Anal. Chem. 1982, 5 4 , 1087-1090 (19) Ampulski, R. S.; Ayers, V. E.; Morell. S. A. Anal. Biochem. 1969, 32, 183-1 69. (20) Taketa, F.; Morell, S. A. Anal. Biochem. 1969, 32, 169-174. (21) Novak, T. J.; Pieva, S. G.; Epstein, J. A n d . Chem. 1980, 52, 1851-1055. (22) Cox, J. A.; Przyjazny, A. Anal. Lett. 1977, 10, 869-085. (23) Kuwata, K.; Uebori, M.; Yamazaki, Y. J . Chromatogr. 1981, 277, 370-302.
1087
(24) Kuwata, K.; Uebori, M.; Yamazaki, Y . J . Chromatogr. Sci. 1979, 17, 264-268. (25) Kuwata, K.; Uebori, M.; Yamazaki, Y . Anal. Chem. 1981, 5 3 , 153 1-1 534.
RECEIVED for review December
8,
Accepted February
22. 1982.
Liquid Chromatographic Determination of Guanadrel in Laboratory Animal Diet as the Fluorescent Acetylacetone Derivative Paul A. Bombardt and Wade J. Adams' Physical and Analytical Chemistry-Drug
Metabolism Research, The Upjohn Company, Kalamazoo, Michigan 4900 1
A hlgh-performance llquld chromatographic method for the rapld determlnatlon of guanadrel In laboratory anlmal dlet Is descrlbed. Derlvatlzatlon of guanadrel wlth acetylacetone In a homogeneous aqueous:organlc solvent permitted the sensltlve and speclflc fluorescence detection of the resuttlng pyrlmldlne wlthout the need for extractlon of the derlvatlve prior to chromatographic analysls. Wlth an excttatlon wavelength of 238 nm, the fluorescence response at 360 nm was linear for drug-dlet mlxtures havlng guanadrel sulfate concentratlons up to 1400 ppm and the llmlt of detectlon was approxlmately 1 ppm (3 ng oncolumn). Assay precision, as estlmated by analyzlng repllcate samples of a laboratory standard, was better than 1.5 % relative standard devlatlon.
The utility of derivatizing guanidino compounds with hexafluoroacetylacetone (1-5) or acetylacetone (6) to form the more volatile corresponding pyrimidines prior to gas chromatographic analysis is well documented. Quantitation of the pyrimidines using highly specific and sensitive electron capture or mass spectrometric detection has afforded sensitivities in t h e nanogram per milliliter range for guanidino compounds in plasma (1-3, 5 ) , urine (3),and tissue (4, 6). Analytical methods based upon the reaction of the amidino moiety with 8-hydroxyquinoline (Sakaguchi reaction) or phenanthrenequinone prior t o colorimetric (7-9) or fluorometric (10, 11) detection, respectively, have also been reported. Separation of the guanidino compounds prior t o derivatization is necessary using these methods since these reagents d o not yield unique derivatives. We report a rapid and specific reversed-phase liquid chromatographic method for the determination of guanadrel. Guanadrel sulfate is a guanidine antihypertensive agent currently under clinical investigation (12). The methodology was used for the quantitation of guanadrel in pelleted drugdiet mixtures administered to mice in a carcinogenicity study in order to comply with the FDA's Good Laboratory Practice regulations (13). Precolumn derivatization of guanadrel with acetylacetone in a homogeneous aqueous:organic solvent permitted the sensitive and specific fluorescence detection of the resulting pyrimidine without the need for extraction prior t o chromatography.
EXPERIMENTAL SECTION Reagents. The reference standard guanidino compounds, guanadrel sulfate (I, [ (1,4-dioxaspiro[4.5]dec-2-yl)methyl]guanidine sulfate) and (cyclohexy1methyl)guanidinesulfate (II), were supplied by the Pharmaceutical Research and Development Laboratories of The Upjohn Company. Acetylacetone was obtained NH I1 CHz-NH-C-NH2 * % H2S04
5
NH
~ C H ~ - N H- -H L~ SNO H ~ ~ I/,
I
n
commercially (Eastman, Rochester, NY, 98% minimum purity; or Aldrich, Milwaukee, WI, 99%+ purity) and used without further purification. Distilled-in-glass spectroscopic grade acetonitrile, methanol and tetrahydrofuran (Burdick and Jackson, Muskegon, MI) were used as received. Inorganic chemicals were analytical reagent grade and were prepared in distilled, deionized water. Apparatus. The high-performance liquid chromatograph used in this study was a modular component system consisting of an Altex Model llOA solvent pump, an in-house designed and fabricated autoinjector (14) fitted with a 40-wL sample loop, a commercially prepared 4.6 rnm i.d. X 250 rnm column packed with 10-wm LiChrosorb RP-8 (E. Merck Laboratories, Elmsford, NY), a dual monochromator spectrofluorometerequipped with a 2 8 4 , flow cell (Model 2 W , Perkin-Elmer, Norwalk, CT), and a variable sensitivity recorder (Model 355,Linear Instruments, Irvine, CA). Automated data acquisition and processing were accomplished using an IBM 1800 computer (15). A two-speed reciprocating shaker (Eberbach and Sons, Ann Arbor, MI) was used for extraction of samples and a block heater (Lab-Line Instruments, Melrose Park, IL) was used for the derivatizations. Mass spectral characterization of the derivatives was accomplished by using a magnetic sector mass spectrometer (Model CH7A, Varian Mat, Bremen, West Germany). Sample Analysis. Internal standard and reference standard solutions containing approximately 700, 210, and 70 pg/mL of the respective compounds were prepared each day samples were analyzed by dissolving the accurately weighed reference standards in 0.05 M ammonium dihydrogen phosphate. Individual pellets of the diet were pulverized using a glass mortar and pestle and 1-g samples weighed into 16 x 125 mm culture tubes fitted with Teflon-lined caps. The standards were prepared by using drug-free pelleted diet. Following addition of 5 mL of 0.05 M ammonium dihydrogn phosphate, the samples were vortexed to expel trapped air and allowed to disintegrate for 15 min. One milliliter of the appropriate internal standard
0003-2700/82/0354-1007$01.25/00 1982 American Chemical Society
1088
ANALYTICAL CHEMISTRY, VOL. 54, NO. 7, JUNE 1982 B
A
Table I. Abbreviated Electron Impact Mass Spectra of the Acetylacetone Derivatives of Guanidino Compounds
I
m/z (% re1 abundance) other ionsa
G
M'.
guanidino compound guanadrel
277 (14)
cyclohexylmethylguanidine t,
219 (22)
2,3-dihydroxypropyl- 197 ( 3 ) guanidineC
234 (85), 179 (loo), 162 (5S), 148 (34), 136 (99), 123 (28) 137 (40), 136 (loo), 124 (13), 1 2 3 (64), 108 ( B ) , 107 (8) 166 (28), 137 (34), 136 (loo), 123 ( 2 1 ) , 108 (14),107 (12)
The six most intense ions for m/z > 100 are reported. Internal standard. A potential degradation product of guanadrel. a
1
0
I
,
6
!
,
, , ,
I
,
12 18 24 0 R E T E N T I O N TIME
,
,
6
,
,
12
,
,
,
18
,
24
(minutes)
-1.15
Flgure 1. Chromatograms of derivatized laboratory animal diet extracts: (A) chromatogram of derivatized diet extract containing 0.0 ppm
guanadrel sulfate. (B)chromatogram of derivatized diet extract containing 70 ppm guanadrel sulfate (G) and internal standard (IS). A potential degradation product of guanadrel, 2,3-dihydroxypropylguanidlne, elutes at approximately 3 min.
- 110
- 1.05
::12 B
z
z tu
.n
-1ooE
10
m z
-095
Y 8
"
3
VI
solution and 3 mL of acetonitrile were then added to all the samples. One milliliter of the appropriate reference standard solution was added to the unknowns. The samples were extracted for 20 min at high speed on a two-speed reciprocatingshaker and centrifuged for 20 min at 2000 rpm. Three-milliliter aliquots of the supernatant or diluted supernatant were transferred to 16 X 125 mm culture tubes fitted with Teflon-lined caps and 1mL of 1 M sodium bicarbonate, 2.5 mL of methanol, and 2 mL of acetylacetone were added t o each sample. The tubes were tightly capped, mixed by inverting, and placed in a block heater for derivatization at 120 f 5 "C for 100 min. The samples were cooled to ambient laboratory temperature following derivatization, mixed by inverting several times, and centrifuged for 10 min at 2000 rpm. High-Performance Liquid Chromatography. Aliquots of the reaction mixtures were chromatographed directly on a microparticulate LiChrosorb RP-8 reversed-phase column using a mobile phase composed of 0.05 M ammonium dihydrogen phosphate:acetonitrile:tetrahydrofuran (5:41) at a flow rate of 0.8 mL/min. Column back-pressurewas approximately 1100 psig. Isolation and Mass Spectral Characterization of Derivatives. The acetylacetone derivatives of guanadrel, 2,3-dihydroxypropylguanidine,and cyclohexylmethylguanidine(internal standard) were isolated by collecting chromatographic fractions corresponding to the retention volumes of the respective compounds. The 2,3-dihydroxypropylguanidine,a degradation product of guanadrel(16),was prepared by hydrolyzing guanadrel in 0.1 M hydrochloric acid, neutralizing the acid with sodium carbonate, and derivatizingthe mixture using reaction conditions identical with those used for the analysis of drug-diet mixtures. Direct inlet electron impact (70 eV) mass spectra were obtained following lyophilization of the chromatographic fractions.
RESULTS AND DISCUSSION Derivatization of guanadrel with acetylacetone in a homogeneous aqueous-organic solvent was particularly convenient for the analysis of guanadrel in laboratory animal diet. The compound was readily extracted from the pulverized diet using 0.05 M ammonium dihydrogen ph0sphate:acetonitrile (7:3), derivatized, and an aliquot of the reaction mixture was assayed directly using reversed-phase liquid chromatography with = 360 nm). fluorescence detection (Xexcitation = 238 nm, Xe-ion Chromatograms of the acetylacetone derivatized extracts of blank feed and feed containing 70 ppm guanadrel sulfate are shown in Figure 1. The guanadrel and internal standard derivatives eluted a t approximately 10.5 and 18 min, respectively, and were well resolved from potential interferences.
-090
E 6
o Guanadrel Derivative 0 Internal Slanderd Derivative h Pnak Height Ratio
24
4!
I
1
2
- 0.85 - 0 80
5
5
D
I
I
3 4 VOLUME ACETYLACETONE (ml)
=
6
1
Flgure 2. Fluorescence response/peak height ratio vs. volume of acetylacetone used for derivatization at 120 f 5 "C.
The compounds eluting a t approximately 16 and 21.5 min were fluorescent impurities present in the acetylacetone. The magnitude of these peaks was independent of the derivatization time under the described reaction conditions. An increase in the acetonitrile or tetrahydrofuran composition of the mobile phase decreased the retention times of both the guanadrel and internal standard derivatives and resulted in poorer resolution from these potential interferences. Malcolm and Marten (2) reported that interferences due to the self-condensation of acetylacetone were produced under both single phase and biphasic derivatization conditions in the analysis of debrisoquin and 4-hydroxydebrisoquin in plasma by gas chromatography/mass spectrometry. No fluorescent interferences corresponding to self-condensation products of acetylacetone were found by using the derivatization and chromatographic conditions described in this report. Mass spectral analysis of lyophilized chromatographic fractions collected a t the elution volumes of guanadrel and the internal standard confirmed the identity of these derivatives (Table I). The chromatographic properties of the acetylacetone derivative of 2,3-dihydroxypropylguanidine,a potential degradation product of guanadrel, were determined by hydrolyzing reference standard guanadrel and chromatographing aliquots of the derivatized mixture. Mass spectral analysis of lyophilized chromatographic fractions collected near the solvent front confirmed that the 2,3-dihydroxypropylguanidine derivative eluted at approximately 3 rnin (Table I). Optimum reaction conditions for derivatization of guanadrel and the internal standard were established by monitoring fluorescence response as a function of the reaction temperature, the quantity of acetylacetone used, and the reaction time. A reaction temperature of 120 k 5 "C was chosen for convenience because derivatization was complete for both com-
ANALYTICAL CHEMISTRY, VOL. 54, NO. 7, JUNE 1982 o Guanadrel Derivative
Table 111. Potency and Content Uniformity of Pelletized Guanadrel Sulfate-Diet Mixturesa
Intnfnal Standard Oerivaliwe A Peak Height Rat10
label (PPm 1 70
lot no. 1
2 3
4
"t// 0
mean (esd) 210 40
80
120
160
240
200
280
REACTION TIME (minulei)
Flgure 3. Fluorescence response/peak height ratio vs. derivatization time at 120 f 5 OC.
mean (esd) 700
response factor ( % RSD)a __..-
70 P P ~
0.02378 ( f 7.8) 2 0.02280 ( i 1 . 8 ) 3 0.02313 ( i 2 . 5 ) 4 0.02362 (.0.9) 5 0.02219 ( r 1 . 4 ) 6 0.02254 ( t 2 . 3 ) 7 0.02181 (10.5) 8 0.02363 ( i 0 . 9 ) 9 0.02238 ( c 0.4) 0.02165 ( t 3 . 2 ) 10 1.1 0.02374 (11.0) mean 0.02283 ( i 3 . 4 ) 1
__
200 pprn
5 6 7 8 9 10 11
Table 11. Assay Precision and Reproducibility assay day
1089
12
-
700 pprn
0.007842 (i 7.9) 0.002326 ( f 0 . 7 ) 0.007741 ( i 1 . 4 ) 0.002297 ( t 4 . 2 ) 0.007524 ( i 1 . 5 ) 0.002255 ( i 2 . 0 ) 0.007861 ( i 1 . 4 ) 0.002319 ( i 0 . 7 ) 0.007311 ( i 1.7) 0.002149 ( i 0 . 8 ) 0.007537 ( + 1 . 7 ) 0.002193 ( i 0 . 9 ) 0.007347 ( i 3 . 1 ) 0.002183 ( i 0 . 4 ) 0.007917 ( f 0 . 5 ) 0.002387 ( t l . 2 ) 0.007487 (i 1.2) 0.002255 (i 1.0) 0.007400 ( i 1 . 6 ) 0.002175 ( i 2 . 0 ) 0.007923 ( f 1 . 3 ) 0.002315 ( f 0 . 4 ) 0.007626 ( i 3 . 1 ) 0.002259 ( t 3 . 4 )
Calculated from three replicate analyses. pounds in a reasonable reaction time using available equipment. Graphs of fluorescent response vs. the volume of acetylacetone used and the reaction time are shown in Figures 2 and 3, respectively. Derivatization was found to be essentially complete in 100 min at 120 "C when 2 mL of acetylacetone was used. Furthermore, the peak height ratio remained constant over the 80-160 min time interval of the reaction using these derivatization conditions. Excellent recoveries of both guanadrel and the internal standard were obtained by using an extraction solvent composed of 0.05 M ammonium dihydrogen phosphate:acetonitrile (7:3) provided the pulverized diet was disintegrated in the ammonium dihydrogen phosphate prior to addition of the acetonitrile. No significant degradation of guanadrel occurred during the period of time required for extraction. A recovery of 100.7 f 1.5% ( n = 18) was calculated relative to reacted standards prepared in the absence of diet. Linear regression analysis of calibration curve data indicated no significant deviations from linearity for drug-diet concentrations up to 1400 ppm guanadrel sulfate (r2 = 0.9984) and intercepts that were not significantly different from zero (p > 0.05). The response was also found to be a linear function of the mass of drug-diet admix analyzed-a correlation coefficient of r2 = 0.9932 was obtained for a nine-point curve constructed by analyzing 0.1-2.0 g samples of thoroughly pulverized 700 ppm dietary admix. The precision and reproducibility of the method were assessed by analyzing replicate samples of drug-diet mixtures containing 70,200, and 700 ppm guanadrel sulfate (Table 11). Intraassay variability was less than i=4.2% RSD (percent relative standard deviation) for samples analyzed on 11 different days, wit,li the mean percent relative standard deviations being *1.5%, fl.6%, and f1.3% for the 70, 200, and 700 ppm samples, respectively. The interassay percent relative standard
mean (esd)
potency, b %
102.6 103.3 98.0 100.3 101.0 (2.4) 104.3 101.5 96.7 97.7 100.0 (3.5) 101.0 102.1 101.8 98.1 100.8 (1.8)
content uniformity,c 5% i 10.4 i 10.1
i10.2
i9.8 i10.1 (0.2) t8.0 t3.4 i 6.9 i6.7 t 6 . 2 (2.0) i5.0 f. 5.7 +5.0 t2.5 c4.6 (1.4)
Six pellets were randomly selected from each lot of dietary admix and analyzed individually by pulverizing and assaying a 1-g sample of each. Percent of label. Standard deviation in potency ( n = 6). deviations, based on data collected over an 18-month interval, were *3.4%, *3.1%, and f3.4% for the 70,200, and 700 ppm samples, respectively. The detection limit of the method ( S N > 3:l) is approximately 1 ppm guanadrel sulfate, which corresponds to an on-column injection of 3 ng of the derivatized compound. The utility of the analytical methodology was demonstrated by analyzing pelleted drug-diet mixtures having label concentrations of 70,210, and 700 ppm guanadrel sulfate (Table 111). Six pellets were randomly selected from each lot of dietary admix and analyzed individually by pulverizing and assaying a 1-g sample of each. Drug potencies determined in this manner were in excellent agreement with label concentrations. Based on the assumption that the errors in the assay and in the inhomogeneity of the formulation are randomly distributed and mutually independent, the total variance associated with the multipellet analysis is equal to the summation of the variances in the assay and in the inhomogeneity of the formulation. Since the variances associated with the multipellet analyses were a t least twice the mean intraassay variances for each of the concentrations, the variances in the multipellet analyses were taken as an estimate of content uniformity. Content uniformities, expressed as percent relative standard deviations, were 10.4% or better for the 12 lots of dietary admix. An inverse relationship between content uniformity and guanadrel sulfate concentration was evident. ACKNOWLEDGMENT The synthesis of cyclohexylmethylguanidine sulfate by S. C. Perricone and the assistance of J. R. Bod in obtaining mass spectra of the derivatized compounds are gratefully acknowledged. We thank S. E. Yoder for preparation of the manuscript. LITERATURE CITED (1) Erdtmansky, P.; Goehi, T. J. Anal. Chem. 1975, 4 7 , 750. (2) Malcolm, S. L.; Marten, T. R. Anal. Chem. 1976, 48, 807. (3) Guerret, M.; Lavene, D.; Longchampt, J.; Kiger, J. L. J. Pharrn. Sci. 1979, 68, 219. (4) Kawabata, T.; Ohshima, H.; Ishibashi T.; Matsui, M.; Kitsuwa. T. J. Chromatogr. 1977, 140, 47. (5) Kaiser, D. G.: VanGiessen, 0. J.; Liggett, W. F.; Thomas, R . C., The Upjohn Company, unpublished work. (6) Mwi, A.; Ichimura, T.; Matsumoto, H. Anal. Bbchem. 1978, 89, 393. (7) Blass, J. P. Biochem. J. 1960, 7 7 , 484. (8) Mori, A.; Hosotani, M.; Tye. L. C. Blochem. M e d . 1974, IO. 8. (9) Matsumoto, M.; Kishikawa, H.; Mori, A. Blochem. Med. 1976, 76, 1. (10) Yamada, S.; Itano, H. A. Blochim. Biophys. Acta 1966, 130, 538.
1090
Anal. Chem. 1982, 5 4 , 1090-1093
(11) Morl, A.; Katayama, Y.; Hlgashldate, S.;Kimura, S. J . Neurochem. 1079, 32. 643. (12) Bloomfield, D. K.; Cangiano, J. L. C u r . Ther. Res. 1060, T I , 727. (13) Fed. Reglst. 1078, 4 3 , 60018. (14) Beyer, W. F.; Gleason, D. D. J . Pharm. Scl. 1975, 6 4 , 3420. (15) Kaiser, D. G.; KO,H.;VanGlessen, G. J.; Zieserl, J. F.; Marks, D.; Kenny, M. D. “Abstracts of Papers”, 174th Natlonal Meeting of the American Chemical Soclety; Chicago, IL, 1977; Amerlcan Chemical Soclety: Washlngton, DC, 1977; COMP 14.
(16) Kaiser, D. G., The Upjohn Company, unpublished work.
RECEIVED for review August 25, 1981. Accepted March 29, 1982. The work reported in this manuscript was presented in part at the 27thAPS Meeting, APhA Academy of Pharmaceutical Sciences, Kansas City, MO, 1979.
Instrumental Aspects of Capillary Supercritical Fluid Chromatography Paul A. Peaden, John C. Fjeldsted, and Milton L. Lee” Departmenf of Chemistry, Brigham Young University, Provo, Utah 84602
Stephen R. Sprlngston and Milos Novotny Department of Chemistry, Indiana University, Bloomington, Indiana 47405
The bask Instrument components requlred for caplllary supercrltlcal fluld chromatographyinclude a high-pressure pump with pressure programmer, a small-volume sample Inlet system, a constant temperature oven, and a small-volume detector. The major stresses applied to the hstrumentatlon are the hlgh pressures and temperatures required to maintain the mobile phase at or above Its crltlcal polnt. Of the dlfferent sample introductlon systems studied, spllt Injection presently appears to be the most effectlve. By use of the described hstrumentatlon, a plate helght of 0.30 mm was obtalned for pyrene ( k = 0.50) on a 100-hm caplllary column contalnlng a bonded poly(methylphenylslloxane)statlonary phase.
Since the first description of supercritical fluid chromatography (SFC) in the separation of metal porphyrins ( I ) , much research has been done to explore the potential advantages of this analytical technique. The unique feature of SFC is that the mobile phase is subjected to pressures and temperatures near its critical point. Under these conditions ita density approaches that of a liquid, while at the same time, solute diffusion coefficients are approximately 2 orders of magnitude greater than those found in liquids. A supercritical fluid possesses solvating properties similar to a liquid, and solute diffusivities intermediate between a gas and a liquid. Therefore, with the favorable mass transfer properties of SFC, higher efficiencies can be obtained in shorter analysis times than can be achieved with capillary liquid chromatography, and comparable efficiencies ( 40 plates/s) to high-performance LC are achievable. I n addition, SFC demonstrates the ability to analyze relatively nonvolatile and thermally labile solutes which cannot be analyzed by gas chromatography or, in many cases, even by liquid chromatography. The density of a supercritical fluid is largely determined by temperature and pressure. When operating at a constant temperature above the critical temperature of the fluid, liquid formation is prevented and the mobile phase density can be easily controlled by adjusting the pressure. Hence, pressure programming (2, 3) gradually increases the mobile phase density and decreases solute retention. This effect is analogous to temperature programming in gas chromatography and gradient elution in liquid chromatography. N
The advantages of using capillary columns in SFC ( 4 , 5 ) are similar to those of usigg capillary columns in liquid or gas chromatography. The low pressure drop across an open bore tube allows higher efficiencies to be achieved than obtainable with a packed column, simply because the column can be made much longer. This low pressure drop is additionally beneficial in capillary SFC because the density of the mobile phase is more uniform throughout the length of the column. Larger pressure drops, and hence density gradients, have been previously shown to be disadvantageous in packed column SFC (6, 7). Another advantage of using capillary columns is the absence of the plate height contributions due to alternate solvent flow paths as are found in packed columns. The discovery of new chromatographic methods has traditionally been followed by the development of suitable instrumentation for their proper utilization. New instrumentation was developed for packed column SFC after it first appeared, and a number of papers have been published on this subject (8-12). Analogous to past trends in which modifications in equipment were needed in converting from packed to capillary columns in gas and liquid chromatography, capillary SFC also requires its own unique instrumentation. The major instrumental modifications are a result of the stresses of high pressure and temperature required to maintain the mobile phase at or above its critical point and the low tolerance of extracolumn volume in sample introduction and detection systems. This has resulted in the use of split injection systems and on-column detection in this study, both of which are new to SFC. This paper describes the successful development of workable instrumentation for capillary SFC. Particular emphasis is placed on sample introduction and detection systems which provide minimum band broadening and maximum sensitivity. Although a number of different mobile phases can be used (13) and are presently under study, this paper was limited to our initial studies employing n-pentane. EXPERIMENTAL SECTION Instrumentation. A general schematic of the total chromatographic system is shown in Figure 1. Each of the principal parts of the SFC system is explained below. Id all cases, n-pentane was used as the mobile phase. A pressure-controlled system was obtained by using a Varian 8500 syringe pump modified for pressure control as published by
0003-2700/82/0354-1090$01.25/00 1982 American Chemical Society