| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
Department of Pathology, Princess Margaret Hospital, Lai Chi Kok, Hong Kong.
a Author for correspondence. Fax 852 23700969.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Key Words: indexing terms: chromatography, reversed-phase • toxicology • abused drugs • drug metabolism and disposition
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
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)(9)(10)(11)(12)(13)(14)(15)(16). The HPLC-DAD 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-phase extractions (SPEs) have been widely used in analyses of combined drug groups (18)(19)(20)(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 solid-phase 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 |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
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
x 2.1 mm (i.d.) precolumn and a 150 x 4.6 mm (i.d.) main
column packed with Hypersil 5 µm (particle size) octadecylsilane
(C18), also from Hewlett-Packard. Column temperature was
maintained at 45 °C.
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 ß-glucuronidase type H-2
(from Helix pomatia) were obtained from Sigma Chemical Co.
(St. Louis, MO). All drugs of pharmaceutical quality—in tablets,
capsules, or injection form—were 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.
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 µL of a mixture 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 µL 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 µL 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-µL extract for sensitive serum analysis.
detection and identification
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.
comparison of drug recovery
Drug-supplemented urine was prepared by adding to urine from a
drug-free subject ~10 µmol/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
nonspecific 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 µL of ß-glucuronidase crude extracts from H.
pomatia (ß-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.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
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 µmol/L. Compared with
the classical solvent-extraction method, 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.
|
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 (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 with respect to the
spectra of their parent drugs are tentatively identified.
|
Our approach to identify the glucuronide metabolites of
benzodiazepines, tricyclic antidepressants, and beta-blockers 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 N-glucuronide 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 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 phentermine—all 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 |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Our experiments with enzymatic digestion of urine samples from overdosed patients proved that the SPE can extract hydroxylated and glucuronide metabolites of midazolam, 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 ß-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 drug's metabolites 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 drug's 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)(34)(35)(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 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 HPLC-DAD 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.
|
Acknowledgments |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  S. H.-K. Lo, K.-L. Mo, K.-S. Wong, S.-P. Poon, C.-K. Chan, C.-K. Lai, and A. Chan Aristolochic acid nephropathy complicating a patient with focal segmental glomerulosclerosis Nephrol. Dial. Transplant., July 1, 2004; 19(7): 1913 - 1915. [Full Text] [PDF]
|
|
|
|
 |
|
 C.-K. Lai and A. Y.-W. Chan Tetrahydropalmatine Poisoning: Diagnoses of Nine Adult Overdoses Based on Toxicology Screens by HPLC with Diode-Array Detection and Gas Chromatography–Mass Spectrometry Clin. Chem., February 1, 1999; 45(2): 229 - 236. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
) indicate the direction of change in peak
absorbance after enzymatic hydrolysis. x-, y-axis
labels for simplified chromatograms and spectra as in Fig. 2