Electrochemical detection of tricyclic antidepressant ... - Website Staff UI

Journal of Electroanalytical Chemistry 521 (2002) 117– 126 www.elsevier.com/locate/jelechem

Electrochemical detection of tricyclic antidepressant drugs by HPLC using highly boron-doped diamond electrodes T.A. Ivandini a, B.V. Sarada a, C. Terashima a, T.N. Rao a, D.A. Tryk a, H. Ishiguro b, Y. Kubota b, A. Fujishima a,* a

Department of Applied Chemistry, School of Engineering, The Uni6ersity of Tokyo, 7 -3 -1, Hongo, Bunkyo-ku, Tokyo 113 -8656, Japan b Department of Urology, School of Medicine, Yokohama City Uni6ersity, 3 -9 Fukuura, Kanazawa-ku, Yokohama 236 -8567, Japan Received 13 August 2001; received in revised form 6 December 2001; accepted 12 December 2001

Abstract Boron-doped diamond (BDD) electrodes have been examined for the electrochemical detection of six tricyclic antidepressant drugs (TCAs): imipramine, desipramine, clomipramine, amitriptyline, nortriptyline, and doxepin. Cyclic voltammetry, flow injection analysis (FIA) and HPLC with electrochemical detection have been used to study the oxidation reactions and to detect the TCAs. At diamond electrodes, well-defined and highly reproducible voltammograms were obtained for all six drugs with a signal to background (S/B) ratio about 2–4 times greater than those at glassy carbon electrodes. Diamond is the first electrode material to show well-defined voltammograms for nortriptyline due to its wide potential window. In the FIA-mode, at an operation potential of 0.85 V versus Ag AgCl, diamond exhibited a background current of 7 nA with rapid stabilization (15 min) conversely to the case of GC, which appeared to stabilize after 1 h, but again increased thereafter. The analytical peaks of HPLC for the TCAs were well resolved. Linear calibration curves were linear over the ranges from 0.05 to 100 mM. The limits of detection (S/N=3) were 3 nM for imipramine and desipramine, 0.5 nM for clomipramine, 163 nM amitriptyline, 1080 nM for nortriptyline and 92 nM for doxepin. The electrodes have shown reproducible results over several days of analysis. This method has been applied for the determination of imipramine and desipramine in plasma samples. The BDD surface was reproducible with no adsorption of blood components during plasma analysis. This work shows the promising use of diamond as an amperometric detector in HPLC, especially for TCA analysis. © 2002 Published by Elsevier Science B.V. Keywords: Tricyclic antidepressant drugs; Boron-doped diamond; HPLC; Plasma sample; Reproducibility

1. Introduction Tricyclic antidepressant drugs (TCAs) are one of the largest groups of drugs for the treatment of psychiatric disorders such as depression, mainly endogenous major depressions. The function of these drugs is to block the reuptake of the neurotransmitters norepinephrine and serotonin in the central nervous system [1]. The chemical structures of these compounds are illustrated in Scheme 1. For this group of drugs, distinct ranges of optimal plasma concentration for therapy are required. Lower concentrations are associated with sub-optimal

* Corresponding author. Tel.: + 81-3-3812-9276; fax: +81-3-38126227. E-mail address: [email protected] (A. Fujishima).

or no therapeutic effect. Furthermore, severe adverse effects and toxicity can appear at high concentrations. Their toxicity results from their central and peripheral anticholinergic effects, their peripheral a-adrenoceptor blocking activity and their membrane-stabilizing activity, resulting in cardiovascular complication, convulsion and coma. Therefore, the analysis of these compounds is important for quality assurance in pharmaceutical preparations and for obtaining optimum therapeutic concentrations to minimize the risk of toxicity. The therapeutic concentration range for most TCAs is approximately from 280 to 850 nM, while toxic effects can occur when plasma concentrations exceed 1.4 mM [2,3]. Several methods are available to determine TCAs, including radioimmunoassay, spectrofluorimetry, gas chromatography and high performance liquid chromatography (HPLC) using UV absorbance, chemilu-

0022-0728/02/$ - see front matter © 2002 Published by Elsevier Science B.V. PII: S 0 0 2 2 - 0 7 2 8 ( 0 2 ) 0 0 6 6 6 - 6

118

T.A. I6andini et al. / Journal of Electroanalytical Chemistry 521 (2002) 117–126

minescence or electrochemical detection [2– 8]. For therapeutic drug monitoring, where specificity is required, chromatographic methods are the most adequate techniques. The advantages of HPLC for the analysis of TCAs are its versatility and simplicity of sample preparation. However, some trialkylamines and related compounds are difficult to detect by use of UV absorption, because they do not adsorb very well in the UV –visible region, since they have low molar absorptivities [2]. The chemiluminescence method is promising, but the limit of detection was found to be relatively high (500 nM) [2,4–6]. Electrochemical detection methods are very advantageous over other methods in terms of simplicity, sensitivity, selectivity and cost. However, these methods have not become as popular as other methods due to certain unavoidable problems such as electrode deactivation, with the necessity of frequent pretreatment and other procedures to reactivate the solid electrodes. Glassy carbon (GC), one of the widely used electrodes for electrochemical detection, due to its relatively wide potential window and low cost, is very susceptible to contamination and fouling [9– 12]. Al-

though few reports are available about the use of GC in the HPLC-ED for the analysis of TCAs, the stability aspects are not mentioned. Surmann and Peter, on the other hand, have reported the influence of pretreatment of carbon fiber electrodes on the analysis of bamipine and imipramine [13]. Therefore, a stable electrode material with sensitive detection capabilities is a prime requirement for wider application of electrochemical detectors. Recently emerging diamond materials meet all these requirements to a great extent. The use of highly boron-doped diamond (BDD) thin films as electrode materials for several applications, including electroanalysis [14– 20], energy storage devices [21], and electrosynthesis [22] has gained great interest in the past few years due to the impressive properties of such films. The superiority of diamond thin films for electroanalysis results from their attractive properties, such as the wide electrochemical potential window in aqueous solution [23,24], very low voltammetric background current ( 1 order of magnitude lower than GC) [25,26] inertness of the surface to adsorption of reaction products [14], extreme electro-

Scheme 1. Chemical structure of six tricyclic antidepressant drugs.


chemical stability [21], and relative insensitivity to dissolved oxygen [27]. Several research groups including our group have previously reported the uniqueness of diamond for the detection of various environmentally, biologically and clinically important compounds using flow injection analysis or liquid chromatography, resulting in sensitive and stable detection [15– 20]. During the use of other electrode materials such as metals, GC and carbon paste, the sensitivity is decreased for the detection of the compounds that oxidize at relatively high potentials due to the interference of the oxygen evolution reaction, or the requirement of frequent surface pretreatment processes due to the strong adsorption of reactants and products on the electrode surface. Recently, Rao et al. demonstrated the ascendancy of diamond electrodes over GC for the detection of sulfa drugs [20]. Therefore, BDD, with its inert surface properties could be a promising electrode for the electrochemical detection of TCAs using HPLC. In the present study, we report the detection of six TCAs, imipramine and its metabolite desipramine, clomipramine, amitriptyline and its metabolite nortriptyline, and doxepin at neutral pH, by the use of BDD electrodes for the electrooxidation. Cyclic voltammetry, flow injection analysis and HPLC with an amperometric detector were used for the detection. Comparison experiments were performed using GC electrodes.

2. Experimental

2.1. Preparation of highly boron-doped diamond electrode The highly BDD films were grown by the microwaveassisted plasma chemical vapor deposition (CVD) technique with a commercial microwave plasma reactor (ASTeX) Corp., Woburn, MA). Films were grown on Si(100) substrates. The details of the preparation have been described previously [14]. A mixture of acetone and methanol in the ratio of 9:1 (v/v) was used as the carbon source. B2O3, the boron source was dissolved in the acetone+methanol mixture at a B/C molar ratio of 1:100. The boron concentration in the film prepared under these conditions was ca. 1.5× 1021 cm − 3 as estimated by Notsu et al., based on nuclear reaction measurements [11B(p,a)8Be], carried out by means of 1-meV proton bombardment and subsequent comparison of the a-spectrum in the 6– 8 MeV region with a BN standard [28]. High purity hydrogen was used as the carrier gas. The bubbling of the acetone+ methanol +B2O3 solution was carried out at 25 °C. The deposition of the film was carried out at a microwave power of 5 kW. A film thickness of 40 mm was achieved after 10 h deposition. The film quality was

119

confirmed by Raman spectroscopy. The results showed the narrow, intense peak at 1332 cm − 1, which is characteristic of diamond and reflects a high degree of crystallinity. In addition, a broad peak centered at approximately 1200 cm − 1 was observed, which is usually attributed to either amorphous diamond or extremely small diamond crystallites [21].

2.2. Voltammetric in6estigation Cyclic voltammetric measurements were carried out in a single compartment cell with a saturated calomel electrode (SCE) as the reference electrode and a platinum wire as the counter electrode. The BDD working electrode along with the conducting Si substrate or GC electrode was pressed against the bottom of the glass cell by a Viton O-ring. The electrical contact for BDD was made through the backside of the scratched Si substrate by contacting the brass current collecting back plate. Before use as a working electrode, BDD film was pretreated by ultrasonication in 2-propanol for about 10 min followed by rinsing with high purity-water. The purpose of this pretreatment is to remove the organic impurities that may have remained or formed during the deposition of diamond in the CVD chamber. The GC electrode (GC-20 plate or GC-20 rod, Tokai Carbon Co., Ltd) was pretreated by polishing with diamond paste (Fujimi), followed by ultrasonication in 2-propanol before the experiment. The supporting electrolyte was a mixture of 0.05 M KH2PO4 + 0.05 M K2HPO4 (KH2PO4/K2HPO4, pH 6.9 90.1). The geometric area of the working electrode was estimated to be 0.09 cm2.

2.3. Flow injection analysis The FIA system used in the present study, consisted of a micro-LC pump (Bioanalytical System, LC-100), an injector (Rheodyne) with a 20-ml injection loop, an amperometric detector (Bioanalytical Systems, LC-4C), and an X–Y recorder (Graphtec, WX400). The flow rate set for the pump was 1 ml min − 1 and was confirmed before every experiment by measuring the volume of the buffer collected at the outlet for 10 min. The wall-jet type flow cell consisted of the Ag AgCl 1 M LiCl reference electrode and a stainless steel tube as the counter electrode, which also served as the tube for the solution outlet. A 0.5-mm-thick silicon rubber gasket was used as a spacer in the cell. The geometric area was estimated to be 0.64 cm2. The cell volume was estimated to be 24 ml by assuming a 25% compression of the gasket. Phosphate buffer (KH2PO4 + K2HPO4, pH 6.99 0.1) was used as the mobile phase. The analyte solutions were prepared with the same buffer. The injection volume of the analyte was 20 ml.

120


Hydrodynamic voltammograms were obtained for each compound prior to amperometric determination and the detection potential was chosen in the limiting current range of the voltammogram.

2.4. HPLC analysis The HPLC system consisted of a reversed-phase C18 column (Inertsil ODS-3 150×4.6 mm ID; particle size, 5 mm) connected to the FIA system. The chromatograms were obtained at an applied potential 0.85 V versus Ag AgCl 1 M LiCl for the imipramine/desipramine couple, at 0.93 versus Ag AgCl for clomipramine and at 1.3 V versus Ag AgCl for the amitriptyline/nortriptyline couple and doxepin. The temperature was maintained at 26 °C. All the experiments were performed with a mixture of acetonitrile and phosphate buffer 0.05 M (KH2PO4/K2HPO4 pH 6.99 0.1) in the ratio of 375:625 (v/v) as the mobile phase for the detection of all TCAs, except for clomipramine (50:50).

cyclic voltammetric behavior of the TCAs. At the BDD electrode, the oxidation of imipramine occurred at slightly higher potentials ( 0.844 V vs. SCE) than that at the GC electrode ( 0.800 V vs. SCE). It is established that diamond exhibits higher overpotentials for inner sphere redox reactions in comparison to GC, with the oxygen evolution and hydrogen evolution reactions being the best examples. At both the electrodes, the voltammogram exhibited similar features, i.e. irreversible oxidation peaks. However, for BDD, the low background current (0.19 nA at 0.844 V vs. SCE) at the oxidation potentials, compared to that at GC (1.39 nA at 0.8 V vs. SCE), resulted in a large S/B ratio. Table 1 presents a comparison of data for S/B ratios obtained from the cyclic voltammetric data for the oxidation of

2.5. Sample preparation A blood sample was collected from a male patient (16 years old, 60 kg), 10 h after intake of an oral dose of 10 mg imipramine (Tofranil®, sugar-coated commercial tablet). This sample was centrifuged for 5 min at 3000g and stored at −20 °C until analysis. A 200-ml portion of 20% HClO4 was added into 200 ml plasma sample and vortex mixed for 5 min and then centrifuged for 5 min. The supernatant was then neutralized using NaOH. The final solutions were filtered using a 0.4 mm pore filter and injected directly into the HPLC column. Dilution was by a factor of 4.

2.6. Materials Imipramine, desipramine, clomipramine, amitriptyline, nortriptyline, doxepin, KH2PO4, K2HPO4, HClO4, NaOH (Wako Chemical Co.) and acetonitrile (Nacalai Tesque) were used without further purification. All of these solutions were prepared using ultrapure (18 MV) water.

Fig. 1. Cyclic voltammograms for 10 mM imipramine in 0.1 M phosphate buffer (pH 6.9) at: (A) at a BDD electrode (B) a GC electrode. Dashed lines show the voltammograms after stirring the solution. The sweep rate was 100 mV s − 1. Table 1 Comparison of S/B ratio obtained from the cyclic voltammetry data for the oxidation of TCAs in 0.1 M phosphate buffer (pH 6.9) for BDD and GC electrode

3. Results and discussion

BDD

3.1. Voltammetric studies Fig. 1 shows the cyclic voltammograms obtained at a potential sweep rate of 100 mV s − 1 for a deaerated solution of 10 mM imipramine in 0.1 M phosphate buffer (pH 6.9) together with the corresponding background voltammogram, at BDD (Fig. 1A) and GC (Fig. 1B). In fact, deaeration has little effect on the

Imipramine Desipramine Clomipramine Amitriptyline Nortriptyline Doxepin

GC

Ep/V vs. SCE

S/B

Ep/V vs. SCE

S/B

0.844 0.833 0.936 1.333 1.350 1.300

52.31 55.84 40.43 19.17 10.23 21.86

0.800 0.833 0.877 0.979 Undetectable 0.895

12.59 33.00 11.85 10.70 8.09


100 mM concentrations of all five TCAs in 0.1 M phosphate buffer at BDD and GC. The results show that the S/B ratios were about 2– 4 times higher for BDD compared to GC. The relatively low background current for diamond was attributed to the predominant hydrogen termination of the surface as well as the compact nonporous surface [16,20]. Films used in this study are predominantly hydrogen terminated, as they were prepared in hydrogen plasma. Such films showed an O/C ratio of 0.03. Earlier, Yagi et al. from our group reported that XPS results of as-deposited BDD showed only a sharp C 1s peak and no discernable O 1s peak, which is typical for the H-terminated surface [29]. However, although the presence of oxygen functional groups on the GC surface is believed to be partly responsible for its high background current [9], our previous studies have shown that surface oxygen does not play a major role in the case of diamond in determining the background current. An increase of O/C ratio from about 0.03 to 0.18 changed its background to about 3 –5 times, which was still much lower than that of glassy carbon [29]. Another important observation is the recovery of the voltammetric peak at a BDD electrode. The oxidation curve of imipramine at BDD could be recovered by stirring the solution for 2 min after the second cycle (Fig. 1A). In the case of GC, the cyclic voltammogram could not be retraced even after vigorous stirring, indicating partial deactivation of the electrode surface probably due to the adsorption of oxidation products (Fig. 1B). The oxidation peak current at GC was about 4 times larger than that obtained at BDD. This may be due to the adsorption of the compound on the electrode

121

surface. At GC, strong adsorption on the surface generally provides high sensitivity for the detection of TCAs, but the adsorbed reaction products form insulating layers on the surface resulting in electrode fouling [20]. The shoulder peak observed at higher potentials for both the electrodes may be due to the oxidation of a reaction product. Similar voltammograms were obtained for desipramine, a metabolite of imipramine. In the case of clomipramine, a higher potential (V0.94 V vs. SCE) was needed to oxidize the compound, since clomipramine contains highly electronegative substituted chlorine that stabilizes the compound and makes oxidation more difficult. At the BDD electrode the peak currents increased linearly with the square root of the sweep rate (w 1/2) within the range of 5–300 mV s − 1 (r 2 0.9996) with a zero intercept. The linearity suggested that the current is limited by diffusion control in the interfacial area of the electrode and the adsorption steps and specific surface interaction can be neglected. The electrochemical oxidation of imipramine, desipramine and clomipramine are believed to occur at the nitrogen atom in cyclohexane involving a two-electron one-proton mechanism, in analogy to that for methyliminobibenzyl [30,31], as illustrated in Scheme 2. The dimer is formed, which is more easily oxidized than the monomer. The redox-peak pair that appeared at 0.037 V after the first cycle may be due to the dimer, since after the first oxidation, the dimer is formed and is available near the electrode surface [32]. A simple experiment was performed to investigate whether this peak was due to the adsorption of reaction products on the electrode surface. After stirring the solution for 2

Scheme 2. Electrooxidation mechanism of imipramine.

122


min, the electrodes were rinsed and the background voltammograms were recorded for both BDD and GC. At BDD, the background voltammogram was identical with that of the fresh electrode. However, at a GC electrode, the peak pair at 0.037 V versus Ag AgCl remained indicating a strong adsorption of the reaction products on the electrode surface (Fig. 2). The large peak currents and change in the background voltammogram indicate the failure of GC for the clinical detection of TCAs. Although dimer formation occurs at the BDD electrode, it is not permanently adsorbed on its surface. Surmann and Peter discussed in detail the effect of electrochemical pretreatment of carbon fibers for the detection of TCAs [13]. Another remote possibility is the chemical oxidation of this dimer by the monomer radical [33] to produce the unoxidized azepine. The shoulder peak in the cyclic voltammogram in Fig. 1 is possibly due to this azepine, although, at the moment, we do not have evidence for this. This mechanism suggests a pH dependence of the overall process. Indeed, decreasing the pH of the solution resulted in a shift of the TCA oxidation peaks toward more positive potentials. Fig. 3 shows the variation of peak potentials for TCA oxidation at a BDD electrode as a function of pH. It can be seen that the peak potentials of imipramine, desipramine and clomipramine shift linearly toward more positive potentials with slopes (dE/dpH) ranging between 23 and 28 mV which is suitable for a mechanism involving two electrons and one proton. Cyclic voltammograms were also observed for amitriptyline, nortriptyline, and doxepin, other types of TCAs, which differ from imipramine, desipramine and clomipramine because there is no nitrogen atom in the cycloheptene ring. At BDD, the oxidation peaks of amitriptyline, nortriptyline and doxepin appeared at higher potential of about 1.3 V vs. SCE, higher than at GC (0.9 V vs. SCE). However, at GC, the electrode surface was easily fouled and the cyclic voltammograms could not be retraced even after vigorous stirring. An important observation was that the GC electrode gave an ill-defined peak for nortriptyline as shown in Fig. 4b, due to the interference of the oxygen evolution reaction together with oxidation of the GC electrode surface itself. Nortriptyline and doxepin, which have no ring nitrogen atoms, were previously reported to be electrochemically inactive [33]. Although doxepin contains an oxygen heteroatom, it does not present attackable lone pairs. Then, Turk et al. found that the electroactivity is achievable by modifying the GC electrode with electropolymers such as polycarbazole or polythiophene. This is due to the presence of electron-rich sulfur or nitrogen atom on the polymer surface, which facilitates the formation of a cation radical [34]. However, at both BDD and GC electrodes the electrooxidation of

Fig. 2. Comparison of background current before and after imipramine oxidation at: (A) a BDD electrode and (B) a GC electrode. Dashed lines show the voltammograms after imipramine oxidation. The sweep rate was 100 mV s − 1.

Fig. 3. Dependence of peak potential on pH of the solution for the oxidation of TCAs.

amitriptyline and doxepin occurred with well-defined oxidation peaks. The advantage of BDD, which is due to its wide potential window, is that a well-defined peak


123

0.1 M phosphate buffer were sigmoidal, as shown in Fig. 5. The potentials for the oxidation currents are in the expected order while the imipramine derivatives containing ring nitrogen oxidized at lower potentials and doxepin, nortriptyline and amitriptyline oxidized at higher potentials, as expected. The amperometric response obtained for a BDD electrode for 20 ml injections of 100 mM imipramine gives reproducible peaks, with a peak variability of 3% (n=18) suggesting no adsorption of oxidation products on the electrode surface. Reproducible peaks were obtained at BDD for concentrations as low as 10 nM. (Fig. 6A), which could be clearly distinguished from the peaks for buffer injec-

Fig. 4. Cyclic voltammograms of 100 mM nortriptyline in 0.1 M phosphate buffer (pH 6.9) at: (a) BDD and (b) GC electrodes. The sweep rate was 100 mV s − 1.

at high potential occurred for nortriptyline, even though the S/B ratio was quite small compared to the other types of TCAs. Due to the high oxidation potential of amines, no reports were available on a voltammetric investigation of these drugs on solid electrodes. However, Greenway and Dolman suggested that the electrooxidation of amitriptyline, nortriptyline and doxepin takes place at the nitrogen atom in the alkylamine, resulting in the formation of a cation radical, followed by deprotonation [2]. The pH dependence of the voltammetric peak current (Fig. 3) showed the same behavior for amitriptyline, nortriptyline and doxepin, i.e. the peak potentials shift linearly to more positive potentials with increasing pH with slopes (dE/dpH) ranging between 58 and 67 mV, which suggests the participation of equal numbers of protons and electrons in the oxidation reaction. Detailed mechanistic studies for these compounds at diamond electrodes will be conducted in the future. The oxidation of amitriptyline and doxepin occurred at lower potentials than that of nortriptyline, since the former contain tertiary amine functional groups, which are easier to oxidize [4].

3.2. Flow injection analysis Hydrodynamic voltammograms obtained at the diamond electrode for 20-ml injections of 10 mM TCAs in

Fig. 5. Hydrodynamic voltammogram for 20-ml injections of 10 mM TCAs in 0.1 M phosphate buffer (pH 6.9) at BDD electrodes.

Fig. 6. FIA-ED results for 20-ml injections of: (A) 10 nM imipramine at a BDD electrode and (B) 10 nM imipramine at a GC electrode. The mobile phase was 0.1 M phosphate buffer pH 6.9. The flow rate was 1 ml min − 1.

124


tions. In the case of GC electrodes, the response for 100 mM decreased to 72% after 18 injections. For 100 nM imipramine, the peak variability was 10%. Amperometric peaks obtained for 10 nM imipramine were highly irregular due to the large noise in the background current (Fig. 6B). The BDD electrode showed a smaller degree of noise in the baseline, with a low background current, compared to those at GC electrodes. Similar results were obtained for other TCAs. A linear dynamic range is obtained from 0.01 to 100 mM (r 2 =0.995), with an experimental detection limit of 0.01 mM for all TCAs, and 0.1 mM for nortriptyline as an exception. At GC, the peak current was saturated at high concentrations and the experimental limit of detection was 0.1 mM for all TCAs, which is one order higher than those achieved at BDD. Nortriptyline was not detectable at GC.

3.3. Chromatographic detection Prior to the chromatographic studies of the TCAs, the stabilities of electrodes were examined and compared over a period of 10 h, for two types of commercial GC electrodes, and a diamond thin film at the detection potentials, i.e. 800 mV versus Ag AgCl for GC and 850 mV versus Ag AgCl for BDD (Fig. 7). At BDD, the background current stabilized within 20 min after switching to the detection potential and the current was very stable during the 10 h of experimentation. However, at both types of GC, longer times (\ 60 min) were required to obtain considerable stability and there was a continuous change in the current during the 10 h of experiment. The continuous change in the current with time was either due to the adsorption of impurities or the oxidation

Fig. 7. Current vs. time profiles for BDD (at 0.85 V vs. Ag AgCl), GC20 Tokai and GC GL Tokai (at 0.80 V vs. Ag AgCl). The mobile phase was acetonitrile +0.5 M phosphate buffer (pH 6.9) = 37.5:62.5. The flow rate was 1 ml min − 1.

of the electrode surface itself. Furthermore, at the end of the experiment, a visible thin film formation on the GC surface was observed, which is probably due to the adsorption of contaminants from the buffer. GC is prone to such contaminant deposition [12]. No visible change was observed on the diamond electrode. Similar observations were made by Sarada et al. for GC at 1.28 V and for BDD at 1.52 V. versus Ag AgCl [35]. The noise in the background current after 10 h of experimentation was about one order of magnitude higher compared to that for BDD. The low noise at BDD shows that very low detection limits can be obtained. For the clinical applications, during therapeutic monitoring, the stability of the electrode surface is highly important and BDD satisfies this requirement. Also, the background current at the BDD electrode was about 3– 5 times less than that for the GC electrodes. This enables the detection of TCAs at very low concentrations. Since desipramine is a metabolite of imipramine and nortriptyline is a metabolite of amitriptyline, the chromatograms were obtained for imipramine and desipramine simultaneously (Fig. 8A) and similarly, for amitriptyline and nortriptyline (Fig. 8B). Chromatograms for clomipramine and doxepin were obtained separately (Fig. 8C and D). The mobile phase was acetonitrile+ phosphate buffer at a ratio of 375:625 for all TCAs except for clomipramine (50:50), since the Cl atom in its structure makes for high polarity. The elution times for desipramine and imipramine were 7.5 and 16.3 min, respectively, and for clomipramine it was 10.55 min. Nortriptyline and amitriptyline eluted at 7.66 and 18.5 min, respectively, and doxepin eluted at 10.39 min. TCA calibration curves obtained for the chromatograms were linear over the range 100–0.05 mM. The limit of detection for imipramine and desipramine was 3 nM (S/N = 3). The lowest limit of detection that can be achieved is 0.5 nM for clomipramine. The linear calibration equation and limits of detection (S/N = 3) are shown in Table 2. BDD electrodes showed good linearity and reproducibility for the detection of TCAs. These detection limits were 5 times less than those reported previously with an electrochemical detector [36]. The results were reproducible from film to film, and with the same film, reproducible peaks were obtained for the 3 months over which the experiments were performed. Ultrasonication in 2-propanol for 10 min was sufficient for cleaning the electrode surface when it was used after a long interval. No mechanical or electrochemical pretreatment was required before each set of experiments, whereas GC required pretreatments including mechanical and electrochemical treatments.


125

Fig. 9. Chromatogram of plasma sample obtained from a patient treated with 10 mg imipramine (Tofranil®). Dilution factor is 4. Other conditions are the same as Fig. 8A.

Fig. 8. Chromatograms obtained from 10 mM concentration of: (A) imipramine and desipramine, (B) clomipramine, (C) amitriptyline and nortriptyline, and (D) doxepin, using an Inertsil ODS-3 column. The mobile phase was acetonitrile + 0.5 M phosphate buffer (pH 6.9) = 37.5:62.5 except for (B) the mobile phase is acetonitrile +0.5 M phosphate buffer (pH) = 50:50. The flow rate was 1 ml min − 1, the oven temperature was 26 °C and the detector was ECD/BDD. The operation potential, + 0.85 V vs. Ag AgCl for (A), + 0.93 V vs. Ag AgCl for (B) and + 1.30 V vs. Ag AgCl for (C) and (D).

3.4. Detection of TCAs in human-blood sample BDD electrodes are most advantageous for the analysis of biological fluids, such as blood due to the lack of adsorption of their components on the electrode surface. At GC, the electrode surface is deactivated due to adsorption of the components even after removal of

several proteins during the extraction process. Fig. 9 shows the chromatogram obtained for supernatant of the extracted blood sample obtained from a patient who was being treated for depression. After intake of an oral dose of 10 mg imipramine, as already mentioned above, imipramine is metabolized to desipramine. Therefore, peaks for imipramine and desipramine were observed in the chromatogram. From the currents obtained for amperometric peaks of HPLC for the diluted blood sample, the average amounts of desipramine and imipramine were found to be 0.126 mM (38.2 mg l − 1) and 0.192 mM (60.8 mg l − 1), respectively (n= 3). The recovery test was done using a blank plasma sample spiked with 1.0 nmol (1.01 and 1.06 mg l − 1) of desipramine and imipramine, respectively, in a blank sample. The average recoveries of desipramine and imipramine were 92.3 and 90.8%, respectively (n=3) as shown in Table 3. The reproducibility is good with a peak variability of 3%. The reproducibility of the BDD electrode surface after plasma analysis was confirmed by comparing the peak currents obtained for a 1.0 mM standard solution concentration of desipramine and imipramine before and after the analysis. Good reproducibility was obtained with peak variabilities of 0.35 and 3.27% (n=4)

Table 2 Calibration linear data for TCAs by using HPLC

Imipramine Desipramine Clomipramine Amitriptyline Nortriptyline Doxepin

Calibration equation

Regression coefficients

LOD/nM based on S/N=3

y= 0.768+12.980x y= 0.345+7.405x y= −1.935+7.735x y= 5.098+5.425x y= 0.922+1.073x y= 4.886+9.347x

0.9976 0.9942 0.9988 0.9996 0.9994 0.9968

3 3 0.5 163 1080 92


126

Table 3 Recoveries from blank plasma spiked with 1.0 nmol of each desipramine and imipramine (n= 3) Compound

Imipramine Desipramine

Recovery Amount found/nmol

Percentage/%

SD/nmol

CV/%

0.9089 0.9283

90.89 92.83

0.0209 0.0252

2.29 2.71

for desipramine and imipramine, respectively. The data indicated that there was no fouling at the BDD electrode after plasma analysis. The result shows that this method can be used for the rapid analysis of imipramine and desipramine in blood samples.

4. Conclusions Conductive boron-doped diamond electrodes exhibited an excellent performance for the electrochemical detection of TCAs. Cyclic voltammetric studies show the superiority of diamond over GC in terms of reproducibility and sensitivity (S/B ratio). Diamond is the first material to yield well-defined cyclic voltammograms for nortriptyline on the unmodified surface due to its wide potential window. FIA and HPLC results indicate that the TCAs examined in this study are detected amperometrically with high sensitivity and reproducibility. The limit of detection (S/N =3) ranges from 0.5 nM for clomipramine to 1.08 mM for nortriptyline. The background current stability test for diamond indicated faster current stabilization after switching to the detection potential (15 min) with a lower background current (7 nA) in comparison to GC electrodes, which exhibited a larger variation in background current during 10 h of experimentation. The result obtained for the detection of imipramine and desipramine in plasma samples showed that the use of a BDD electrode with HPLC for the detection of TCAs is rapid and reproducible with no adsorption of the electroactive component of blood on the electrode surface. These results demonstrated the possibility of the practical utility of diamond in amperometric detection coupled with FIA and HPLC systems especially for TCAs.

Acknowledgements This research was supported by the Japan Society for the Promotion of Science (JSPS), Research for the Future Program, ‘Exploratory Research on Novel Artificial Materials and Substances for Next Generation Industries’.

References [1] J. Wang, M. Bonakdar, C. Morgan, Anal. Chem. 58 (1986) 1024. [2] G.M. Greenway, S.J.L. Dolman, Analyst 124 (1999) 759. [3] S. Ulrich, J. Materns, J. Chromatogr. B 696 (1997) 217. [4] A. Mason, K.M. Rowan, B. Law, A.C. Moffat, Analyst 109 (1984) 1213. [5] A. Bakkali, E. Corta, J.I. Ciria, L.A. Berrueta, B. Gallo, Talanta 49 (1999) 773. [6] J.L. Paz, A. Townshen, Anal. Commun. 33 (1996) 31. [7] F. Pommier, A. Sioufi, J. Godbillon, J. Chromatogr. B 703 (1997) 147. [8] R.F. Suckow, T.B. Cooper, J. Pharm. Sci. 70 (1981) 257. [9] D.T. Fagan, I.-F. Hu, T. Kuwana, Anal. Chem. 57 (1985) 2759. [10] I.-F. Hu, D.H. Karweik, T. Kuwana, J. Electroanal. Chem. 188 (1985) 59. [11] J. Xu, Q. Chen, G.M. Swain, Anal. Chem. 70 (1998) 3146. [12] S. Ranganathan, T.-C. Kuo, R.L. McCreery, Anal. Chem. 71 (1999) 3574. [13] P. Surmann, B. Peter, Electroanalysis 8 (1996) 692. [14] T.N. Rao, I. Yagi, T. Miwa, D.A. Tryk, A. Fujishima, Anal. Chem. 71 (1999) 2506. [15] B.V. Sarada, T.N. Rao, I. Yagi, T. Miwa, D.A. Tryk, A. Fujishima, Anal. Chem. 72 (2000) 1632. [16] M.D. Koppang, M. Witek, J. Blau, G.M. Swain, Anal. Chem. 71 (1999) 1188. [17] J. Xu, G.M. Swain, Anal. Chem. 70 (1998) 1502. [18] S. Jolley, M. Koppang, T. Jackson, G.M. Swain, Anal. Chem. 69 (1997) 4409. [19] M.C. Granger, J. Xu, J.W. Stojek, G.M. Swain, Anal. Chim. Acta 397 (1999) 145. [20] T.N. Rao, B.V. Sarada, D.A. Tryk, A. Fujishima, J. Electroanal. Chem. 491 (2000) 175. [21] K. Honda, T.N. Rao, D.A. Tryk, A. Fujishima, M. Watanabe, K. Yasui, H. Masuda, J. Electrochem. Soc. 147 (2000) 659. [22] F. Okino, H. Shibata, S. Kawasaki, H. Taouhara, T. Momota, M. Nishitani-Garno, I. Sakaguchi, T. Ando, J. Electrochem. Solid-State Lett. 2 (1999) 382. [23] N. Vinokur, B. Miller, Y. Avyigal, R. Kalish, J. Electrochem. Soc. 143 (1996) L238. [24] H.B. Martin, A. Argoitia, U. Landau, A.B. Anderson, J. Angus, J. Electrochem. Soc. 143 (1996) L133. [25] R. Tenne, C. Levy-Clement, Isr. J. Chem. 38 (1998) 57. [26] G.M. Swain, J. Electrochem. Soc. 141 (1994) 3382. [27] T. Yano, D.A. Tryk, K. Hashimoto, A. Fujishima, J. Electrochem. Soc. 145 (1998) 1870. [28] H. Notsu, I. Yagi, T. Tatsuma, D.A. Tryk, A. Fujishima, J. Electroanal. 492 (2000) 31. [29] I. Yagi, H. Notsu, T. Kondo, D.A. Tryk, A. Fujishima, J. Electroanal. 473 (1999) 173. [30] S.N. Frank, A. Bard, J. Electrochem. Soc. 122 (1975) 898. [31] M. Neptune, R.L. McCreery, A. Mania, J. Med. Chem. 22 (1979) 196. [32] I. Biryol, B. Uslu, Z. Kucukyavuz, Pharm. Biomed. Anal. 15 (1996) 371. [33] E. Bishop, W. Hussein, Analyst 109 (1984) 73. [34] D.J. Turk, S.A. McClintock, W.C. Purdy, Anal. Lett. 18 (1985) 2605. [35] B.V. Sarada, T.N. Rao, N. Spataru, D.A. Tryk, A. Fujishima, submitted for publication. [36] A.G. Chen, Y.K. Wing, H. Chiu, S. Lee, C.N. Chen, K. Chan, J. Chromatogr. B 693 (1997) 153.