed by a second technique based on a different physico-chemical principle at least as sensitive as the screening test. The preferred conrmatory techniques are either gas or liquid chromatography with mass spectrometry (GCMS, LCMS).

Corresponding author. Department of Pharmacology, Therapeutics & Toxicology, University of Wales College of Medicine, Heath Park, Cardiff CF14 4XN, UK. Tel.: 44-29-2074-2068; fax: 44-29-2074-8316. E-mail address: wilsonjf@cardiff.ac.uk (J.F. Wilson).

Since the sensitivity of available immunoassays is generally higher than of chromatographic procedures, it is normal to purify and concentrate the urine sample by extraction before chromatography. There are two types of extraction system in use, liquidliquid and solid phase [2]. They extract drugs by virtue of differing aspects of their physico-chemical properties. A commercial variant of a liquidliquid system called the Toxi-Tube1 forms part of the Toxi-Lab1 TLC system (Microgen Bioproducts Ltd., Camberley, UK). Whilst much is known regarding the relative merits of the various chromatographic and immunoassay detection systems, the contribution of the extraction procedure has received little attention. We1 report an evaluation of extraction systems for amfetamine, morphine and benzoylecgonine based on the performance of laboratories participating in the United Kingdom National External Quality Assessment Scheme (UKNEQAS) for drugs of abuse in urine.
1 Steering Committee for the United Kingdom National External Quality Assessment Scheme for Drug Assays.

0379-0738/01/$ see front matter # 2001 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 3 7 9 - 0 7 3 8 ( 0 0 ) 0 0 3 8 3 - 2

24

J.F. Wilson et al. / Forensic Science International 119 (2001) 2327

2. Materials and methods A questionnaire requesting details of the extraction technique(s) used for drugs of abuse analysis was circulated to members of the UKNEQAS for drugs of abuse in urine in November 1997. Laboratories were asked to complete a separate questionnaire for each different type of extraction procedure used. Details were requested of sample volume, internal standards, hydrolysis and other pre-extraction procedures, of the extraction, back extraction, concentration and derivatisation steps, of the detection technique and the lower limit of detection of the overall procedure. Samples for drug analysis were 25 ml freeze-dried aliquots of drug-free urine spiked by quantitative addition of drugs including D-amfetamine 0.4 (4) and 0.8 (3) mg/l, morphine 0.4 (4) and 0.8 (4) mg/l, and benzoylecgonine 0.15/0.2 (2) and 0.45/0.5 (4) mg/l [4]. In parentheses is the number of samples distributed containing each drug concentration. The drug concentrations were chosen to probe performance in the range between the EU workplace testing thresholds [3] of 0.2, 0.2 and 0.15 mg/l for amfetamine, morphine and benzoylecgonine, respectively, and the UKNEQAS clinical thresholds [4] of 1.0, 1.0 and 0.3 mg/ l, respectively. In addition, data from samples negative for the study analytes and structurally related compounds were evaluated to check for possible false positive results. The number of negative samples were ve amfetamine, nine morphine and 14 benzoylecgonine. Samples included in the study were circulated between February 1997 and August 1998 by post (airmail overseas). Laboratories were requested to reconstitute samples by addition of 25 ml distilled water and to analyse the samples immediately using their standard protocols. Approximately 1 month was allowed from the date of sample shipment for return of analytical data. Laboratories made separate reports on the one or more tests used in analysis of each sample. For each test, they gave a quantitative result or a report from the range positive, negative and uncertain. Only results from the same chromatographic test as reported in the questionnaire were considered. Laboratories were classied according to details of their extraction technique and the performance of sub-groups compared by chi-square analysis of the frequency of positive and of negative plus uncertain test results for each drug concentration. 3. Results A total of 187 participants returned analytical data for the set of samples that accompanied the questionnaire. One or more chromatographic technique was used by 74 laboratories in their analyses and 69 of these laboratories returned at least one completed questionnaire, a return rate of 93% for the relevant chromatographic techniques. Questionnaires from laboratories using exclusively immunological

techniques were excluded. There were 59 questionnaires detailing information about extraction of D-amfetamine, 63 morphine and 38 benzoylecgonine. The percentage of laboratories that included various procedures in their extraction technique is summarised in Table 1. Five laboratories used Toxi-Tube A for extraction but went on to use GC, GCMS or HPLC detection techniques instead of the ToxiLab TLC system. For analysis of extraction technique, the latter were grouped with the in-house liquidliquid procedures. The in-house liquidliquid extraction systems were diverse. Virtually all protocols extracted from an alkalinised urine sample of pH > 8 but the range of solvents was large with only the commonest being listed in Table 1. Those reported were n-butyl acetate, 1-chlorobutane, chloroform, cyclohexane, dichloromethane (DCM), diethyl ether, ethyl acetate, hexane, isooctane, methyl tert-butyl ether and toluene. Chloroform and DCM were often modied by addition of isopropanol or butan-2-ol, and ethyl acetate and hexane by addition of acetonitrile, diethyl ether or isoamyl alcohol. One laboratory used a mixture of methyl tert-butyl ether, DCM and isopropanol. Back extraction was used in some liquidliquid protocols (Table 1). Excluding laboratories making direct injection from the organic phase, separation of phases was by centrifugation (one laboratory used phase separator lter paper). Variability in solid phase systems was less marked as most laboratories followed the protocols recommended by the manufacturers of the various columns. Urine pH was adjusted to be within a specied range before application to a suitably primed column. The most popular columns were Bond Elut Certify1 and Tox Elut1 from Varian (Varian Inc., Palo Alto, CA, USA) and Isolute-Conrm1 HCX from International Sorbent Technology (Jones Chromatography, Hengoed, UK) (Table 1). Other column types were used by single laboratories, one laboratory used a disk extraction system and one the Supelco microextraction bre system (SPME, Supelco Inc., Bellefonte, PA, USA). Approximately equal numbers used gravity and vacuum methods to elute the column using the washing and elution steps recommended by the manufacturer. Data on the lower limit of detection and volume of urine extracted are presented in Tables 2 and 3. The values quoted by the manufacturer of the Toxi-Lab system are indicated in the latter two tables by superscripts. They refer to detection by the commercial TLC separation and detection system after extraction by Toxi-Tube A. 3.1.
D-amfetamine

There were 37 (20%) negative or uncertain reports for amfetamine out of a total of 185 reports in the samples containing 0.4 mg/l amfetamine and 11 (13%) out of 152 reports in the samples containing 0.8 mg/l. When broken down by either extraction technique or nal detection method (Table 4), there was a signicant excess of negative and uncertain results (P < 0:01) by Toxi-Lab in the 0.4 mg/l

J.F. Wilson et al. / Forensic Science International 119 (2001) 2327 Table 1 Percentage of laboratories using different extraction protocolsa Extraction step Pre-extraction steps Internal standard Deuterium standard Hydrolysis Enzyme pH adjustment Extraction type Liquidliquid Chloroform modifier n-Butyl acetate DCM isopropanol Toxi-Tube1 A Solid phase Bond Elut1 Tox Elut1 Isolute1 Post-extraction steps Back-extraction Acidification before drying Concentration step With air With nitrogen Toxi-Disc1 Derivatisation For GCMS Detection Technique TLC HP-TLC GC GCMS HPLC Total laboratories reporting
a

25

Amfetamine (%) 37 10 7 5 58 39 12 7 3 37 24 15 2 2 9 15 76 12 20 34 31 29 44 15 37 3 59

Morphine (%) 33 11 48 24 54 16 3 2 25 59 24 14 10 0 3 100 27 32 27 41 41 49 6 3 41 63

Benzoylecgonine (%) 42 16 13 11 53 16 5 3 3 26 58 32 5 8 3 0 92 18 34 26 50 47 37 5 5 53 38

DCM, dichloromethane; HP-TLC, high-performance TLC.

samples where the Toxi-Lab system was operating below its declared cut-off of 0.5 mg/l. True positives fell from 93 to 62%. Laboratories using TLC after either liquidliquid or solid phase extractions are presented in Table 4 as in-house

TLC. They performed signicantly less well (P < 0:01) at both amfetamine concentrations. Performing an acidication step before sample concentration yielded a signicant improvement in the percentage

Table 2 Percentage of laboratories reporting different lower limits of detection Detection limit (mg/l) 0.0050.05 0.10.4 0.5 1.0 Total laboratories reporting
a

Amfetamine (%) 23 23 31a 23 39

Morphine (%) 26 29 12 33a 42

Benzoylecgonine (%) 42 21 8 29a 24

Toxi-Lab A value.

26

J.F. Wilson et al. / Forensic Science International 119 (2001) 2327

Table 3 Percentage of laboratories using different extraction volumes Urine volume (ml) <1.0 1.02.5 3.04.5 5.0 6.020.0 Total laboratories reporting
a

Amfetamine (%) 10 17 17 47a 8 59

Morphine (%) 3 17 10 51a 19 63

Benzoylecgonine (%) 21 11 55a 13 38

Toxi-Lab A value.

of true positive results at both amfetamine concentrations (P < 0:05). True positives rose from 82 and 83% both to 100% in the 0.4 and 0.8 mg/l concentration samples, respectively. Before considering the volume of urine used for extraction, those laboratories using 5.0 ml were split into Toxi-Lab and non-Toxi-Lab groups so that the inuence of the poorer performance by Toxi-Lab in the 0.4 mg/l samples was isolated. There was a paradoxical signicantly poorer performance in both 0.4 mg/l (P < 0:01) and 0.8 mg/l (P < 0:05) samples by the group of laboratories that extracted >5 ml volumes of urine by liquidliquid and solid phase procedures (Table 4).

There were no false positive results out of a total of 153 reported for the ve samples in which amfetamine and related compounds were not present. 3.2. Morphine There were 51 (23%) negative or uncertain reports for morphine out of a total of 221 reports in the samples containing 0.4 mg/l morphine and 29 (13%) out of 230 reports in the samples containing 0.8 mg/l. The laboratories using GCMS as the nal detection technique performed signicantly better than other technique groups (P < 0:001) with a true positive rate of 98% at both morphine concen-

Table 4 Percentage of true positive results Group Concentration mg/l Amfetamine (%) 0.4 Extraction type Liquidliquid Toxi-Lab A Solid phase Detection technique GC GCMS HPLC Toxi-Lab In-house TLC HP-TLC Sample volume (ml) <4.5 5.0 Toxi-Lab 5.0 non-Toxi-Lab >5.0
* **

Morphine (%) 0.8 92 93 94 92 98 100 92 66** 94 87 100 75* 0.4 65 51 92**** 98**** 48*** 78 75 84 51*** 82 95 0.8 72 79 95**** 98**** 74 87 81 89 79 87 96

Benzoylecgonine (%) 0.15/0.2 80 50 67 50 78 50 25 75 50 73 0.45/0.5 100 73 92 100 95 73** 100 63** 100 73* 88 83

89 62** 93 92 96 88 61*** 57*** 92 60** 100 67**

P < 0:05 (signicantly lower, chi-square test). P < 0:01 (signicantly lower, chi-square test). *** P < 0:001 (signicantly lower, chi-square test). **** P < 0:001 (signicantly higher, chi-square test).

J.F. Wilson et al. / Forensic Science International 119 (2001) 2327

27

trations studied (Table 4). The Toxi-Lab group had a signicantly lower true positive rate of 48% in samples containing 0.4 mg/l morphine, well below the 1.0 mg/l cut-off for the technique. When the three methods of sample extraction were compared, solid phase was signicantly better (P < 0:001) than liquidliquid or Toxi-Lab systems at both drug concentrations (Table 4). The latter difference might, however, result from the widespread use of solid phase extraction before the more sensitive GCMS detection. An analysis with the solid phase users sub-divided into GCMS and non-GCMS groups showed the true positive rates to be 97% for the solid phase/GCMS combination at both morphine concentrations and for solid phase/non-GCMS users, 87% at 0.4 mg/l morphine and 92% at 0.8 mg/l morphine. The latter values were signicantly greater (P < 0:05) than for liquidliquid and Toxi-Lab extraction. Solid phase extraction therefore outperformed the other extraction techniques for morphine regardless of the detection technique used. An analysis of the non-Toxi-Lab data categorised by inclusion of pre-extraction hydrolysis procedures failed to show a signicant difference in true positive frequency when comparing enzymatic hydrolysis, acid hydrolysis or no hydrolysis (P > 0:05). The Toxi-Lab data were excluded to avoid the poorer performance of this technique at 0.4 mg/l from introducing bias to the no hydrolysis group. Similarly, there was no signicant difference resulting from use of different volumes of urine for extraction (P > 0:1) when the non-Toxi-Lab groups were compared (Table 4). There were ve (1.6%) false positive reports for morphine out of a total of 309 reports in the nine samples not containing morphine or related compounds. The ve reports were from three laboratories all using solid phase systems, two with GCMS and one high-performance TLC. 3.3. Benzoylecgonine There were 12 (34%) negative or uncertain reports for benzoylecgonine out of a total of 35 reports in the two samples containing 0.15/0.2 mg/l benzoylecgonine and nine (10%) out of 94 reports in the samples containing 0.45/ 0.5 mg/l. The data for the lower concentration samples were so sparse as to make inter-group comparisons insensitive. No signicant differences were detected between extraction techniques (P > 0:05, Table 4). In the comparison between detection techniques, both high-performance TLC and ToxiLab gave signicantly fewer true positive results (P < 0:01) than other techniques in data from samples with the higher concentrations of benzoylecgonine (Table 4). Comparison between laboratories based on the volume of urine analysed demonstrated no signicant difference (P > 0:05) after the signicantly poorer performance of the Toxi-Lab system at higher drug concentrations had been isolated (Table 4).

There was one false positive report for benzoylecgonine by solid phase extraction with GCMS out of a total of 89 reports in the 14 samples not containing cocaine metabolites. 4. Discussion A number of the signicant differences demonstrated in the study were traced to the performance of the detection technique used to identify drugs rather than to differences in the preliminary extraction technique. The Toxi-Lab system was signicantly less sensitive than other techniques for all three analytes. However, the drug concentrations selected for study were chosen as likely to challenge the limits of detection often claimed for techniques. They were lower in several cases than the declared thresholds for Toxi-Lab. There was thus no evidence that Toxi-Lab was not performing to specication. In-house TLC techniques performed poorly for detection of amphetamine as did high-performance TLC for benzoylecgonine. There were two signicant ndings related to the extraction of amfetamine. First, acidication of the sample before reducing the sample volume was benecial, probably by reducing the volatility of amfetamine. Secondly, protocols that use urine volumes of greater than 5 ml are paradoxically at an analytical disadvantage and should be updated. For the extraction of morphine, solid phase systems were superior to liquidliquid systems and this could be demonstrated as being independent from the better performance observed for GCMS analysis. It should be noted that the spiked samples used in the study contained no glucuronide metabolites. Thus, the above nding refers specically to morphine. We have no information about the effectiveness of the various hydrolysis procedures in use except that their inclusion had no adverse effects on the extraction. No signicant differences between the extraction techniques were identied for benzoylecgonine.

References
[1] R. Badia, R. de la Torre, S. Corcione, J. Segura, Analytical approaches of European Union laboratories to drugs of abuse analysis, Clin. Chem. 44 (1998) 790799. [2] O.H. Drummer, Chromatographic screening techniques in systematic toxicological analysis, J. Chromatogr. B Biomed. Sci. Appl. 733 (1999) 2745. [3] R. de la Torre, J. Segura, R. de Zeeuw, J. Williams, Recommendations for the reliable detection of illicit drugs in urine in the European Union, with special attention to the workplace, Ann. Clin. Biochem. 34 (1997) 339344. [4] J.F. Wilson, B.L. Smith, Evaluation of detection techniques and laboratory prociency in testing for drugs of abuse in urine: an external quality assessment scheme using clinically realistic urine samples, Ann. Clin. Biochem. 36 (1999) 592600.