Clinical Chemistry 43:2 312325 (1997)

Drug Monitoring and Toxicology

Uniform solid-phase extraction procedure for toxicological drug screening in serum and urine by HPLC with photodiode-array detection
Chi-Kong Lai,* Ting Lee, Kam-Ming Au, and Albert Yan-Wo Chan
In this HPLC diode-array detection method for toxicological drug screening, a mixed-mode solid-phase extraction procedure is optimized for isolation of a broad range of drugs from serum and urine. Basic, neutral, and weakly acidic drugs are uniformly recovered. The extract from the solid-phase cartridge is readily injected to a reversed-phase HPLC column for separation by gradient elution. Unknown drugs and metabolites in urine and serum samples from acute drug poisoning cases are rapidly identified by matching their retention times and ultraviolet spectra with hundreds of reference compounds in the library. Urine metabolites of common toxicants from various medications and drugs of abuse are recorded, with their changes of retention times and ultraviolet spectra as related to their metabolic transformations. Glucuronide conjugates of common benzodiazepines, tricyclic antidepressants, and beta-blockers are examined directly without chemical or enzymatic hydrolysis. The system is reliable for diverse clinical investigations of drug overdoses, drug-induced psychoses, and substance abuse.
INDEXING TERMS: chromatography, reversed-phase toxicology abused drugs drug metabolism and disposition

Acute poisoning in patients, usually with therapeutic drugs or abused substances, is an important cause of admission to hospitals [1-4]. The utility and reliability of emergency toxicological testing have been reviewed [5, 6]. Comprehensive urine and serum drug screens are advocated, but their impact on clinical diagnosis and management is often limited by slow reporting time. A systematic procedure for simultaneous detection of as many toxicants as possible in biosamples, with a good balance

Department of Pathology, Princess Margaret Hospital, Lai Chi Kok, Hong Kong. *Author for correspondence. Fax 852 23700969. Received July 1, 1996; revised October 21, 1996; accepted October 31, 1996.

between speed, universality, and sensitivity, is thus highly desirable. With the availability of lower-cost instruments and more-streamlined approaches to sample preparation [7], GC-MS has become the gold standard for comprehensive toxicology screening. However, its use as an emergency screening tool is often restricted by cost, workload, or expertise. Traditional toxicology screening methods adopted by many clinical laboratories, primarily a combination of immunoassay and thin-layer or gas chromatography, are time-consuming, labor-intensive, and highly dependent on technologist interpretation. A comprehensive drug screen usually involves multiple procedures in conjugate cleavage, solvent extraction, and (or) derivatization; moreover, different chromatographic conditions may be required to cover several essential classes of drugs or to analyze different matrixes. Given the time pressure for rapid analyses of a wide range of substances, any way to unify drug screening procedures is of great value. With the introduction of the on-line photodiode-array detector (DAD), the development of HPLC methods for broad-spectrum drug screening has attracted great interest in clinical and forensic laboratories [8-16]. The HPLCDAD coupling fulfills the norm of toxicological investigation for confirming the identity of unknowns by obtaining secondary chemical information (i.e., retention time plus ultraviolet spectrum). Although additional information is usually required for absolute identification (because of the lower resolution of HPLC and the less-unique results of ultraviolet spectroscopy), HPLC-DAD has been ranked second only to GC-MS as a single system for systematic toxicological analysis [17]. In particular, reversed-phase HPLC operating at low pH with gradient elution offers the potential for separating acidic, neutral, and basic drugs in a single run [13]. However, HPLC-DAD methods reported for human samples are so far dedicated for use with specific matrixes and limited drug groups only. Speed and efficiency of drug isolation from biological fluids are the limiting factors for clinical use of chromatographic screens. Rapid solid-

312

Clinical Chemistry 43, No. 2, 1997

313

phase extractions (SPEs) have been widely used in analyses of combined drug groups [18-21]. A unified procedure for acidic, neutral, and basic drugs in urine, plasma, and whole blood has been proposed for gas-chromatographic analysis [22, 23]. Mixed-phase sorbent, combining nonpolar and cation-exchange packing materials, is advocated for drug-class-independent screening because its use offers higher recoveries for a broader range of drugs [24]. To date, we have seen no report of a single solidphase extraction method applicable for broad-spectrum drug screening by HPLC-DAD for both serum and urine samples. Here we describe a unified procedure for the simultaneous SPE of stimulants, narcotics, sedative-hypnotics, antidepressants, antipsychotics, antihistamines, cardiovascular drugs, analgesics, and many of the metabolites of these drugs in serum and urine. After the extraction, HPLC-DAD is used for subsequent separation and detection. Our operational routine allows a quick start-up of assays for emergency investigation, and a compilation of urine metabolite profiles for common drugs increases the reliability of a quick urine screen. The power of our SPE-HPLC-DAD system extends to direct identification of glucuronide metabolites of many drugs in urine.

Materials and Methods clinical samples

Serum and urine samples were obtained from hospital emergency wards, psychiatric units, and substance abuse clinics. We also included urine toxicology external quality-control specimens (distributed for qualitative analysis by subscription from RCPA Quality Assurance Programs, Bedford Park, Australia) for periodic checks. The development and application of our method in this study were approved by the Ethical Committee of the Hospital Authority of Hong Kong.

qualityin tablets, capsules, or injection formwere kindly supplied by the pharmacy of the Princess Margaret Hospital, Hong Kong; stock solutions were prepared at 10 g/L in 800 mL/L acetonitrile and stored at 4 C. Bond Elut Certify columns, packed with 130 mg of sorbent in a 3-mL solvent reservoir, were obtained from Varian Sample Preparation Products (Harbor City, CA). SPE was done with a VacElut reduced-pressure manifold from Analytichem International (Harbor City, CA). For solvent extraction, we used Toxi-TubeTM A (Toxi-Lab, Irvine, CA); these tubes, designed for use with 5-mL urine samples, contain sodium carbonate and bicarbonate to give a pH of 9.0 in a mixture of dichloromethane and dichloroethane. Commercial controls for serum therapeutic drug monitoring were obtained form Ciba-Corning Diagnostics (Irvine, CA). These controls, available at three concentrations in lyophilized form, are prepared from human serum with added drugs from all of the classes mentioned above. Controls for urine toxicology were obtained from Bio-Rad Labs. (ECS, Anaheim, CA). The urine controls, lyophilized human urine supplemented with drugs or their metabolites from various classes of drugs of abuse, included amphetamines, barbiturates, benzodiazepines, cannabinoids, cocaine, methadone, methaqualone, opiates, phencyclidine, and propoxyphene.

chromatographic conditions
The analytical flow rate was 1.0 mL/min. We used gradient elution generated by the proportional mixing of two solvents: Solvent A contained 50 mL/L acetonitrile and solvent B contained 500 mL/L acetonitrile, both in 50 mmol/L phosphate buffer, pH 3.0, containing 375 mg/L sodium octyl sulfate and 3 mL/L triethylamine. Solvent gradient conditions changed linearly from 15% B to 90% B in 20 min, stayed at 90% B for 5 min, and returned to 15% B in 3 min. Total run time was 30 min, the analysis time being 28 min. To ensure stabilized analytical conditions, we waited 6 min between runs.

equipment
We used a Model HP1090 A liquid chromatograph (Hewlett-Packard, Waldbronn, Germany) equipped with a built-in DAD and an automated sampler. On-line printing of ultraviolet spectra, wavelength ratios, and peak purity data was generated by an HP2225 ThinkJet printer interfaced with the instrument. The analytical column consisted of a 10 2.1 mm (i.d.) precolumn and a 150 4.6 mm (i.d.) main column packed with Hypersil 5 m (particle size) octadecylsilane (C18), also from HewlettPackard. Column temperature was maintained at 45 C.

sample preparation
To 3 mL of urine or serum in a plastic tube, we added an equal volume of sample diluent (0.4 mol/L phosphate buffer, pH 6.0) and vortex-mixed the samples for 1 min. SPE columns were inserted into the reduced-pressure manifold and conditioned by washing once with 3 mL of methanol followed by 3 mL of conditioning solution (0.2 mol/L phosphate buffer, pH 6.0). We made sure the columns did not dry out before the specimens were applied. We poured the samples into each column reservoir, from which they were drawn slowly through the column at a flow rate of 1 mL/min. The columns were washed with 3 mL of wash buffer (50 mL/L methanol in 0.2 mol/L phosphate buffer, pH 6.0) and the fluid was drained by passing air through the columns for 10 s. Concentrated drugs were eluted with 750 L of a mixture

reagents and materials

Potassium dihydrogen phosphate, triethylamine buffer solution, and phosphoric acid (all SupraPure grades) and acetonitrile and methanol (both HPLC grade) were purchased from Merck (Darmstadt, Germany). Sodium octyl sulfate, ammonium acetate, and -glucuronidase type H-2 (from Helix pomatia) were obtained from Sigma Chemical Co. (St. Louis, MO). All drugs of pharmaceutical

314

Lai et al.: Uniform solid-phase extraction for toxicological drug screening by HPLC

of 100% methanol and 10% ammonia (5:1 by vol) at a slower flow rate, 0.5 mL/min. For urine analysis in cases of acute poisoning, we injected 25 L of the extract directly into the chromatograph. To enhance the sensitivity of drug detection at lower concentrations, we evaporated the SPE extract under nitrogen gas after adding a few drops of 1 mol/L HCl to reduce the volatility of some drugs (e.g., sympathomimetic amines) [9]; the dried residue was then reconstituted in 150 L of solvent B for HPLC injection. In case of insufficient sample volume, we adjusted the reconstitution volume according to initial loading volume of the specimen, to give a fixed concentration factor of 20-fold. That is, a 1-mL serum sample can be concentrated to a 50-L extract for sensitive serum analysis. The retention and spectral data of 300 toxicologically relevant compounds (therapeutic and illicit drugs, their metabolites, and endogenous substances) determined by the gradient HPLC procedure have been recorded. The basic library for pure compounds was initially established by analyses of injections of diluted drug solution (1:200 dilution of drug stock solution in solvent B). During each run, the absorbance was recorded at 210 nm, with 8-nm bandwidth; the reference wavelength was 550 nm, with 40-nm bandwidth. Ultraviolet spectra over the range 210 350 nm were automatically extracted if the peak signal was 3 mA (threshold setting) and the peak covered 0.1 min (peak width setting). The built-in microprocessor output processed peak signals immediately to the on-line printer, plotting overlaid ultraviolet curves for apex and downslope, and reporting both the A210 nm/A230 nm ratio and the factor of peak purity (ideal 1.000). To construct a spectral library, we used an Excel-5TM (Microsoft, Redmond, WA) electronic workbook on a 486-PC, digitalizing ultraviolet spectra for pure products, for drug metabolites, and for matrix peaks in serum or urine processed under the same analytical conditions. Arranged according to retention time, each entry was assigned a view area that also documented relevant information. Each library entry includes identification parameters, metabolite profiles, chemical structure and properties, pharmaco- and toxicokinetic data, clinical presentation of overdose, and a brief note on the disposition of the drug in humans. In addition, we have included three indexing lists for quick searches according to (a) tentative identification according to retention times, absorption maxima, and A210 nm/A230 nm ratios; (b) computer search by name to view the relevant information by use of the Excel-5 ViewManager function; (c) cross-reference to the bench log books for detailed inspection. The log books compile actual printouts of chromatogram and ultraviolet spectra of parent drugs and metabolites in real cases of drug overdose/drug abuse. The manual but systematic search routine allows quick identification in urgent cases.

express start up and unattended shutdown

To meet the emergency needs for acute overdose investigations, we set up instrument programs for quick starts and automatic shutdowns. We found the following instrumental routine allowed consistent performance and long-term stability of the dedicated column: 1) Start up: Wash with 150 mL/L acetonitrile for 15 min and mobile phase for 10 min, and then inject samples.The HPLC column is thermostated at 45 C to ensure fast solvent equilibration. Typical instrument run time is 30 min, during which SPE is carried out in parallel. 2) Shut down: Wash with deionized water for 30 min, with 0 700 mL/L acetonitrile gradient for 30 min, and with 700 mL/L acetonitrile for 30 min. This unattended shutdown allows proper conditioning of the analytical column for the next operation.

detection and identification

comparison of drug recovery

Drug-supplemented urine was prepared by adding to urine from a drug-free subject 10 mol/L each of 10 drugs from common therapeutic classes. A reference solution was also prepared by adding the same amount of drugs to solvent B at one-fourth the volume of the urine. The drugs used (and the classes they represent) included fluoxetine (antidepressants), nortriptyline (tricyclic antidepressants), verapamil (calcium channel blockers), naproxen (nonsteroidal antiinflammatory drug), midazolam (benzodiazepines), metoprolol (beta-blockers), furosemide (diuretics), ephedrine (decongestants), ranitidine (histamine-R2 receptor antagonists), and morphine (opiates). These urine samples were processed by Bond Elut Certify as described above. For solvent extraction by Toxi-Tube A, we added 5 mL of urine to the tube and vortex-mixed the contents for 5 min. After centrifugation, the upper organic phase was aspirated, dried under nitrogen gas, and reconstituted in 1.25 mL of solvent B. Extracts of Bond Elut Certify and Toxi-Tube thus prepared were injected into the HPLC (calibrated against the reference solution). The extraction efficiency of each drug by either mode of extraction was calculated from the integrated peak areas at 210 nm.

estimation of detection limits for qualitative identification

Quantitative control materials for therapeutic drug monitoring and drugs of abuse detection were obtained from commercial sources. Serum or urine controls were processed in three separate runs in the same manner as the routine samples. SPE extracts were injected into the HPLC without further enrichment. For drugs that were positively identified, the mean absorption signals of the peaks were related to its assigned concentration. To assess linearity of detection, we assayed the serum controls at all three target concentrations. We calculated the theoretical detection limit as the concentration of drug giving a signal above the peak threshold of 3 mA, which is the minimum signal for reliable spectral recognition against the nonspe-

Clinical Chemistry 43, No. 2, 1997

315

cific background of 0.6 mA (i.e., signal-to-noise ratio 5). The estimated detection limit for qualitative identification is thus a function of extraction efficiency, chromatographic peak symmetry, and photometric absorptivity of individual drugs.

urine drug metabolites

Profile records. The method has been used routinely in our laboratory to screen 300 urine or serum specimens. Urine metabolite profiles were established from positive drug overdose cases for which both serum and urine were submitted for analyses and in which an unmetabolized drug was unequivocally identified. The metabolites, which usually eluted faster than their parent compound, were readily recognized if their characteristic spectra resembled those of their parent compounds. To prepare a simplified chromatogram of a metabolite, we reconstructed only signals of the related peaks (according to retention time and relative absorbance). Spectra of special metabolites that give chromogenic shifts were established by study of recurrent cases. Studies with glucuronidase/sulfatase digestion. Urines from positive cases involving overdoses of benzodiazepines, tricyclic antidepressants, or beta-blockers were further investigated by glucuronide hydrolysis. Paired 3-mL urine specimens were diluted in 3 mL of SPE diluent and then incubated at 56 C for 2 h with or without 50 L of -glucuronidase crude extracts from H. pomatia (-glucuronidase 89 400 kU/L; sulfatase 3300 kU/L). After incubation, the samples were cooled to room temperature for SPE. By comparing the chromatograms obtained with and without enzyme digestion, we could readily identify hydroxylated metabolites (which emerge after digestion) and glucuronide metabolites (which disappear after digestion). Conjugated metabolites were related to their free counterparts on the basis of their characteristic ultraviolet spectra.

Fig. 1. Comparison of the mean extraction efficiency of solid-phase extraction (Bond Elut Certify) and solvent extraction (Toxi-Tube A) methods as determined from duplicate determinations of 10 tested drugs added to urines.

the SPE procedure gave superior recoveries of acidic drugs (e.g., furosemide, naproxen), amphoteric compounds (e.g., morphine), and polar substances (e.g., ephedrine). Detection limit for qualitative drug identification. The expected drug detection ability of the system for 28 drugs, representing various drug classes, is summarized in Table 1. The assessment covers most relevant therapeutic agents that require quantitative determination in serum. Also documented is the expected detection ability for popular illicit drugs at critical urine concentrations. The estimated detection limit, as defined earlier, is the theoretically lowest concentration of a compound yielding a spectrum that can be matched against the spectral library. The detector responses for the tri-level serum controls were generally proportional to their target concentrations, supporting our estimation of detection limit by extrapolation. As expected, the estimated detection limits for various compounds varied according to differences in extraction efficiency, chromatographic peak symmetry, and molar absorptivity.

Results assay performance

Quick start and reproducibility of retention times. The reproducibility of the retention times for five compounds was measured for 20 initial runs of the same drug mixtures over a 4-month period. The mean SD retention times (minutes) were: codeine, 5.0 0.7; carbamazepine, 15.0 0.3; alprazolam, 18.0 0.4; imipramine, 21.0 0.4; and chlorpromazine, 23.5 0.4. These results validated our quick start-up procedure, which ensures short reporting time with little compromise of retention time reproducibility. Extraction efficiency. Fig. 1 demonstrates the wide extraction ability of the mixed-mode SPE. Recoveries of probe drugs from diverse classes ranged from 71% to 99% (duplicate analyses) at concentrations of 10 mol/L. Compared with the classical solvent-extraction method,

316

Lai et al.: Uniform solid-phase extraction for toxicological drug screening by HPLC

Table 1. Assessment of detection limits for positive identifications of drugs.

Target concn., mol/L High Medium Low High Absorbance, mAa Medium Low LOD, mol/Lb

Serum control Acetaminophen Amitriptyline Carbamazepine Chloramphenicol Clonazepam Diazepam Imipramine Lidocaine Phenobarbital Phenytoin Procainamide Quinidine Salicylate Theophylline Urine control Amobarbital Amphetamine Benzoylecgonine Codeine Hydroxy-alprazolam Methadone Methamphetamine Methaqualone Morphine Oxazepam Phencyclidine Phenobarbital Propoxyphene Secobarbital
a b

887.00 1.57 73.00 135.00 0.28 21.80 1.66 38.80 211.00 105.00 37.00 18.20 3400.0 166.00

351.00 0.84 44.00 46.00 0.13 5.30 0.81 16.60 108.00 57.00 23.00 11.10 1300.0 89.00 8.50 23.00 3.00 6.27 6.00 6.00 21.00 7.50 5.37 6.50 0.64 8.00 5.50 8.00

99.00 0.33 15.00 23.00 0.06 1.80 0.71 7.30 45.00 23.00 8.00 4.30 200.00 29.00

248.0 5.6 297.8 93.0 ND 116.6 5.6 114.3 195.5 178.0 92.8 149.2 24.0 125.0

80.8 3.1 187.4 44.3 ND 27.3 3.0 50.2 101.8 107.3 56.2 87.2 7.8 61.6 10.0 53.3 8.0 38.0 40.2 29.3 30.0 48.4 76.5 56.2 ND 9.5 21.3 8.7

26.8 ND 79.9 22.9 ND 9.2 ND 21.5 46.3 50.5 18.2 38.6 ND 17.2

11.60 0.83 0.67 2.65 / 0.58 0.85 1.00 3.11 1.57 1.24 0.36 462.00 4.46 2.58 1.29 1.12 0.49 0.45 0.61 2.10 0.47 0.21 0.35 / 2.53 0.78 2.76

Absorbance of peak apex, mean of triplicate analyses. Limit of detection estimated as drug concentration yielding signal 3 mA threshold. ND, not detected.

The quick SPE procedure yields a fourfold concentration of drugs from biological fluids, which is sufficient for overdose detection of most drugs in urine. Further concentration of the SPE extract allows enhancement of sensitivity for drug detection, which is required for detecting certain drugs present at low concentrations in serum (e.g., clonazepam) and for confirming the presence of some drugs of abuse in urine (e.g., phencyclidine).

complete list of drugs

The drugs, metabolites, and matrix interferences recorded as reference for identification are listed in order of retention times (Table 2). For simplicity, only retention time and ultraviolet absorption maxima are given. The list is by no mean exhaustive; the drugs included were selected because of their toxic potentials and local poisoning epidemiology [4]. Also included are drugs known to inhibit liver enzymes, whose presence may potentiate the toxicity of other drugs via drug interactions [25].

Major nondrug matrix substances were recognized and registered in the library. The serum extracts were generally clean, with three small peaks (usually 6 mA) frequently seen at 2.9, 11.7, and 21.6 min. The number and extent of urine matrix interferences varied somewhat and frequently resulted in noisy background at 8-13 min into the HPLC chromatogram (210 nm). The only consistent urine interference was a very sharp peak at 11.8 min, although two additional peaks at 9.6 and 17.5 min were very often present. These three urine matrix peaks, when present as substantial signals in concentrated urine, may overlap with the peaks for adjacent drugs and affect the identification potential for those drugs.

urine drug metabolite profile

Figures 2 4 reproduce the urine metabolite patterns of 10 common drugs selected from positive urine drug profiles of drug-intoxicated patients admitted to our hospital. The drugs shown are representatives of diverse drug classes: benzodiazepines (midazolam), tricyclic antidepressants

Clinical Chemistry 43, No. 2, 1997

317

Table 2. List of drugs, metabolites, and matrix interferences examined.

RT, min UV maxima, nm Name Typea RT, min UV maxima, nm Name Typea

2.4 2.5 2.7 2.7 2.8 2.8 2.9 3.1 3.3 3.5 3.5 3.5 3.6 3.9 4.0 4.0 4.0 4.3 4.4 4.6 4.6 4.8 4.8 4.9 4.9 5.0 5.0 5.1 5.2 5.5 5.7 6.0 6.2 6.2 6.2 6.2 6.3 6.4 6.5 6.6 6.7 6.7 6.7 6.9 7.2 7.3 7.6 7.7 7.7 8.0 8.0 8.1 8.4 8.4 8.5

275, 215 270 247 286, 212 270 280, 211 280, 215 269.5, 211.5 270 304, 260, 211 270 219 280, 210 272 284, 211.5 293, 213 309, 270, 215 275, 230 342, 250 280, 211 315, 227 282, 211 250 258 268 282, 209 280, 211 298, 240 320, 242 212 293, 213 290, 230, 215 NA 257 257 298, 212 291 NA 300, 234 272, 219 275, 224 278, 235 280 265 276, 224 240, 211.5 254 254 266 269.5, 211.5 332, 240, 212 297, 250 270, 231.5 300, 264, 243 295, 211

Dextromethophan Theophylline Acetaminophen Morphine Caffeine M2 Codeine M2 Serum matrix 1 Sulfamethoxazole M2 Caffeine M1*b Hydralazine Caffeine Cimetidine Procainamide Trimethoprim M1 Pholcodine Sulpiride Metoclopramine M2 Atenolol Quinidine M4 Codeine M1* Ranitidine Dihydrocodeine Trimipramine M8 Phenylpropanolamine N-Acetylprocainamide Naloxone Codeine Clozapine M5 Quinidine M3 Barbital Sulpiride M1 Propranolol M3 Primidone Ephedrine Pseudoephedrine Clozapine M4 Doxepin M6 Phenobarbital M1 Salicylate Nadolol Guaifenesin Benzoylecgonine Oxymorphone Clonidine Metoprolol M2* Dothiepin M3 Amphetamine M2 Methylephedrine Lidocaine M2 Sulfamethoxazole Quinidine M2 Promethazine M8 Naproxen M7 Metoclopramine M1* Doxepin M5

P P P P M M I M M P P P P M P P M P C M P P C P P P P M M P M M P P P C M M P P P P P P M C M P M P M M M C G

8.6 8.7 8.8 9.1 9.1 9.2 9.4 9.4 9.5 9.6 9.7 9.9 9.9 9.9 9.9 10.1 10.2 10.3 10.3 10.6 10.6 10.6 10.7 10.9 10.9 11.0 11.2 11.3 11.4 11.6 11.6 11.7 11.8 11.8 11.8 11.9 11.9 11.9 11.9 12.0 12.0 12.1 12.3 12.3 12.4 12.4 12.4 12.5 12.5 12.5 12.5 12.5 12.7 12.7 12.8

297, 250 NA 276 327, 295.5, 230 264 302, 259, 219 240, 211 297, 250 327, 295.5, 230 288 254 280 278 NA 275, 230 211 268 224 280, 223 283, 215 277 297, 250 250, 211 272 310, 255 250 254 260 275, 223 300, 260, 216 330, 240, 212 NA 251, 211 327.5, 239 305, 218 211.5 249 270, 231.5 251 266 254 251, 211 305, 218 240, 211 252, 211 318, 231.5 287, 235 254 263, 223 310, 255 219 270, 231.5 270 270, 231.5 326, 235

Promethazine M7 M Carbamazepine M5* C Tocainide P Ofloxacin M1 M Sulfamethoxazole M1 C Phenylephrine P Amitriptyline M8 M Promethazine M6 M Ofloxacin P Urine matrix 1 I Amphetamine M1 M Oxycodone P Mefenamic acid M7 M Phenobarbital P Verapamil M4 C Butabarbital P Carbamazepine M4 C Alprazolam M2* G Mefenamic acid M6 M Carbamazepine M3 M Chloramphenicol P Promethazine M5* M Imipramine M5 M Trimethoprim P Chlorpromazine M8 M Flumazemil P Amphetamine P Atropine P Metoprolol M1 M Clozapine M3 C Quinidine M1 M Serum matrix 2 I Trimipramine M7 M Urine matrix 2 I Zopiclone M1 M Carbamazepine M2 C Imipramine M4 C Naproxen M6 M Trimipramine M6* M Lidocaine M1* M Methamphetamin P Trimipramine M5 M Zopiclone P Amitriptyline M7 G Chlormethiazole P Lorazepam M1 G MDA P Phentermine P Chlorpheniramine M7 M Chlorpromazine M7 M Midazolam M2* G Naproxen M5 M Ketamine P Naproxen M4* M Acebutolol P (continued on next page)

318

Lai et al.: Uniform solid-phase extraction for toxicological drug screening by HPLC

Table 2. Continued.
RT, min UV maxima, nm Name Typea RT, min UV maxima, nm Name Typea

12.8 12.8 12.9 12.9 13.0 13.0 13.0 13.0 13.2 13.2 13.2 13.2 13.3 13.3 13.5 13.5 13.5 13.6 13.6 13.7 13.8 13.8 13.8 13.9 13.9 14.1 14.1 14.1 14.3 14.4 14.4 14.5 14.6 14.7 14.7 14.7 14.7 14.8 14.8 14.9 14.9 14.9 15.0 15.1 15.1 15.1 15.1 15.2 15.2 15.2 15.3 15.3 15.3 15.5 15.5

248, 212 310, 274, 235, 215 263, 223 310, 231 311, 275, 215 260 287, 235 260 344, 275, 235 291 290, 230, 215 231.5, 312 235 270, 231.5 263, 223 248, 212 260, 219 301, 256, 211 290, 230, 215 266 275, 223 291 330, 240, 212 310, 255 280, 230 263, 223 314, 265 238 290, 240, 211 318 247, 223 337, 293, 269, 234 283, 215 NA 278, 235 291 281, 219.5 345, 302, 278, 240 250, 211 211 250 260 260 283, 215 303, 245 295, 240 NA 240, 211 250, 211 314, 265 211 NA 297, 250 250, 211 NA

Promethazine M4 Thioridazine M6* Chlorpheniramine M6 Oxazepam M1 Metoclopramine Benzhexol M1 MDMA Pheniramine Furosemide Doxepin M4 Propranolol M2* Temazepam M1 Bromazepam Naproxen M3 Chlorpheniramine M5 Promethazine M3 Terfenadine M2 Molindine Propranolol M1 Lidocaine Metoprolol Doxepin M3* Quinidine Chlorpromazine M6* Verapamil M3 Chlorpheniramine M4 Thioridazine M5 Meclobemide Clozapine M2* Doxepin M2 Haloperidol M1 Promethazine M2 Carbamazepine M1 Aprobarbital Cocaine Doxepin M1 Mefenamic acid M5 Chlorpromazine M5 Imipramine M3* Pentobarbital Trazodone Disopyramide Methylphenidate Carbamazepine Chlordiazepoxide Ketoconazole M3* Pethidine M1 Amitriptyline M6 Imipramine M2 Thioridazine M4 Amobarbital Phenytoin Promethazine M1 Imipramine M1 Pethidine

C C M G P M P P P M M G P M M C M P G P P M P M M M M P M M M C M P P C M C M P P P P P P M M M M M P P M M P

15.6 15.6 15.6 15.6 15.7 15.7 15.8 15.8 15.8 15.9 15.9 16.0 16.2 16.2 16.2 16.3 16.3 16.5 16.5 16.5 16.7 16.7 16.7 16.7 16.8 16.8 17.0 17.0 17.0 17.1 17.1 17.2 17.2 17.2 17.2 17.3 17.4 17.4 17.5 17.5 17.5 17.6 17.6 17.6 17.8 17.9 18.1 18.3 18.4 18.4 18.4 18.5 18.5 18.7 18.8

263, 303, 281, 270, 303 240, NA 259 250, 280, 248 211 263, 290, 302, 224 310, 250 263, 280, 240, 345, 280 276, 250 278, 290, 234 274, 240 240, 290, 273, 263 263, 263, 240, 220 225 313, 248 310, 310, 314, 265 314, 308, 211 318, 262 340, 310, 295, 288, 295,

223 263, 231, 211 219.5 231.5 211

211 230

223 211 239, 211 260, 219 223 230 211 302, 278, 240 236 231 211 249 211 230, 215 220 227 223 211

253, 220 255 229 265 265 267, 227 231.5 265, 239.5 240, 211 240 238 240

Chlorpheniramine M3 M Dothiepin M2* M Mefenamic acid M4 M Naproxen M2 M Labetalol P Amitriptyline M5* M Glutethimide P Propoxyphene M3 M Trimipramine M4 M Verapamil M2 M Piroxicam P Secobarbital P Chlorpheniramine M2* M Clozapine M1 M Buspirone P Alprazolam M1 H Nitrazepam P Trimipramine M3 M Chlorpheniramine M1 M Verapamil M1* M Amitriptyline M4 M Chlorpromazine M4 C Pentazocine P Risperidone P Trimipramine M2 M Phenolphthalein P Clozapine P Oxazolam P Propyphenazone P Phenazopyridine P Amitriptyline M3 H Propranolol P Mesoridazine P Fenfluramine P Brompheniramine P Chlorpheniramine P Amitriptyline M2 M Midazolam M1 H Alprazolam P Flunitrazepam P Urine matrix 3 I Chlorpromazine M3 M Oxazepam P Thioridazine M3 M Phencyclidine P Thioridazine M2 M Methaqualone P Erythromycin M2 M Lorazepam P Fentanyl P Cloxazolam P Clonazepam P Ketoconazole M2 M Thiopentone P Ketoconazole M1 M (continued on next page)

Clinical Chemistry 43, No. 2, 1997

319

Table 2. Continued.
RT, min UV maxima, nm Name Typea RT, min UV maxima, nm Name Typea

18.9 19.0 19.1 19.1 19.2 19.2 19.3 19.3 19.3 19.3 19.4 19.4 19.5 19.5 19.5 19.6 19.6 19.6 19.6 19.6 19.7 19.8 19.8 19.8 20.1 20.2 20.5 20.5 20.5 20.6 20.6 20.8 20.8 20.8 21.0 21.2 21.2 21.2 21.3 21.4

228 239 291 296, 211.5 242, 211.5 281, 219.5 221 223 268 280 284, 213 258 314, 265 299 231.5, 312 269, 232 245, 220 281, 219.5 270, 231.5 260, 219 310, 231 295, 250 259 295, 235 303, 263, 231, 211 270, 231.5 292, 231 250 315, 231 315, 231 272 295, 240 251 335, 235 240 310, 255 231 250, 211 240 290, 211

Tolbutamide Citalopram Doxepin Amoxapine Diltiazem Mefenamic acid M3* Midazolam Triazolam Amineptine Mianserin Buprenorphine Diphenhydramine Thioridazine M1 Flecainide Temazepam Pericyazine Haloperidol Mefenamic acid M2 Naproxen M1 Terfenadine M1 Nordiazepam Loxapine Propoxyphene M2 Paroxetine Dothiepin M1 Naproxen Clobazam Desipramine Lormetazepam Flurazepam Phenyltoloxamine Ketoconazole Fluvoxamine Nifedipine Nortriptyline Chlorpromazine M2 Hydroxyzine Trimipramine M1 Methadone M1 Protriptyline

P P P P P M P P P P P P M P P P P M M M P P M P M P P P P P P P P P P M P M C P

21.4 21.5 21.6 21.6 21.6 21.7 21.8 22.0 22.0 22.0 22.0 22.2 22.3 22.3 22.5 22.6 22.6 22.7 22.9 23.0 23.2 23.3 23.6 23.7 23.9 23.9 24.5 24.5 24.7 25.1 25.3 25.4 25.5 25.8 26.2 26.9 27.2 27.4 27.9 28.0

262, 228 303, 263, 231, 211 260 250, 211 280, 230 297, 250 259 308, 259, 231 325, 267, 227 310, 255 272 320, 255, 210 240, 211 211 260 313, 229 262, 228 240, 211 257, 223 NA 273 211 250, 211 310, 257 250, 211 310, 255 240 314, 265 259 311, 231.5 280 349.5, 281, 223 310, 227.5 308, 257 238 250 268, 231 260, 219 299, 240 263, 219.5

Norfluoxetine Dothiepin Serum matrix 3 Imipramine Verapamil Promethazine Propoxyphene Thiothixene Zuclopenthixol Chlorpromazine M1 Maprotline Perphenazine Amitriptyline Erythromycin M1 Benzhexol Diazepam Fluoxetine Amitriptyline M1 Benzotropine Methadone Sertraline Erythromycin Clomipramine Fluphenazine Trimipramine Chlorpromazine Spironalactone Thioridazine Propoxyphene M1* Prazepam Pimozide Mefenamic acid M1 Pinazepam Triflupromazine Cloperastine Trimeprazine Fluphenthixol Terfenadine Indomethacin Ibuprofen

P P I P P P P P P M P P P M P P P M P P P P P P P P P P M P P M P P P P P P P P

a Type of substances: P, pure product; M, metabolite with similar UV spectrum as its parent drug; G, metabolite identified as glucuronide; C, metabolite with major change of UV spectrum; H, metabolite identified as hydroxylated compound; I, matrix interference. b Metabolite frequently detected as the most prominent compound in patients urine is marked with *. RT, retention time; UV, ultraviolet; NA, no absorption peak in the wavelength range of 210 350 nm; MDA, 3,4-methylenedioxyamphetamine; MDMA, 3,4-methylenedioxymethamphetamine.

(doxepin), beta-blockers (propranolol), opioids (methadone), histamine-R1 receptor antagonist (promethazine), anticonvulsants (carbamazepine), nonsteroidal antiinflammatory drugs (napoxen), antiarrhythmics (quinidine), sedative-hypnotics (zopiclone), and antipsychotics (clozapine). The urine metabolite patterns of some commonly seen drugs are listed in Fig. 2, in which the metabolites are readily identified from their characteristic ultraviolet

spectra, which resemble those of their parent drugs. Presumably, these metabolites originate from side-chain transformations, such as dealkylation or hydroxylation, which do not affect their ultraviolet chromophores. Urine metabolites of other drugs have distinct spectral shifts because of metabolic transformation. Fig. 3 illustrates the chromatographic and spectral change of some common drugs with well-established metabolic transformation. Well-known metabolites with ultraviolet spectral shifts

320

Lai et al.: Uniform solid-phase extraction for toxicological drug screening by HPLC

Fig. 2. Urine metabolite patterns of selected drugs.

Metabolites were identified on the basis of their structure-preserving ultraviolet spectra. Signals are reconstructed from patients cases as described in Materials and Methods. Parent drugs in the simplified chromatogram (210 nm) are marked with *, and their ultraviolet spectra are shown on the right. Simplified chromatograms: x-axis, retention time (min); y-axis, absorbance relative to that of the highest peak. Ultraviolet spectra: x-axis, wavelength (nm); y-axis, absorbance in normalized scale.

with respect to the spectra of their parent drugs are tentatively identified. Our approach to identify the glucuronide metabolites of benzodiazepines, tricyclic antidepressants, and betablockers by enzyme digestion is illustrated in Fig. 4. Deconjugation after glucuronidase/sulfatase hydrolysis resulted in the release of free metabolite with little change of ultraviolet spectra but significant chromatographic shifts to higher retention times. The identities of the pairs of metabolites, responding to enzyme cleavage as a reciprocal change in signals, can be inferred with reference to the established drug metabolism and disposition [26]. The glucuronide conjugate of 1-hydroxy-midazolam is the major urine product of midazolam. Propranolol glucuronide and naphthoxylactic acid account for the major

urine metabolites of propranolol [26]. The metabolic transformation of doxepin is more complicated, the Nglucuronide conjugate of the parent drug being the major metabolite, together with nordoxepin, doxepin-N-oxide, dinordoxepin, and O-glucuronides of the ring-hydroxylated metabolites [27]. Similar experiments have been applied to studies of the glucuronide metabolites of alprazolam, oxazepam, lorazepam, and amitriptyline; identification characteristics for these metabolites are detailed in Table 2.

validation study
The performance of our method in 10 consecutive runs of the external quality-control program for urine toxicology is summarized in Table 3. The program covers drugs of

Clinical Chemistry 43, No. 2, 1997

321

Fig. 3. Urine metabolite patterns of some drugs exhibiting chromogenic change on their passage through the metabolic transformation pathway.
x-, y-axis labels for simplified chromatograms and spectra as in Fig. 2.

abuse, drug compliance, and drug overdose. Urine specimens were generally collected from human subjects; however, drug-supplemented urines were also distributed. Drugs known to be present were established by reference laboratories using GC-MS. The program organizers also reported in each end-of-cycle assessment the medication history or amount of drug added when appropriate. Overall, our rapid method correctly identified 27 of 37 target drugs. Correctly identified were drugs from common drug classes such as analgesics (acetaminophen, salicylate, naproxen), opioids (dextropropoxyphene, pethidine, methadone, morphine, codeine), cardiovascular drugs (verapamil), antipsychotics (clozapine), sedative-hypnotics (various benzodiazepines, pentobarbital), sympathomimetic amines (amphetamine, pseudoephedrine), and some over-the-counter drugs (ranitidine, trimethoprim, sulfamethoxazole). Our method also detected some compounds with low identification rates overall (confidential program data, not shown): chloramphenicol, pentobarbital, phenytoin, ranitidine, and sulfamethoxazole.

We made three misidentifications: pholcodine was erroneously reported for a codeine metabolite, imipramine for trimipramine, and amphetamine instead of phentermineall structurally related compounds with almost identical ultraviolet spectra and close retention times. Investigation of the reasons for failing to identify other target drugs gave the following results: drugs present but below the detection limit (n 2), erythromycin and acetaminophen; drugs not yet established in the library entries at the time of analysis (n 5), phenolphthalein, metoclopramide (2 occasions), cannabinoids, and dexchlorpheniramine. In summary, for those drugs established in our drug library, the method positively identified 84% (27 of 32).

Discussion
The present HPLC method for drug screening is based on a mixed-phase bonded silica-gel cartridge claimed by the manufacturer to retain basic, neutral, and acidic drugs under the proper extraction conditions. Chen et al. [22, 23] derived a two-step elution procedure to recover drugs

322

Lai et al.: Uniform solid-phase extraction for toxicological drug screening by HPLC

Fig. 4. Effect of glucuronidase treatment on urine drug metabolite patterns of some benzodiazepines, tricyclic antidepressants, and beta-blockers.
Arrows (12) indicate the direction of change in peak absorbance after enzymatic hydrolysis. x-, y-axis labels for simplified chromatograms and spectra as in Fig. 2.

into two solvent fractions (an acidicneutral and a basic fraction) for gas-chromatographic analyses. In the modified procedures presented here, a single-step elution in ammoniamethanol is introduced to disrupt both ionic and nonpolar interaction. As a result, all drugs and metabolites are quickly and efficiently recovered in a minimum volume of eluate suitable for direct injection into the HPLC. With full compatibility of solvent strength and analyte solubility, HPLC analysis can gain the full benefit of the mixed-mode SPE procedure for fast analysis. Compared with traditional solvent extraction, the mixed-mode SPE gave superior extraction efficiencies for representative drugs from a wide range of therapeutic classes. The detection limits for a group of common therapeutic drugs in serum and drugs of abuse in urine were established by commercial controls with quantitative target values. Thus, the SPE system validly isolated various classes of drugs from both urine and serum. Our experiments with enzymatic digestion of urine samples from overdosed patients proved that the SPE can extract hydroxylated and glucuronide metabolites of mi-

dazolam, lorazepam, oxazepam, alprazolam, temazepam, propranolol, doxepin, and amitriptyline. Hydroxylation and subsequent glucuronic acid conjugation are common metabolic pathways for many drugs. The resulting metabolites are usually amphoteric and more water-soluble than the parent drug, being readily excreted in the urine as prominent compounds. Many of these amphoteric metabolites are poorly extracted by conventional methods of sample preparation, and reference materials for them are rarely commercially available. The detection of newer benzodiazepines with rapid renal clearance poses problems for immunochemical and chromatographic methods. Treatment of urine samples with -glucuronidase enhances the positive rate of immunoassays greatly, as confirmed by HPLC [28] and GC-MS [29]. Our HPLC method, however, allows direct identification of benzodiazepine glucuronide while precluding the problem of lengthy enzymatic hydrolysis and the accompanying risk of structure-destroying chemical hydrolysis. Documentation of the retention times, relative abundances, and ultraviolet spectra for each drugs metabolites

Clinical Chemistry 43, No. 2, 1997

323

Table 3. Prospective assessment of quick urine screen by external quality-control program for urine toxicology.
Program cycle no. Target drug and dose HPLC result

3 - 07 Inpatient 3 - 08 Inpatient

3 - 09 Lab. staff 3 - 10 Inpatient 3 - 11 Volunteer

3 - 12 ED patient

4 - 01 Addicts

4 - 02 Overdose 4 - 03 Lab. worker 4 - 04 Volunteer

Dextropropoxyphene Acetaminophen Verapamil, 80 mg tds Ranitidine, 150 mg bd Phenolphthalein Chloramphenicol, 500 mg qid Salicylate, 100 mg Naproxen, 250 mg bd Pentobarbital (9 mg/L)a Salicylate (240 mg/L) Clozapine, 100 mg mane Erythromycin, 250 mg qid Acetaminophen Oxazepam (3 mg/L) Temazepam (2 mg/L) Nordiazepam (3 mg/L) Diazepam (1 mg/L) Morphine (1 mg/L) Amphetamine (1 mg/L) Codeine Acetaminophen Pethidine Ranitidine Metoclopramine Methadone Cannabinoid Codeine Acetaminophen Trimipramine Phenytoin Metoclopramine, trace Phentermine (3 mg/L) Trimethoprim Sulfamethoxazole Oxazepam (0.25 mg/L) Pseudoephedrine Dexchlorpheniramine

Dextropropoxyphene Acetaminophen Verapamil Ranitidine / Chloramphenicol Salicylate Naproxen Pentobarbital Salicylate Clozapine / Acetaminophen Oxazepam Temazepam Nordiazepam Diazepam Morphine Amphetamine Codeine Acetaminophen Pethidine Ranitidine Unidentified Methadone / Pholcodine / Imipramine Phenytoin / Amphetamine Trimethoprim Sulfamethoxazole Oxazepam Pseudoephedrine/ ephedrine /

Values in parentheses indicate concentration of added drug to prepare drug-containing urine sample.

in urine is advantageous. Recognition of multiple metabolites enhances identification certainty, reduces problems of interpretation arising from coeluted peaks or look-alike spectra, and extends the period after exposure in which the drugs can be detected in urine. Use of ultraviolet spectra and chromatographic retention data to aid metabolite identification was also attempted by Law and Stafford [30]. A wide range of common metabolic transformation, such as ring hydroxylation, O-demethylation, N-dealkylation, and formation of N- and S-oxides, could change the characteristic ultraviolet spectra. More work is needed to explore DAD data for metabolite identification.

The significance of metabolite detection in clinical toxicology includes: (a) identification of drugs metabolized rapidly, resulting in low serum or urine concentrations of the parent drugs, e.g., benzodiazepines; (b) detection of toxic metabolites related to clinical toxicity, e.g., 10-hydroxy-nortriptyline in a slow-metabolizing subject [31]; and (c) estimation of metabolite/parent drug ratios to assess the route, dose, and time of poisoning and to differentiate between recent drug use and use in the more distant past [32]. Importantly, the interpretation of urine or serum metabolite profiles requires knowledge of the relation between a drugs metabolism and its pharmacological or toxicological effects. Separation of hundreds of drugs and metabolites by HPLC requires gradient elution with wide polarity change. We initially adopted the mobile-phase system suggested by Bogusz and Wu [13], who used low-pH phosphate buffer containing triethylamine as an organic amine modifier. In urine samples, however, we found peaks congregating in the early part of the chromatogram, probably from extra drug metabolites extracted by our SPE method. Therefore, we added a constant amount of the anionic pair ion octyl sulfate, which selectively enhanced the retention of basic compounds. The modification resulted in a better spread of the drugs of interest over the entire run. Coupled with the universal recovery of weakly acidic, neutral, and basic drugs by mixed-mode SPE, a single HPLC run can detect various compounds that may be present simultaneously in multiple overdose, polydrug abuse, or drug interaction cases. Issues of standardization and interlaboratory comparability of HPLC data for drug screening have been extensively investigated [13, 33-36]. Much emphasis has been put on the calibration of HPLC retention time indexes, but there has been little assessment of ultraviolet spectra. Unlike GC-MS, the HPLC retention and ultraviolet-absorbing properties are more subject to variations in local conditions, such that precise execution of every technical detail of a reference HPLC-DAD method does not necessarily yield comparable databases. Reference series of retention time calibrations may fail to compensate for compounds with atypical functionalities (e.g., urine drug metabolites), given the inherent problems of inconsistent column manufacturing. Likewise, interlaboratory sharing of ultraviolet spectral data is difficult because of the diversity of optical design and diode resolution in the various detectors available. Our experience, as well as that of others [9, 10], suggests that an individual laboratory planning to establish an HPLC drug screen may have to develop its own search library. We stress that the express start-up and unattended shutdown steps are essential parts for running the stat HPLC so as to meet the random but urgent demand for overdose investigation. In initial screening for drug exposure, urine is the best sample for finding the greatest number of drugs in easily detectable quantities [6]. The SPE procedure, synchronized with the quick start-up

324

Lai et al.: Uniform solid-phase extraction for toxicological drug screening by HPLC

column equilibration within 30 min, yields a urine extract in which most drugs and metabolites are detectable without a lengthy procedure of extract evaporation. Thus, for a rapid screen for acute overdoses, the fast, qualitative urine screen almost always gives sufficient information for overdose management. In serum analysis, however, one must concentrate the SPE extract to avoid missing low-dose drugs. This special treatment may be reserved for in-depth investigation of cases in which a fast initial urine/serum screen shows discordance with the clinical presentation. In the nonurgent use of the method to confirm substance abuse, enrichment of the urine SPE extract is also required to match the detection cutoffs for those initial screens by immunoassays. The strength as well as the potential pitfall of the drug-screening system is well elucidated in the prospective assessment of our rapid method in the external quality control program for urine toxicology. Our HPLCDAD method is limited to its reference library entries; it is not yet established for analysis of cannabinoids (e.g.), which may require a dedicated procedure for sample preparation, chromatographic separation, and detection. The problem of misidentification in the external assessment program was restricted to structurally related compounds with similar ultraviolet spectra and small retention time difference. Compounds with such risks of mismatching should be noted; their correct differentiation will require more-stringent calibration of retention time (e.g., inclusion of an appropriate internal standard) or reanalysis of a sample supplemented with a known drug standard. Finally, the application of our method is validated by the high identification rate of drugs commonly encountered in typical cases of acute overdose, psychiatric therapy, or drug abuse. Some potentially toxic drugs that pose identification problems in other screening systems are readily identified by our HPLC method: e.g., chloramphenicol, pentobarbital, and phenytoin. The strength of our system in detecting these weakly acidic drugs is attributable to the universal recovery of the SPE procedure. The single SPE-HPLC-DAD procedure did not always provide clear-cut identification of every compound in routine cases. Considering the potential coverage of 300 substances in a gradient of 30 min, with an average peak width of 0.1 min, the chance of incomplete separation is obvious. Ultraviolet spectra provide relatively better information for drug compounds; the diversity of a normalized ultraviolet curve within the wavelength interval 210 350 nm permits positive identification of several drugs. However, a minor subset of compounds do not have characteristic ultraviolet curves. The worst scenario we observed involves the barbiturates. Because of their absence of ultraviolet absorption beyond 235 nm in the acidic mobile-phase environment, confirmation of their identity at low concentrations against the noisy background of a urine matrix could be difficult. Simple complementary methods such as immunoassays were helpful

in clarifying an identification. In other cases, metabolite patterns contributed to a positive identification. For instance, methadone has a low characteristic ultraviolet curve and its identification might be difficult should it coelute with other compounds; however, identifying the major metabolite of methadone, which shows a characteristic absorption band with maxima at 240 nm (Fig. 3), is fairly straightforward. Previous reports on clinical application of HPLC-DAD for toxicology screening have addressed identification problems from a different perspective. The routine for checking peak purity and spectral overlay described by Turcant et al. [10] is no different from what we have presented here, given that similar instruments were used in both studies. Other chemometric tools and computer algorithms applied include background subtraction [11], derivative spectroscopy [9], and multiple component analysis [16]. More-complicated mathematical treatment of DAD signals has been suggested for identifying incompletely resolved components within a chromatographic peak, even when unique spectral features are lacking [37]. Although we focused the single SPE-HPLC-DAD procedure on a wider coverage of drug classes and an exploration of additional information from urine metabolites, the procedure could readily be incorporated into advanced instrumentation and sophisticated computer data system for automated systematic analysis. Used in combination with other immunochemical or chromatographic techniques, HPLC-DAD can be a highly practicable and informative tool for drug screening and confirmation. In conclusion, we have proposed a single chromatographic process for most drugs of toxicological interest in clinical settings. The HPLC method optimizes drug recovery, chromatographic separation, and spectral information for broad-spectrum drug identification. It is simple, fast, and highly informative and easily lends itself to automation. The experimental approach detailed here should allow clinical laboratories to develop a cost-effective customized HPLC system to meet local demand.

We thank Y.S. Wong and T.W. Chan of the Hong Kong Government Laboratory for professional advice, C.W. Lam for sharing his book collections, and W.I. Lo for clinical evaluation of overdosed patients. This work was supported in part by the Research Fund of the Hospital Authority of Hong Kong.

References
1. Litovitz TL, Felberg L, Soloway RA, Ford M, Gellar R. 1994 annual report of the American Association of Poison Control Centers toxic exposure surveillance system. Am J Emerg Med 1995:13:551 97. 2. Lo DST, Chao TC, Tay MK. Toxicology services in Singapore. Ann Acad Med 1993;22:37 42. 3. Henderson A, Wright M, Pond SM. Experience with 732 acute

Clinical Chemistry 43, No. 2, 1997

325

4.

5.

6. 7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18. 19.

20.

overdose patients admitted to an intensive care unit over six years. Med J Aust 1993;158:28 30. Chan TYK, Critchley JAJH, Chan MTV, Yu CM. Drug overdose and other poisoning in Hong KongThe Prince of Wales Hospital (Shatin) experience. Hum Exp Toxicol 1994;13:5125. Widdop B. Hospital toxicology & drug abuse screening. In: Moffat AC, ed. Clarkes isolation and identification of drugs, 2nd ed. London, UK: Pharmaceutical Press, 1986:334. Osterloh JD. Utility and reliability of emergency toxicologic testing. Emerg Med Clin North Am 1990;8:693723. Maurer HH. Methods. In: Pfleger K, Maurer HH, Weber A, eds. Mass spectral and GC data of drugs, poisons, pesticides, pollutants and their metabolites, 2nd ed. Weinheim, Germany: VCH, 1992:332. Binder SR, Regalia M, Biaggi-McEachern M, Mazhar M. Automated liquid chromatographic analysis of drugs in urine by on-line sample clean-up and isocratic multi-column separation. J Chromatogr 1989;473:325 41. Logan BK, Stafford DT, Tebbett IR, Moore CM. Rapid screening for 100 basic drugs and metabolites in urine using cation exchange solid-phase extraction and high-performance liquid chromatography with diode array detection. J Anal Toxicol 1990;14:154 9. Turcant A, Premel-Cabic A, Cailleux A, Allain P. Toxicological screening of drugs by microbore high-performance liquid chromatography with photodiode-array detection and ultraviolet spectral library searches. Clin Chem 1991;37:1210 5. Puopolo PR, Volpicelli SA, Johnson DK, Flood JG. Emergency toxicology testing (detection, confirmation, and quantification) of basic drugs in serum by liquid chromatography with photodiode array detection. Clin Chem 1991;37:2124 30. Puopolo PR, Pothier ME, Volpicelli SA, Flood JG. Single procedure for detection, confirmation, and quantification of benzodiazepines in serum by liquid chromatography with photodiode array detection. Clin Chem 1991;37:701 6. Bogusz M, Wu M. Standardized HPLC/DAD system, based on retention indices and spectral library, applicable for systematic toxicological screening. J Anal Toxicol 1991;15:188 97. Drummer OH, Kotsos A, Mclntyre IM. A class-independent drug screen in forensic toxicology using a photodiode array detector. J Anal Toxicol 1993;17:2259. Demedts P, Wauters A, Franck F, Neels H. Evaluation of the REMEDi drug profiling system. Eur J Clin Chem Clin Biochem 1994;32:409 17. Koves EM. Use of high-performance liquid chromatography diode array detection in forensic toxicology. J Chromatogr A 1995;692: 10319. de Zeeuw RA, Hartstra J, Franke JP. Potential and pitfalls of chromatographic techniques and detection modes in substance identification for systematic toxicological analysis. J Chromatogr A 1994;674:313. Wolff K, Sanderson MJ, Hay AWM. A rapid horizontal TLC method for detecting drugs of abuse. Ann Clin Biochem 1990;27:482 8. Lillsunde P, Korte T. Comprehensive drug screening in urine using SPE and combined TLC and GC-MS identification. J Anal Toxicol 1991;15:71 81. Ferrara SD, Tedeschi L, Frison G, Castagna F. Solid-phase extraction and HPLC-UV confirmation of drugs of abuse in urine. J Anal Toxicol 1992;16:21722.

21. Sollans A, Carnicero M, Torre R, Segura J. Comprehensive screening procedure for detection of stimulants, narcotics, adrenergic drugs, and their metabolites in human urine. J Chromatogr 1995;19:104 14. 22. Chen XH, Wijsbeek J, Franke JP, de Zeeuw RA. A single column procedure on Bond Elut Certify for systematic toxicological analysis of drugs in plasma and urine. J Forensic Sci 1992;37:6171. 23. Chen XH, Franke JP, Wijsbeek J, de Zeeuw RA. Isolation of acidic, neutral and basic drugs from whole blood using a single mixedmode solid-phase extraction column. J Anal Toxicol 1992;16: 3515. 24. Chen XH, Franke JP, de Zeeuw RA. Principles of solid-phase extraction. In: Liu RH, Goldberger BA, eds. Handbook of workplace drug testing. Washington, DC: AACC Press, 1995:122. 25. Stockley IH, ed. Drug interactions, 3rd ed. Oxford, UK: Blackwell Scientific Publishers, 1995:932pp. 26. Baselt RC, Crabey RH. Disposition of toxic drugs and chemicals in man, 4th ed. Foster City, CA: Chemical Toxicology Institute, 1995:802pp. 27. Luo H, Hawes EM, Mckay G, Korchinski ED, Midha KK. The quaternary ammonium-linked glucuronide of doxepin: a major metabolite in depressed patients treated with doxepin. Drug Metab Dispos 1991;19:722 4. 28. Simonsson P, Lide n A, Lindberg S. Effect of -glucuronidase on urinary benzodiazepine concentrations determined by fluorescence polarization immunoassay. Clin Chem 1995;41:920 3. 29. Meatherall R. Benzodiazepine screening using EMIT II and TDx: urine hydrolysis pretreatment required. J Anal Toxicol 1994;18: 38590. 30. Law B, Stafford LE. The use of ultraviolet spectra and chromatographic retention data as an aid to metabolite identification. J Pharm Biochem Anal 1993;11:729 36. 31. Kalow W, Grant DM. Pharmacogenetics. In: Scriver CR, Beaudet AL, Sly WS, Valle D, eds. The metabolic and molecular basis of inherited disease, 7th ed. New York: McGraw Hill, 1995:293 326. 32. Tilstone WJ, Stead AH. Pharmacokinetics, metabolism, and the interpretation of results. In: Moffat AC, ed. Clarkes isolation and identification of drugs, 2nd ed. London, UK: Pharmaceutical Press, 1986:276 305. 33. Hill DW, Kind AJ. Reversed-phase solvent-gradient HPLC retention indexes of drugs. J Anal Toxicol 1994;18:233 42. 34. Bogusz M, Erkens M. Influence of biological matrix on chromatographic behavior and detection of selected acidic, neutral, and basic drugs examined by means of a standardized HPLC-DAD system. J Anal Toxicol 1995;19:49 55. 35. Maier RD, Bogusz M. Identification power of a standardized HPLC-DAD system for systematic toxicological analysis. J Anal Toxicol 1995;19:79 83. 36. Bogusz M, Hill DW, Rehorek A. Comparability of RP-HPLC detection indices of drugs in three databases. J Liq Chromatogr Rel Technol 1996;19:1291316. 37. Strasters JK. Chemometrics and photodiode array detection. In: Huber L, George SA, eds. Diode array detection in HPLC (Chromatographic Science series, Vol. 62). New York: Marcel Dekker, 1993:127 62.