BIOLOGICAL MASS SPECTROMETRY, VOL. 20, 709-716 (1991)
Determination of the Antidepressant Levoprotiline and its N-Desmethyl Metabolite in Biological Fluids by Gas Chromatography/Mass Spectrometry R. Ackermann,? G. Kaiser, F. Schueller and W. Dieterle Research and Development Department, Pharmaceuticals Division, CIBA-GEIGY Ltd, CH-4002 Bade, Switzerland
A specific and sensitive gas chromatographic/mass spectrometric method was developed and validated for the determination of the antidepressant levoprotiline in blood, plasma and urine and the simultaneous determination of
levoprotiline and its desmethyl metabolite in urine. Deuterium-labelled analogues were used as internal standards. The compounds were isolated from the biological fluids by liquid-liquid extraction under basic conditions. Following derivatization with perlluoropropionic anhydride, the samples were analysed by capillary column gas chromatogaphy/electron impact mass spectrometry with selected ion monitoring. The analysis of spiked samples demonstrated the high accuracy and precision of the method. Blood concentrations of levoprotiline down to 0.7 nmol I-' (1 ml used for analysis) could be quantified with a coefficient of variation of 10% or less. The method is suitable for use in pharmacokinetic and bioavailability studies of levoprotiline in humans.
H
INTRODUCTION
I /N--R
The new antidepressant levoprotiline, (R)-a-[(methyl10H)-ethanol, amino)methyl]-9,10-ethanoanthracene-9( is chemically the ( -)-R-enantiomer of racemic oxaprotiline. It is administered perorally in the form of a hydrochloride (levoprotiline.HC1, Fig. 1). In experimental animals and man, racemic oxaprotiline and its (+)S-enantiomer are potent inhibitors of noradrenaline uptake, while levoprotiline has no such activity. However, in preliminary clinical trials, levoprotiline has shown comparable antidepressive activity to traditional antidepressants, but distinctly better tolerability.'6 The basic pharmacokinetics and metabolic characteristics of oxaprotiline and levoprotiline have already been reported.7-'0 After peroral administration to man, oxaprotiline and levoprotiline are almost completely absorbed and extensively metabolized. Direct 0-glucuronidation is the major metabolic pathway; N-desmethyl-levoprotiline is a minor metabolite. Several methods for the analysis of oxaprotiline and levoprotiline in biological fluids have been described. High-performance liquid chromatography (HPLC) and packed column gas chromatography/mass spectrometry (GC/MS) were used for the determination of oxaprotiline in blood, plasma and urine."-13 HPLC and thinlayer chromatography (TLC) were applied to measure levoprotiline in plasma or b l ~ o d . ' ~ ,We ' ~ have now developed a highly sensitive capillary GC/MS method which allows the accurate and precise determination of levoprotiiine in blood down to concentrations of less than 1 nmol 1-'. The method is also suitable for -f Author to whom correspondence should be addressed.
1052-9306/91/11070948 $05.00
0 1991 by John Wiley & Sons, Ltd.
R = CH3
Levoprotiline
R - H
D e s m e t hy I -1evoprot i I ine levoprotiline.HCI and desmethyl-
1 . Structure of levoprotiline.HCI. Figure
determining levoprotiline and its desmethyl metabolite simultaneously in urine. In this paper we describe the details of the method and demonstrate their application by assaying blood and urine samples of healthy volunteers dosed orally with levoprotiline.HC1.
EXPERIMENTAL Reagents and materials Levoprotiline.HC1 and N-desmethyl-levoprotiline.HC1 were available from CIBA-GEIGY (Bade, Switzerland). The deuterated internal standards, (2H,)oxaprotiline.HC1 ('H,-methyl moiety, isotopic purity: 'H, 5.0%, 'H, 93.7%, 'H, 1.2%) and ('H,)N-desmethyloxaprotiline.HC1 (deuterated in the anthracene ring system, isotopic purity: 'H, 26.6%, 2H,71.9%, 2H,o 1.4%), were synthesized in the Isotope Laboratory of CIBA-GEIGY. All solvents and chemicals were of anaReceived 12 June 1991 Revised 29 July 1991
710
R. ACKERMANN, G. KAISER, F. SCHUELLER A N D W. DIETERLE
lytical grade and were obtained from Fluka (Buchs, Switzerland) or E. Merck (Darmstadt, Germany). A 2 mol 1-' sodium carbonate solution was used as buffer (pH 10.8). Water used in the experiments was bidistilled. Human blank blood and plasma were obtained from Blutspendezentrum, SRK, Basle, Switzerland, and human blank urine from healthy volunteers. Biological fluids were stored below -18°C until used for preparing spiked samples. The extraction steps were performed in one-way glass ampoules. The samples or solutions were dosed with automatic pipettes from Gilson (Villier-le-Bel, France) and weighed on analytical balances, type ME 30 or AE 50 (Mettler, Greifensee, Switzerland). Shaking was done either on a horizontal shaker, type TRl (Infors, Basle, Switzerland), or on a Vortex evaporator (HaakleBuchler, Saddle Brook, New Jersey, USA). The samples were centrifuged on a Multex centrifuge (MSE Scientific Instruments, Crawley, Sussex, UK). The clinical samples were ultrasonicated using an ultrasonic desintegrator, type Sonifer B 12 (Branson Sonic Power Company, Danbury, USA). A Speed Vac combined evaporator/ concentrator (Savant Instruments Inc., Farmingdale, New York, USA) was used to concentrate solutions, or the solvents were evaporated under a stream of nitrogen at 40"C. Stock and working solutions Stock solutions of levoprotiline.HC1, desmethyllevoprotiline.HC1 and of the corresponding internal standards were prepared in methanol and were stored at 6 "C. Although the solutions were found to be stable for at least 6 months, new stock solutions were prepared every 3 months since small losses in volume may occur by evaporation if the solutions are used every day. For each assay, working solutions were obtained by dilution of the corresponding stock solution with 0.01 mol 1-' hydrochloric acid. For the simultaneous determination of levoprotiline and desmethyl-levoprotiline in urine, a working solution containing both compounds was prepared. The concentrations of the compounds in the stock and working solutions depended on the concentration range to be covered by the spiked samples. Approximate concentrations of stock and working solutions of levoprotiline.HC1 were 30 and 0.08 pmol 1-', respectively, for assays of blood or plasma, and 350 and 2 pmol 1-', respectively, for assays of urine. To prepare samples with extremely low concentrations of levoprotiline.HC1, the stock solution was diluted twice, resulting in a concentration of the working solution of about 0.004 pmol I-'. The concentrations of desmethyllevoprotiline.HC1 in the stock and working solutions were 40 and 0.25 pmol 1-', respectively. About two times higher concentrations were used for the stock and working solutions of the deuterated standards.
To 0.5 or 1 ml of the biological fluid were added: 10-500 pl of the working solution, an aliquot (100-200 pl) of the internal standard working solution and aliquots of 0.01 mol 1-1 hydrochloric acid to obtain equal volumes in all samples. The samples were shaken on a vibration shaker at speed 7 for 5 min and subsequently centrifuged. Routinely used concentration ranges of calibration and validation samples were: 0.6 to 40 pmol per sample for the determination of unchanged levoprotiline in blood and plasma (internal standard: 30 pmol per sample); 20-1000 pmol per sample for the determination of unchanged levoprotiline in urine (internal standard: 800 pmol per sample); 2-120 pmol per sample for the determination of desmethyl-levoprotiline in urine (internal standard: 110 pmol per sample). To determine extremely low concentrations of levoprotifine in blood, a calibration and validation range of 0.04-2.1 pmol per sample was used (internal standard: 1.7 pmol per sample).
Preparation of clinical samples Clinical samples (i.e. samples from clinical trial with levoprotiline.HC1) were ultrasonicated for 10 s at 50 W, then 0.5 or 1 ml were weighed into 10 ml glass ampoules. For samples whose concentrations were expected to lie above the calibration range, smaller volumes were used and blank blood, plasma or urine was added to obtain the same volume for all samples. The exact sample amount used was always determined by weighing. To each sample were added: an aliquot (100-200 pl) of the internal standard working solution and aliquots of 0.01 mol 1-' hydrochloric acid to obtain the same total volume as for the spiked samples. Sample working-up procedure Both spiked and clinical samples were equilibrated for 15 min at 200 rev min-' on a horizontal shaker. Then, 0.5 ml of buffer pH 10.8 and 4.5 ml of toluene were added. Each ampoule was sealed and shaken on the horizontal shaker for 10 min at 300 rev min-' followed by centrifugation at 1250 x g for 10 min. After opening the ampoule, the two phases were separated by freezing the aqueous phase in a dry iceethanol mixture and transferring the organic phase into a new ampoule. One ml of 0.1 mol 1-' hydrochloric acid was added to the organic phase, and the mixture was shaken, centrifuged and frozen as described above. The organic phase was discarded. The aqueous phase was then made alkaline by addition of 0.5 ml of buffer (pH 10.Q and extracted with 2 ml of hexane as described above. The organic phase was transferred into a 5 ml ampoule and evaporated to dryness.
Preparation of spiked samples
Derivatization procedure
With each assay, six calibration samples and six validation samples were prepared by spiking human blank, blood, plasma or urine in 10 ml glass ampoules.
The residue from the above extraction was dissolved in 0.2 ml of toluene and 20 pl of perfluoropropionic anhydride were added. After vortexing for 5 s at speed 10,
LEVOPROTILINE AND A METABOLITE BY G U M S
the mixture was heated to 100°C for 0.5 h. Then, 1 ml of a mixture of methanol-water (1 :2, v/v) and 1.4 ml of hexane were added and the sample vortexed for 1 min at speed 9, followed by a short centrifugation at 1250 x g. The aqueous phase was frozen and the organic phase transferred into a 1.5 ml conical vial. The residue was concentrated successively with 400,200 and 80 pl of hexane, reconstituted in 1&20 pl of toluene, and 1-2 pl were injected into the gas chromatograph/ mass spectrometer. GC/MS Analyses were performed either on a Hewlett-Packard 5987 B GC/MS system (equipped with an H P lo00 multiuser/multitasking computer) or on an H P 5890 A gas chromatograph interfaced with an HP 5970 mass selective detector (MSD; equipped with a Pascal Workstation). At both instruments, the gas chromatograph was equipped with an H P 7673 automatic sample injector and a split/splitless capillary inlet system. A fused-silica capillary column (12 m, 0.2 mm id.), coated with cross-linked methyl silicone (film thickness 0.33 pm, purchased from Hewlett-Packard) was inserted directly into the ion source of the mass spectrometer and connected with the injector port of the gas chromatograph by an uncoated, deactivated fused silica precolumn (1.2 m, 0.32 mm i.d.), acting as a retention gap. A sylanized glass liner containing some glass wool at the lower end was used in the injector port. The injector temperature was maintained at 270 "C. The carrier gas was helium at an inlet pressure of 100 kPa, resulting in a gas linear velocity of 50 cm s-'. The corresponding flow was approximately 1 ml min-'. Injections were made in splitless mode (0.5 min splitless period) with a split flow of 50 ml min-' and a septum purge of 3 ml min-'. The column oven was operated isothermally at 100°C for 0.5 min after injection, heated at a rate of 30°C min-' to 290°C and then held at this temperature until the end of the run. The total run time was about 6.5-7.0 min. The GC/MS interface was maintained at 260 "C (280°C for the MSD); the ion source temperature was set to 200°C. The mass spectrometer was operated under EI ionization conditions. At the beginning of each analysis day, the mass spectrometer was calibrated with the Autotune program. The ion source parameters were then adjusted to obtain the highest signal on the fragment ion m/z 502 of the calibration compound perfluorotributylamine. The peak width at one-half of peak height was about 1 Da for the H P 5987 system and 0.7 Da for the MSD. The peak dwell time was set to 20 ms. Selected ion monitoring (SIM) was performed on the following fragment ions: m/z 557 and 560 for the derivative of levoprotiline and its trideuterated standard, respectively, and m/z 543 and 552 for the derivatives of N-desmethyl-levoprotiline and its nonadeuterated standard, respectively. Quantitative evaluation Quantification was based on the peak-area ratio (= y) of the substance and its deuterated standard. The linear least-squares regression line y = a + bx was calculated,
71 1
where x was the concentration of levoprotiline or desmethyl-levoprotiline in the calibration samples. For the levoprotiline concentration range 0.04 to 2.1 pmol per sample, weighted linear regression with a weighting factorA = l/(x,)' and a power factor k = 1.2 was used to calculate the calibration line. Subsequently, 'unknown' concentrations in validation and clinical samples were calculated from the calibration line. For clinical samples which had been weighed, the concentrations were corrected for the sample weight yielding molar 'concentrations' based on 1 g of the biological fluid (e.g. pmol g-'). To report the results in SI units (i.e. nmol 1-I), the values may be multiplied by the density of the biological fluid (i.e. 1.0595 g ml-' for blood, 1.0269 g ml-' for plasma, 1.015 g ml- for urine).16
RESULTS AND DISCUSSION Stock solutions Different stock solutions of levoprotiline or desmethyllevoprotiline were used for the preparation of calibration and validation samples The two stock solutions of each compound were first compared in a separate experiment, where aliquots of the solutions were mixed with the internal standards, derivatized and analysed by GC/MS. If the average peak-area ratio of substance to internal standard differed by less than 1% between two solutions, these solutions were considered as equivalent and were used for actual drug assays. Each of the two stock solutions was alternately used to prepare calibration or validation samples. Isolation procedure and derivatization The isolation of levoprotiline and its desmethyl metabolite from biological fluids was achieved by conventional liquid-liquid extraction with toluene at pH values above 10. Endogenous compounds which may affect the chromatographic resolution were removed by two further extraction steps: first, back-extraction of the analytes into acidified aqueous solution (pH < 4); second, re-extraction from the aqueous phase after alkalization into hexane. Although the extraction yield of the last step could be slightly increased by using toluene, we preferred hexane since this solvent was easier to evaporate. The overall yield of the three-step isolation procedure was about 75% for both compounds. The background in the gas chromatograms was comparable after extraction of the compounds from 1 to 3 ml of blood, plasma or urine. Consequently, the sensitivity of the method can be improved by using more than 1 ml of the biological fluids for analysis. The 0,N-bis-pentafluoropropionyl derivatives of levoprotiline and desmethyl-levoprotilinewere obtained by reaction with perfluoropropionic anhydride in toluene at 100°C. The reaction was complete after 30 min at the latest. The N-desmethyl compound yielded only one product ; no tris-pentafluoropropionyl derivative was found. Furthermore, no 'H/'H-exchange was
L
- 0
- 0 -0
- 0
- m
-In
- 0
2
N
&+ -
0
p. 0
N .
,
-
-
I
I
I
0 c
gq 0
F=: m
p
0 (D
m
t
I I
I
I
I
I
I
0
0
In
713
LEVOPROTILINE AND A METABOLITE BY GC/MS
graphic resolution and-as a consequence-the sensitivity by a factor of 2 to 5, dependent on the calibration range. The mass spectra of the derivatives of levoprotiline, desmethyl-levoprotiline and their deuterated internal standards are shown in Fig. 2. Molecular ions of low abundance were observed at m/z 585 and 588 for the derivatives of levoprotiline and its standard, respectively, and at m/z 571 and 580 for the derivatives of desmethyl-levoprotiline and its standard, respectively. For SIM, the intensive fragment ions resulting from the loss of the ethylene bridge, i.e. the [M - 28]+' ions, at m/z 551, 560, 543 and 552, respectively, were chosen. Typical SIM chromatograms for levoprotiline in blood are shown in Fig. 3. A characteristic SIM chromatogram for levoprotiline and its desmethyl metabolite in urine is shown in Fig. 4. No interference peaks derived from endogenous components were observed. When using the MSD equipped with the Pascal workstation, we had to take into account that the data system has no multitasking capabilities. If in the data acquisition mode the chromatogram is 'simultaneously' drawn on the screen, the peak resolution suffered. The data sytem required about 120 ms per ion and scan for graphic data processing. Thus, the actual scan circle time in the SIM mode with two ions was >240 ms instead of a dwell time of 20 ms selected in the SIM parameter input. Due to the small peak width of capillary column peaks, the number of scans was then not sufficient to characterize the peak precisely. Thus, we
observed for the deuterated compounds used as internal standards. All extraction steps were done in one-way glass ampoules. This technique offers several advantages : any contact of organic solvents with plastic material is avoided which reduces the risk of chromatographic interferences; furthermore, the ampoules are used once, then discarded. Thus, there is no risk of a contamination from preceding analyses. The separation of aqueous (heavier) and organic phase (lighter) and the transfer of the organic phase from one to another ampoule is very convenient: after freezing the aqueous phase, the neck of the ampoule is broken off, then the organic solvent is simply poured into a new ampoule.
GCWS Packed column GC was previously used in the GC/MS method for racemic oxaprotiline.' This technique could also be applied to levoprotiline. However, the sensitivity of this method did not allow accurate and precise determinations during the terminal elimination of levoprotiline from the body. Here we have replaced the packed columns by micro-bore (0.2 mm id.) fusedsilica capillary columns which improved the chromato-
r
VJ
c
C
3
2ooj,, I,, ,,I
0 0 I
6.00
, ,
I
I
I
I
,
6.1 0
,
,
,
6.20
6.30
a l l
I
6.00
6.10
i
,, I I , , , , , , , , , , , , , , ,
6.20
6.30
I
m / z 543
1
m/z 5 5 2
I
1
I
20000
j , i L 5.8
6.0
6.2
6.4
6.6
6.8
5.8
6.0
6.2
6.4
6.6
6.8
5.8
6.0
6.2
6.4
6.6
6.8
all
I
Retention time [mini
Figure 3. Typical selected ion chromatograms of a derivatized extract of 1 ml of human blood spiked with (A) 73 pmol of ('H,)oxaprotiline (internal standard) only, and (6)0.9 pmol of levoprotiline and 73 pmol of ('H,)oxaprotiline. SIM was performed on the ions m b 557 for levoorotiline and mlz 560 for the labelled internal standard.
Retention t i m e [rninl Figure 4. Typical selected ion chromatograms of a derivatired extract of 1 ml of human urine spiked with 2.6 pmol of desmethyllevoprotiline (m/z 543). 2 0 pmol of (ZH,)desmethyl-oxaprotiline (m/z 552). 20 Dmol of IevoDrotiline (.m,h 557) and 200 Dmol of ('H,)oxaprotiline (m/z 560):
714
R. ACKERMANN, G. KAISER, F. SCHUELLER AND W. DIETERLE
selected the ‘trace-no’ mode, meaning that the data system did not draw the chromatographic data on the screen during data acquisition. Then, the scan circle time was small enough to achieve sufficient resolution of the peaks. Calibration, assay linearity The calibration curves showed a linear response in the used concentration ranges (see Experimental; Preparation of spiked samples). Typical parameters for - a calibration curve of levoprotiline in blood (concentration range: 0.6-41 pmol per sample; amount of internal standard: 40 pmol) were: y = 0.03 0.003238 x (where y is the peak area ratio and x the concentration), standard error of estimate sy = 0.00968 and coefficient of correlation R = 0.99975.
+
Specificity The described method is not enantiospecific, i.e. it does not differentiate between the (-)-R- (i.e. levoprotiline) and the ( +)-S-enantiomer of racemic oxaprotiline. Thus, the results can only be expressed in terms of levoprotiline if there is no in uiuo inversion of the configuration. In fact, we have previously demonstrated that Ievoprotiline is not converted to the (+)-S-enantiomer in man. Urine samples collected after single and repeated dosing of levoprotiline.HC1 were analysed for the main biotransformation products of racemic oxaportiline, i.e. the diastereoisomeric 0-glucuronides. If any configurational inversion occurred in man, the 0glucuronide of ( + )-S-oxaprotiline should be detected in urine. In previous studies with 14C-labelled oxaprotiline, the diastereoisomeric oxaprotiline glucuronides were completely separated by HPLC.’ Urine from the above trial was subjected to the same HPLC separation and the fractions corresponding to the positions of the glucuronides of (-)-R- and (+)-S-oxaprotiline were collected. After acidic treatment to hydrolyse the conjugates, these fractions were individually analysed by GC/MS13 for oxaprotiline. In all cases, no oxaprotiline was detected in the (+)-S-fraction, whereas the amount found in the (-)-R-fraction agreed well with the total amount measured directly in the corresponding urine sample. Hence, a conversion from levoprotiline to its (+)-S-enantiomer can be excluded. Based on these results we conclude that there is also no inversion of desmethyl-levoprotiline. Validation, Accuracy, precision The method was validated by analysis of spiked biological samples. Six validation samples with different concentrations were analysed. Recoveries were calculated in percentage of each given concentration. The intra-day accuracy and precision were determined by averaging the recoveries found on any given day, without considering samples whose concentrations were below the limit of quantification (see the following section). Typical results of intra-day validation are listed in Table 1 for levoprotiline in blood. The mean recov-
Table 1. Intra-day validation for IevoprotiC i w in blood Five spiked samples in the concentration range of 6-40 pmol per sample were adysed at each of 11 analysis days and the recoveries (in percentage of tbe given concentrations) were averaged on each day RWOV6IV
Dav
Mean
cv
1 2 3 4 5 6 7 8 9 10 11
98.5 101.3 97.2 107.1 100.2 102.7 100.0 98.9 101.2 101.5 102.7
4.4 3.2 5.3 1.7 2.0 1.3 3.2 3.9 3.3 5.1 1.6
CV: coefficient of variation (%).
eries ranged between 97.2 and 107.1%, the coefficients of variation (CV) between 1.3 and 5.3%. The inter-day accuracy and precision were determined by averaging the recoveries found on different days for the samples in a given concentration range. Results of inter-day validation are listed in Table 2 for levoprotiline in blood, plasma and urine and in Table 3 for desmethyl-levoprotiline in urine. With the exception of the lowest concentration each, the mean recoveries were in the range 96.3-103.2% and demonstrated the good accuracy of the method. In the low validation range of levoprotiline in blood, a good precision was achieved for concentrations above 0.7 pmol per sample (CV: 5.5-9.3%). In the normal validation ranges of levoprotiline and desmethyl-levoprotiline, a highprecision CV (1.8-6.1%) was obtained for all concentrations except the lowest ones.
Limits of detection and quantification The lowest concentration of levoprotiline which resulted in a signal-to-noise ratio > 3 was 0.1 pmol per sample (1 ml of biological fluid used). In actual assays, the limit of detection (LOD) was dependent on the range of concentrations covered by the calibration and validation samples. The lowest LOD was reached only in blood assays using the concentration range of 0.04 to 2.1 pmol per sample. For the other concentration ranges, we used the inter-day coefficient of variation (CV) to estimate retrospectively LOD, since the lowest concentrations of the spiked samples were already too high to estimate a signal-to-noise ratio (see Figs 3 and 4). The LOD was estimated as that concentration at which the CV was about 100%. LOD of levoprotiline in blood (normal concentration range), plasma and urine was 0.2, 0.3 and 2 pmol per sample, respectively. For desmethyl-levoprotiline in urine, LOD was 1 pmol per sample.
715
LEVOPROTILINE AND A METABOLITE BY GC/MS
Table 2 Interday validation for levoprotiline in blood, plasma and urine. Six spiked samples with different concentrations were analysed in each assay and tbe recoveries (in percentage of the given concentrations) were averaged in the indicated concentration ranges Blood, low concentration range Range of given conc. (pmol per sample) Mean reovery
cv (Yo) N
0.04
0.36
to
to
0.05 47.6 142.0 6
0.42 96.4 19.5 6
0.72 to 0.85 103.1 7.8 6
1.07 to 1.26 99.2 5.5 6
1.43 to 1.68 102.0 9.3 6
1.78 to 2.10 102.? 6.6 6
25.2
Blood, normal concentration range Range of given conc. (pmol per sample) Mean recovery
cv (%) N
0.6 to 0.9 146.5 21.8 11
6.2 to 8.1 103.2 3.8 11
12.4 to 16.2 100.1 4.5 11
18.9 to 24.4 99.5 3.7 11
32.6 101.7 3.2 11
31.3 to 40.7 100.8 4.2 11
12.3 to 12.4 100.9 1.8 3
18.1 to 19.1 100.3 4.9 3
23.6 to 24.3 98.6 3.1 3
29.7 to 30.1 98.2 6.1 2
Urine 185 368 to to 203 404 99.8 100.0 1.9 2.7 7 7
521 to 574 100.3 2.1 7
678 to 768 99.8 2.1 7
852 to 962 99.8 2.1 7
to
Plasma Range of given conc. (pmol per sample) Mean recovery cv (%) N
0.6 96.7 41.8 3
Range of given conc. (pmol per sample) Mean recovery cv (%) N
17.5 to 19.2 108.4 11.1 7
0.6 to
6.1 to 6.2 96.3 6.1 3
Table 3. Interday validation for desmethyl-levoprotiline in urine. Six spiked samples with different concentratiow were analysed in each assay and the recoveries (in percentage of the given concentrations) were averaged in tbe indicated concentration ranges Range of given conc. (pmol per sample) Mean recovery cv (%)
N
87.1 111.3 2.2 24.0 42.3 64.1 to to to to to to 2.8 27.4 52.7 76.2 101.0 129.8 115.7 98.3 98.2 97.9 98.9 97.7 38.9 3.1 2.9 3.1 2.7 2.5 7 7 7 7 7 7
By analogy, the limit of quantification (LOQ) was defined as that concentration at which the CV was about 10%. For levoprotiline in blood, LOQ was 0.7 pmol per sample in the low concentration range and 2 pmol per sample in the normal concentration range. In plasma and urine, LOQ was 3 and 20 pmol per sample, respectively. For desmethyl-levoprotiline in urine, LOQ was 10 pmol per sample. Using 1 ml of the biological fluid for analysis the above values in pmol per sample can be equated with pmol ml-' or nmol I-'. If larger volumes of the biological fluids (2 or 3 ml) were used for analysis, then the actual limits in nmol I-' were correspondingly lower, i.e. the sensitivity was increased.
Stability of biological samples We demonstrated that racemic oxaprotiline was stable in blood, plasma and urine samples for at least 3 months when deep-frozen at - 18"C.The stability was tested both in spiked and in clinical samples. For example, 13 clinical plasma samples were analysed three times. The second analysis was done 9 days and the third analysis 87 days after the first analysis. Compared with the first analysis, the mean recovery for oxaprotiline concentrations above 4 nmol 1-' was 99.0% in analysis 2 and 100.6% in analysis 3. Conclusively, levoprotiline, the ( -)-R-enantiomer of oxaprotiline, has the same stability characteristics as the racemate. It is worthwhile to mention that a possible racemization during storage would not affect the validity of the data since the assay measures both enantiomers. Application The described method was applied to the determination of unchanged levoprotiline in blood, plasma and urine and of desmethyl-levoprotiline in urine of healthy volunteers dosed orally with 75 mg of levoprotiline.HC1. For illustration, the blood concentration-time curve of levoprotiline is shown in Fig. 5 and the cumulative urinary excretion of unchanged and desmethyllevoprotiline is depicted in Fig. 6. The drug was rapidly
R. ACKERMANN, G. KAISER, F. SCHUELLER A N D W. DIETERLE
(u
0
0.6
Levoprotiline Desmethyl-levoprotiline
0.5
0
u
LC
-
0.4
a7
0 ._ c y
0.3
0.2
0 X
I d
0.1
0.0 0
12
24
36
48
60
72
84
0
96
12
T i m e a f t e r dosing [ h ]
24
36
48
60
72
Time a f t e r dosing [h]
Figure 5. Blood concentrations of unchanged levoprotiline in healthy volunteers after a single oral dose of 75 mg of levoprotiline.HCI; mean f SD, N = 12.
Figure 6. Cumulative urinary excretion of levoprotiline and desmethyl-levoprotiline in healthy volunteers after a single oral dose of 75 mg of levoprotiline.HC1; mean f SD, N = 6.
absorbed (t,,, = 4 h) and eliminated from blood with a terminal half-life of about 20 h. Less than 2% of the dose was excreted in urine as unchanged drug and its desmethyl metabolite, confirming previous findings that levoprotiline and desmethyl-levoprotiline are mainly excreted in the form of their gl~curonides.~
blood, plasma and urine and the simultaneous determination of unchanged and desmethyl-levoprotiline in urine. The method is suitable for use in pharmacokinetic and bioavailability studies of the antidepressant levoprotiline.HC1.
_
_
~
CONCLUSION Acknowledgements The GC/MS method described here permits the specific and highly sensitive determination of levoprotiline in
The authors are indebted to Dr W. Kiing and his collaborators for the synthesis of the deuterated standards.
REFERENCES 1. P. C. Waldmeier, P. A. Baumann, M. Wilhelm, R. Bernasconi and L. Maitre, Eur. J. Pharmacol. 66,387(1977). 2. P. C. Waldmeier, P. A. Baumann, K. Hauser, L. Maitre and A. Storni, Biochem. Pharmacol. 31,2169 (1982). 3. S. A. Checkley, C. Thompson, S. Burton, C. Franey and J. Arendt, Psychopharmacology 87,116(1985). 4. M. Wolfersdorf, G. Wendt, U. Binz, B. Steiner and G. Hole, Pharmacopsychiatry 21,203(1988). 5. G. Wendt and U. Binz, Munch. Med. Wschr. 132 (Suppl. 1). 544 (1990). 6. K. NBhunek, J. Svestka, E. Ceskovh, R. RysBnek, V. Obrovska, J. Misurec and G. Wendt, Munch. Med. Weschr. 132 (Suppl. l ) , S50 (1990). 7. W. Dieterle, J. W. Faigle, P. Imhof, M. Sulc and J. Wagner, Xenobiotica 14,303 (1984). 8. W. Dieterle, J. W. Faigle, H.-P. Kriemler and T. Winkler, Xenobiotica 14,31 1 (1984). 9.W. Dieterle, J. W. Faigle, W. Kung and W. Theobald, Biopharm. Drug. Dispos. 5,377(1984).
10. W. Dieterle, R. Ackermann, G. Kaiser and W. Theobald, Munch. Med. Wschr. 132 (Suppl. l),S39 (1990). 11. H. J. Kuss and E. Feistenauer, J. Chromafogr. 204, 349 (1981 ). 12. U. Breyer-Pfaff, R. Wiatr and K. Nill, J. Chromatogr. 309, 107 (1984). 13. R. Ackermann, G. Kaiser, W. Dieterle and J. P. Dubois, in Proceedings of the Third European Congress of Biopharmaceutics and Pharmacokinetics, ed. by J. M. Aiche and J. Hirtz, vol. I, p. 278. lmprimerie de I'Universit6, Clermont-Ferrand, France
(1987). 14. M. Jedrychowski, E. Hoffmann and P. R. Bieck, J. Pharm. Biomed. Anal. 7, 1 897 ( 1 989). 15. C. Horne, H. Spahn and E. Mutschler, Arzneim.-Forsch./Drug Res. 37, 1 1 79 (1987). 16. Geigy Scientic Tables, ed. C. Lentner, vol. 1 (1981 ) and vol. 3 (1984).Ciba-Geigy, Basle.
Determination of the Antidepressant Levoprotiline and its N-Desmethyl Metabolite in Biological Fluids by Gas Chromatography/Mass Spectrometry R. Ackermann,? G. Kaiser, F. Schueller and W. Dieterle Research and Development Department, Pharmaceuticals Division, CIBA-GEIGY Ltd, CH-4002 Bade, Switzerland
A specific and sensitive gas chromatographic/mass spectrometric method was developed and validated for the determination of the antidepressant levoprotiline in blood, plasma and urine and the simultaneous determination of
levoprotiline and its desmethyl metabolite in urine. Deuterium-labelled analogues were used as internal standards. The compounds were isolated from the biological fluids by liquid-liquid extraction under basic conditions. Following derivatization with perlluoropropionic anhydride, the samples were analysed by capillary column gas chromatogaphy/electron impact mass spectrometry with selected ion monitoring. The analysis of spiked samples demonstrated the high accuracy and precision of the method. Blood concentrations of levoprotiline down to 0.7 nmol I-' (1 ml used for analysis) could be quantified with a coefficient of variation of 10% or less. The method is suitable for use in pharmacokinetic and bioavailability studies of levoprotiline in humans.
H
INTRODUCTION
I /N--R
The new antidepressant levoprotiline, (R)-a-[(methyl10H)-ethanol, amino)methyl]-9,10-ethanoanthracene-9( is chemically the ( -)-R-enantiomer of racemic oxaprotiline. It is administered perorally in the form of a hydrochloride (levoprotiline.HC1, Fig. 1). In experimental animals and man, racemic oxaprotiline and its (+)S-enantiomer are potent inhibitors of noradrenaline uptake, while levoprotiline has no such activity. However, in preliminary clinical trials, levoprotiline has shown comparable antidepressive activity to traditional antidepressants, but distinctly better tolerability.'6 The basic pharmacokinetics and metabolic characteristics of oxaprotiline and levoprotiline have already been reported.7-'0 After peroral administration to man, oxaprotiline and levoprotiline are almost completely absorbed and extensively metabolized. Direct 0-glucuronidation is the major metabolic pathway; N-desmethyl-levoprotiline is a minor metabolite. Several methods for the analysis of oxaprotiline and levoprotiline in biological fluids have been described. High-performance liquid chromatography (HPLC) and packed column gas chromatography/mass spectrometry (GC/MS) were used for the determination of oxaprotiline in blood, plasma and urine."-13 HPLC and thinlayer chromatography (TLC) were applied to measure levoprotiline in plasma or b l ~ o d . ' ~ ,We ' ~ have now developed a highly sensitive capillary GC/MS method which allows the accurate and precise determination of levoprotiiine in blood down to concentrations of less than 1 nmol 1-'. The method is also suitable for -f Author to whom correspondence should be addressed.
1052-9306/91/11070948 $05.00
0 1991 by John Wiley & Sons, Ltd.
R = CH3
Levoprotiline
R - H
D e s m e t hy I -1evoprot i I ine levoprotiline.HCI and desmethyl-
1 . Structure of levoprotiline.HCI. Figure
determining levoprotiline and its desmethyl metabolite simultaneously in urine. In this paper we describe the details of the method and demonstrate their application by assaying blood and urine samples of healthy volunteers dosed orally with levoprotiline.HC1.
EXPERIMENTAL Reagents and materials Levoprotiline.HC1 and N-desmethyl-levoprotiline.HC1 were available from CIBA-GEIGY (Bade, Switzerland). The deuterated internal standards, (2H,)oxaprotiline.HC1 ('H,-methyl moiety, isotopic purity: 'H, 5.0%, 'H, 93.7%, 'H, 1.2%) and ('H,)N-desmethyloxaprotiline.HC1 (deuterated in the anthracene ring system, isotopic purity: 'H, 26.6%, 2H,71.9%, 2H,o 1.4%), were synthesized in the Isotope Laboratory of CIBA-GEIGY. All solvents and chemicals were of anaReceived 12 June 1991 Revised 29 July 1991
710
R. ACKERMANN, G. KAISER, F. SCHUELLER A N D W. DIETERLE
lytical grade and were obtained from Fluka (Buchs, Switzerland) or E. Merck (Darmstadt, Germany). A 2 mol 1-' sodium carbonate solution was used as buffer (pH 10.8). Water used in the experiments was bidistilled. Human blank blood and plasma were obtained from Blutspendezentrum, SRK, Basle, Switzerland, and human blank urine from healthy volunteers. Biological fluids were stored below -18°C until used for preparing spiked samples. The extraction steps were performed in one-way glass ampoules. The samples or solutions were dosed with automatic pipettes from Gilson (Villier-le-Bel, France) and weighed on analytical balances, type ME 30 or AE 50 (Mettler, Greifensee, Switzerland). Shaking was done either on a horizontal shaker, type TRl (Infors, Basle, Switzerland), or on a Vortex evaporator (HaakleBuchler, Saddle Brook, New Jersey, USA). The samples were centrifuged on a Multex centrifuge (MSE Scientific Instruments, Crawley, Sussex, UK). The clinical samples were ultrasonicated using an ultrasonic desintegrator, type Sonifer B 12 (Branson Sonic Power Company, Danbury, USA). A Speed Vac combined evaporator/ concentrator (Savant Instruments Inc., Farmingdale, New York, USA) was used to concentrate solutions, or the solvents were evaporated under a stream of nitrogen at 40"C. Stock and working solutions Stock solutions of levoprotiline.HC1, desmethyllevoprotiline.HC1 and of the corresponding internal standards were prepared in methanol and were stored at 6 "C. Although the solutions were found to be stable for at least 6 months, new stock solutions were prepared every 3 months since small losses in volume may occur by evaporation if the solutions are used every day. For each assay, working solutions were obtained by dilution of the corresponding stock solution with 0.01 mol 1-' hydrochloric acid. For the simultaneous determination of levoprotiline and desmethyl-levoprotiline in urine, a working solution containing both compounds was prepared. The concentrations of the compounds in the stock and working solutions depended on the concentration range to be covered by the spiked samples. Approximate concentrations of stock and working solutions of levoprotiline.HC1 were 30 and 0.08 pmol 1-', respectively, for assays of blood or plasma, and 350 and 2 pmol 1-', respectively, for assays of urine. To prepare samples with extremely low concentrations of levoprotiline.HC1, the stock solution was diluted twice, resulting in a concentration of the working solution of about 0.004 pmol I-'. The concentrations of desmethyllevoprotiline.HC1 in the stock and working solutions were 40 and 0.25 pmol 1-', respectively. About two times higher concentrations were used for the stock and working solutions of the deuterated standards.
To 0.5 or 1 ml of the biological fluid were added: 10-500 pl of the working solution, an aliquot (100-200 pl) of the internal standard working solution and aliquots of 0.01 mol 1-1 hydrochloric acid to obtain equal volumes in all samples. The samples were shaken on a vibration shaker at speed 7 for 5 min and subsequently centrifuged. Routinely used concentration ranges of calibration and validation samples were: 0.6 to 40 pmol per sample for the determination of unchanged levoprotiline in blood and plasma (internal standard: 30 pmol per sample); 20-1000 pmol per sample for the determination of unchanged levoprotiline in urine (internal standard: 800 pmol per sample); 2-120 pmol per sample for the determination of desmethyl-levoprotiline in urine (internal standard: 110 pmol per sample). To determine extremely low concentrations of levoprotifine in blood, a calibration and validation range of 0.04-2.1 pmol per sample was used (internal standard: 1.7 pmol per sample).
Preparation of clinical samples Clinical samples (i.e. samples from clinical trial with levoprotiline.HC1) were ultrasonicated for 10 s at 50 W, then 0.5 or 1 ml were weighed into 10 ml glass ampoules. For samples whose concentrations were expected to lie above the calibration range, smaller volumes were used and blank blood, plasma or urine was added to obtain the same volume for all samples. The exact sample amount used was always determined by weighing. To each sample were added: an aliquot (100-200 pl) of the internal standard working solution and aliquots of 0.01 mol 1-' hydrochloric acid to obtain the same total volume as for the spiked samples. Sample working-up procedure Both spiked and clinical samples were equilibrated for 15 min at 200 rev min-' on a horizontal shaker. Then, 0.5 ml of buffer pH 10.8 and 4.5 ml of toluene were added. Each ampoule was sealed and shaken on the horizontal shaker for 10 min at 300 rev min-' followed by centrifugation at 1250 x g for 10 min. After opening the ampoule, the two phases were separated by freezing the aqueous phase in a dry iceethanol mixture and transferring the organic phase into a new ampoule. One ml of 0.1 mol 1-' hydrochloric acid was added to the organic phase, and the mixture was shaken, centrifuged and frozen as described above. The organic phase was discarded. The aqueous phase was then made alkaline by addition of 0.5 ml of buffer (pH 10.Q and extracted with 2 ml of hexane as described above. The organic phase was transferred into a 5 ml ampoule and evaporated to dryness.
Preparation of spiked samples
Derivatization procedure
With each assay, six calibration samples and six validation samples were prepared by spiking human blank, blood, plasma or urine in 10 ml glass ampoules.
The residue from the above extraction was dissolved in 0.2 ml of toluene and 20 pl of perfluoropropionic anhydride were added. After vortexing for 5 s at speed 10,
LEVOPROTILINE AND A METABOLITE BY G U M S
the mixture was heated to 100°C for 0.5 h. Then, 1 ml of a mixture of methanol-water (1 :2, v/v) and 1.4 ml of hexane were added and the sample vortexed for 1 min at speed 9, followed by a short centrifugation at 1250 x g. The aqueous phase was frozen and the organic phase transferred into a 1.5 ml conical vial. The residue was concentrated successively with 400,200 and 80 pl of hexane, reconstituted in 1&20 pl of toluene, and 1-2 pl were injected into the gas chromatograph/ mass spectrometer. GC/MS Analyses were performed either on a Hewlett-Packard 5987 B GC/MS system (equipped with an H P lo00 multiuser/multitasking computer) or on an H P 5890 A gas chromatograph interfaced with an HP 5970 mass selective detector (MSD; equipped with a Pascal Workstation). At both instruments, the gas chromatograph was equipped with an H P 7673 automatic sample injector and a split/splitless capillary inlet system. A fused-silica capillary column (12 m, 0.2 mm id.), coated with cross-linked methyl silicone (film thickness 0.33 pm, purchased from Hewlett-Packard) was inserted directly into the ion source of the mass spectrometer and connected with the injector port of the gas chromatograph by an uncoated, deactivated fused silica precolumn (1.2 m, 0.32 mm i.d.), acting as a retention gap. A sylanized glass liner containing some glass wool at the lower end was used in the injector port. The injector temperature was maintained at 270 "C. The carrier gas was helium at an inlet pressure of 100 kPa, resulting in a gas linear velocity of 50 cm s-'. The corresponding flow was approximately 1 ml min-'. Injections were made in splitless mode (0.5 min splitless period) with a split flow of 50 ml min-' and a septum purge of 3 ml min-'. The column oven was operated isothermally at 100°C for 0.5 min after injection, heated at a rate of 30°C min-' to 290°C and then held at this temperature until the end of the run. The total run time was about 6.5-7.0 min. The GC/MS interface was maintained at 260 "C (280°C for the MSD); the ion source temperature was set to 200°C. The mass spectrometer was operated under EI ionization conditions. At the beginning of each analysis day, the mass spectrometer was calibrated with the Autotune program. The ion source parameters were then adjusted to obtain the highest signal on the fragment ion m/z 502 of the calibration compound perfluorotributylamine. The peak width at one-half of peak height was about 1 Da for the H P 5987 system and 0.7 Da for the MSD. The peak dwell time was set to 20 ms. Selected ion monitoring (SIM) was performed on the following fragment ions: m/z 557 and 560 for the derivative of levoprotiline and its trideuterated standard, respectively, and m/z 543 and 552 for the derivatives of N-desmethyl-levoprotiline and its nonadeuterated standard, respectively. Quantitative evaluation Quantification was based on the peak-area ratio (= y) of the substance and its deuterated standard. The linear least-squares regression line y = a + bx was calculated,
71 1
where x was the concentration of levoprotiline or desmethyl-levoprotiline in the calibration samples. For the levoprotiline concentration range 0.04 to 2.1 pmol per sample, weighted linear regression with a weighting factorA = l/(x,)' and a power factor k = 1.2 was used to calculate the calibration line. Subsequently, 'unknown' concentrations in validation and clinical samples were calculated from the calibration line. For clinical samples which had been weighed, the concentrations were corrected for the sample weight yielding molar 'concentrations' based on 1 g of the biological fluid (e.g. pmol g-'). To report the results in SI units (i.e. nmol 1-I), the values may be multiplied by the density of the biological fluid (i.e. 1.0595 g ml-' for blood, 1.0269 g ml-' for plasma, 1.015 g ml- for urine).16
RESULTS AND DISCUSSION Stock solutions Different stock solutions of levoprotiline or desmethyllevoprotiline were used for the preparation of calibration and validation samples The two stock solutions of each compound were first compared in a separate experiment, where aliquots of the solutions were mixed with the internal standards, derivatized and analysed by GC/MS. If the average peak-area ratio of substance to internal standard differed by less than 1% between two solutions, these solutions were considered as equivalent and were used for actual drug assays. Each of the two stock solutions was alternately used to prepare calibration or validation samples. Isolation procedure and derivatization The isolation of levoprotiline and its desmethyl metabolite from biological fluids was achieved by conventional liquid-liquid extraction with toluene at pH values above 10. Endogenous compounds which may affect the chromatographic resolution were removed by two further extraction steps: first, back-extraction of the analytes into acidified aqueous solution (pH < 4); second, re-extraction from the aqueous phase after alkalization into hexane. Although the extraction yield of the last step could be slightly increased by using toluene, we preferred hexane since this solvent was easier to evaporate. The overall yield of the three-step isolation procedure was about 75% for both compounds. The background in the gas chromatograms was comparable after extraction of the compounds from 1 to 3 ml of blood, plasma or urine. Consequently, the sensitivity of the method can be improved by using more than 1 ml of the biological fluids for analysis. The 0,N-bis-pentafluoropropionyl derivatives of levoprotiline and desmethyl-levoprotilinewere obtained by reaction with perfluoropropionic anhydride in toluene at 100°C. The reaction was complete after 30 min at the latest. The N-desmethyl compound yielded only one product ; no tris-pentafluoropropionyl derivative was found. Furthermore, no 'H/'H-exchange was
L
- 0
- 0 -0
- 0
- m
-In
- 0
2
N
&+ -
0
p. 0
N .
,
-
-
I
I
I
0 c
gq 0
F=: m
p
0 (D
m
t
I I
I
I
I
I
I
0
0
In
713
LEVOPROTILINE AND A METABOLITE BY GC/MS
graphic resolution and-as a consequence-the sensitivity by a factor of 2 to 5, dependent on the calibration range. The mass spectra of the derivatives of levoprotiline, desmethyl-levoprotiline and their deuterated internal standards are shown in Fig. 2. Molecular ions of low abundance were observed at m/z 585 and 588 for the derivatives of levoprotiline and its standard, respectively, and at m/z 571 and 580 for the derivatives of desmethyl-levoprotiline and its standard, respectively. For SIM, the intensive fragment ions resulting from the loss of the ethylene bridge, i.e. the [M - 28]+' ions, at m/z 551, 560, 543 and 552, respectively, were chosen. Typical SIM chromatograms for levoprotiline in blood are shown in Fig. 3. A characteristic SIM chromatogram for levoprotiline and its desmethyl metabolite in urine is shown in Fig. 4. No interference peaks derived from endogenous components were observed. When using the MSD equipped with the Pascal workstation, we had to take into account that the data system has no multitasking capabilities. If in the data acquisition mode the chromatogram is 'simultaneously' drawn on the screen, the peak resolution suffered. The data sytem required about 120 ms per ion and scan for graphic data processing. Thus, the actual scan circle time in the SIM mode with two ions was >240 ms instead of a dwell time of 20 ms selected in the SIM parameter input. Due to the small peak width of capillary column peaks, the number of scans was then not sufficient to characterize the peak precisely. Thus, we
observed for the deuterated compounds used as internal standards. All extraction steps were done in one-way glass ampoules. This technique offers several advantages : any contact of organic solvents with plastic material is avoided which reduces the risk of chromatographic interferences; furthermore, the ampoules are used once, then discarded. Thus, there is no risk of a contamination from preceding analyses. The separation of aqueous (heavier) and organic phase (lighter) and the transfer of the organic phase from one to another ampoule is very convenient: after freezing the aqueous phase, the neck of the ampoule is broken off, then the organic solvent is simply poured into a new ampoule.
GCWS Packed column GC was previously used in the GC/MS method for racemic oxaprotiline.' This technique could also be applied to levoprotiline. However, the sensitivity of this method did not allow accurate and precise determinations during the terminal elimination of levoprotiline from the body. Here we have replaced the packed columns by micro-bore (0.2 mm id.) fusedsilica capillary columns which improved the chromato-
r
VJ
c
C
3
2ooj,, I,, ,,I
0 0 I
6.00
, ,
I
I
I
I
,
6.1 0
,
,
,
6.20
6.30
a l l
I
6.00
6.10
i
,, I I , , , , , , , , , , , , , , ,
6.20
6.30
I
m / z 543
1
m/z 5 5 2
I
1
I
20000
j , i L 5.8
6.0
6.2
6.4
6.6
6.8
5.8
6.0
6.2
6.4
6.6
6.8
5.8
6.0
6.2
6.4
6.6
6.8
all
I
Retention time [mini
Figure 3. Typical selected ion chromatograms of a derivatized extract of 1 ml of human blood spiked with (A) 73 pmol of ('H,)oxaprotiline (internal standard) only, and (6)0.9 pmol of levoprotiline and 73 pmol of ('H,)oxaprotiline. SIM was performed on the ions m b 557 for levoorotiline and mlz 560 for the labelled internal standard.
Retention t i m e [rninl Figure 4. Typical selected ion chromatograms of a derivatired extract of 1 ml of human urine spiked with 2.6 pmol of desmethyllevoprotiline (m/z 543). 2 0 pmol of (ZH,)desmethyl-oxaprotiline (m/z 552). 20 Dmol of IevoDrotiline (.m,h 557) and 200 Dmol of ('H,)oxaprotiline (m/z 560):
714
R. ACKERMANN, G. KAISER, F. SCHUELLER AND W. DIETERLE
selected the ‘trace-no’ mode, meaning that the data system did not draw the chromatographic data on the screen during data acquisition. Then, the scan circle time was small enough to achieve sufficient resolution of the peaks. Calibration, assay linearity The calibration curves showed a linear response in the used concentration ranges (see Experimental; Preparation of spiked samples). Typical parameters for - a calibration curve of levoprotiline in blood (concentration range: 0.6-41 pmol per sample; amount of internal standard: 40 pmol) were: y = 0.03 0.003238 x (where y is the peak area ratio and x the concentration), standard error of estimate sy = 0.00968 and coefficient of correlation R = 0.99975.
+
Specificity The described method is not enantiospecific, i.e. it does not differentiate between the (-)-R- (i.e. levoprotiline) and the ( +)-S-enantiomer of racemic oxaprotiline. Thus, the results can only be expressed in terms of levoprotiline if there is no in uiuo inversion of the configuration. In fact, we have previously demonstrated that Ievoprotiline is not converted to the (+)-S-enantiomer in man. Urine samples collected after single and repeated dosing of levoprotiline.HC1 were analysed for the main biotransformation products of racemic oxaportiline, i.e. the diastereoisomeric 0-glucuronides. If any configurational inversion occurred in man, the 0glucuronide of ( + )-S-oxaprotiline should be detected in urine. In previous studies with 14C-labelled oxaprotiline, the diastereoisomeric oxaprotiline glucuronides were completely separated by HPLC.’ Urine from the above trial was subjected to the same HPLC separation and the fractions corresponding to the positions of the glucuronides of (-)-R- and (+)-S-oxaprotiline were collected. After acidic treatment to hydrolyse the conjugates, these fractions were individually analysed by GC/MS13 for oxaprotiline. In all cases, no oxaprotiline was detected in the (+)-S-fraction, whereas the amount found in the (-)-R-fraction agreed well with the total amount measured directly in the corresponding urine sample. Hence, a conversion from levoprotiline to its (+)-S-enantiomer can be excluded. Based on these results we conclude that there is also no inversion of desmethyl-levoprotiline. Validation, Accuracy, precision The method was validated by analysis of spiked biological samples. Six validation samples with different concentrations were analysed. Recoveries were calculated in percentage of each given concentration. The intra-day accuracy and precision were determined by averaging the recoveries found on any given day, without considering samples whose concentrations were below the limit of quantification (see the following section). Typical results of intra-day validation are listed in Table 1 for levoprotiline in blood. The mean recov-
Table 1. Intra-day validation for IevoprotiC i w in blood Five spiked samples in the concentration range of 6-40 pmol per sample were adysed at each of 11 analysis days and the recoveries (in percentage of tbe given concentrations) were averaged on each day RWOV6IV
Dav
Mean
cv
1 2 3 4 5 6 7 8 9 10 11
98.5 101.3 97.2 107.1 100.2 102.7 100.0 98.9 101.2 101.5 102.7
4.4 3.2 5.3 1.7 2.0 1.3 3.2 3.9 3.3 5.1 1.6
CV: coefficient of variation (%).
eries ranged between 97.2 and 107.1%, the coefficients of variation (CV) between 1.3 and 5.3%. The inter-day accuracy and precision were determined by averaging the recoveries found on different days for the samples in a given concentration range. Results of inter-day validation are listed in Table 2 for levoprotiline in blood, plasma and urine and in Table 3 for desmethyl-levoprotiline in urine. With the exception of the lowest concentration each, the mean recoveries were in the range 96.3-103.2% and demonstrated the good accuracy of the method. In the low validation range of levoprotiline in blood, a good precision was achieved for concentrations above 0.7 pmol per sample (CV: 5.5-9.3%). In the normal validation ranges of levoprotiline and desmethyl-levoprotiline, a highprecision CV (1.8-6.1%) was obtained for all concentrations except the lowest ones.
Limits of detection and quantification The lowest concentration of levoprotiline which resulted in a signal-to-noise ratio > 3 was 0.1 pmol per sample (1 ml of biological fluid used). In actual assays, the limit of detection (LOD) was dependent on the range of concentrations covered by the calibration and validation samples. The lowest LOD was reached only in blood assays using the concentration range of 0.04 to 2.1 pmol per sample. For the other concentration ranges, we used the inter-day coefficient of variation (CV) to estimate retrospectively LOD, since the lowest concentrations of the spiked samples were already too high to estimate a signal-to-noise ratio (see Figs 3 and 4). The LOD was estimated as that concentration at which the CV was about 100%. LOD of levoprotiline in blood (normal concentration range), plasma and urine was 0.2, 0.3 and 2 pmol per sample, respectively. For desmethyl-levoprotiline in urine, LOD was 1 pmol per sample.
715
LEVOPROTILINE AND A METABOLITE BY GC/MS
Table 2 Interday validation for levoprotiline in blood, plasma and urine. Six spiked samples with different concentrations were analysed in each assay and tbe recoveries (in percentage of the given concentrations) were averaged in the indicated concentration ranges Blood, low concentration range Range of given conc. (pmol per sample) Mean reovery
cv (Yo) N
0.04
0.36
to
to
0.05 47.6 142.0 6
0.42 96.4 19.5 6
0.72 to 0.85 103.1 7.8 6
1.07 to 1.26 99.2 5.5 6
1.43 to 1.68 102.0 9.3 6
1.78 to 2.10 102.? 6.6 6
25.2
Blood, normal concentration range Range of given conc. (pmol per sample) Mean recovery
cv (%) N
0.6 to 0.9 146.5 21.8 11
6.2 to 8.1 103.2 3.8 11
12.4 to 16.2 100.1 4.5 11
18.9 to 24.4 99.5 3.7 11
32.6 101.7 3.2 11
31.3 to 40.7 100.8 4.2 11
12.3 to 12.4 100.9 1.8 3
18.1 to 19.1 100.3 4.9 3
23.6 to 24.3 98.6 3.1 3
29.7 to 30.1 98.2 6.1 2
Urine 185 368 to to 203 404 99.8 100.0 1.9 2.7 7 7
521 to 574 100.3 2.1 7
678 to 768 99.8 2.1 7
852 to 962 99.8 2.1 7
to
Plasma Range of given conc. (pmol per sample) Mean recovery cv (%) N
0.6 96.7 41.8 3
Range of given conc. (pmol per sample) Mean recovery cv (%) N
17.5 to 19.2 108.4 11.1 7
0.6 to
6.1 to 6.2 96.3 6.1 3
Table 3. Interday validation for desmethyl-levoprotiline in urine. Six spiked samples with different concentratiow were analysed in each assay and the recoveries (in percentage of the given concentrations) were averaged in tbe indicated concentration ranges Range of given conc. (pmol per sample) Mean recovery cv (%)
N
87.1 111.3 2.2 24.0 42.3 64.1 to to to to to to 2.8 27.4 52.7 76.2 101.0 129.8 115.7 98.3 98.2 97.9 98.9 97.7 38.9 3.1 2.9 3.1 2.7 2.5 7 7 7 7 7 7
By analogy, the limit of quantification (LOQ) was defined as that concentration at which the CV was about 10%. For levoprotiline in blood, LOQ was 0.7 pmol per sample in the low concentration range and 2 pmol per sample in the normal concentration range. In plasma and urine, LOQ was 3 and 20 pmol per sample, respectively. For desmethyl-levoprotiline in urine, LOQ was 10 pmol per sample. Using 1 ml of the biological fluid for analysis the above values in pmol per sample can be equated with pmol ml-' or nmol I-'. If larger volumes of the biological fluids (2 or 3 ml) were used for analysis, then the actual limits in nmol I-' were correspondingly lower, i.e. the sensitivity was increased.
Stability of biological samples We demonstrated that racemic oxaprotiline was stable in blood, plasma and urine samples for at least 3 months when deep-frozen at - 18"C.The stability was tested both in spiked and in clinical samples. For example, 13 clinical plasma samples were analysed three times. The second analysis was done 9 days and the third analysis 87 days after the first analysis. Compared with the first analysis, the mean recovery for oxaprotiline concentrations above 4 nmol 1-' was 99.0% in analysis 2 and 100.6% in analysis 3. Conclusively, levoprotiline, the ( -)-R-enantiomer of oxaprotiline, has the same stability characteristics as the racemate. It is worthwhile to mention that a possible racemization during storage would not affect the validity of the data since the assay measures both enantiomers. Application The described method was applied to the determination of unchanged levoprotiline in blood, plasma and urine and of desmethyl-levoprotiline in urine of healthy volunteers dosed orally with 75 mg of levoprotiline.HC1. For illustration, the blood concentration-time curve of levoprotiline is shown in Fig. 5 and the cumulative urinary excretion of unchanged and desmethyllevoprotiline is depicted in Fig. 6. The drug was rapidly
R. ACKERMANN, G. KAISER, F. SCHUELLER A N D W. DIETERLE
(u
0
0.6
Levoprotiline Desmethyl-levoprotiline
0.5
0
u
LC
-
0.4
a7
0 ._ c y
0.3
0.2
0 X
I d
0.1
0.0 0
12
24
36
48
60
72
84
0
96
12
T i m e a f t e r dosing [ h ]
24
36
48
60
72
Time a f t e r dosing [h]
Figure 5. Blood concentrations of unchanged levoprotiline in healthy volunteers after a single oral dose of 75 mg of levoprotiline.HCI; mean f SD, N = 12.
Figure 6. Cumulative urinary excretion of levoprotiline and desmethyl-levoprotiline in healthy volunteers after a single oral dose of 75 mg of levoprotiline.HC1; mean f SD, N = 6.
absorbed (t,,, = 4 h) and eliminated from blood with a terminal half-life of about 20 h. Less than 2% of the dose was excreted in urine as unchanged drug and its desmethyl metabolite, confirming previous findings that levoprotiline and desmethyl-levoprotiline are mainly excreted in the form of their gl~curonides.~
blood, plasma and urine and the simultaneous determination of unchanged and desmethyl-levoprotiline in urine. The method is suitable for use in pharmacokinetic and bioavailability studies of the antidepressant levoprotiline.HC1.
_
_
~
CONCLUSION Acknowledgements The GC/MS method described here permits the specific and highly sensitive determination of levoprotiline in
The authors are indebted to Dr W. Kiing and his collaborators for the synthesis of the deuterated standards.
REFERENCES 1. P. C. Waldmeier, P. A. Baumann, M. Wilhelm, R. Bernasconi and L. Maitre, Eur. J. Pharmacol. 66,387(1977). 2. P. C. Waldmeier, P. A. Baumann, K. Hauser, L. Maitre and A. Storni, Biochem. Pharmacol. 31,2169 (1982). 3. S. A. Checkley, C. Thompson, S. Burton, C. Franey and J. Arendt, Psychopharmacology 87,116(1985). 4. M. Wolfersdorf, G. Wendt, U. Binz, B. Steiner and G. Hole, Pharmacopsychiatry 21,203(1988). 5. G. Wendt and U. Binz, Munch. Med. Wschr. 132 (Suppl. 1). 544 (1990). 6. K. NBhunek, J. Svestka, E. Ceskovh, R. RysBnek, V. Obrovska, J. Misurec and G. Wendt, Munch. Med. Weschr. 132 (Suppl. l ) , S50 (1990). 7. W. Dieterle, J. W. Faigle, P. Imhof, M. Sulc and J. Wagner, Xenobiotica 14,303 (1984). 8. W. Dieterle, J. W. Faigle, H.-P. Kriemler and T. Winkler, Xenobiotica 14,31 1 (1984). 9.W. Dieterle, J. W. Faigle, W. Kung and W. Theobald, Biopharm. Drug. Dispos. 5,377(1984).
10. W. Dieterle, R. Ackermann, G. Kaiser and W. Theobald, Munch. Med. Wschr. 132 (Suppl. l),S39 (1990). 11. H. J. Kuss and E. Feistenauer, J. Chromafogr. 204, 349 (1981 ). 12. U. Breyer-Pfaff, R. Wiatr and K. Nill, J. Chromatogr. 309, 107 (1984). 13. R. Ackermann, G. Kaiser, W. Dieterle and J. P. Dubois, in Proceedings of the Third European Congress of Biopharmaceutics and Pharmacokinetics, ed. by J. M. Aiche and J. Hirtz, vol. I, p. 278. lmprimerie de I'Universit6, Clermont-Ferrand, France
(1987). 14. M. Jedrychowski, E. Hoffmann and P. R. Bieck, J. Pharm. Biomed. Anal. 7, 1 897 ( 1 989). 15. C. Horne, H. Spahn and E. Mutschler, Arzneim.-Forsch./Drug Res. 37, 1 1 79 (1987). 16. Geigy Scientic Tables, ed. C. Lentner, vol. 1 (1981 ) and vol. 3 (1984).Ciba-Geigy, Basle.
