Adsorption of Local Anesthetics on Activated Carbon: Freundlich Adsorption Isotherms l K U 0 ABE'*§, HIROSHIKAMAYA**, AND ISSAKU uEOA**"
Received April 4,1988, from the * Depafiment of Anesthesia, University of Utah School of Medicine, Salt Lake City, UT 84732, and *Anesthesia Service, Veterans Administration Medical Center, Salt Lake City, UT84748. Accepted for publication July 7, 1989. §On leave from the Osaka Municipal Technical Research Institute, 1-6-50Morinomiya, Joto-Ku, Osaka 536, Japan. Abstract 0 We have shown that adsorbability of local anesthetics onto activated carbon, expressed by the partition coefficients at infinite dilution,correlated well with the pharmacologicalactivity. However,there is no parameter that can singly express the tendency to be adsorbed. Adsorbability is a loosely defined term and its meaning varies with the adsorption model. This study showed that the logarithm of the adsorbed amount of drugs was linearly related to the logarithm of the free drug concentration, in conformity to the Freundlich adsorption isotherms.The slope of the double logarithmic plot is expressed by 1lNin the Freundlich equation and is considered to be inversely related to the drug affinity to the adsorbent. The slope was used to evaluate the tendency to be adsorbed, or "adsorbability"of seven aromatic amine local anesthetics. Phenobarbital was included to compare anionic drugs in contrast to the cationic local anesthetics. The slopes were nearly equal between the cationic and neutral local anesthetics. Apparently, the lower hydrophobicity of the cationic forms is compensated by the electrostatic attraction from the negative charges present on the activated carbon surface. With phenobarbital, the slope value of the anionic form was larger than the neutral form. The lower affinity of the anionic form may be caused by the electrostatic repulsion. The molecular size parameters (i.e., molecular weight, molar refraction, and parachor) showed a linear relationship to the slope values. It may be possible to estimate the affinity-related slope values from these parameters.
Activated carbon is now widely used in environmental engineering for waste water treatments and as an adsorbent for hazardous gases. In health sciences, its applications include artificial liver,l,z treatment for drug poisoning,3-7 and complex formation with anticancer drugs to be deposited into malignant tissues for their prolonged release.8 Recently, we9 have successfully correlated activated carbon adsorbability (expressed by the partition coefficient of drugs a t infinite dilution) with drug potencies as a parameter for the quantitative structureactivity relationship (QSAR).Activated carbon adsorbability is particularly effective in analyzing the QSAR of charged drugs where the conventional oil-water partition coefficients fail to correlate with the drug activities. The adsorbability represents the amphipathic property of drugs that have limited solubility into the organic phase. We hope that the activated carbon system will contribute to future drug design. Studies on drug adsorption onto the interfacial region may contribute to an understanding of the action mechanisms. Although the term "adsorbability" intuitively conveys the concept on the tendency of a solute to be adsorbed onto activated carbon surfaces, it is not well defined. There is no ideal parameter that can singly represent this property. Hence, adsorbability can be expressed in a number of ways, and each has different meaning. In our previous communications we expressed carbon adsorbability by extrapolating the adsorption isotherms to infinite dilution of the ligand concentration in the bulk solution. Because therapeutic drug concentrations are not high, the infinite dilution parameter 354 I Journal of Pharmaceutical Sciences Vol. 79, No. 4, April 7990
meets the purpose of QSAR analyses. The ratio between the adsorbed and free drug concentrations converges into a constant value a t dilute concentrations. However, the partition coefficient starts to vary when the drug concentrations are increased. The change in the partition coefficient by the increase in the concentration varies with each drug. As will be shown in this study, the order of the adsorbed amounts of two drugs reverses a t higher drug concentrations. For problems in the high concentration range, such as for the treatment of drug poisoning, the infinite dilution parameter may not be suitable for describing the adsorbability . The present study is aimed at elucidating the adsorption pattern a t finite anesthetic concentrations. Judislo averaged the adsorbed amounts a t various drug:carbon ratios. Standard methods, however, consist of constructing adsorption isotherms and then analyzing the adsorption curve by an appropriate model. Carbon surface adsorption often follows the Freundlich adsorption isotherms,llJZ in which the logarithm of the adsorbed amount bears a linear relationship with the logarithm of the free drug concentrations. In this study, the linearity was confirmed, and the slope of the double logarithmic plot was compared among aromatic amine local anesthetics. Phenobarbital was included because the drug has nerve-blocking activity and ionizes into an anionic form, in contrast to the standard local anesthetics that ionize into a cationic form. In alkaline media, the ester-type local anesthetics undergo nonenzymatic hydrolysis. Despite the early demonstrations of this alkaline hydrolysis,13the problem tends to be overlooked recently. Hence, we included the effect of pH on the hydrolysis of procaine in some detail.
Experimental Section The activated carbon was the same as previously reported9 and was a single batch of 200-350 mesh CAL-activated carbon (Calgon, Pittsburgh, PA) with a surface area of 1010 cm2/g of carbon9 when measured by the adsorption isotherms of nitrogen gas at 77 K according to the Brunauer-Emmett-Teller theory with a Shibata P-700 Surface-Area Analyzer (Shibata Chemical Instruments, Tokyo, Japan). The carbon was washed with hot distilled water and dried at 110 "Cfor 2 days and then was kept in a desiccator containing silica gel. Water was purified by distillation, followed by treatment with activated carbon and ion-exchanger columns, and ultrafiltration using a Millipore system (Bedford, MA). The specific resistivity was maintained above 16 Mohm * cm-'. The following local anesthetics were obtained from Sigma Chemical (St. Louis, MO) in a crystalline form of hydrochloride salts: procaine, tetracaine, and dibucaine. Lidocaine, mepivacaine, and chloroprocaine were gifts from Astra Pharmaceutical Products (Worcester, MA), Sterling Drug (Rensselaer, NY), and Pennwalt (Rochester, NY), respectively. These anesthetics were also in crystalline forms of hydrochloride salts. Anesthetic adsorption was estimated at neutral and alkaline pH.
0022-3549/90/0400-0354$0 l.OO/O 0 1990, American Pharmaceutical Association
The pH was adjusted with a 0.1 M phosphate buffer (KH,PO, and Na,HPO,). The strongly alkaline pH 11was adjusted by the addition ofO.1 M NaOH. The presence of the buffer did not affect the adsorption of local anesthetics on the activated carbon surface. The bulk pH was measured by a Radiometer Ion 85 Analyzer (Westlake, OH) and a combination glass electrode. The kinetics of the nonenzymatic alkaline hydrolysis of the ester type local anesthetic, procaine, was studied at 1 mM concentration, adjusting the pH to various values at 25 "C. The adsorption isotherms were obtained by adding various quantities of CALactivated carbon to 50-mL sample tubes. An aliquot of 40 mL of the solution of known solute concentration was added to the sample tube and tightly capped. The tubes were shaken in a water bath maintained a t 25 "C for 20 h. Preliminary experimenta showed that adsorption was essentially complete within 20 h. After equilibration, a sample was taken from each tube and was filtered through a 0.45-pm pore size membrane to remove any suspended carbon. The anesthetic content in the filtrate solution was measured by ultraviolet absorbance using a Perkin-Elmer 554 spectrophotometer (Norwalk, CT) in a stoppered 10.0-mmlight path cuvette. The cuvette temperature was maintained at 25 "C by an electronically controlled Peltier heat exchanger. Because the molar absorptivity of aromatic amine local anesthetics varies according to the pH, calibration curves were constructed at each pH value. The amount of solute adsorbed on the carbon a t equilibrium was then calculated as the difference between the original and the filtrate concentrations.
Results and Discussion Nonenzymatic Hydrolysis of Procaine-Figure 1 shows the time-dependent spectral changes of procaine in a pH 11 solution a t 25 "C. There were two isosbestic points. The extent of hydrolysis was estimated from the change in the main absorbance peak at 287 nm, and the hydrolysis rate is expressed by the following equations:
(1) t112
(2)
= In (2/k)
where C , is the concentration of unhydrolyzed procaine in the solution after t hours, C , is the initial procaine concentration, and k is the rate constant. Figure 2 shows the hydrolysis rate of procaine at several pH values a t 25 "C. The rate constant (k)and the half-life period (t,,z)at each pH are shown in Table I. Below pH 8, procaine hydrolysis was almost negligible. Above pH 8,however, the hydrolysis rate steeply increased with the increase in pH values. Adsorption Rate-The time required to obtain equilibrium was estimated from the adsorption rate of procaine onto activated carbon a t pH 7.0.Table I1 shows the degree of attainment of equilibrium with time. A 1-h reaction time was found to be sufficient for assessment of the adsorption isotherm a t high pH. Table I shows that 6%ofthe initial procaine content is hydrolyzed in 1 h at pH 11. The adsorption data were corrected for the hydrolysis. Adsorption Isotherms-Adsorption isotherms of eight drugs from aqueous solution onto CAL-activated carbon were measured a t neutral and alkaline pH. The adsorption isotherms were expressed by the double logarithmic plot according to the following Freundlich model:
or
log X = log K + (l/N) log C
(4)
where X is the amount of solute adsorbed (mg per g of carbon), C is the equilibrium solute concentration in the bulk (mg/L), and K and 1/N are the Freundlich constants. The anesthetic
0.4
0
0.3
-
w
0 2 U m U
%m
0.2
n
-
0
e
G -
a
Y
-1
0)
0
220 240 260 280 300 320 340 WAVE LENGTH (nm) Flgure 1-Alkaline hydrolysis of ester type local anesthetics. Timedependent ultraviolet spectra of procaine at pH 11 and 25 "C.The spectrum was measured after the solution was diluted to 0.02 mM with pH 7 buffer solution. The initial procaine concentrationwas 1 .O mM. Key: 23 h; and (E) 70 h. Two isosbestic (A) control (0 h); (6)2.5 h; (C) 5.5 h; (0) points are apparent.
-2 0
20
40
TIME
60
(hour)
Flgure 2-Effect of pH on the hydrolysis of procaine at 25°C. The procaine concentration was 1.0 mM. The data are plotted on the semilogarithmicscale according to eq 2. The hydrolysis rates are shown in Table II. Key to pH values: (A) 5.98; (0)6.96; (A)7.99; (V)8.95; (V) 9.57; and (0)10.96. Journal of Pharmaceutical Sciences I 355 Vol. 79, No. 4, April 1990
Table CNonenzymatlc Alkallne Hydrolysls of Procalne at 25 "C
PH
Rate Constant (k), h-'
Half-life (t,,2), h
5.98 6.96 7.99 8.95 9.57 10.95
0 0.000 0.002 0.018 0.039 0.060
-
3001
400 39 18 12 h
0) \
Table Il-Adsorption pH 7.00 and 25 "C
Rate of Procaine onto Activated Carbon' at
E"
Y
X
Adsorption Time
c, -c,/c,-C,b
10 min 30 min l h
0.938 0.976 0.987 0.996 0.999 1.ooo
2h 4h 24 h
a The activated carbon concentration was 1250 mg1L. C,: the initial concentration of procaine (1.O mM); C,:the concentration of procaine remaining in the solution at time 1; C,: the equilibrium concentration of procaine.
concentration was expressed by the free base weight. Figure 3 is the log-log plot for the adsorption isotherms measured a t neutral pH. The adsorption constants and correlation coefficients were estimated by linear regression analysis and are shown in Table 111. The goodness of the fit was evaluated by the absolute relative percent deviation ( R D P according to the following equation, and is included in the table:
where Xcelis the calculated adsorbed value according to eq 4 and X,,, is the experimental value. In this table, dibucaine data a t pH 11.0 are omitted because the data points were not enough to estimate the parameters. The Freundlich adsorption constant, lIN, is a dimensionless parameter and is said to be related to the intensity of drug adsorption.15 The value of 1/N decreases with increasing intensity. Figure 4 shows a linear relation between 11N and molecular weight, where N is related to the affinity and increases with the increase in the molecular weight. The plot for phenobarbital, however, deviated from this relationship.
I
I
I
I
0.1
1
10
100
C (rng/L) Flgure >Adsorption isotherms of local anesthetics on CAL-activated carbon at pH 7.00 and 25 "C. Key: (0)procaine; (0)mepivacaine; (A) lidocaine; (A)chloroprocaine; (V) tetracaine; (V)phenobarbital; (0) benzocaine; and (B) dibucaine. The adsorption isotherms of mepivacaine were measured at pH 6.60. The abscissa is the equilibrium bulk anesthetic concentration, and the ordinate is the amount adsorbed to the carbon.
This deviation is probably caused by the difficulty of adhesion of the phenobarbital molecules to the carbon surface because the barbiturate ring and benzene ring are not aligned on the same plane. Abe et a1.16-18 demonstrated that there exist linear relationships between 11N and molecular weight or molecular refraction in homologous series, such as alkanols, esters, amino acids, saccharides, etc. Table IV lists these physical parameters (@). The correlation between these parameters and 1/Nis analyzed by the following equation and the result is shown in Table V:
11N = a@+ p
(6)
In this table, s is the standard deviation, F represents the overall goodness of the fit, and t is the significance level of the
Table Ill-Adsorptlon Constants for the Freundllch Equatlon
Drug 1 Procaine
Procaine 3 Lidocaine 4 Tetracaine 5 Dibucaine 0 Dibucainea 7 Mepivacaine 8 Mepivacaine 9 Chloroprocaine 10 Benzocaine 11 Phenobarbital 12 Phenobarbital 2
PHb
nc
7.00
8 6 8 7 5 2 13 9 7 7 6 9
1 1.oo
7.00 7.00 7.00 11 .oo 6.60 8.60 7.00 7.00 7.00 9.00
1INe
r'
RD, %g
143 204 167 229 245
0.0994 0.1 05 0.117 0.0649 0.0230
0.998 0.991 0.997 0.998 0.987
0.751 2.49 0.748 0.578 0.472
145 185 195 165 117 59.6
0.101 0.0963 0.0838 0.157 0.192 0.236
0.995 0.995 0.990 0.998 0.995 1.ooo
1.35 1.10 1.70 1.64 1.86 0.705
K, mglgd
-
-
-
-
Dibucaine data at pH 11.OO were not estimated because the number of the data points was insufficient for calculation. The pH values where the experiments were performed. Number of data points. Freundlich constant (mg/g). Freundlich constant (dimensionless). 'Correlation coefficient. g Absolute relative percent deviation. 356 I Journal of Pharmaceutical Sciences Vol. 79, No. 4, April 1990
0.~~1 0.20
5
t'
0 12
/
0 11
0.15
l -
2000.10 ' 150-
0.05
0' 100
I
300
400
Mw Flgum &Relation between the Freundlich adsorption constant, 1 IN, and the molecular weight, Mw. The numbers correspond to the compounds listed in Table 111. Table IV-Physlcal Constants of Local Anesthetlcs
Drug
Procaine Chloroprocaine Lidocaine Tetracaine Dibucaine Mepivacaine Benzocaine Phenobarbital
3
100
I
200
0.1
4)'
Mwh
Mr' Pr'
(Ib
Flgure 5-Effect of pH on adsorption of local anesthetics on activated carbon. Key: (0)procaine; (V) dibucaine; and (0)mepivacaine. The closed symbols are pH 1 1 .O and the open symbols are pH 7.00.For mepivacaine, the closed symbols are pH 8.60and the open symbols are
pH 6.60.
PKa
MwC
Mrd
-
9.05' 8.97' 7.92' 8.46' 8.73' 7.77' 2.65' 7.46
236.3 270.8 234.3 264.4 243.5 246.3 165.2 232.2
67.680 72.514 70.734 77.166 101.532 73.326 45.294 59.128
569.8 609,4 595.7
PP
g: :; 604.9 382.7 463.4
that of the uncharged form10.20 because the charged molecules are more soluble in water. The Freundlich constant, K, of the charged species was lower than that of the uncharged species (Table 111). On the other hand, the 1IN values were nearly equal for both types. The nearly equal affinity between charged and uncharged forms, despite their difference in hydrophobicity, may be
250
of the Regrecrslon Analysis of Equatlon 6 p"
rd
se
100
C (mg/L)
'Reference 19. Reference 22. Molecular weight. Molar refraction. Parachor. Table V-Reeutts
10
1
F'
-7.68*10-4 0.286 0.975 8.67~10'~135 -2.43. 0.270 0.956 0.0114 74.9 0.298 0.952 0.0120 -3.43. 67.3
1
200
-
150
-
100
-
P
re
11.6 8.65 8.20
'Physical constants. Regression coefficient. Constant. Correlation constant. Standard deviation. 'Overall goodness of the fit. 0 Significance level of the regression coefficienta. Molecular weight. ' Molar refraction. Parachor. regression coefficient a. A high correlation coefficient is demonstrated for the molecular size parameters and adsorption. These results indicate that a linear relationship holds between the physical parameters and 11N for local anesthetics. It may be possible to estimate the slope value of the adsorption isotherms from these physical parameters.
0)
\
E"
Y
X
50
I
I
Journal of Pharmaceutical Sciences I 357 Vol. 79, No. 4, April 1990
prompted by the negative charges present at the activated carbon surface that electrostatically attract cationic drugs. The presence of small amounts of oxides in activated carbons is responsible for the surface negative charges. Conversely, the adsorption of negatively charged molecules, such as barbiturates, may be suppressed by the presence of surface negative charges. Phenobarbital is negatively ionized in alkaline solution because of lactam (“keto”) and lactim (“enol”)tautomerization. Figure 6 shows the effect of pH on the adsorption of phenobarbital. As expected, the adsorbed amount of phenobarbital a t high pH was lower than that at low pH in the limit of the present experimental condition. The 1/N value of the anionic form (high pH) was larger than the uncharged form (low pH), indicating that the affinity of the anionic form is weaker than that of the uncharged form. Similar phenomena were observed for the adsorption of anionic and cationic surfactants of similar molecular weight.21 The cationic surfactants showed lower 1/N values than the anionic surfactants.21
References and Notes Kawanishi, H.; Nishiki, M.; Ezaki, H.; Yamane, S.; Tsuchiya, T.; Sugiyama, M.; Cho, T.; Kimura, S. Jpn. J.Art. Organs 1984,13, 614. Dunlop, E. H.; Hughes, R. D.; Williams, R. Med. Biol. Eng. Comput. 1978,16,343. Berg, M. J.; Berlinger, W. G.; Goldberg, M. J.; Spector, R.; Johnson, G. F. N. Eng. J . Med. 1982,307,642. Neuvonen, P. J.; Olkkola, K. T. Eur. J. Clin. Pharmacol. 1984, 26, 761.
358 I Journal of Pharmaceutical Sciences Vol. 79, No. 4, April 1990
5. Hillman, R. J.; Prescott, L. F. Br. Med. J . 1985,291, 1472. 6. Adler, L. J.; Waters, D. H.; Gwilt, P. R. Biopharm. DrugDispos. 1986, 7, 421. 7. Honda, Y.;Nakano, M.; Nakano, N. I. Chem. Pharm. BuZ1.1986, 34, 4385. 8. Ha iwara, A.; Takahashi, T.; Lee, R.; Ueda, T.; Takeda, M.; Itoh, T. i k i t a J. Med. (Japan) 1985,11, 581. 9. Abe, I.; Kamaya, H.; Ueda, I. J . Pharm. Sci. 1988, 77, 166. 10. Judis, J. J. Pharm. Sci. 1985, 74, 476. 11. Abe, I.; Hirashima, T. Kagaku To Kogyo (Japan) 1987,61,82. 12. Gessner, P. K.; Hasan, M. M. J . Pharm. Sci. 1987, 76, 319. 13. Higuchi, T.; Havinga, A.; Busse, L. W. J . Am. Pharm. Assoc. 1950,39,405. 14. Jossens, L.; Prausnitz, J. M.; Fritz, W.; Schliinder, E. U.; Meyers, A. L. Chem. Eng. Sci. 1978,33, 1097. 15. Adamson, A. W. Physical Chemistty of SuTfaces,4th ed.; Wiley: New York, 1982; p 373. 16. Abe, I.; Hayashi, K.; Kitagawa, M. Bull. Chem. SOC.Jpn. 1981, 54,2819. 17. Abe, I.; Hayashi, K.; Kitagawa, M. Bull. Chem. Soc. Jpn. 1982, 55, 687. 18. Abe, I.; Hayashi, K.; Kitagawa, M. Carbon 21, 1983, 189. 19. Kamaya, H.; Hayes, J. J., Jr.; Ueda, I. Anesth. Analg. 1983,62, 1025. 20. Andersen, A. H. Acta. Pharmacol. 1947,3, 199. 21. Abe, I. Kagaku To Kogyo (Japan) 1981,55,165. 22. Kakemi, K.; Arita, T.; Hori, R.; Konishi, R. Chem. Pharm. Bull. 1967,15, 1705.
Acknowledgments This stud was supported by the Medical Research Service of the Veterans Azministration and NIH grants GM25716, GM26950, and GM27670.
Received April 4,1988, from the * Depafiment of Anesthesia, University of Utah School of Medicine, Salt Lake City, UT 84732, and *Anesthesia Service, Veterans Administration Medical Center, Salt Lake City, UT84748. Accepted for publication July 7, 1989. §On leave from the Osaka Municipal Technical Research Institute, 1-6-50Morinomiya, Joto-Ku, Osaka 536, Japan. Abstract 0 We have shown that adsorbability of local anesthetics onto activated carbon, expressed by the partition coefficients at infinite dilution,correlated well with the pharmacologicalactivity. However,there is no parameter that can singly express the tendency to be adsorbed. Adsorbability is a loosely defined term and its meaning varies with the adsorption model. This study showed that the logarithm of the adsorbed amount of drugs was linearly related to the logarithm of the free drug concentration, in conformity to the Freundlich adsorption isotherms.The slope of the double logarithmic plot is expressed by 1lNin the Freundlich equation and is considered to be inversely related to the drug affinity to the adsorbent. The slope was used to evaluate the tendency to be adsorbed, or "adsorbability"of seven aromatic amine local anesthetics. Phenobarbital was included to compare anionic drugs in contrast to the cationic local anesthetics. The slopes were nearly equal between the cationic and neutral local anesthetics. Apparently, the lower hydrophobicity of the cationic forms is compensated by the electrostatic attraction from the negative charges present on the activated carbon surface. With phenobarbital, the slope value of the anionic form was larger than the neutral form. The lower affinity of the anionic form may be caused by the electrostatic repulsion. The molecular size parameters (i.e., molecular weight, molar refraction, and parachor) showed a linear relationship to the slope values. It may be possible to estimate the affinity-related slope values from these parameters.
Activated carbon is now widely used in environmental engineering for waste water treatments and as an adsorbent for hazardous gases. In health sciences, its applications include artificial liver,l,z treatment for drug poisoning,3-7 and complex formation with anticancer drugs to be deposited into malignant tissues for their prolonged release.8 Recently, we9 have successfully correlated activated carbon adsorbability (expressed by the partition coefficient of drugs a t infinite dilution) with drug potencies as a parameter for the quantitative structureactivity relationship (QSAR).Activated carbon adsorbability is particularly effective in analyzing the QSAR of charged drugs where the conventional oil-water partition coefficients fail to correlate with the drug activities. The adsorbability represents the amphipathic property of drugs that have limited solubility into the organic phase. We hope that the activated carbon system will contribute to future drug design. Studies on drug adsorption onto the interfacial region may contribute to an understanding of the action mechanisms. Although the term "adsorbability" intuitively conveys the concept on the tendency of a solute to be adsorbed onto activated carbon surfaces, it is not well defined. There is no ideal parameter that can singly represent this property. Hence, adsorbability can be expressed in a number of ways, and each has different meaning. In our previous communications we expressed carbon adsorbability by extrapolating the adsorption isotherms to infinite dilution of the ligand concentration in the bulk solution. Because therapeutic drug concentrations are not high, the infinite dilution parameter 354 I Journal of Pharmaceutical Sciences Vol. 79, No. 4, April 7990
meets the purpose of QSAR analyses. The ratio between the adsorbed and free drug concentrations converges into a constant value a t dilute concentrations. However, the partition coefficient starts to vary when the drug concentrations are increased. The change in the partition coefficient by the increase in the concentration varies with each drug. As will be shown in this study, the order of the adsorbed amounts of two drugs reverses a t higher drug concentrations. For problems in the high concentration range, such as for the treatment of drug poisoning, the infinite dilution parameter may not be suitable for describing the adsorbability . The present study is aimed at elucidating the adsorption pattern a t finite anesthetic concentrations. Judislo averaged the adsorbed amounts a t various drug:carbon ratios. Standard methods, however, consist of constructing adsorption isotherms and then analyzing the adsorption curve by an appropriate model. Carbon surface adsorption often follows the Freundlich adsorption isotherms,llJZ in which the logarithm of the adsorbed amount bears a linear relationship with the logarithm of the free drug concentrations. In this study, the linearity was confirmed, and the slope of the double logarithmic plot was compared among aromatic amine local anesthetics. Phenobarbital was included because the drug has nerve-blocking activity and ionizes into an anionic form, in contrast to the standard local anesthetics that ionize into a cationic form. In alkaline media, the ester-type local anesthetics undergo nonenzymatic hydrolysis. Despite the early demonstrations of this alkaline hydrolysis,13the problem tends to be overlooked recently. Hence, we included the effect of pH on the hydrolysis of procaine in some detail.
Experimental Section The activated carbon was the same as previously reported9 and was a single batch of 200-350 mesh CAL-activated carbon (Calgon, Pittsburgh, PA) with a surface area of 1010 cm2/g of carbon9 when measured by the adsorption isotherms of nitrogen gas at 77 K according to the Brunauer-Emmett-Teller theory with a Shibata P-700 Surface-Area Analyzer (Shibata Chemical Instruments, Tokyo, Japan). The carbon was washed with hot distilled water and dried at 110 "Cfor 2 days and then was kept in a desiccator containing silica gel. Water was purified by distillation, followed by treatment with activated carbon and ion-exchanger columns, and ultrafiltration using a Millipore system (Bedford, MA). The specific resistivity was maintained above 16 Mohm * cm-'. The following local anesthetics were obtained from Sigma Chemical (St. Louis, MO) in a crystalline form of hydrochloride salts: procaine, tetracaine, and dibucaine. Lidocaine, mepivacaine, and chloroprocaine were gifts from Astra Pharmaceutical Products (Worcester, MA), Sterling Drug (Rensselaer, NY), and Pennwalt (Rochester, NY), respectively. These anesthetics were also in crystalline forms of hydrochloride salts. Anesthetic adsorption was estimated at neutral and alkaline pH.
0022-3549/90/0400-0354$0 l.OO/O 0 1990, American Pharmaceutical Association
The pH was adjusted with a 0.1 M phosphate buffer (KH,PO, and Na,HPO,). The strongly alkaline pH 11was adjusted by the addition ofO.1 M NaOH. The presence of the buffer did not affect the adsorption of local anesthetics on the activated carbon surface. The bulk pH was measured by a Radiometer Ion 85 Analyzer (Westlake, OH) and a combination glass electrode. The kinetics of the nonenzymatic alkaline hydrolysis of the ester type local anesthetic, procaine, was studied at 1 mM concentration, adjusting the pH to various values at 25 "C. The adsorption isotherms were obtained by adding various quantities of CALactivated carbon to 50-mL sample tubes. An aliquot of 40 mL of the solution of known solute concentration was added to the sample tube and tightly capped. The tubes were shaken in a water bath maintained a t 25 "C for 20 h. Preliminary experimenta showed that adsorption was essentially complete within 20 h. After equilibration, a sample was taken from each tube and was filtered through a 0.45-pm pore size membrane to remove any suspended carbon. The anesthetic content in the filtrate solution was measured by ultraviolet absorbance using a Perkin-Elmer 554 spectrophotometer (Norwalk, CT) in a stoppered 10.0-mmlight path cuvette. The cuvette temperature was maintained at 25 "C by an electronically controlled Peltier heat exchanger. Because the molar absorptivity of aromatic amine local anesthetics varies according to the pH, calibration curves were constructed at each pH value. The amount of solute adsorbed on the carbon a t equilibrium was then calculated as the difference between the original and the filtrate concentrations.
Results and Discussion Nonenzymatic Hydrolysis of Procaine-Figure 1 shows the time-dependent spectral changes of procaine in a pH 11 solution a t 25 "C. There were two isosbestic points. The extent of hydrolysis was estimated from the change in the main absorbance peak at 287 nm, and the hydrolysis rate is expressed by the following equations:
(1) t112
(2)
= In (2/k)
where C , is the concentration of unhydrolyzed procaine in the solution after t hours, C , is the initial procaine concentration, and k is the rate constant. Figure 2 shows the hydrolysis rate of procaine at several pH values a t 25 "C. The rate constant (k)and the half-life period (t,,z)at each pH are shown in Table I. Below pH 8, procaine hydrolysis was almost negligible. Above pH 8,however, the hydrolysis rate steeply increased with the increase in pH values. Adsorption Rate-The time required to obtain equilibrium was estimated from the adsorption rate of procaine onto activated carbon a t pH 7.0.Table I1 shows the degree of attainment of equilibrium with time. A 1-h reaction time was found to be sufficient for assessment of the adsorption isotherm a t high pH. Table I shows that 6%ofthe initial procaine content is hydrolyzed in 1 h at pH 11. The adsorption data were corrected for the hydrolysis. Adsorption Isotherms-Adsorption isotherms of eight drugs from aqueous solution onto CAL-activated carbon were measured a t neutral and alkaline pH. The adsorption isotherms were expressed by the double logarithmic plot according to the following Freundlich model:
or
log X = log K + (l/N) log C
(4)
where X is the amount of solute adsorbed (mg per g of carbon), C is the equilibrium solute concentration in the bulk (mg/L), and K and 1/N are the Freundlich constants. The anesthetic
0.4
0
0.3
-
w
0 2 U m U
%m
0.2
n
-
0
e
G -
a
Y
-1
0)
0
220 240 260 280 300 320 340 WAVE LENGTH (nm) Flgure 1-Alkaline hydrolysis of ester type local anesthetics. Timedependent ultraviolet spectra of procaine at pH 11 and 25 "C.The spectrum was measured after the solution was diluted to 0.02 mM with pH 7 buffer solution. The initial procaine concentrationwas 1 .O mM. Key: 23 h; and (E) 70 h. Two isosbestic (A) control (0 h); (6)2.5 h; (C) 5.5 h; (0) points are apparent.
-2 0
20
40
TIME
60
(hour)
Flgure 2-Effect of pH on the hydrolysis of procaine at 25°C. The procaine concentration was 1.0 mM. The data are plotted on the semilogarithmicscale according to eq 2. The hydrolysis rates are shown in Table II. Key to pH values: (A) 5.98; (0)6.96; (A)7.99; (V)8.95; (V) 9.57; and (0)10.96. Journal of Pharmaceutical Sciences I 355 Vol. 79, No. 4, April 1990
Table CNonenzymatlc Alkallne Hydrolysls of Procalne at 25 "C
PH
Rate Constant (k), h-'
Half-life (t,,2), h
5.98 6.96 7.99 8.95 9.57 10.95
0 0.000 0.002 0.018 0.039 0.060
-
3001
400 39 18 12 h
0) \
Table Il-Adsorption pH 7.00 and 25 "C
Rate of Procaine onto Activated Carbon' at
E"
Y
X
Adsorption Time
c, -c,/c,-C,b
10 min 30 min l h
0.938 0.976 0.987 0.996 0.999 1.ooo
2h 4h 24 h
a The activated carbon concentration was 1250 mg1L. C,: the initial concentration of procaine (1.O mM); C,:the concentration of procaine remaining in the solution at time 1; C,: the equilibrium concentration of procaine.
concentration was expressed by the free base weight. Figure 3 is the log-log plot for the adsorption isotherms measured a t neutral pH. The adsorption constants and correlation coefficients were estimated by linear regression analysis and are shown in Table 111. The goodness of the fit was evaluated by the absolute relative percent deviation ( R D P according to the following equation, and is included in the table:
where Xcelis the calculated adsorbed value according to eq 4 and X,,, is the experimental value. In this table, dibucaine data a t pH 11.0 are omitted because the data points were not enough to estimate the parameters. The Freundlich adsorption constant, lIN, is a dimensionless parameter and is said to be related to the intensity of drug adsorption.15 The value of 1/N decreases with increasing intensity. Figure 4 shows a linear relation between 11N and molecular weight, where N is related to the affinity and increases with the increase in the molecular weight. The plot for phenobarbital, however, deviated from this relationship.
I
I
I
I
0.1
1
10
100
C (rng/L) Flgure >Adsorption isotherms of local anesthetics on CAL-activated carbon at pH 7.00 and 25 "C. Key: (0)procaine; (0)mepivacaine; (A) lidocaine; (A)chloroprocaine; (V) tetracaine; (V)phenobarbital; (0) benzocaine; and (B) dibucaine. The adsorption isotherms of mepivacaine were measured at pH 6.60. The abscissa is the equilibrium bulk anesthetic concentration, and the ordinate is the amount adsorbed to the carbon.
This deviation is probably caused by the difficulty of adhesion of the phenobarbital molecules to the carbon surface because the barbiturate ring and benzene ring are not aligned on the same plane. Abe et a1.16-18 demonstrated that there exist linear relationships between 11N and molecular weight or molecular refraction in homologous series, such as alkanols, esters, amino acids, saccharides, etc. Table IV lists these physical parameters (@). The correlation between these parameters and 1/Nis analyzed by the following equation and the result is shown in Table V:
11N = a@+ p
(6)
In this table, s is the standard deviation, F represents the overall goodness of the fit, and t is the significance level of the
Table Ill-Adsorptlon Constants for the Freundllch Equatlon
Drug 1 Procaine
Procaine 3 Lidocaine 4 Tetracaine 5 Dibucaine 0 Dibucainea 7 Mepivacaine 8 Mepivacaine 9 Chloroprocaine 10 Benzocaine 11 Phenobarbital 12 Phenobarbital 2
PHb
nc
7.00
8 6 8 7 5 2 13 9 7 7 6 9
1 1.oo
7.00 7.00 7.00 11 .oo 6.60 8.60 7.00 7.00 7.00 9.00
1INe
r'
RD, %g
143 204 167 229 245
0.0994 0.1 05 0.117 0.0649 0.0230
0.998 0.991 0.997 0.998 0.987
0.751 2.49 0.748 0.578 0.472
145 185 195 165 117 59.6
0.101 0.0963 0.0838 0.157 0.192 0.236
0.995 0.995 0.990 0.998 0.995 1.ooo
1.35 1.10 1.70 1.64 1.86 0.705
K, mglgd
-
-
-
-
Dibucaine data at pH 11.OO were not estimated because the number of the data points was insufficient for calculation. The pH values where the experiments were performed. Number of data points. Freundlich constant (mg/g). Freundlich constant (dimensionless). 'Correlation coefficient. g Absolute relative percent deviation. 356 I Journal of Pharmaceutical Sciences Vol. 79, No. 4, April 1990
0.~~1 0.20
5
t'
0 12
/
0 11
0.15
l -
2000.10 ' 150-
0.05
0' 100
I
300
400
Mw Flgum &Relation between the Freundlich adsorption constant, 1 IN, and the molecular weight, Mw. The numbers correspond to the compounds listed in Table 111. Table IV-Physlcal Constants of Local Anesthetlcs
Drug
Procaine Chloroprocaine Lidocaine Tetracaine Dibucaine Mepivacaine Benzocaine Phenobarbital
3
100
I
200
0.1
4)'
Mwh
Mr' Pr'
(Ib
Flgure 5-Effect of pH on adsorption of local anesthetics on activated carbon. Key: (0)procaine; (V) dibucaine; and (0)mepivacaine. The closed symbols are pH 1 1 .O and the open symbols are pH 7.00.For mepivacaine, the closed symbols are pH 8.60and the open symbols are
pH 6.60.
PKa
MwC
Mrd
-
9.05' 8.97' 7.92' 8.46' 8.73' 7.77' 2.65' 7.46
236.3 270.8 234.3 264.4 243.5 246.3 165.2 232.2
67.680 72.514 70.734 77.166 101.532 73.326 45.294 59.128
569.8 609,4 595.7
PP
g: :; 604.9 382.7 463.4
that of the uncharged form10.20 because the charged molecules are more soluble in water. The Freundlich constant, K, of the charged species was lower than that of the uncharged species (Table 111). On the other hand, the 1IN values were nearly equal for both types. The nearly equal affinity between charged and uncharged forms, despite their difference in hydrophobicity, may be
250
of the Regrecrslon Analysis of Equatlon 6 p"
rd
se
100
C (mg/L)
'Reference 19. Reference 22. Molecular weight. Molar refraction. Parachor. Table V-Reeutts
10
1
F'
-7.68*10-4 0.286 0.975 8.67~10'~135 -2.43. 0.270 0.956 0.0114 74.9 0.298 0.952 0.0120 -3.43. 67.3
1
200
-
150
-
100
-
P
re
11.6 8.65 8.20
'Physical constants. Regression coefficient. Constant. Correlation constant. Standard deviation. 'Overall goodness of the fit. 0 Significance level of the regression coefficienta. Molecular weight. ' Molar refraction. Parachor. regression coefficient a. A high correlation coefficient is demonstrated for the molecular size parameters and adsorption. These results indicate that a linear relationship holds between the physical parameters and 11N for local anesthetics. It may be possible to estimate the slope value of the adsorption isotherms from these physical parameters.
0)
\
E"
Y
X
50
I
I
Journal of Pharmaceutical Sciences I 357 Vol. 79, No. 4, April 1990
prompted by the negative charges present at the activated carbon surface that electrostatically attract cationic drugs. The presence of small amounts of oxides in activated carbons is responsible for the surface negative charges. Conversely, the adsorption of negatively charged molecules, such as barbiturates, may be suppressed by the presence of surface negative charges. Phenobarbital is negatively ionized in alkaline solution because of lactam (“keto”) and lactim (“enol”)tautomerization. Figure 6 shows the effect of pH on the adsorption of phenobarbital. As expected, the adsorbed amount of phenobarbital a t high pH was lower than that at low pH in the limit of the present experimental condition. The 1/N value of the anionic form (high pH) was larger than the uncharged form (low pH), indicating that the affinity of the anionic form is weaker than that of the uncharged form. Similar phenomena were observed for the adsorption of anionic and cationic surfactants of similar molecular weight.21 The cationic surfactants showed lower 1/N values than the anionic surfactants.21
References and Notes Kawanishi, H.; Nishiki, M.; Ezaki, H.; Yamane, S.; Tsuchiya, T.; Sugiyama, M.; Cho, T.; Kimura, S. Jpn. J.Art. Organs 1984,13, 614. Dunlop, E. H.; Hughes, R. D.; Williams, R. Med. Biol. Eng. Comput. 1978,16,343. Berg, M. J.; Berlinger, W. G.; Goldberg, M. J.; Spector, R.; Johnson, G. F. N. Eng. J . Med. 1982,307,642. Neuvonen, P. J.; Olkkola, K. T. Eur. J. Clin. Pharmacol. 1984, 26, 761.
358 I Journal of Pharmaceutical Sciences Vol. 79, No. 4, April 1990
5. Hillman, R. J.; Prescott, L. F. Br. Med. J . 1985,291, 1472. 6. Adler, L. J.; Waters, D. H.; Gwilt, P. R. Biopharm. DrugDispos. 1986, 7, 421. 7. Honda, Y.;Nakano, M.; Nakano, N. I. Chem. Pharm. BuZ1.1986, 34, 4385. 8. Ha iwara, A.; Takahashi, T.; Lee, R.; Ueda, T.; Takeda, M.; Itoh, T. i k i t a J. Med. (Japan) 1985,11, 581. 9. Abe, I.; Kamaya, H.; Ueda, I. J . Pharm. Sci. 1988, 77, 166. 10. Judis, J. J. Pharm. Sci. 1985, 74, 476. 11. Abe, I.; Hirashima, T. Kagaku To Kogyo (Japan) 1987,61,82. 12. Gessner, P. K.; Hasan, M. M. J . Pharm. Sci. 1987, 76, 319. 13. Higuchi, T.; Havinga, A.; Busse, L. W. J . Am. Pharm. Assoc. 1950,39,405. 14. Jossens, L.; Prausnitz, J. M.; Fritz, W.; Schliinder, E. U.; Meyers, A. L. Chem. Eng. Sci. 1978,33, 1097. 15. Adamson, A. W. Physical Chemistty of SuTfaces,4th ed.; Wiley: New York, 1982; p 373. 16. Abe, I.; Hayashi, K.; Kitagawa, M. Bull. Chem. SOC.Jpn. 1981, 54,2819. 17. Abe, I.; Hayashi, K.; Kitagawa, M. Bull. Chem. Soc. Jpn. 1982, 55, 687. 18. Abe, I.; Hayashi, K.; Kitagawa, M. Carbon 21, 1983, 189. 19. Kamaya, H.; Hayes, J. J., Jr.; Ueda, I. Anesth. Analg. 1983,62, 1025. 20. Andersen, A. H. Acta. Pharmacol. 1947,3, 199. 21. Abe, I. Kagaku To Kogyo (Japan) 1981,55,165. 22. Kakemi, K.; Arita, T.; Hori, R.; Konishi, R. Chem. Pharm. Bull. 1967,15, 1705.
Acknowledgments This stud was supported by the Medical Research Service of the Veterans Azministration and NIH grants GM25716, GM26950, and GM27670.
