BIOMEDICAL CHROMATOGRAPHY, VOL. 6 , 205-299 (1992)
Determination of MDL 201,012 at Femtomole/Millilitre Levels in Human Plasma by Liquid Chromatography with Electrochemical Detection? Donald L. Reynolds,* Larry S. Eichmeier a n d Dennis H. Giesing Clinical Pharmacology Department, Marion Merrcll Dow, Inc.. Kansas City, Missouri 64134, USA
A sensitive and selective liquid chromatographic method to quantitate MDL 201,012 in human plasma was developed and validated. MDL 201,012 (I), diethyl-MDL 201,012 (internal standard, 11) and desmethyldiolMDL 201,012 (masking agent, Ill) were isolated from basified plasma (2 mL) by solid phase extraction using Bond-Elut@C-18 cartridges. Endogenous components were selectively removed prior to eluting the analytes from the sorbent. Components were separated using on-line LC column switching with a cyanopropyl precolumn and a phenyl analytical column, The analytical column effluent was monitored electrochemically at a glassy carbon electrode at a potential of 1025 mV vs. Ag/AgCl. Peak-height ratios were proportional to the amount of MDL 201,012 added to plasma over the range 125-7500 pg/mL MDL 201,012. Absolute recovery of MDL 201,012 from human plasma was >94% across the calibration range. The minimum quantitation limit was 125 pg/mL. Assay precision (%RSD) ranged from 5.2 to 13% based on the analysis of quality control standards containing 125, 250,500, 1000,2500,5000 and 7500 pg/mL MDL 201,012. Corresponding assay accuracy (YOrelative error) was +8.5%. The method has been successfully used to quantitate MDL 201,012 in samples from acute dose tolerance studies in human volunteers.
+
INTRODUCTION
MDL 201,012 (I) [l-cyclobuty1-1-hydroxy-l-phenyl-7(N,N-dimethylamino)hept-5-yn-2-onehydrochloride] is a new compound under development for treatment of symptoms associated with uninhibited and reflux neurogenic bladder. Physicochemical properties of this compound include an aqueous solubility >lo00 mg/ mL, a pK, value of 8.0 and an apparent octanol/water partition coefficient (log P) of 0.24. A sensitive and selective assay was required to quantitate MDL, 201,012 in human plasma to support clinical safety studies. This report describes the development and validation of a liquid chromatographic method to quantitate MDL 201,012 in human plasma at concentrations of 1257500 pg/mL MDL 201,012. The method involves selective solid phase extraction of drug, internal standard (diethyl-MDL 201,012, IS, 11) and masking agent (desmethyldiol-MDL 201,012, 111) from plasma, liquid chromatographic separation of the analytes using online column switching, and quantitation by amperometric detection.
EXPERIMENTAL Reagents. MDL 201,012 (I), IS (11) [ l-cyclobutyl-l-hydroxyl-phenyl-7-(N,N-diethylamino)-hept-5-yn-2-one hydrochloride] and masking agent (111) [l-cyclobutyl-l,2-dihydroxy-l-
’Author to whom correspondence \hould
be dddresed part, at the 3rJ Inteinational Symposium on Pharmaceutical and BiomediLal Analysis. B ( x t o n , MA. USA, Mdy 1991
t This work was presented,
in
02h9-387YIY2/0h0295-0S $07.50 01992 by John Wiley & Sons. Ltd
0
R3
0
R4
phenyl-7-(N-methylamino)-hept-5-ync],were obtained from Marion Merrell Dow, Inc. (Kansas City, MO, USA). Acetonitrilc (J. T. Baker, Phillipsburg, NJ, USA) and methano1 (Burdick & Jackson, Muskegon, MI, USA) were HPLC grade. Deionized water was passed through was passed through a NANOpure I1 system (BarnsteadiThermolyne, Debuque, I A , USA) prior to use. Ethyl acetate (Baker) was Resi-Analyzed grade. Pooled, heparinized human plasma was obtained from Biological Specialities Corporation (Lansdale, PA, USA). Bond-Elut C-18 cartridges (3 m L capacity; Analytichem International, Harbor City, CA, USA) were used as received. All other reagents were of analytical reagent grade or better. Apparatus. The chromatographic system (Fig. 1) consisted of two pumps (M-511) and M-6000; Waters Assoc., Milford, MA, USA), an autosampler (WISP 712; Waters) and an Received 3 December 1991 Accepted I2 June 1992
DONALD L. REYNOLDS. LARRY S . EICHMEIER AND DENNIS H.
296
GIESING
concentrations of SO ngimL and 3.33 pg/mL, respectively. Solutions were stable for at least 4 months at 4 "C. I
I I
1
PUMP1
]
II I -1
I
I
ECDETECTOR
1
Figure 1. Schematic diagram of t h e chromatographic system. Key: Inject position = (-1; Load position = (- - -).
amperometric detector (EC 400; EG&G Princeton Applied Research, Princeton. NJ, USA). On-line column switching was performed using an electrically actuated valve (EQC6W; Valco Instrument Co., Houston, T X , USA), controlled by a data acquisition system (Waters 840). Separations were obtained using a cyanopropyl precolumn (5 pm, Sphcri-5, 2.1 x 3 0 m m ; Brownlee Labs., Santa Clara, C A , USA) and a phenyl analytical column ( 5 pm, Spheri-5, 2.1 x 100 mm; Brownlee) maintained at ambient temperature. Peak-height ratios were determined using a Waters 840 data acquisition system. Solid phase extractions were performed in a batch mode using a 144-place vacuum manifold (12 x 12 matrix; Territorial Tool, Dexter, MI, USA). Solvent evaporation was achieved at 40 "C using a Speed-Vac Concentrator (Savant Instruments, Inc., Farmingdale, NY, USA). Chromatographic conditions. The switching valve was initialized so that the cffluent from the autosampler was directed to waste (Fig. 1; inject position). A t the time of injection, the valve was rotated to the Load position where analytes were isolated on the precolumn. A t 0.6 min after injection, the valve was returned to the Inject position and analytes were transferred to the analytical column for final separation. Isocratic solvent mixtures consisting of acetonitrilc and aqueous buffer containing 15 mM sodium acetate, 15 mM monobasic potassium phosphate and 75 mM tetramethylammonium hydroxide (TMAOH) were used for both Pumps 1 and 2. The mobile phase for Pump 1 containcd 20X acetonitrile in buffer (pH 6.2) and the solvent used for Pump 2 was 45% acetonitrile in buffer (pH 5.9). Flow rates werc maintained at 0.5 and 0.6 mL/min, respectively, throughout the analysis. Analytes werc detected amperometrically at a dual glassy carbon electrode. The upstream electrode was maintained at a potential of + 850 mV and the working electrode was held at + 1025 mV, each vs. a Ag/AgCI reference. The detector was operated at a range of 20 nA full scale. Electrode surfaces were polished daily using an alumina slurry followed by ultrasonication in water for =5 min. Electrodes were electrochemically preconditioned prior to use by applying three full scans ( + 1.5 V to - 1.5 V, 20 mV/s) followed by application of a + 1.5 V potential for 30 min. Standard solutions. Stock solutions. Stock solutions of drug (56.0 mg MDL. 201,012/50 mL) and IS (5.0 mg diethyl-MDL 201,012150 mL) were prepared in water. The masking agent stock solution (5.0 mg desmethyldiol-MDL 201,012/50 mL) was prepared in acetonitri1e:water (1: 1). Spiking standard solutions containing 1.25, 2 . 5 , 5 .O, 7.0,25.O. 50.0 and 75.0 n d mL M D L 201,012 were prepared by diluting the M D L 201,012 stock solution with water. Internal standard and masking agent stock solutions were diluted with water to final
Mobile phases. Mobile phase buffer was prcpared by dissolving sodium acetate (2.0 g), monobasic potassium phosphate (2.0 g) and T M A O H (13.6 g) in water and diluting to a final volume of 1000 mL. Mobile phase buffer solutions were adjusted to the desired pH with concentrated sulphuric acid prior to addition of acetonitrile. The precolumn mobile phase was prepared by mixing acetonitrile (100 mL) and mobile phase buffer (pH 6.2,400 mL). A combination of acetonitrile (450 mL) and mobile phase buffer (pFi 5.9,550 mL) was used for the analytical column mobile phase. Mobile phases were prepared daily and vacuum filtered prior to use. Solid-phase wash buffus (pH5.9 and 12). Wash buffer (pH 12) was prepared by dissolving sodium acetate (0.52 g), monobasic potassium phosphate (0.52 g) and T M A O H (6.8 g) in water and diluting to a final volume of 500 mL. Wash buffer (pH 5.9) was prepared from wash buffer (pH 12) solution by adjusting to pH 5.9 with concentrated sulphuric acid. Quality control standards. Quality control standards containing 125, 250, 5M. 1000, 2500, 5000, and 7500pdmL MDL 201,012 were preparcd by diluting 0.0416, 0.0883, 0.167, 0.333, 0.833, 1.667 and 2 . 5 m L aliquots, respectively, of a 300 ng/mL M D L 201,012 stock solution t o a final volume of 100 mL with pooled blank human plasma. Standards were subdivided into 2.0 mL aliquots and were stored in glass tubes at -20°C. Sample preparation. To a 2.0 mL human plasma sample in a 13 X 100 m m disposable tube was added 0.20 mL IS solution, 0.25 tnL masking agent solution, and 0.20 mL water (or M D L 201,012 spiking solution for calibration standards). Samples containing greater than 7.5 ng/mL M D L 201,012 were appropriately diluted with pooled blank human plasma prior to analysis. Tubes were vortexcd and 1 N NaOH (0.2 mL) was added to each tubc. Tube contents were well mixed and samples wcrc randomized prior to solid phase extraction. Bond-Elut C-18 cartridges were preconditioned by rinsing with methanol (3 mL) followed by water (3 mL). Plasma mixtures were introduced and pulled through the cartridge with a gentle vacuum (approximately 5-10 in Hg). Cartridges were rinsed sequentially with water (3 mL), p H 5.9 wash buffer (1 mL), water (2mL), 0.5% acetic acid (1 mL), acetonitrile (1 mL), water (1 mL), 0.2 N N a O H (1 mL), 25% acetonitrile in 0.2 N NaOH (1 mL), 0 . 2 N N a O H (1 mL), p H 1 2 wash buffer (1 mL) and 0 . 2 N N a O H (2mL). Cartridges were dried one row at a time for 10 min each under full vacuum (=15inHg). Cartridges in other rows were sealed with a four-fold thickncss of Parafilm during the drying step. After all the cartridges were dried, the analytes were eluted with ethyl acetate (3 x 1 mL) under gravity flow conditions. A n aliquot of 10% acetonitrile in precolumn mobile phase buffer (0.2mL) was added to each tubc, the tube contents were vortexed and samples were evaporated to dryness at 40°C under reduced vacuum. The residue was reconstituted with 10% acetonitrile in water (0.2 mL) and an aliquot (165 pL) was analysed by liquid chromatography. Extraction efficiency from plasma. Absolute recovery of MDL 201,012 from human plasma was determined at concentrations of 500,2500 and 5000 p d m L M D L 201,012 by assaying five samples at each concentration. IS recovery was detcrmined at a concentration of 5.0 n d m L ( n = 15). Peak-heights of extracted standards were compared to peakheights of identical standards added to plasma blank extracts.
I.C/ED OF MDL 201,012
Assay validation. Thc assay was validated over the concentration range 125-7500 pg/mL MDL 201.012 by analysing seven calibration standards ( n = 2) and seven quality control standards (n = 5) on five separate days. The best-fit straight line was determined daily by least-squares linear regression analysis using a weighting factor of reciprocal concentration. Concentrations of MDL 201,012 in quality control standards were determined using the regression equation.
Plasma storage stability. The stability of MDL 201,012 plasma samples stored in glass or polypropylene tubes was evaluated at temperatures of - 20 "C and - 70 "C. Stability was determined at concentrations of 500 and 5000pghL MDL 201,012.
RESULTS AND DISCUSSION
Analysis of MDL 201,012 in human plasma consists of isolation of drug and TS from plasma by solid phase extraction, separation by reversed phase liquid chromatography with on-line column switching and quantitation using amperometric detection. The validity of this procedure was established by investigation of extraction efficiency, selectivity, linearity, precision and accuracy of calibration curves, and assay precision and accuracy.
297
provided increased recovery but also increased the co-elution of polar endogenous components. The best compromise between maximal anal yte recovery and minimal potential interference was obtained using ethyl acetate. Preliminary experiments, however, suggested an approximate recovery of =60% for MDL 201,012 from plasma using ethyl acetate. The low recovery resulted from inadequate elution of analytes from the C-18 sorbent, most likely due to ion-dipole interactions between analytes and ionized silanols on the C-18 surface. These interactions were eliminated by masking the silanol groups with a basic wash solution containing tetramethylammonium ions. Absolute recoveries of >94% were found for MDL 201,012 and IS under these conditions. Prior to elution, cartridges must be completely dried of aqueous solvents. Analyte recovery decreased dramatically with insufficient cartridge drying. Generally, drying for 10min under full vacuum (-13-15 in Hg) was necessary to provide sufficient drying and reproducible recovery. However, air flow through the 3cc Bond-Elut cartridges was so unrestricted that sufficient vacuum could only be achieved when 12 cartridges or less were dried simultaneously. Drying cartridges one row at a time while covering the remaining rows with Parafilm maintained the desired vacuum and provided reproducible analyte recovery. Liquid chromatography
Solid phase isolation
The low detection limits required for MDL 201,012 analyses coupled with the use of amperometric detection at high working potentials necessitated extensive sample cleanup prior to MDL 201,012 detection. Solid phase extraction techniques have been widely used to provide rapid selective cleanup from complex matrices and were consequently evaluated for isolating MDL 201,012 and IS from plasma. Sample cleanup was achieved using a previously reported strategy (Johnson et uf., 1988). Influences of silanophilic and pH effects on the tertiary amine group (pK, = 8.0) were used to optimize separation selectivity. Analytes were adsorbed from basified plasma onto C-18 cartridges in the non-ionized form. Proteins and hydrophilic components were removed with water and aqueous buffer washes. Interactions between the protonated analytes and residual silanols under acidic conditions were used to provide initial sample cleanup. These interactions were strong enough to allow the use of pure acetonitrile (1 mL) as a wash solvent. Acids and strongly retained non-ionic endogenous components were removed with acetonitrile; however, cleanup was not sufficient to provide MDL 201,012 quantitation at the desired levels. Additional cleanup was provided through use of an aqueous acetonitrile wash solvent (pH 12) which removed hydrophilic bases prior to analyte elution. To facilitate subsequent solvent evaporation and concentration of the analytes, organic solvents were screened as elution solvents. Heptane, pentane, chloroform, methylene chloride, ethyl acetate, acetonitrile and methanol were evaluated. No recovery of MDL 201,012 was observed with the non-polar solvents heptane and pentane, whereas the more polar solvents
Sensitive and selective dctection was required to quantitate MDL 201,012 at concentrations down to 125 pg/ mL. UV detection of MDL 201,012 was not possible due to low wavelength requirements (205 nm) and insufficient MDL 201,012 molar absorptivity. In addition, facile derivatization did not appear feasible. Amperometric detection, based on oxidation of the tertiary amine group at a glassy carbon electrode, was then investigated. Maximizing MDL 201,012 detection while minimizing background current was achieved by optimizing mobile phase pH, buffer type and concentration, and applied electrode potentials. Increasing mobile phase pH, phosphate concentration and working electrode potential resulted in increased MDL 201,012 response; however, the background current also increased. Mobile phases at pHB6.5 and working electrode potentials > 1 .0S V could not be used due to excessive background current. Background current was decreased by operating a second, upstream electrode at a potential of + 850 mV vs. AglAgCl, where some mobile phase components were selectively oxidized. Sensitivity was also increased by adding 75 mM tetramethylammonium ions to the mobile phase, which provided a two-fold increase in MDL 201,012 response without further increasing the background current. The best compromise between MDL 201,012 response and background current was obtained using acetonitrile: buffer mobile phases containing 15 mM sodium acetate, 15 mM monobasic potassium phosphate and 75 mM TMAOH (adjusted to p H 5.9) with upstream and working electrodes maintained at +850 mV and 1025 mV vs. Ag/AgCI, respectively. Although selective sample cleanup was achieved through solid phase extraction, extensive baseline dis-
+
298
DONALD L. REYNOLDS, LARRY S. EICHMEIER AND DENNIS H. GIESING
turbances were observed when liquid chromatography was performed using a single column. These disturbances primarily resulted from mobile phase composition/equilibria changes following injection due to high detector sensitivity and working potentials. These effects were minimized through use of on-line column switching techniques which directed the injection solvent and first eluent fraction to waste. Separation selectivity was achieved using a cyano precolumn and a phenyl analytical column to maximize the polarity difference between the columns. Millibore columns (2.1 mm i d . ) were used to enhance sensitivity due to the higher analyte concentrations in the effluent as compared to traditional (4.6 mm i.d.) columns (Guiochon and Colin, 1984). Analytes were adsorbed on the precolumn using a weak acetonitrile: buffer mobile phase and baseline disturbances were minimized by directing only the volume of mobile phase contained in the precolumn to the analytical column (Fig. 1). No band broadening was observed upon switching due to analyte preconcentration on the analytical column. Subsequent cleanup a n d o r reequilibration procedures were not required for the precolumn using this procedure. Final separation was achieved using a phenyl column with a buffered acetonitrile eluent containing 75 mM tetramethylammonium ion adjusted to p H 5.9. A combination of phosphate and acetate buffers was used to increase the buffer capacity due to the intermediate mobile phase p H used. Excellent selectivity with acceptable detector response and background current were obtained at this pH; however, significant silanol effects were observed for MDL 201,012 and IS, even with 75 mM tetramethylammonium ion. Peak tailing and non-linear detector response were evident and became more severe with decreasing MDL 201,012 concentrations. These effects were eliminated by using desmethyl-MDL 201,012 as a masking agent. The masking agent was added to plasma prior to extraction rather than adding additional competing cations to the mobile phase to avoid additional detection complications. When added at concentrations of =400 ng/mL, the masking agent effectively competed with MDL 201,012 for silanol sites on the stationary phase surface, which decreased peak tailing and provided linearity down to concentrations of 125 pg/mL MDL 201,012. Extraction efficiency from plasma Absolute recovery of MDL 201,012 (+SD) (n= 5 ) from human plasma was determined at concentrations of 500, 2500 and 7500pg/mL MDL 201,012. Recoveries were 94.l(f17)%, 100(*2.9)% and 95.3(f8.0)%, respectively. Recovery of IS (n = 15) was 96.3( *6.8)% at a concentration of 5 ng/mL.
I
I
I
I
I
2
0
I
I
6
4
Time (minutes)
Figure 2. Chromatograms of plasma samples. (a) Blank pooled human plasma; (b) quality control standard containing 125 pg/ mL MDL 201,012; (c) plasma sample from a human subject 3 h after receiving a 4 mg oral dose of MDL 201,012. I=MDL 201,012; II= internal standard.
ponents and masking agent (Fig. 2a,b). In patient samples, neither plasma components nor metabolites eluted near MDL 201,012 or IS (Fig. 2c), illustrating excellent specificity. System specificity for MDL 201,012 in the presence of potential metabolites (MDL 201,012 N-oxide, MDL 201,012 diol, desmethyl-MDL 201,012, and desmethyldiol-MDL 201,012) was also determined. All potential metabolites were chromatographically resolved from MDL 201,012 and IS (data not shown). In addition, none of the possible metabolites interfered with solid phase isolation of MDL 201,012 and IS or quantitation when present in plasma at concentrations up to lOOOng/ml. The assay is therefore specific for MDL 201,012. Greater than 250 plasma samples from acute dose tolerance studies have been assayed using this method. No apparent interferences were observed throughout the analyses. Figure 3 shows the MDL 201,012 plasma concentration-time profile observed in a subject fol25
1
Specificity Retention times for MDL 201,012 and IS were 3.8 and 5.0 min, respectively. Chromatograms of blank human plasma, a 125pg/mL MDL 201,012 quality control standard and a plasma sample from a human subject receiving MDL 201,012 are shown in Fig. 2. Drug and IS are suitably resolved from endogenous plasma com-
0
3
6
9
12
Time Postdose (hr)
Figure 3. Plasma concentration-time profile of MDL 201,012 in a human subject after receiving a 4 mg oral dose of MDL 201.01 2.
LUED OF MDL 201,012
299
Table 1. Precision and accuracy of MDL 201,012 calibration standards in human plasma assayed during a fiveday period
Table 2. Precision and accuracy of MDL 201,012 quality control standards in human plasma assayed during a five-day period
Conc. added (pglmL)
Conc. added (pg/mL)
N
Conc. found (pglmL1
125 9 140 250 9 245 500 10 463 700 9 688 2500 10 2480 5000 8 4970 7500 9 7590 aRelative standard deviation.
RSD'
I%)
16 5.7 11 6.7 3.9
3.5 2.5
Relative error (%)
12 - 1.9 - 7.4 - 1.7 - 0.64 - 0.58 1.2
lowing administration of a 4 mg dose of MDL 201,012 as an oral solution. The assay is sufficiently sensitive to monitor MDL 201,012 plasma concentrations for at least five half-lives in subjects receiving this dose.
Linearity, precision and accuracy of the calibration curves The assay was validated by assaying seven calibration standards ( n = 2) and five replicates of seven quality control standards on five separate days. Peak-height ratios were proportional to the amount of MDL 201,012 added tot plasma over the concentration range 125-7500 pg/mL. The best-fit straight line was determined daily by least-squares linear regression analysis. A weighting factor of reciprocal concentration was used to ensure homoscedasticity over the concentration range (Johnson et al. 1988). Mean (+SD) regression results obtained during assay validation are described by the following equation: peak-height ratio = 0.331 (rt0.0024) x [MDL 201,012 concentration] - 0.0003 (k0.006) (correlation coefficient = 0.9985). The reproducibility of calibration curves was determined as the variation of calibration standards from the regression line. Calibration curve precision (YO relative standard deviation) ranged from 2.5 to 16% over five days, with corresponding accuracies ( O h relative errors) ranging from - 7.4 to 12% (Table 1).
N
Conc. found (pg/mLl
125 21 128 250 21 266 500 22 545 1000 22 1040 2500 24 2470 5000 25 5180 7500 25 7700 aRelative standard deviation.
(%)
Relative error (%)
13 10 7.0 5.4 8.5 5.2 7.2
2.6 6.5 9.0 3.6 - 1.1 3.5 2.7
RSDa
Assay precision and accuracy Assay precision and accuracy was determined by analysing five replicates of seven quality control standards on five separate days with the calibration standards. Assay precision was f13%, based on relative standard deviations ranging from 5.2 to 13% for quality control standards containing 125 to 7500 pg/mL MDL 201,012 (Table 2). Corresponding assay accuracy ranged from -1.1 to 8.9%. The minimum quantitation limit was 125 pg/mL MDL 201,012 using a 2 mL sample. Plasma storage stability MDL 201,012 plasma stability under storage conditions was evaluated by assaying five replicate samples at each concentration stored in glass and polypropylene at - 20 and - 70 "C. MDL 201,012 concentrations in samples assayed after nine months storage were within +lo% of theoretical values for all storage conditions evaluated. These data suggest that MDL 201,012 patient plasma samples can be stored in polypropylene tubes at - 20 "C for up to nine months without significant drug loss. In summary, a sensitive and selective liquid chromatographic method has been developed and validated to quantify MDL 201,012 in human plasma over the concentration range 125-7500 ng/mL MDL 201,012. The method has been successfully used to quantitate MDL 201,012 in samples from clinical acute dose tolerance studies.
REFERENCES Guiochon, G. and Colin, H. (1984). In Microcolumn High Performance Liquid Chromatography, ed. by Kucera, P., Chap. 1. Journal of Chromatography Library, Vol. 28, Elsevier, New York.
Johnson, E. L., Reynolds, D. L., Wright, D. S . and Pachla, L. A. (1988). J. Chrorn. Sci. 26,372.
Determination of MDL 201,012 at Femtomole/Millilitre Levels in Human Plasma by Liquid Chromatography with Electrochemical Detection? Donald L. Reynolds,* Larry S. Eichmeier a n d Dennis H. Giesing Clinical Pharmacology Department, Marion Merrcll Dow, Inc.. Kansas City, Missouri 64134, USA
A sensitive and selective liquid chromatographic method to quantitate MDL 201,012 in human plasma was developed and validated. MDL 201,012 (I), diethyl-MDL 201,012 (internal standard, 11) and desmethyldiolMDL 201,012 (masking agent, Ill) were isolated from basified plasma (2 mL) by solid phase extraction using Bond-Elut@C-18 cartridges. Endogenous components were selectively removed prior to eluting the analytes from the sorbent. Components were separated using on-line LC column switching with a cyanopropyl precolumn and a phenyl analytical column, The analytical column effluent was monitored electrochemically at a glassy carbon electrode at a potential of 1025 mV vs. Ag/AgCl. Peak-height ratios were proportional to the amount of MDL 201,012 added to plasma over the range 125-7500 pg/mL MDL 201,012. Absolute recovery of MDL 201,012 from human plasma was >94% across the calibration range. The minimum quantitation limit was 125 pg/mL. Assay precision (%RSD) ranged from 5.2 to 13% based on the analysis of quality control standards containing 125, 250,500, 1000,2500,5000 and 7500 pg/mL MDL 201,012. Corresponding assay accuracy (YOrelative error) was +8.5%. The method has been successfully used to quantitate MDL 201,012 in samples from acute dose tolerance studies in human volunteers.
+
INTRODUCTION
MDL 201,012 (I) [l-cyclobuty1-1-hydroxy-l-phenyl-7(N,N-dimethylamino)hept-5-yn-2-onehydrochloride] is a new compound under development for treatment of symptoms associated with uninhibited and reflux neurogenic bladder. Physicochemical properties of this compound include an aqueous solubility >lo00 mg/ mL, a pK, value of 8.0 and an apparent octanol/water partition coefficient (log P) of 0.24. A sensitive and selective assay was required to quantitate MDL, 201,012 in human plasma to support clinical safety studies. This report describes the development and validation of a liquid chromatographic method to quantitate MDL 201,012 in human plasma at concentrations of 1257500 pg/mL MDL 201,012. The method involves selective solid phase extraction of drug, internal standard (diethyl-MDL 201,012, IS, 11) and masking agent (desmethyldiol-MDL 201,012, 111) from plasma, liquid chromatographic separation of the analytes using online column switching, and quantitation by amperometric detection.
EXPERIMENTAL Reagents. MDL 201,012 (I), IS (11) [ l-cyclobutyl-l-hydroxyl-phenyl-7-(N,N-diethylamino)-hept-5-yn-2-one hydrochloride] and masking agent (111) [l-cyclobutyl-l,2-dihydroxy-l-
’Author to whom correspondence \hould
be dddresed part, at the 3rJ Inteinational Symposium on Pharmaceutical and BiomediLal Analysis. B ( x t o n , MA. USA, Mdy 1991
t This work was presented,
in
02h9-387YIY2/0h0295-0S $07.50 01992 by John Wiley & Sons. Ltd
0
R3
0
R4
phenyl-7-(N-methylamino)-hept-5-ync],were obtained from Marion Merrell Dow, Inc. (Kansas City, MO, USA). Acetonitrilc (J. T. Baker, Phillipsburg, NJ, USA) and methano1 (Burdick & Jackson, Muskegon, MI, USA) were HPLC grade. Deionized water was passed through was passed through a NANOpure I1 system (BarnsteadiThermolyne, Debuque, I A , USA) prior to use. Ethyl acetate (Baker) was Resi-Analyzed grade. Pooled, heparinized human plasma was obtained from Biological Specialities Corporation (Lansdale, PA, USA). Bond-Elut C-18 cartridges (3 m L capacity; Analytichem International, Harbor City, CA, USA) were used as received. All other reagents were of analytical reagent grade or better. Apparatus. The chromatographic system (Fig. 1) consisted of two pumps (M-511) and M-6000; Waters Assoc., Milford, MA, USA), an autosampler (WISP 712; Waters) and an Received 3 December 1991 Accepted I2 June 1992
DONALD L. REYNOLDS. LARRY S . EICHMEIER AND DENNIS H.
296
GIESING
concentrations of SO ngimL and 3.33 pg/mL, respectively. Solutions were stable for at least 4 months at 4 "C. I
I I
1
PUMP1
]
II I -1
I
I
ECDETECTOR
1
Figure 1. Schematic diagram of t h e chromatographic system. Key: Inject position = (-1; Load position = (- - -).
amperometric detector (EC 400; EG&G Princeton Applied Research, Princeton. NJ, USA). On-line column switching was performed using an electrically actuated valve (EQC6W; Valco Instrument Co., Houston, T X , USA), controlled by a data acquisition system (Waters 840). Separations were obtained using a cyanopropyl precolumn (5 pm, Sphcri-5, 2.1 x 3 0 m m ; Brownlee Labs., Santa Clara, C A , USA) and a phenyl analytical column ( 5 pm, Spheri-5, 2.1 x 100 mm; Brownlee) maintained at ambient temperature. Peak-height ratios were determined using a Waters 840 data acquisition system. Solid phase extractions were performed in a batch mode using a 144-place vacuum manifold (12 x 12 matrix; Territorial Tool, Dexter, MI, USA). Solvent evaporation was achieved at 40 "C using a Speed-Vac Concentrator (Savant Instruments, Inc., Farmingdale, NY, USA). Chromatographic conditions. The switching valve was initialized so that the cffluent from the autosampler was directed to waste (Fig. 1; inject position). A t the time of injection, the valve was rotated to the Load position where analytes were isolated on the precolumn. A t 0.6 min after injection, the valve was returned to the Inject position and analytes were transferred to the analytical column for final separation. Isocratic solvent mixtures consisting of acetonitrilc and aqueous buffer containing 15 mM sodium acetate, 15 mM monobasic potassium phosphate and 75 mM tetramethylammonium hydroxide (TMAOH) were used for both Pumps 1 and 2. The mobile phase for Pump 1 containcd 20X acetonitrile in buffer (pH 6.2) and the solvent used for Pump 2 was 45% acetonitrile in buffer (pH 5.9). Flow rates werc maintained at 0.5 and 0.6 mL/min, respectively, throughout the analysis. Analytes werc detected amperometrically at a dual glassy carbon electrode. The upstream electrode was maintained at a potential of + 850 mV and the working electrode was held at + 1025 mV, each vs. a Ag/AgCI reference. The detector was operated at a range of 20 nA full scale. Electrode surfaces were polished daily using an alumina slurry followed by ultrasonication in water for =5 min. Electrodes were electrochemically preconditioned prior to use by applying three full scans ( + 1.5 V to - 1.5 V, 20 mV/s) followed by application of a + 1.5 V potential for 30 min. Standard solutions. Stock solutions. Stock solutions of drug (56.0 mg MDL. 201,012/50 mL) and IS (5.0 mg diethyl-MDL 201,012150 mL) were prepared in water. The masking agent stock solution (5.0 mg desmethyldiol-MDL 201,012/50 mL) was prepared in acetonitri1e:water (1: 1). Spiking standard solutions containing 1.25, 2 . 5 , 5 .O, 7.0,25.O. 50.0 and 75.0 n d mL M D L 201,012 were prepared by diluting the M D L 201,012 stock solution with water. Internal standard and masking agent stock solutions were diluted with water to final
Mobile phases. Mobile phase buffer was prcpared by dissolving sodium acetate (2.0 g), monobasic potassium phosphate (2.0 g) and T M A O H (13.6 g) in water and diluting to a final volume of 1000 mL. Mobile phase buffer solutions were adjusted to the desired pH with concentrated sulphuric acid prior to addition of acetonitrile. The precolumn mobile phase was prepared by mixing acetonitrile (100 mL) and mobile phase buffer (pH 6.2,400 mL). A combination of acetonitrile (450 mL) and mobile phase buffer (pFi 5.9,550 mL) was used for the analytical column mobile phase. Mobile phases were prepared daily and vacuum filtered prior to use. Solid-phase wash buffus (pH5.9 and 12). Wash buffer (pH 12) was prepared by dissolving sodium acetate (0.52 g), monobasic potassium phosphate (0.52 g) and T M A O H (6.8 g) in water and diluting to a final volume of 500 mL. Wash buffer (pH 5.9) was prepared from wash buffer (pH 12) solution by adjusting to pH 5.9 with concentrated sulphuric acid. Quality control standards. Quality control standards containing 125, 250, 5M. 1000, 2500, 5000, and 7500pdmL MDL 201,012 were preparcd by diluting 0.0416, 0.0883, 0.167, 0.333, 0.833, 1.667 and 2 . 5 m L aliquots, respectively, of a 300 ng/mL M D L 201,012 stock solution t o a final volume of 100 mL with pooled blank human plasma. Standards were subdivided into 2.0 mL aliquots and were stored in glass tubes at -20°C. Sample preparation. To a 2.0 mL human plasma sample in a 13 X 100 m m disposable tube was added 0.20 mL IS solution, 0.25 tnL masking agent solution, and 0.20 mL water (or M D L 201,012 spiking solution for calibration standards). Samples containing greater than 7.5 ng/mL M D L 201,012 were appropriately diluted with pooled blank human plasma prior to analysis. Tubes were vortexcd and 1 N NaOH (0.2 mL) was added to each tubc. Tube contents were well mixed and samples wcrc randomized prior to solid phase extraction. Bond-Elut C-18 cartridges were preconditioned by rinsing with methanol (3 mL) followed by water (3 mL). Plasma mixtures were introduced and pulled through the cartridge with a gentle vacuum (approximately 5-10 in Hg). Cartridges were rinsed sequentially with water (3 mL), p H 5.9 wash buffer (1 mL), water (2mL), 0.5% acetic acid (1 mL), acetonitrile (1 mL), water (1 mL), 0.2 N N a O H (1 mL), 25% acetonitrile in 0.2 N NaOH (1 mL), 0 . 2 N N a O H (1 mL), p H 1 2 wash buffer (1 mL) and 0 . 2 N N a O H (2mL). Cartridges were dried one row at a time for 10 min each under full vacuum (=15inHg). Cartridges in other rows were sealed with a four-fold thickncss of Parafilm during the drying step. After all the cartridges were dried, the analytes were eluted with ethyl acetate (3 x 1 mL) under gravity flow conditions. A n aliquot of 10% acetonitrile in precolumn mobile phase buffer (0.2mL) was added to each tubc, the tube contents were vortexed and samples were evaporated to dryness at 40°C under reduced vacuum. The residue was reconstituted with 10% acetonitrile in water (0.2 mL) and an aliquot (165 pL) was analysed by liquid chromatography. Extraction efficiency from plasma. Absolute recovery of MDL 201,012 from human plasma was determined at concentrations of 500,2500 and 5000 p d m L M D L 201,012 by assaying five samples at each concentration. IS recovery was detcrmined at a concentration of 5.0 n d m L ( n = 15). Peak-heights of extracted standards were compared to peakheights of identical standards added to plasma blank extracts.
I.C/ED OF MDL 201,012
Assay validation. Thc assay was validated over the concentration range 125-7500 pg/mL MDL 201.012 by analysing seven calibration standards ( n = 2) and seven quality control standards (n = 5) on five separate days. The best-fit straight line was determined daily by least-squares linear regression analysis using a weighting factor of reciprocal concentration. Concentrations of MDL 201,012 in quality control standards were determined using the regression equation.
Plasma storage stability. The stability of MDL 201,012 plasma samples stored in glass or polypropylene tubes was evaluated at temperatures of - 20 "C and - 70 "C. Stability was determined at concentrations of 500 and 5000pghL MDL 201,012.
RESULTS AND DISCUSSION
Analysis of MDL 201,012 in human plasma consists of isolation of drug and TS from plasma by solid phase extraction, separation by reversed phase liquid chromatography with on-line column switching and quantitation using amperometric detection. The validity of this procedure was established by investigation of extraction efficiency, selectivity, linearity, precision and accuracy of calibration curves, and assay precision and accuracy.
297
provided increased recovery but also increased the co-elution of polar endogenous components. The best compromise between maximal anal yte recovery and minimal potential interference was obtained using ethyl acetate. Preliminary experiments, however, suggested an approximate recovery of =60% for MDL 201,012 from plasma using ethyl acetate. The low recovery resulted from inadequate elution of analytes from the C-18 sorbent, most likely due to ion-dipole interactions between analytes and ionized silanols on the C-18 surface. These interactions were eliminated by masking the silanol groups with a basic wash solution containing tetramethylammonium ions. Absolute recoveries of >94% were found for MDL 201,012 and IS under these conditions. Prior to elution, cartridges must be completely dried of aqueous solvents. Analyte recovery decreased dramatically with insufficient cartridge drying. Generally, drying for 10min under full vacuum (-13-15 in Hg) was necessary to provide sufficient drying and reproducible recovery. However, air flow through the 3cc Bond-Elut cartridges was so unrestricted that sufficient vacuum could only be achieved when 12 cartridges or less were dried simultaneously. Drying cartridges one row at a time while covering the remaining rows with Parafilm maintained the desired vacuum and provided reproducible analyte recovery. Liquid chromatography
Solid phase isolation
The low detection limits required for MDL 201,012 analyses coupled with the use of amperometric detection at high working potentials necessitated extensive sample cleanup prior to MDL 201,012 detection. Solid phase extraction techniques have been widely used to provide rapid selective cleanup from complex matrices and were consequently evaluated for isolating MDL 201,012 and IS from plasma. Sample cleanup was achieved using a previously reported strategy (Johnson et uf., 1988). Influences of silanophilic and pH effects on the tertiary amine group (pK, = 8.0) were used to optimize separation selectivity. Analytes were adsorbed from basified plasma onto C-18 cartridges in the non-ionized form. Proteins and hydrophilic components were removed with water and aqueous buffer washes. Interactions between the protonated analytes and residual silanols under acidic conditions were used to provide initial sample cleanup. These interactions were strong enough to allow the use of pure acetonitrile (1 mL) as a wash solvent. Acids and strongly retained non-ionic endogenous components were removed with acetonitrile; however, cleanup was not sufficient to provide MDL 201,012 quantitation at the desired levels. Additional cleanup was provided through use of an aqueous acetonitrile wash solvent (pH 12) which removed hydrophilic bases prior to analyte elution. To facilitate subsequent solvent evaporation and concentration of the analytes, organic solvents were screened as elution solvents. Heptane, pentane, chloroform, methylene chloride, ethyl acetate, acetonitrile and methanol were evaluated. No recovery of MDL 201,012 was observed with the non-polar solvents heptane and pentane, whereas the more polar solvents
Sensitive and selective dctection was required to quantitate MDL 201,012 at concentrations down to 125 pg/ mL. UV detection of MDL 201,012 was not possible due to low wavelength requirements (205 nm) and insufficient MDL 201,012 molar absorptivity. In addition, facile derivatization did not appear feasible. Amperometric detection, based on oxidation of the tertiary amine group at a glassy carbon electrode, was then investigated. Maximizing MDL 201,012 detection while minimizing background current was achieved by optimizing mobile phase pH, buffer type and concentration, and applied electrode potentials. Increasing mobile phase pH, phosphate concentration and working electrode potential resulted in increased MDL 201,012 response; however, the background current also increased. Mobile phases at pHB6.5 and working electrode potentials > 1 .0S V could not be used due to excessive background current. Background current was decreased by operating a second, upstream electrode at a potential of + 850 mV vs. AglAgCl, where some mobile phase components were selectively oxidized. Sensitivity was also increased by adding 75 mM tetramethylammonium ions to the mobile phase, which provided a two-fold increase in MDL 201,012 response without further increasing the background current. The best compromise between MDL 201,012 response and background current was obtained using acetonitrile: buffer mobile phases containing 15 mM sodium acetate, 15 mM monobasic potassium phosphate and 75 mM TMAOH (adjusted to p H 5.9) with upstream and working electrodes maintained at +850 mV and 1025 mV vs. Ag/AgCI, respectively. Although selective sample cleanup was achieved through solid phase extraction, extensive baseline dis-
+
298
DONALD L. REYNOLDS, LARRY S. EICHMEIER AND DENNIS H. GIESING
turbances were observed when liquid chromatography was performed using a single column. These disturbances primarily resulted from mobile phase composition/equilibria changes following injection due to high detector sensitivity and working potentials. These effects were minimized through use of on-line column switching techniques which directed the injection solvent and first eluent fraction to waste. Separation selectivity was achieved using a cyano precolumn and a phenyl analytical column to maximize the polarity difference between the columns. Millibore columns (2.1 mm i d . ) were used to enhance sensitivity due to the higher analyte concentrations in the effluent as compared to traditional (4.6 mm i.d.) columns (Guiochon and Colin, 1984). Analytes were adsorbed on the precolumn using a weak acetonitrile: buffer mobile phase and baseline disturbances were minimized by directing only the volume of mobile phase contained in the precolumn to the analytical column (Fig. 1). No band broadening was observed upon switching due to analyte preconcentration on the analytical column. Subsequent cleanup a n d o r reequilibration procedures were not required for the precolumn using this procedure. Final separation was achieved using a phenyl column with a buffered acetonitrile eluent containing 75 mM tetramethylammonium ion adjusted to p H 5.9. A combination of phosphate and acetate buffers was used to increase the buffer capacity due to the intermediate mobile phase p H used. Excellent selectivity with acceptable detector response and background current were obtained at this pH; however, significant silanol effects were observed for MDL 201,012 and IS, even with 75 mM tetramethylammonium ion. Peak tailing and non-linear detector response were evident and became more severe with decreasing MDL 201,012 concentrations. These effects were eliminated by using desmethyl-MDL 201,012 as a masking agent. The masking agent was added to plasma prior to extraction rather than adding additional competing cations to the mobile phase to avoid additional detection complications. When added at concentrations of =400 ng/mL, the masking agent effectively competed with MDL 201,012 for silanol sites on the stationary phase surface, which decreased peak tailing and provided linearity down to concentrations of 125 pg/mL MDL 201,012. Extraction efficiency from plasma Absolute recovery of MDL 201,012 (+SD) (n= 5 ) from human plasma was determined at concentrations of 500, 2500 and 7500pg/mL MDL 201,012. Recoveries were 94.l(f17)%, 100(*2.9)% and 95.3(f8.0)%, respectively. Recovery of IS (n = 15) was 96.3( *6.8)% at a concentration of 5 ng/mL.
I
I
I
I
I
2
0
I
I
6
4
Time (minutes)
Figure 2. Chromatograms of plasma samples. (a) Blank pooled human plasma; (b) quality control standard containing 125 pg/ mL MDL 201,012; (c) plasma sample from a human subject 3 h after receiving a 4 mg oral dose of MDL 201,012. I=MDL 201,012; II= internal standard.
ponents and masking agent (Fig. 2a,b). In patient samples, neither plasma components nor metabolites eluted near MDL 201,012 or IS (Fig. 2c), illustrating excellent specificity. System specificity for MDL 201,012 in the presence of potential metabolites (MDL 201,012 N-oxide, MDL 201,012 diol, desmethyl-MDL 201,012, and desmethyldiol-MDL 201,012) was also determined. All potential metabolites were chromatographically resolved from MDL 201,012 and IS (data not shown). In addition, none of the possible metabolites interfered with solid phase isolation of MDL 201,012 and IS or quantitation when present in plasma at concentrations up to lOOOng/ml. The assay is therefore specific for MDL 201,012. Greater than 250 plasma samples from acute dose tolerance studies have been assayed using this method. No apparent interferences were observed throughout the analyses. Figure 3 shows the MDL 201,012 plasma concentration-time profile observed in a subject fol25
1
Specificity Retention times for MDL 201,012 and IS were 3.8 and 5.0 min, respectively. Chromatograms of blank human plasma, a 125pg/mL MDL 201,012 quality control standard and a plasma sample from a human subject receiving MDL 201,012 are shown in Fig. 2. Drug and IS are suitably resolved from endogenous plasma com-
0
3
6
9
12
Time Postdose (hr)
Figure 3. Plasma concentration-time profile of MDL 201,012 in a human subject after receiving a 4 mg oral dose of MDL 201.01 2.
LUED OF MDL 201,012
299
Table 1. Precision and accuracy of MDL 201,012 calibration standards in human plasma assayed during a fiveday period
Table 2. Precision and accuracy of MDL 201,012 quality control standards in human plasma assayed during a five-day period
Conc. added (pglmL)
Conc. added (pg/mL)
N
Conc. found (pglmL1
125 9 140 250 9 245 500 10 463 700 9 688 2500 10 2480 5000 8 4970 7500 9 7590 aRelative standard deviation.
RSD'
I%)
16 5.7 11 6.7 3.9
3.5 2.5
Relative error (%)
12 - 1.9 - 7.4 - 1.7 - 0.64 - 0.58 1.2
lowing administration of a 4 mg dose of MDL 201,012 as an oral solution. The assay is sufficiently sensitive to monitor MDL 201,012 plasma concentrations for at least five half-lives in subjects receiving this dose.
Linearity, precision and accuracy of the calibration curves The assay was validated by assaying seven calibration standards ( n = 2) and five replicates of seven quality control standards on five separate days. Peak-height ratios were proportional to the amount of MDL 201,012 added tot plasma over the concentration range 125-7500 pg/mL. The best-fit straight line was determined daily by least-squares linear regression analysis. A weighting factor of reciprocal concentration was used to ensure homoscedasticity over the concentration range (Johnson et al. 1988). Mean (+SD) regression results obtained during assay validation are described by the following equation: peak-height ratio = 0.331 (rt0.0024) x [MDL 201,012 concentration] - 0.0003 (k0.006) (correlation coefficient = 0.9985). The reproducibility of calibration curves was determined as the variation of calibration standards from the regression line. Calibration curve precision (YO relative standard deviation) ranged from 2.5 to 16% over five days, with corresponding accuracies ( O h relative errors) ranging from - 7.4 to 12% (Table 1).
N
Conc. found (pg/mLl
125 21 128 250 21 266 500 22 545 1000 22 1040 2500 24 2470 5000 25 5180 7500 25 7700 aRelative standard deviation.
(%)
Relative error (%)
13 10 7.0 5.4 8.5 5.2 7.2
2.6 6.5 9.0 3.6 - 1.1 3.5 2.7
RSDa
Assay precision and accuracy Assay precision and accuracy was determined by analysing five replicates of seven quality control standards on five separate days with the calibration standards. Assay precision was f13%, based on relative standard deviations ranging from 5.2 to 13% for quality control standards containing 125 to 7500 pg/mL MDL 201,012 (Table 2). Corresponding assay accuracy ranged from -1.1 to 8.9%. The minimum quantitation limit was 125 pg/mL MDL 201,012 using a 2 mL sample. Plasma storage stability MDL 201,012 plasma stability under storage conditions was evaluated by assaying five replicate samples at each concentration stored in glass and polypropylene at - 20 and - 70 "C. MDL 201,012 concentrations in samples assayed after nine months storage were within +lo% of theoretical values for all storage conditions evaluated. These data suggest that MDL 201,012 patient plasma samples can be stored in polypropylene tubes at - 20 "C for up to nine months without significant drug loss. In summary, a sensitive and selective liquid chromatographic method has been developed and validated to quantify MDL 201,012 in human plasma over the concentration range 125-7500 ng/mL MDL 201,012. The method has been successfully used to quantitate MDL 201,012 in samples from clinical acute dose tolerance studies.
REFERENCES Guiochon, G. and Colin, H. (1984). In Microcolumn High Performance Liquid Chromatography, ed. by Kucera, P., Chap. 1. Journal of Chromatography Library, Vol. 28, Elsevier, New York.
Johnson, E. L., Reynolds, D. L., Wright, D. S . and Pachla, L. A. (1988). J. Chrorn. Sci. 26,372.
