Development of a Radioimmunoassay for the Determination of Zolpidem in Biological Samples Isabel De Clerck and P. Daenens* Laboratory of Toxicology, Katholieke Universiteit Leuven, E. Van Evenstraat, 4, B-3000 Leuven, Belgium The development of a specific and sensitive radioimmunoassay for the detection of zolpidem and its metabolites in urine and serum samples is described. The assay can be used to pre-screen forensic and emergency samples. 6-Methyl-2-(4-methylphenyl)imidazo[1,2-a]- pyridine-3-acetic acid was prepared as a hapten and was coupled directly and indirectly to bovine serum albumin.The immunization of rabbits with the hapten–bovine serum albumin conjugates resulted in the production of highly specific antibodies, showing no significant cross-reactivities towards existing drugs. The fourparameter logistic model was used to process the calibration data into a fitting curve (r2 = 0.9764). Intraand inter-assay relative standard deviations were < 6.30 and < 12.28%, respectively.The limit of quantification was 0.1 ng ml21. Using this assay, zolpidem levels were determined in urine and serum, and could be easily detected up to 48 h after oral intake of 10 mg of zolpidem. Keywords: Radioimmunoassay; zolpidem; imidazo[1,2-a]pyridine Zolpidem [Fig. 1(a)], N,N,6-trimethyl-2-(4-methylphenyl)- imidazole[1,2-a]pyridine-3-acetamide l-(+)-hemitartrate, is a non-benzodiazepine hypnotic drug with an imidazopyridine backbone, which acts in the brain principally at receptors of the w1-receptor subtype belonging to the g-aminobutyric acid- (GABA)ergic system.Its main features are a fast onset of action and a short elimination half-life (2.4 h), unlike benzodiazepines. Stilnoct is administered orally (therapeutic dose = 10 mg). Therapeutic plasma concentrations are in the low nanogram range with average peak plasma levels of about 140 ng ml21 after 0.5–3 h.1 The drug is extensively metabolized and excreted in the urine as pharmacologically inactive metabolites.The biotransformation proceeds through three different pathways including (i) methyl oxidation on the phenyl moiety of the molecule (metabolite I, the main metabolite in urine that accounts for 52% of the administered dose), (ii) methyl oxidation on the imidazopyridine moiety leading to the corresponding carboxylic acid (metabolite II, that accounts for Å 10% of the administered dose) and (iii) hydroxylation on the imidazopyridine moiety (metabolite X, that accounts for < 10% of the administered dose).Another minor pathway, involving the hydroxylation of one of the methyl groups of the substituted amide (metabolite XI), is observed.2 Unchanged zolpidem has been observed only in trace amounts in urine ( < 10 ng ml21).3 Zolpidem has little potential for abuse because higher doses are associated with increased incidence of nausea and vomiting.4 Nevertheless, some cases of misuse have been reported.5,6 Fatal cases were always in combination with other psychotropic drugs and/or alcohol and could not be directly linked to zolpidem.7 Techniques used for the determination of zolpidem in biological samples are high-performance liquid chromatography (HPLC) combined with fluorimetric and UV detection8 and capillary gas chromatography with thermoionic detection (NPD).9 Because these methods are time consuming, they are not appropriate for the pre-screening of large numbers of samples.The aim of this work was to develop a sensitive and specific immunoassay for the detection of zolpidem and its metabolites. 6-Methyl-2-(4-methylphenyl)imidazo[1,2-a]pyridine- 3-acetic acid was chosen as the hapten and was coupled directly to bovine serum albumin (BSA) ( = immunogen 1). Alternatively, a spacer, consisting of three carbon atoms, was introduced on the carboxylic acid, prior to coupling to BSA ( = immunogen 2). Two series of four New Zealand rabbits were immunized with the two immunogens. N,N-Didemethylzolpidem- N-{2-{3-(4-hydroxy-3-[125I]iodophenyl)}methyl propionate} was synthesized and purified to serve as a radiotracer.The antisera obtained with immunogens 1 and 2 were compared and the optimum conditions for the radioimmunoassay (RIA) were determined. The test was validated and applied to urine and serum. Experimental Materials and Equipment 6-Methyl-2-(4-methylphenyl)imidazo[1,2-a]pyridine-3-acetic acid was synthesized in our laboratory by a slightly adapted procedure as described in the patent literature.10–12 Standard amounts of this product and of metabolites I, II and XI were obtained as a gift from Synth�elabo (Paris, France).Other reagents were obtained from the following sources: N,N’-carbonyldiimidazole from Aldrich (Steinheim, Germany); b-alanine methyl ester hydrochloride and l-tyrosine methyl ester from Fluka Chemika (Buchs, Switzerland); N,N-dimethylformamide (DMF) from UCB (Brussels, Belgium); Nhydroxysuccinimide (NHS), 1-(3-dimethylaminopropyl)- 3-ethylcarbodiimide hydrochloride (EDC) and N,NA-dicyclohexylcarbodiimide (DCC) from Acros Chimica Fig. 1 Structural formulae of zolpidem, (a); hapten 1, (b); hapten 2, (c); and the radiotracer, (d). Analyst, October 1997, Vol. 122 (1119–1124) 1119(Geel, Belgium); and sodium [125I]iodide IMS 30 from Amersham International (Amersham, Buckinghamshire, UK). Bovine serum albumin (BSA) (fraction 5), goat antiserum to rabbit g-globulin (GARGG), normal rabbit serum (NRS) and Freund’s complete and incomplete adjuvant were purchased from Calbiochem Biochemicals and Immunochemicals (San Diego, CA, USA).Norit Supra A was a gift from Norit (Amersfoort, The Netherlands). The Spectra/Por molecular porous cellulose membranes [relative molecular mass (Mr 12 000–14 000)] for the dialysis of the BSA conjugate were obtained from Spectrum Medical Industries (Los Angeles, CA, USA).All other reagents were obtained from Merck (Darmstadt, Germany). The acidified iodoplatinate reagent was prepared by adding 2 ml of hydrochloric acid (32%) to a solution of 0.25 g of hexachloroplatinic(iv) acid hexahydrate (40% Pt) and 5 g of potassium iodide in 100 ml of water. Thin-layer chromatography (TLC) was carried out using Polygram Sil G/UV254 plates (Machery-Nagel, D�uren, Germany). Liquid surface-assisted secondary ion mass spectrometry (L-SIMS) was performed with a Kratos Concept 1 H instrument (Kratos, Manchester, UK) using a 6 keV Cs+ beam and a thioglycerol matrix.Nuclear magnetic resonance (NMR) spectra were registered in deuteriated methanol, with addition of D2O and NaOD for hapten 1, and were taken with a Unity 500 MHz instrument (Varian, Palo Alto, CA, USA). The radioactivity of the tracer was counted on a g-counter (Berthold BF 5300, Wildbad, Germany). The degree of incorporation of the hapten was determined on a Lambda 16 UV/VIS spectrometer (Perkin- Elmer, Norwalk, CT, USA).HPLC was carried out by using a Merck–Hitachi Model L-6002 pump, equipped with a Rheodyne injector (Model 7125, Berkeley, CA, USA), supplied with a 200 ml sample loop. The analysis and purification of the radioligand were performed on an analytical LiChrospher Si-60 5 mm column (125-4) (Merck). The column eluates were monitored by using a g-counter detector (Canberra Industries, Mereden, CT, USA). Preparation of the Hapten To a suspension of 6-methyl-2-(4-methylphenyl)imidazo[1,2- a]pyridine-3-acetic acid (500 mg, 1.8 mmol) in tetrahydrofuran (THF) (12.5 ml), N,NA-carbonyldiimidazole (690 mg, 4.2 mmol) was added.The mixture was sonicated at room temperature. After 25, 40, 75 and 90 min, aliquots of the reaction mixture were taken and examined by TLC on silica plates with toluene– acetone–methanol (70 + 20 + 10, v/v/v). The reaction product (RF = 0.77) and the parent product (RF = 0.23) were localized using short-wave UV light (254 nm) and by spraying the plates with the acidified iodoplatinate reagent (brown spots).After 90 min, the reaction was complete. A suspension of b-alanine methyl ester hydrochloride (300 mg, 2.2 mmol) and triethylamine (220 mg, 2.2 mmol) in THF was added to the reaction mixture and sonicated at room temperature. The reaction was followed by TLC (RF = 0.67), using the same solvent mixture. The rean was complete after 30 min and the solvent was evaporated.Purification was carried out by adding 9 ml of water and 1 ml of saturated hydrogencarbonate solution to the residue, followed by extraction with diethyl ether (15 ml). After drying with anhydrous sodium sulfate, the solvent was evaporated under a stream of nitrogen. The ester (50 mg, 0.14 mmol) was suspended in distilled water and the pH was adjusted to 9 with sodium hydroxide (1 m). The mixture was sonicated until dissolution and concentrated hydrochloric acid (2 ml) was added.The mixture was allowed to stand at 4 °C until precipitation of the acid (RF = 0.08). The precipitate was filtered and dried. The structure of the hapten was confirmed by NMR and mass spectrometry ( = hapten 2) [Fig. 1(c)]. Alternatively, 6-methyl-2-(4-methylphenyl)imidazo[1,2- a]pyridine-3-acetic acid was used as a hapten ( = hapten 1) [Fig. 1(b)]. Preparation of the Immunogens Identical reactions were carried out for both haptens 1 and 2. The hapten (0.1 mmol) was dissolved in 7 ml of dimethyl sulfoxide (DMSO) and 11.5 mg (0.1 mmol) of NHS was added to this solution.A 20 mg (0.1 mmol) amount of EDC was dissolved in 200 ml of distilled water and slowly added to the hapten–NHS solution. The reactions were followed with the same TLC system as described above: RF hapten 1 = 0.23, RF activated ester 1 = 0.38; RF hapten 2 = 0.08 and RF activated ester 2 = 0.17. After 4 h, the reaction was almost complete and the solution was slowly added to a solution of BSA (67 mg) in 10 ml of phosphate buffer (pH 7.0, 0.05 mol l21).The mixture was stirred continuously and allowed to react for 24 h at room temperature. (60% DMSO was needed to keep the hapten in solution). The low molecular mass compounds were removed from the solution by dialysis, using a cellulose membrane with a cut-off value of 12 000–14 000 Da. The mixture was dialysed in phosphate buffer (pH 7.0, 0.05 mol l21), changing the buffer three times during the first day.After the first day, the buffer concentration was gradually decreased to 0.001 mol l21 and changed twice a day. The dialysis was stopped after 3 d and the hapten–BSA solution divided into portions of 2 ml each and stored at 220 °C. Degree of Incorporation The degree of incorporation of the hapten was estimated by UV spectrometry.13 The hapten has a maximum absorption at 308 nm. BSA does not interfere at this wavelength. Synthesis of the Tracer14 A 52 mg (0.45 mmol) amount of NHS and 93 mg (0.45 mmol) of DCC were added to a solution of 100 mg (0.36 mmol) of 6-methyl-2-(4-methylphenyl)-imidazo[1,2-a]pyridine-3-acetic acid, dissolved in 9 ml of DMF.The mixture was stirred for 1 h at room temperature and placed in a refrigerator (4 °C) overnight. The precipitate was removed by filtration and 58 mg (0.3 mmol) of l-tyrosine methyl ester were added to the filtrate. After mixing, the solution was allowed to stand at room temperature for 15 min.Purification was carried out by column chromatography. To a solution of 50 ml of methanol and 30 ml of phosphate buffer, 10 ml of l-tyrosine methyl ester conjugate solution (1 mg per 10 ml of methanol = 2.18 nmol) and 10 ml (1 mCi, 0.5 nmol) of sodium [125I]iodide solution were added. The temperature was kept at 0 °C. A 10 ml (0.75 nmol) portion of a freshly prepared chloramine-T solution was then added and the solution was vortex-mixed. Another 10 ml aliquot of the chloramine-T solution was added at 3 and 6 min.After 10 min, 20 ml of an aqueous solution of sodium metabisulfite (8.7 nmol) were added. Immunization Antisera were raised in two series of four New Zealand White Rabbits (immunogens 1 and 2). Aliquots (2 ml) of the dialysed hapten–BSA conjugate (90 nmol immunogen 1, 100 nmol immunogen 2) were emulsified with 3 ml of Freund’s complete adjuvant. The rabbits were immunized by subcutaneous injection of 1.2 ml of the water–oil emulsion.The injection volume was divided along 4–6 places on the back of the rabbit. The first two booster injections were given at a 2 week interval. The 1120 Analyst, October 1997, Vol. 122following booster solutions were made up with Freund’s incomplete adjuvant and injected at 4 week intervals. Small blood samples (10–15 ml) were collected every month from the lateral ear vein, starting 2 weeks after the second injection. Six months after the first injection, 2 weeks after the final booster, 50 ml blood samples were taken by cardiac puncture.Titration of Antisera Antisera were diluted (1 + 99 to 1 + 49 999) with phosphate buffer (pH 7.4, 0.05 mol l21) for titration experiments. The titres were determined by adding 100 ml of the antiserum dilution and 100 ml of the tracer (12 500 counts min21) to 400 ml of BSA matrix solution (2% BSA in phosphate buffer of pH 7.4). The mixtures were allowed to equilibrate at room temperature for 90 min.Bound and free radioligand were separated by a second antibody method using GARGG as described below. The serum dilution, able to bind 50% of the tracer, was calculated by constructing binder dilution curves. Optimization of the Assay Parameters that possibly influence the immunoassay such as pH, incubation time and concentration of BSA and tracer were tested. The following incubation conditions were kept constant for all experiments: to 300 ml of BSA solution (2% in phosphate buffer, pH 7.4, 0.05 mol l21) were added 100 ml of tracer, 100 ml of sample or standard solution and 100 ml of antiserum dilution.The mixture was allowed to equilibrate at room temperature for 90 min. For the optimization of the immunoassay conditions two methods for the separation of bound and free ligand were studied. First, the non-specific adsorption of the tracer onto charcoal was tested but was found to give excessive nonspecific (NSB) values (7–9%). As a second separation technique, a second antibody method using GARGG was tested.To optimize this procedure, several parameters were investigated, such as the incubation time and the optimum amounts of GARGG and NRS. After the initial incubation (90 min) at room temperature, bound and free radioligand were separated by adding 50 ml of NRS (5% in phosphate buffer) and 100 ml of GARGG (8% in phosphate buffer). After mixing, incubating for a further 20 h and centrifuging for 15 min at 3000g, both the supernatant (i.e., free fraction) and the pellet (i.e., bound fraction) were counted on a g-counter for 1 min.Calibration Graph The second antibody separation technique (GARGG) was selected for the construction of the calibration graph. Zolpidem was diluted in drug-free urine to produce a concentration range of 100–10 000 pg ml21 and 100 ml of the spiked samples were analysed with the RIA procedure. Non-specific binding was determined by replacing the antiserum by an equal volume of buffer.Calibration graphs were constructed by plotting B/B0, representing counts bound above non-specific, relative to counts bound above non-specific for zero dose of analyte, against the concentration of unlabelled ligand on a semilogarithmic scale. Human Samples To one healthy volunteer (female, 26 years), 10 mg of zolpidem (Stilnoct) were administered orally and urine and serum samples were collected over a 48 h period. Results Preparation of the Hapten, Immunogen and Tracer The main goal of our assay is to separate positive from negative urine samples and to have an idea of the concentration range of zolpidem and zolpidem-related material. 6-Methyl-2-(4-methylphenyl) imidazo[1,2-a]pyridine-3-acetic acid is the molecule of choice for realizing the synthesis of the hapten and the conjugation to BSA since the imidazo[1,2-a]pyridine part is the most specific part and the carboxylic acid can be used as the conjugation site. For the synthesis of hapten 2, 6-methyl- 2-(4-methylphenyl)imidazo[1,2-a]pyridine-3-acetic acid was first reacted with carbonyldiimidazole to form an acylimidazole derivative and the latter was subsequently reacted with b- alanine methyl ester to form an amide.Acidic hydrolysis of the ester function resulted in hapten 2. The structure was confirmed by mass spectrometry and NMR. The major fragments (m/z) of hapten 1 are 65, 92, 219, 235 and 280. For hapten 1, the assignment of the 13C peaks was based on an APT-spectrum (attached proton test) and on a heteronuclear-correlation experiment (HETCOR). 13C NMR (CD3OD/D2O + NaOD): d 177.6 (CO2H), 144.4 (C-2), 142.6 (C-1A), 138.8 (C-9), 131.8 (C-6), 130.2 (C-2A), 129.5 (C-7), 128.9 (C-3A), 123.7 (C-4A), 122.7 (C-5), 117.8 (C- 3), 115.8 (C-8), 33.6 (3-CH2), 21.2 (4A-Me), 18.3 (6-Me). 1H NMR (CD3OD/D2O + NaOD): d 2.17 (s, 3 H, 4A-Me), 2.23 (s, 3 H, 6-Me), 3.64 (s, 2 H, 3-CH2), 7.04 (dd, 1 H, J = 9.0/ 1.4 Hz, H-7), 7.15 (d, 2 H, J = 8.0 Hz, H-2A/6A), 7.31 (d, 1 H, J = 9.0 Hz, H-8), 7.41 (d, 2 H, J = 8.0 Hz, H-3A/5A), 7.71 (s, 1 H, H-5).The major fragments (m/z) of hapten 2 are : 65, 92, 219, 235 and 351. For hapten 2, the assignment of the peaks was based mainly on the assignments for hapten 1. 13C NMR (CD3OD): d 175.1 (CO2H), 169.7 (CONH), 142.4 (C-1A), 139.7 (C-9), 137.3 (C-7), 135.6 (C-2), 131.3 (C-2A), 129.6 (C-3A), 129.4 (C-6), 125.9 (C-5), 124.8 (C-4A), 118.5 (C- 3), 112.1 (C-8), 36.9 (C-b-alanyl), 34.5 (3-CH2), 30.9 (C-a- alanyl), 21.4 (4A-Me), 18.1 (6-Me). 1H NMR (CD3OD): d 2.43 (s, 3 H, 4A-Me), 2.52 (s, 3 H, 6-Me), 2.58 (t, 2 H, 3J = 6.45 Hz, CH2-a-alanyl), 3.52 (t, 2 H, J = 6.45 Hz, CH2-b-alanyl), 4.14 (s, 2 H, 3-CH2), 4.91 (br, NH, CO2H), 7.42 (d, 2 H, J = 7.8 Hz, H-2A/6A, 7.56 (d, 2 H, J = 8.1 Hz, H-3A/5A), 7.82 (d, 1 H, J = 9.2 Hz, H-8), 7.86 (dd, 1 H, J = 9.2/1.1 Hz, H-7), 8.5 (s, 1 H, H-5). The hapten–BSA conjugate was prepared by the carbodiimide technique, which was first described by Sheehan and Hess.15 By using UV spectrometry, a hapten : protein ratio (mmol) of 21.2 : 0.96 and 16.5 : 0.96 was calculated for immunogens 1 and 2, respectively, corresponding to the coupling of about 22 and 17 mol, respectively, of hapten with 1 mol of BSA.An average of 15 molecules of hapten per molecule of BSA has been recommended for an optimum immuno-response.16 The synthesis and purification of N,N-didemethylzolpidem- N-{2-{3-(4-hydroxy-3-[125I]-iodophenyl)}methyl propionate} [Fig. 1(d)] have already been described in detail.14 The freshly prepared tracer could be used for about 6 weeks without any loss of binding to the antibody. Titre of the Antibodies After the examination of the serial dilutions of all the antisera from the different bleeds, the titres were derived from binder dilution curves. The evolution of the titres of the different rabbits, immunized with immunogens 1 and 2, respectively, is shown in Fig. 2. As could be expected, great differences exist in immunogenic responses between immunogens 1 and 2, with titres ranging from 1/75 to 1/3200 for the immunogen without spacer and from 1/12 800 to 1/25 000 for the immunogen with spacer. Because of the low titres, the sera of rabbits 1.1 and 1.4 were not considered for further experiments. The antisera of the last bleeds of all the rabbits were lyophilized and stored at 4 °C until use. The optimization and specificity experiments were performed for all the antisera except for R1.1 and R1.4.All Analyst, October 1997, Vol. 122 1121other experiments were only carried out with the antiserum of R2.1. Determination of Optimum Conditions for the RIA As a separation technique, a second antibody method was evaluated.17 The optimum ratio of GARGG antiserum to NRS was investigated. Final dilutions of GARGG (1 + 124) and NRS (1 + 999) resulted in the highest concentrations of tracer. The optimum second incubation time for the precipitation of the antibody-bound tracer was 20 h at room temperature.In comparison with the adsorption technique, this method resulted in lower NSB values (1–3%) and in higher dilutions of the antiserum to reach 50% binding of the tracer (1 + 12 799). The performance characteristics of the immunoassay are demonstrated in a calibration graph obtained with the serum of R2.1 for the concentration range 100–10 000 pg ml21 (Fig. 3), using the second antibody method as the separation technique.The curve of best fit was obtained by applying the four-parametric logistic model to the experimental data.18 Limit of Quantification The limit of quantification (LOQ) of the RIA can be defined as the lowest concentration on the calibration graph that can be measured with acceptable precision and accuracy. The precision at this level has to be less than or equal to 20%. To determine the LOQ, five standard solutions of increasing concentration were used and the relative standard deviation was determined.The LOQ was 0.1 ng ml21. Precision and Recovery The precision and recovery of the method could be calculated after replicate analyses of independently prepared quality control urine samples. The concentration levels of the samples ranged from 0.25 to 1000 ng ml21. The intra- and inter-assay relative standard deviations were 4.01–6.30 and 9.13–12.28%, respectively, and the accuracy was about 100% (Table 1). Specificity The antibody specificity was assessed by measuring the crossreactivity to other hypnotics and to some widely used anxiolytics, antidepressants, analgesics and analeptics. The following compounds were found not to be detectable at a level of 10 mg ml21: chlordiazepoxide, diazepam, clobazam, nitrazepam, flunitrazepam, bromazepam, lormetazepam, triazolam, meprobamate, methaqualone, hexobarbital, fenobarbital, buspirone, hydroxyzine, chlorcyclizine, codeine, cocaine, caffein, ibuprofen, trazodone, zopiclone, chlorpromazine and acetylsalicylic acid.None of the above was found to cross-react ( < 0.005%) with the immunoassay. To rule out cross-reactivity with endogenous compounds, several urine samples of volunteers were screened and no crossreactivity could be detected. To select one antiserum out of the pool, the specificity towards the metabolites was examined. The two major metabolites of zolpidem, metabolites I and II, and a minor metabolite (metabolite XI) were tested for their cross-reactivity.The degree of cross-reactivity of the metabolites was expressed as the relative dose required for 50% displacement of the maximum tracer binding. The major metabolite (I) showed a low cross-reactivity, ranging from 2.3 3 1027 to 7.7 3 1023%. This could be expected considering that the position of the Fig. 2 Evolution of the serum titres of the two series of rabbits, immunized with the immunogens 1 and 2, respectively. (a) represents the antisera from rabbits R1.1 (/), R1.2 (~), 1.3 (-) and R1.4 (3) and (b) shows the titres of R2.1 (/), R2.2 (~), R2.3 (-) and R2.4 (3).Fig. 3 Calibration graph for zolpidem, together with 95% confidence intervals. The graph follows the equation y = d [(a 2 d)/1 + (x/c)b], where a = 0.8972 ± 0.01006; b = 9.9390 ± 0.4498; c = 6.8896 ± 0.0280; d = 0.02248 ± 0.01195; and r2 = 0.9764. Table 1 Intra- and inter-assay variance characteristics and recovery Added/ng RSD (%) n Recovery (%) Intra-assay 0.5 4.01 6 103.4 1.75 4.48 6 101.5 7.5 6.30 6 98.1 1000 5.90 6 99.3 Inter-assay 0.25 12.28 10 100.9 1 9.30 9 99.4 7.5 7.40 10 101.3 1000 9.13 9 104.0 1122 Analyst, October 1997, Vol. 122carboxylic acid function is close to the binding site with the antibodies. The cross-reactivity towards metabolite II was higher and varied from 0.9 to 6.9%. In this metabolite, the substituent is positioned further away from the binding site. Metabolite XI, which has an identical structure with zolpidem, except for the hydroxylic function on the acetamide side-chain in position 3 of the imidazopyridine skeleton, showed a very high cross-reactivity (91–116%).Because of the pre-screening purpose of this RIA, the antiserum with the highest crossreactivity for the main metabolite was selected, i.e., the antiserum of rabbit 2.1 (6th bleed). The results are shown in Table 2. Linearity To confirm the linearity of the response throughout the range of the calibration graph, samples were checked to see whether they diluted in parallel to the calibration graph.An unknown urine sample, containing approximately 4.6 ng ml21 of the drug, was analysed three times, both undiluted and diluted with blank urine (1 + 1, 1 + 3, 1 + 7 and 1 + 15). The observed (y) and the expected (x) values yielded the following linear regression equation : y = 0.0115 + 0.9703x (r = 0.9938). Human Urine and Serum Samples A set of 14 urine samples (including a blank) were collected from a volunteer during a 48 h period after the intake of 10 mg of zolpidem (Stilnoct).The samples were analysed using the described RIA procedure and the results were expressed in terms of zolpidem concentration, although there was also a contribution from the metabolites. The highest value (994 ng ml21) was reached 3 h after the ingestion of one tablet of Stilnoct. After 48 h, the hypnotic could still easily be detected (about 6.3 ng ml21).From the same volunteer and at specific time intervals, 13 serum samples (including a blank) were collected and analysed by RIA. The serum levels measured covered a concentration from 1.3 to 122 ng ml21. The highest value was recorded 2 h after administration (about 122 ng ml21). Thereafter, the level of zolpidem decreased to 1.3 ng ml21 after 48 h. Fig. 4 shows the serum concentration–time profile of zolpidem. Discussion This paper describes the development of an RIA for zolpidem and its metabolites in urine and serum. Two immunogens were synthesized to study the difference in affinity and specificity between an immunogen with and without a spacer.The antibodies from the rabbits immunized with the immunogen with a spacer showed a much higher affinity than those immunized with the other immunogen. A difference in specificity could not be seen. The synthesis and purification of N,N-didemethylzolpidem-N-{2-{3-(4-hydroxy-3-[125I]-iodophenyl)} methyl propionate} resulted in a high radiochemical purity and specific activity of the radioligand,14 thereby providing a very sensitive immunoassay (LOQ = 0.1 ng ml21).Owing to the nature of the synthesized hapten and immunogens, the detected cross-reactivity of the three metabolites is to be expected. The important fact is that zolpidem and its metabolites can be effectively detected in urine and serum up to 48 h after the intake of one tablet of Stilnoct (10 mg of zolpidem), without interference from other drugs.The proposed RIA can be used in the pre-screening of toxicological samples. Its applica- Table 2 Specificity of the antisera towards three metabolites of zolpidem in comparison with the unmetabolised drug Cross-reactivity (%) Compound R1.2 R1.3 R2.1 R2.2 R2.3 R2.4 Zolpidem 100 100 100 100 100 100 7.7 3 1023 2.3 3 1023 2.3 3 1027 4.2 3 1023 3 3 1024 1.6 3 1023 2.53 2.4 5.55 0.9 6.9 2.9 91 116 116 114 104 92 Analyst, October 1997, Vol. 122 1123bility to the analysis of blood and plasma samples is under study. Dr. R. Busson, Dr. G. Janssen and Dr. J. Rozensky are gratefully acknowledged for recording the NMR and mass spectra. References 1 Th�enot, J. P., Hermann, P., Durand, A., Burke, J. T., Allen, J., Garrigou, D., Vajta, S., Albin, H., Th�ebault, J. J., Olive, G., and Warrington, S. J., in Imidazopyridines in Sleep Disorders, ed. Sauvanet, J. P., Langer, S. Z., and Morselli, P. L., Raven Press, New York, 1988, pp. 139–153. 2 Durand, A., Th�enot, J. P., Bianchetti, G., and Morselli, P. L., Drug Metab. Rev., 1992, 24, 239. 3 Augsburger, M., Giroud, C., Lucchini, P., and Rivier, L., in Contributions to Forensic Toxicology. The Proceedings of the 31st International Meeting of the International Association of Forensic Toxicologists (TIAFT), 1993, ed. Mueller, R. K., Molinapress, Leipzig, 1994, pp. 18–22. 4 Mendelson, W. B., and Jain, B., Drug Saf., 1995, 13, 257. 5 Debailleul, G., Abi Khalil, F., and Lheureux, P., J.Anal. Toxicol., 1991, 15, 35. 6 Khodasevitch, T., and Volgram, J., Bull. Int. Assoc. Forensic Toxicol., 1996, 26(2), 37. 7 Garnier, R., Guerault, E., Muzard, D., Azoyan, P., Chaumet- Riffaud, A., and Efthymiou, M., Clin. Toxicol., 1994, 32, 391. 8 Guinebault, P., Dubruc, C., Hermann, P., and Th�enot, J. P., J. Chromatogr., 1986, 383, 206. 9 Debruyne, D., Lacotte, J., Hurault De Ligny, B., and Moulin, M., J. Pharm. Sci., 1991, 80, 71. 10 Almirante, L., and Murwann, W., Br.Pat., 1 076 089, 1967. 11 Kaplan, J. P., and George, P., Eur. Pat., 0 050 563, 1982. 12 Kaplan, J. P., and George, P., US Pat., 4 501 745, 1985. 13 Erlanger, B. F., Methods in Enzymology, Academic Press, New York, 1980, vol. 70, p. 85. 14 De Clerck, I., and Daenens, P., J. Radiolabelled Comp. Radiopharm., 1997, 39, 195. 15 Sheehan, J. C., and Hess, G. P., J. Am. Chem. Soc., 1955, 77, 1067. 16 Erlanger, B. F., Pharmacol. Rev., 1973, 25, 271. 17 Utiger, C.R., Parker, M. L., and Daughaday, W. H., J. Clin. Invest., 1962, 41, 254. 18 Dudley, R., Edwards, P., Ekins, R. P., Finney, D. J., McKenzie, I. G. M., Raab, G. M., Rodbard, D., and Rodgers, R. P. C., Clin. Chem. (Winston-Salem, N.C.), 1985, 31, 1264. Paper 7/02869E Received April 28, 1997 Accepted June 6, 1997 Fig. 4 Serum levels, expressed in terms of zolpidem concentration, in one healthy volunteer after a single oral intake of one tablet (10 mg) of Stilnoct. 1124 Analyst, October 1997, Vol. 122 Development of a Radioimmunoassay for the Determination of Zolpidem in Biological Samples Isabel De Clerck and P. Daenens* Laboratory of Toxicology, Katholieke Universiteit Leuven, E. Van Evenstraat, 4, B-3000 Leuven, Belgium The development of a specific and sensitive radioimmunoassay for the detection of zolpidem and its metabolites in urine and serum samples is described. The assay can be used to pre-screen forensic and emergency samples. 6-Methyl-2-(4-methylphenyl)imidazo[1,2-a]- pyridine-3-acetic acid was prepared as a hapten and was coupled directly and indirectly to bovine serum albumin. The immunization of rabbits with the hapten–bovine serum albumin conjugates resulted in the production of highly specific antibodies, showing no significant cross-reactivities towards existing drugs.The fourparameter logistic model was used to process the calibration data into a fitting curve (r2 = 0.9764). Intraand inter-assay relative standard deviations were < 6.30 and < 12.28%, respectively.The limit of quantification was 0.1 ng ml21. Using this assay, zolpidem levels were determined in urine and serum, and could be easily detected up to 48 h after oral intake of 10 mg of zolpidem. Keywords: Radioimmunoassay; zolpidem; imidazo[1,2-a]pyridine Zolpidem [Fig. 1(a)], N,N,6-trimethyl-2-(4-methylphenyl)- imidazole[1,2-a]pyridine-3-acetamide l-(+)-hemitartrate, is a non-benzodiazepine hypnotic drug with an imidazopyridine backbone, which acts in the brain principally at receptors of the w1-receptor subtype belonging to the g-aminobutyric acid- (GABA)ergic system.Its main features are a fast onset of action and a short elimination half-life (2.4 h), unlike benzodiazepines. Stilnoct is administered orally (therapeutic dose = 10 mg). Therapeutic plasma concentrations are in the low nanogram range with average peak plasma levels of about 140 ng ml21 after 0.5–3 h.1 The drug is extensively metabolized and excreted in the urine as pharmacologically inactive metabolites.The biotransformation proceeds through three different pathways including (i) methyl oxidation on the phenyl moiety of the molecule (metabolite I, the main metabolite in urine that accounts for 52% of the administered dose), (ii) methyl oxidation on the imidazopyridine moiety leading to the corresponding carboxylic acid (metabolite II, that accounts for Å 10% of the administered dose) and (iii) hydroxylation on the imidazopyridine moiety (metabolite X, that accounts for < 10% of the administered dose).Another minor pathway, involving the hydroxylation of one of the methyl groups of the substituted amide (metabolite XI), is observed.2 Unchanged zolpidem has been observed only in trace amounts in urine ( < 10 ng ml21).3 Zolpidem has little potential for abuse because higher doses are associated with increased incidence of nausea and vomiting.4 Nevertheless, some cases of misuse have been reported.5,6 Fatal cases were always in combination with other psychotropic drugs and/or alcohol and could not be directly linked to zolpidem.7 Techniques used for the determination of zolpidem in biological samples are high-performance liquid chromatography (HPLC) combineduorimetric and UV detection8 and capillary gas chromatography with thermoionic detection (NPD).9 Because these methods are time consuming, they are not appropriate for the pre-screening of large numbers of samples.The aim of this work was to develop a sensitive and specific immunoassay for the detection of zolpidem and its metabolites. 6-Methyl-2-(4-methylphenyl)imidazo[1,2-a]pyridine- 3-acetic acid was chosen as the hapten and was coupled directly to bovine serum albumin (BSA) ( = immunogen 1). Alternatively, a spacer, consisting of three carbon atoms, was introduced on the carboxylic acid, prior to coupling to BSA ( = immunogen 2).Two series of four New Zealand rabbits were immunized with the two immunogens. N,N-Didemethylzolpidem- N-{2-{3-(4-hydroxy-3-[125I]iodophenyl)}methyl propionate} was synthesized and purified to serve as a radiotracer. The antisera obtained with immunogens 1 and 2 were compared and the optimum conditions for the radioimmunoassay (RIA) were determined. The test was validated and applied to urine and serum. Experimental Materials and Equipment 6-Methyl-2-(4-methylphenyl)imidazo[1,2-a]pyridine-3-acetic acid was synthesized in our laboratory by a slightly adapted procedure as described in the patent literature.10–12 Standard amounts of this product and of metabolites I, II and XI were obtained as a gift from Synth�elabo (Paris, France).Other reagents were obtained from the following sources: N,N’-carbonyldiimidazole from Aldrich (Steinheim, Germany); b-alanine methyl ester hydrochloride and l-tyrosine methyl ester from Fluka Chemika (Buchs, Switzerland); N,N-dimethylformamide (DMF) from UCB (Brussels, Belgium); Nhydroxysuccinimide (NHS), 1-(3-dimethylaminopropyl)- 3-ethylcarbodiimide hydrochloride (EDC) and N,NA-dicyclohexylcarbodiimide (DCC) from Acros Chimica Fig. 1 Structural formulae of zolpidem, (a); hapten 1, (b); hapten 2, (c); and the radiotracer, (d). Analyst, October 1997, Vol. 122 (1119–1124) 1119(Geel, Belgium); and sodium [125I]iodide IMS 30 from Amersham International (Amersham, Buckinghamshire, UK).Bovine serum albumin (BSA) (fraction 5), goat antiserum to rabbit g-globulin (GARGG), normal rabbit serum (NRS) and Freund’s complete and incomplete adjuvant were purchased from Calbiochem Biochemicals and Immunochemicals (San Diego, CA, USA). Norit Supra A was a gift from Norit (Amersfoort, The Netherlands). The Spectra/Por molecular porous cellulose membranes [relative molecular mass (Mr 12 000–14 000)] for the dialysis of the BSA conjugate were obtained from Spectrum Medical Industries (Los Angeles, CA, USA).All other reagents were obtained from Merck (Darmstadt, Germany). The acidified iodoplatinate reagent was prepared by adding 2 ml of hydrochloric acid (32%) to a solution of 0.25 g of hexachloroplatinic(iv) acid hexahydrate (40% Pt) and 5 g of potassium iodide in 100 ml of water. Thin-layer chromatography (TLC) was carried out using Polygram Sil G/UV254 plates (Machery-Nagel, D�uren, Germany). Liquid surface-assisted secondary ion mass spectrometry (L-SIMS) was performed with a Kratos Concept 1 H instrument (Kratos, Manchester, UK) using a 6 keV Cs+ beam and a thioglycerol matrix.Nuclear magnetic resonance (NMR) spectra were registered in deuteriated methanol, with addition of D2O and NaOD for hapten 1, and were taken with a Unity 500 MHz instrument (Varian, Palo Alto, CA, USA). The radioactivity of the tracer was counted on a g-counter (Berthold BF 5300, Wildbad, Germany). The degree of incorporation of the hapten was determined on a Lambda 16 UV/VIS spectrometer (Perkin- Elmer, Norwalk, CT, USA).HPLC was carried out by using a Merck–Hitachi Model L-6002 pump, equipped with a Rheodyne injector (Model 7125, Berkeley, CA, USA), supplied with a 200 ml sample loop. The analysis and purification of the radioligand were performed on an analytical LiChrospher Si-60 5 mm column (125-4) (Merck). The column eluates were monitored by using a g-counter detector (Canberra Industries, Mereden, CT, USA).Preparation of the Hapten To a suspension of 6-methyl-2-(4-methylphenyl)imidazo[1,2- a]pyridine-3-acetic acid (500 mg, 1.8 mmol) in tetrahydrofuran (THF) (12.5 ml), N,NA-carbonyldiimidazole (690 mg, 4.2 mmol) was added. The mixture was sonicated at room temperature. After 25, 40, 75 and 90 min, aliquots of the reaction mixture were taken and examined by TLC on silica plates with toluene– acetone–methanol (70 + 20 + 10, v/v/v). The reaction product (RF = 0.77) and the parent product (RF = 0.23) were localized using short-wave UV light (254 nm) and by spraying the plates with the acidified iodoplatinate reagent (brown spots).After 90 min, the reaction was complete. A suspension of b-alanine methyl ester hydrochloride (300 mg, 2.2 mmol) and triethylamine (220 mg, 2.2 mmol) in THF was added to the reaction mixture and sonicated at room temperature. The reaction was followed by TLC (RF = 0.67), using the same solvent mixture. The reaction was complete after 30 min and the solvent was evaporated.Purification was carried out by adding 9 ml of water and 1 ml of saturated hydrogencarbonate solution to the residue, followed by extraction with diethyl ether (15 ml). After drying with anhydrous sodium sulfate, the solvent was evaporated under a stream of nitrogen. The ester (50 mg, 0.14 mmol) was suspended in distilled water and the pH was adjusted to 9 with sodium hydroxide (1 m). The mixture was sonicated until dissolution and concentrated hydrochloric acid (2 ml) was added.The mixture was allowed to stand at 4 °C until precipitation of the acid (RF = 0.08). The precipitate was filtered and dried. The structure of the hapten was confirmed by NMR and mass spectrometry ( = hapten 2) [Fig. 1(c)]. Alternatively, 6-methyl-2-(4-methylphenyl)imidazo[1,2- a]pyridine-3-acetic acid was used as a hapten ( = hapten 1) [Fig. 1(b)]. Preparation of the Immunogens Identical reactions were carried out for both haptens 1 and 2.The hapten (0.1 mmol) was dissolved in 7 ml of dimethyl sulfoxide (DMSO) and 11.5 mg (0.1 mmol) of NHS was added to this solution. A 20 mg (0.1 mmol) amount of EDC was dissolved in 200 ml of distilled water and slowly added to the hapten–NHS solution. The reactions were followed with the same TLC system as described above: RF hapten 1 = 0.23, RF activated ester 1 = 0.38; RF hapten 2 = 0.08 and RF activated ester 2 = 0.17. After 4 h, the reaction was almost complete and the solution was slowly added to a solution of BSA (67 mg) in 10 ml of phosphate buffer (pH 7.0, 0.05 mol l21).The mixture was stirred continuously and allowed to react for 24 h at room temperature. (60% DMSO was needed to keep the hapten in solution). The low molecular mass compounds were removed from the solution by dialysis, using a cellulose membrane with a cut-off value of 12 000–14 000 Da. The mixture was dialysed in phosphate buffer (pH 7.0, 0.05 mol l21), changing the buffer three times during the first day.After the first day, the buffer concentration was gradually decreased to 0.001 mol l21 and changed twice a day. The dialysis was stopped after 3 d and the hapten–BSA solution divided into portions of 2 ml each and stored at 220 °C. Degree of Incorporation The degree of incorporation of the hapten was estimated by UV spectrometry.13 The hapten has a maximum absorption at 308 nm. BSA does not interfere at this wavelength.Synthesis of the Tracer14 A 52 mg (0.45 mmol) amount of NHS and 93 mg (0.45 mmol) of DCC were added to a solution of 100 mg (0.36 mmol) of 6-methyl-2-(4-methylphenyl)-imidazo[1,2-a]pyridine-3-acetic acid, dissolved in 9 ml of DMF. The mixture was stirred for 1 h at room temperature and placed in a refrigerator (4 °C) overnight. The precipitate was removed by filtration and 58 mg (0.3 mmol) of l-tyrosine methyl ester were added to the filtrate. After mixing, the solution was allowed to stand at room temperature for 15 min.Purification was carried out by column chromatography. To a solution of 50 ml of methanol and 30 ml of phosphate buffer, 10 ml of l-tyrosine methyl ester conjugate solution (1 mg per 10 ml of methanol = 2.18 nmol) and 10 ml (1 mCi5 nmol) of sodium [125I]iodide solution were added. The temperature was kept at 0 °C. A 10 ml (0.75 nmol) portion of a freshly prepared chloramine-T solution was then added and the solution was vortex-mixed. Another 10 ml aliquot of the chloramine-T solution was added at 3 and 6 min.After 10 min, 20 ml of an aqueous solution of sodium metabisulfite (8.7 nmol) were added. Immunization Antisera were raised in two series of four New Zealand White Rabbits (immunogens 1 and 2). Aliquots (2 ml) of the dialysed hapten–BSA conjugate (90 nmol immunogen 1, 100 nmol immunogen 2) were emulsified with 3 ml of Freund’s complete adjuvant. The rabbits were immunized by subcutaneous injection of 1.2 ml of the water–oil emulsion.The injection volume was divided along 4–6 places on the back of the rabbit. The first two booster injections were given at a 2 week interval. The 1120 Analyst, October 1997, Vol. 122following booster solutions were made up with Freund’s incomplete adjuvant and injected at 4 week intervals. Small blood samples (10–15 ml) were collected every month from the lateral ear vein, starting 2 weeks after the second injection.Six months after the first injection, 2 weeks after the final booster, 50 ml blood samples were taken by cardiac puncture. Titration of Antisera Antisera were diluted (1 + 99 to 1 + 49 999) with phosphate buffer (pH 7.4, 0.05 mol l21) for titration experiments. The titres were determined by adding 100 ml of the antiserum dilution and 100 ml of the tracer (12 500 counts min21) to 400 ml of BSA matrix solution (2% BSA in phosphate buffer of pH 7.4). The mixtures were allowed to equilibrate at room temperature for 90 min.Bound and free radioligand were separated by a second antibody method using GARGG as described below. The serum dilution, able to bind 50% of the tracer, was calculated by constructing binder dilution curves. Optimization of the Assay Parameters that possibly influence the immunoassay such as pH, incubation time and concentration of BSA and tracer were tested. The following incubation conditions were kept constant for all experiments: to 300 ml of BSA solution (2% in phosphate buffer, pH 7.4, 0.05 mol l21) were added 100 ml of tracer, 100 ml of sample or standard solution and 100 ml of antiserum dilution.The mixture was allowed to equilibrate at room temperature for 90 min. For the optimization of the immunoassay conditions two methods for the separation of bound and free ligand were studied. First, the non-specific adsorption of the tracer onto charcoal was tested but was found to give excessive nonspecific (NSB) values (7–9%). As a second separation technique, a second antibody method using GARGG was tested. To optimize this procedure, several parameters were investigated, such as the incubation time and the optimum amounts of GARGG and NRS.After the initial incubation (90 min) at room temperature, bound and free radioligand were separated by adding 50 ml of NRS (5% in phosphate buffer) and 100 ml of GARGG (8% in phosphate buffer). After mixing, incubating for a further 20 h and centrifuging for 15 min at 3000g, both the supernatant (i.e., free fraction) and the pellet (i.e., bound fraction) were counted on a g-counter for 1 min.Calibration Graph The second antibody separation technique (GARGG) was selected for the construction of the calibration graph. Zolpidem was diluted in drug-free urine to produce a concentration range of 100–10 000 pg ml21 and 100 ml of the spiked samples were analysed with the RIA procedure. Non-specific binding was determined by replacing the antiserum by an equal volume of buffer.Calibration graphs were constructed by plotting B/B0, representing counts bound above non-specific, relative to counts bound above non-specific for zero dose of analyte, against the concentration of unlabelled ligand on a semilogarithmic scale. Human Samples To one healthy volunteer (female, 26 years), 10 mg of zolpidem (Stilnoct) were administered orally and urine and serum samples were collected over a 48 h period.Results Preparation of the Hapten, Immunogen and Tracer The main goal of our assay is to separate positive from negative urine samples and to have an idea of the concentration range of zolpidem and zolpidem-related material. 6-Methyl-2-(4-methylphenyl) imidazo[1,2-a]pyridine-3-acetic acid is the molecule of choice for realizing the synthesis of the hapten and the conjugation to BSA since the imidazo[1,2-a]pyridine part is the most specific part and the carboxylic acid can be used as the conjugation site.For the synthesis of hapten 2, 6-methyl- 2-(4-methylphenyl)imidazo[1,2-a]pyridine-3-acetic acid was first reacted with carbonyldiimidazole to form an acylimidazole derivative and the latter was subsequently reacted with b- alanine methyl ester to form an amide. Acidic hydrolysis of the ester function resulted in hapten 2. The structure was confirmed by mass spectrometry and NMR. The major fragments (m/z) of hapten 1 are 65, 92, 219, 235 and 280.For hapten 1, the assignment of the 13C peaks was based on an APT-spectrum (attached proton test) and on a heteronuclear-correlation experiment (HETCOR). 13C NMR (CD3OD/D2O + NaOD): d 177.6 (CO2H), 144.4 (C-2), 142.6 (C-1A), 138.8 (C-9), 131.8 (C-6), 130.2 (C-2A), 129.5 (C-7), 128.9 (C-3A), 123.7 (C-4A), 122.7 (C-5), 117.8 (C- 3), 115.8 (C-8), 33.6 (3-CH2), 21.2 (4A-Me), 18.3 (6-Me). 1H NMR (CD3OD/D2O + NaOD): d 2.17 (s, 3 H, 4A-Me), 2.23 (s, 3 H, 6-Me), 3.64 (s, 2 H, 3-CH2), 7.04 (dd, 1 H, J = 9.0/ 1.4 Hz, H-7), 7.15 (d, 2 H, J = 8.0 Hz, H-2A/6A), 7.31 (d, 1 H, J = 9.0 Hz, H-8), 7.41 (d, 2 H, J = 8.0 Hz, H-3A/5A), 7.71 (s, 1 H, H-5).The major fragments (m/z) of hapten 2 are : 65, 92, 219, 235 and 351. For hapten 2, the assignment of the peaks was based mainly on the assignments for hapten 1. 13C NMR (CD3OD): d 175.1 (CO2H), 169.7 (CONH), 142.4 (C-1A), 139.7 (C-9), 137.3 (C-7), 135.6 (C-2), 131.3 (C-2A), 129.6 (C-3A), 129.4 (C-6), 125.9 (C-5), 124.8 (C-4A), 118.5 (C- 3), 112.1 (C-8), 36.9 (C-b-alanyl), 34.5 (3-CH2), 30.9 (C-a- alanyl), 21.4 (4A-Me), 18.1 (6-Me). 1H NMR (CD3OD): d 2.43 (s, 3 H, 4A-Me), 2.52 (s, 3 H, 6-Me), 2.58 (t, 2 H, 3J = 6.45 Hz, CH2-a-alanyl), 3.52 (t, 2 H, J = 6.45 Hz, CH2-b-alanyl), 4.14 (s, 2 H, 3-CH2), 4.91 (br, NH, CO2H), 7.42 (d, 2 H, J = 7.8 Hz, H-2A/6A, 7.56 (d, 2 H, J = 8.1 Hz, H-3A/5A), 7.82 (d, 1 H, J = 9.2 Hz, H-8), 7.86 (dd, 1 H, J = 9.2/1.1 Hz, H-7), 8.5 (s, 1 H, H-5). The hapten–BSA conjugate was prepared by the carbodiimide technique, which was first described by Sheehan and Hess.15 By using UV spectrometry, a hapten : protein ratio (mmol) of 21.2 : 0.96 and 16.5 : 0.96 was calculated for immunogens 1 and 2, respectively, corresponding to the coupling of about 22 and 17 mol, respectively, of hapten with 1 mol of BSA.An average of 15 molecules of hapten per molecule of BSA has been recommended for an optimum immuno-response.16 The synthesis and purification of N,N-didemethylzolpidem- N-{2-{3-(4-hydroxy-3-[125I]-iodophenyl)}methyl propionate} [Fig. 1(d)] have already been described in detail.14 The freshly prepared tracer could be used for about 6 weeks without any loss of binding to the antibody. Titre of the Antibodies After the examination of the serial dilutions of all the antisera from the different bleeds, the titres were derived from binder dilution curves. The evolution of the titres of the different rabbits, immunized with immunogens 1 and 2, respectively, is shown in Fig. 2. As could be expected, great differences exist in immunogenic responses between immunogens 1 and 2, with titres ranging from 1/75 to 1/3200 for the immunogen without spacer and from 1/12 800 to 1/25 000 for the immunogen with spacer. Because of the low titres, the sera of rabbits 1.1 and 1.4 were not considered for further experiments. The antisera of the last bleeds of all the rabbits were lyophilized and stored at 4 °C until use.The optimization and specificity experiments were performed for all the antisera except for R1.1 and R1.4. All Analyst, October 1997, Vol. 122 1121other experiments were only carried out with the antiserum of R2.1. Determination of Optimum Conditions for the RIA As a separation technique, a second antibody method was evaluated.17 The optimum ratio of GARGG antiserum to NRS was investigated. Final dilutions of GARGG (1 + 124) and NRS (1 + 999) resulted in the highest concentrations of tracer.The optimum second incubation time for the precipitation of the antibody-bound tracer was 20 h at room temperature. In comparison with the adsorption technique, this method resulted in lower NSB values (1–3%) and in higher dilutions of the antiserum to reach 50% binding of the tracer (1 + 12 799). The performance characteristics of the immunoassay are demonstrated in a calibration graph obtained with the serum of R2.1 for the concentration range 100–10 000 pg ml21 (Fig. 3), using the second antibody method as the separation technique. The curve of best fit was obtained by applying the four-parametric logistic model to the experimental data.18 Limit of Quantification The limit of quantification (LOQ) of the RIA can be defined as the lowest concentration on the calibration graph that can be measured with acceptable precision and accuracy. The precision at this level has to be less than or equal to 20%. To determine the LOQ, five standard solutions of increasing concentration were used and the relative standard deviation was determined.The LOQ was 0.1 ng ml21. Precision and Recovery The precision and recovery of the method could be calculated after replicate analyses of independently prepared quality control urine samples. The concentration levels of the samples ranged from 0.25 to 1000 ng ml21. The intra- and inter-assay relative standard deviations were 4.01–6.30 and 9.13–12.28%, respectively, and the accuracy was about 100% (Table 1).Specificity The antibody specificity was assessed by measuring the crossreactivity to other hypnotics and to some widely used anxiolytics, antidepressants, analgesics and analeptics. The following compounds were found not to be detectable at a level of 10 mg ml21: chlordiazepoxide, diazepam, clobazam, nitrazepam, flunitrazepam, bromazepam, lormetazepam, triazolam, meprobamate, methaqualone, hexobarbital, fenobarbital, buspirone, hydroxyzine, chlorcyclizine, codeine, cocaine, caffein, ibuprofen, trazodone, zopiclone, chlorpromazine and acetylsalicylic acid.None of the above was found to cross-react ( < 0.005%) with the immunoassay. To rule out cross-reactivity with endogenous compounds, several urine samples of volunteers were screened and no crossreactivity could be detected. To select one antiserum out of the pool, the specificity towards the metabolites was examined. The two major metabolites of zolpidem, metabolites I and II, and a minor metabolite (metabolite XI) were tested for their cross-reactivity. The degree of cross-reactivity of the metabolites was expressed as the relative dose required for 50% displacement of the maximum tracer binding.The major metabolite (I) showed a low cross-reactivity, ranging from 2.3 3 1027 to 7.7 3 1023%. This could be expected considering that the position of the Fig. 2 Evolution of the serum titres of the two series of rabbits, immunized with the immunogens 1 and 2, respectively.(a) represents the antisera from rabbits R1.1 (/), R1.2 (~), 1.3 (-) and R1.4 (3) and (b) shows the titres of R2.1 (/), R2.2 (~), R2.3 (-) and R2.4 (3). Fig. 3 Calibration graph for zolpidem, together with 95% confidence intervals. The graph follows the equation y = d [(a 2 d)/1 + (x/c)b], where a = 0.8972 ± 0.01006; b = 9.9390 ± 0.4498; c = 6.8896 ± 0.0280; d = 0.02248 ± 0.01195; and r2 = 0.9764. Table 1 Intra- and inter-assay variance characteristics and recovery Added/ng RSD (%) n Recovery (%) Intra-assay 0.5 4.01 6 103.4 1.75 4.48 6 101.5 7.5 6.30 6 98.1 1000 5.90 6 99.3 Inter-assay 0.25 12.28 10 100.9 1 9.30 9 99.4 7.5 7.40 10 101.3 1000 9.13 9 104.0 1122 Analyst, October 1997, Vol. 122carboxylic acid function is close to the binding site with the antibodies. The cross-reactivity towards metabolite II was higher and varied from 0.9 to 6.9%. In this metabolite, the substituent is positioned further away from the binding site.Metabolite XI, which has an identical structure with zolpidem, except for the hydroxylic function on the acetamide side-chain in position 3 of the imidazopyridine skeleton, showed a very high cross-reactivity (91–116%). Because of the pre-screening purpose of this RIA, the antiserum with the highest crossreactivity for the main metabolite was selected, i.e., the antiserum of rabbit 2.1 (6th bleed). The results are shown in Table 2. Linearity To confirm the linearity of the response throughout the range of the calibration graph, samples were checked to see whether they diluted in parallel to the calibration graph.An unknown urine sample, containing approximately 4.6 ng ml21 of the drug, was analysed three times, both undiluted and diluted with blank urine (1 + 1, 1 + 3, 1 + 7 and 1 + 15). The observed (y) and the expected (x) values yielded the following linear regression equation : y = 0.0115 + 0.9703x (r = 0.9938). Human Urine and Serum Samples A set of 14 urine samples (including a blank) were collected from a volunteer during a 48 h period after the intake of 10 mg of zolpidem (Stilnoct).The samples were analysed using the described RIA procedure and the results were expressed in terms of zolpidem concentration, although there was also a contribution from the metabolites. The highest value (994 ng ml21) was reached 3 h after the ingestion of one tablet of Stilnoct. After 48 h, the hypnotic could still easily be detected (about 6.3 ng ml21).From the same volunteer and at specific time intervals, 13 serum samples (including a blank) were collected and analysed by RIA. The serum levels measured covered a concentration from 1.3 to 122 ng ml21. The highest value was recorded 2 h after administration (about 122 ng ml21). Thereafter, the level of zolpidem decreased to 1.3 ng ml21 after 48 h. Fig. 4 shows the serum concentration–time profile of zolpidem. Discussion This paper describes the development of an RIA for zolpidem and its metabolites in urine and serum.Two immunogens were synthesized to study the difference in affinity and specificity between an immunogen with and without a spacer. The antibodies from the rabbits immunized with the immunogen with a spacer showed a much higher affinity than those immunized with the other immunogen. A difference in specificity could not be seen. The synthesis and purification of N,N-didemethylzolpidem-N-{2-{3-(4-hydroxy-3-[125I]-iodophenyl)} methyl propionate} resulted in a high radiochemical purity and specific activity of the radioligand,14 thereby providing a very sensitive immunoassay (LOQ = 0.1 ng ml21).Owing to the nature of the synthesized hapten and immunogens, the detected cross-reactivity of the three metabolites is to be expected. The important fact is that zolpidem and its metabolites can be effectively detected in urine and serum up to 48 h after the intake of one tablet of Stilnoct (10 mg of zolpidem), without interference from other drugs.The proposed RIA can be used in the pre-screening of toxicological samples. Its applica- Table 2 Specificity of the antisera towards three metabolites of zolpidem in comparison with the unmetabolised drug Cross-reactivity (%) Compound R1.2 R1.3 R2.1 R2.2 R2.3 R2.4 Zolpidem 100 100 100 100 100 100 7.7 3 1023 2.3 3 1023 2.3 3 1027 4.2 3 1023 3 3 1024 1.6 3 1023 2.53 2.4 5.55 0.9 6.9 2.9 91 116 116 114 104 92 Analyst, October 1997, Vol. 122 1123bility to the analysis of blood and plasma samples is under study. Dr. R. Busson, Dr. G. Janssen and Dr. J. Rozensky are gratefully acknowledged for recording the NMR and mass spectra. References 1 Th�enot, J. P., Hermann, P., Durand, A., Burke, J. T., Allen, J., Garrigou, D., Vajta, S., Albin, H., Th�ebault, J. J., Olive, G., and Warrington, S. J., in Imidazopyridines in Sleep Disorders, ed. Sauvanet, J. P., Langer, S. Z., and Morselli, P. L., Raven Press, New York, 1988, pp. 139–153. 2 Durand, A., Th�enot, J. P., Bianchetti, G., and Morselli, P. L., Drug Metab. Rev., 1992, 24, 239. 3 Augsburger, M., Giroud, C., Lucchini, P Rivier, L., in Contributions to Forensic Toxicology. The Proceedings of the 31st International Meeting of the International Association of Forensic Toxicologists (TIAFT), 1993, ed. Mueller, R. K., Molinapress, Leipzig, 1994, pp. 18–22. 4 Mendelson, W. B., and Jain, B., Drug Saf., 1995, 13, 257. 5 Debailleul, G., Abi Khalil, F., and Lheureux, P., J. Anal. Toxicol., 1991, 15, 35. 6 Khodasevitch, T., and Volgram, J., Bull. Int. Assoc. Forensic Toxicol., 1996, 26(2), 37. 7 Garnier, R., Guerault, E., Muzard, D., Azoyan, P., Chaumet- Riffaud, A., and Efthymiou, M., Clin. Toxicol., 1994, 32, 391. 8 Guinebault, P., Dubruc, C., Hermann, P., and Th�enot, J. P., J. Chromatogr., 1986, 383, 206. 9 Debruyne, D., Lacotte, J., Hurault De Ligny, B., and Moulin, M., J. Pharm. Sci., 1991, 80, 71. 10 Almirante, L., and Murwann, W., Br. Pat., 1 076 089, 1967. 11 Kaplan, J. P., and George, P., Eur. Pat., 0 050 563, 1982. 12 Kaplan, J. P., and George, P., US Pat., 4 501 745, 1985. 13 Erlanger, B. F., Methods in Enzymology, Academic Press, New York, 1980, vol. 70, p. 85. 14 De Clerck, I., and Daenens, P., J. Radiolabelled Comp. Radiopharm., 1997, 39, 195. 15 Sheehan, J. C., and Hess, G. P., J. Am. Chem. Soc., 1955, 77, 1067. 16 Erlanger, B. F., Pharmacol. Rev., 1973, 25, 271. 17 Utiger, C. R., Parker, M. L., and Daughaday, W. H., J. Clin. Invest., 1962, 41, 254. 18 Dudley, R., Edwards, P., Ekins, R. P., Finney, D. J., McKenzie, I. G. M., Raab, G. M., Rodbard, D., and Rodgers, R. P. C., Clin. Chem. (Winston-Salem, N.C.), 1985, 31, 1264. Paper 7/02869E Received April 28, 1997 Accepted June 6, 1997 Fig. 4 Serum levels, expressed in terms of zolpidem concentration, in one healthy volunteer after a single oral intake of one tablet (10 mg) of Stilnoct. 1124 Analyst, October 1997, Vol. 1