- Email: [email protected]
N-Methyl-2-Pyrrolidone as a Cosolvent: Relationship of Cosolvent Effect with Solute Polarity and the Presence of Proton-Donating Groups on Model Drug Compounds
BMethyl-2-pyrrolidone as a Cosolvent: Relationship of Cosolvent Effect with Solute Polarity and the Presence of Proton-Donating Groups on Model Drug Compounds RALPHTAR ANTI NO^, EDMUNDBISHOP,FANG-CHUNG CHEN, KHURSHIDIQBAL',
AND
A. WASEEM MALICK
Received March 1, 1993, from the Pharmaceutical Research and Development, Hofhann-La Roche, Inc., Nutley, NJ 07110. for publication May 4, 1994@. *Current address: R. W. Johnson Pharmaceutical Research Institute, Raritan, NJ 08876. Abstract 0 KMethyl-2-pyrrolidone(methylpyrrolidone), a cosolvent which has been used in veterinary medicine and in transdermal delivery devices, was investigated as a cosolvent for model drug compounds of widely varying polarity. These compounds were digoxin, sulfamethoxazole, hydrocortisone acetate, theophylline, phenytoin, and reserpine. Methylpyrrolidonewas found to be an extremely efficient cosolvent for low solubility polar drugs such as digoxin or drugs containing multiple proton-donating groups such as phenytoin. The increase in solubility observed in aqueous solutions of digoxin and phenytoin to which 0.2 volume fraction of methylpyrrolidone was added was 500x and 65x, respectively. Significant deviations from log-linear solubilization were observed with digoxin, sulfamethoxazole, phenytoin, and reserpine, indicating significant water-solute-cosolvent interactions.
Introduction Methylpyrrolidone is a water-miscible, relatively polar, aprotic solvent that is thermally stable and has a high boiling point (202 "C).' Due to these properties, its major commercial use has been as a chemical reaction medium. It has been used in pharmaceuticals to enhance the percutaneous absorption of drugs2v3and in veterinary injections of tetracycline derivat i v e ~ Methylpyrrolidone .~ has an acute oral LD50 in rats of 7 cm3/kg.5 The intraperitoneal LDsOin the mouse and rat have been reported as 2.4 and 4.3 gkg, respectively.6 Lee et al. has provided a comprehensive summary of the toxicological studies of methylpyrr~lidone.~It is apparent from these studies that 1 mL of a solution with 0.2 volume fraction methylpyrrolidone administered to a rat of average weight (250 g) is within the guideline proposed by Bartsch and so may also be suitable for use as a vehicle for pharmacokinetic or toxicological evaluation of drugs in animals. A methylpyrrolidone volume fraction of 0.2 is therefore considered to be a "benchmark" concentration for evaluating the usefulness of the cosolvent in this investigation. The use of methylpyrrolidone as a cosolvent for several model drugs of widely varying polarity was investigated. These drugs were digoxin, sulfamethoxazole, hydrocortisone acetate, theophylline, phenytoin, and reserpine. All of the compounds except theophylline are practically insoluble in water. The solubility of theophylline in water is approximately 10 mg/mL.8 It has previously been showngthat for many solutes a log-linear increase in solubility occurs with increasing volume fraction of cosolvent. Equation 1describes this log-linear relationship: log
s, = fllog S,) + (1 - fllog s,
(1)
In eq 1,S , is the solubility of drug in a water-cosolvent mixture where the volume fraction of the cosolvent is f. S, and S, are the solubilities of the drug in the cosolvent and ~~~
@Abstractpublished in Advance ACS Abstracts, August 15,1994.
0 1994, American Chemical Society and American Pharmaceutical Association
Accepted
water respectively. Yalkowskygshowed that by assuming the properties of a solute-water-cosolvent system can be determined by a linear combination of the calculated solubilities of a solute in the cosolvent and water separately, eq 1takes an the form
log S, = log S,
+ af
(2)
where al is the solubilizing power of the cosolvent. Furthermore, he found that for many drugs exhibiting nonlinear solubility curves, the empirical relationship log
s, = log b, + b f +
(3)
where bl and b2 are constants, appears to be valid. Solubility data for the model drug compounds in this study were fit to eqs 2 and 3. log PCd, values have been successfully used for a number of compounds to estimate aqueous solubility1° and the slopes of solubilization curves have been shown to be related t o PC,,,,.9 Estimated log PC,,w values were calculated by the method of Nys and Rekkerl1J2and used as a polarity index.
Experimental Section Materials-N-Methyl-2-pyrrolidone(99.5%)(M-PYROL) was supplied by GAF (Wayne, NJ). Reserpine (percent compound purity, 99%; maximum absorbance wavelength, 216 nm), digoxin (97%,220 nm), and phenytoin (99%,226 nm) were obtained from Aldrich Chemical Co. (Milwaukee, WI). Theophylline (99%,270 nm), hydrocortisone acetate (98%,244 nm), and sulfamethoxazole (99%, 256 nm) were obtained from Sigma Chemical Co. (St. Louis, MO). All materials were reagent grade and used as received. Solubility Determinations-Methylpyrrolidone and water components of the mixtures were measured as separate volumes using volumetric glass pipettes and dispensed into borosilicate glass vials. The volume fraction of methylpyrrolidone in the test solutions were 0,0.2,0.4,0.6,0.8,or 1.0. Each drug was added to methylpyrrolidone/ water mixtures in 15-mL vials until complete saturation occurred and there was an excess of undissolved material. Saturated equilibrium was found to occur a t varying times of 2-10 h depending on the drug used and the volume fraction of methylpyrrolidondwater mixture. To assure that all drugs were at saturated equilibria and handled in a consistent manner, the saturated solutions were equilibrated for 24 h at 25 4z 0.5 "C in a shaking water bath (Tecator 1024, Fisher Scientific, Fairlawn, NJ). Each sample was then filtered through a 0.45-pm Millipore Durapore filter (Millipore Products, Inc., Bedford, MA). Ultraviolet absorbance (Hewlett-Packard Diode Array Spectrophotometer 8425A, Hewlett-Packard Co., San Diego, CA) was used to determine drug concentration in the filtrate after dilution in a miscible solvent that did not compete with the ultraviolet maximum absorbance range for the drug of interest. These diluted samples were compared against linear standard solutions of the drugs dissolved at appropriate concentrations in the miscible solvents to be within the W detectable range. Methylpyrrolidone/water mixtures used (corresponding to the proper volume fractions) without drug and diluted into solvent were used as blanking controls. Samples were run in triplicate.
0022-3549/94/1200-1213$04.50/0
Journal of Pharmaceutical Sciences / 1213 Vol. 83, No. 9, September 1994
PC and Relevant Solubility Data for Model Drug Compounds in MethylpyrrolidoneMTater Mixtures
Table 1-log
estimated log PC
drug
-3.94 -1.42 1.31 1.85 2.14 5.25
Digoxin Sulfamethoxazole Hydrocortisoneacetate Theophylline Phenytoin Reserpine a
log SdS, at vt = 0.2
SdSaat
vt = 0.2
Proton-Donating Groups per Molecule
Proton-DonatingGroups per mg x 10-l8
bi
1.85 0.63 -0.08 0.097 1.03 -0.36
500 14 8 2 65 2
6 3 3 1 2 1
4.62 7.14 4.47 3.34 4.76 0.90
17.2 5.5 1.9 1.o 7.4 0.1
S, = predicted solubility, So = observed solubility, S, = aqueous solubility. 1000
T
loooo
T
100
10
E r L 1
0.1
-I
0.01
0.0
I
,
I
1
0.2
0.4
0.6
0.8
I 1 .o
Volume Fraction Methylpyrrolidone Figure 1-Solubility curve for digoxin in methylpyrrolidone/water mixtures. Error bars represent standard deviation (n = 3).
'Oo0
0.0
c-
0.2
0.0
0.2
I I I I 0.4 0.6 0.8 1 .o Volume Fraction Methylpyrrolidone Figure 3-Solubility curve for hydrocortisone acetate in methylpyrrolidone/water mixtures. Error bars represent standard deviation (n = 3).
0.01
I I
T
0.1
0.0
0.2
0.4 0.6 0.8 Volume Fraction Methylpyrrolidone
1 .o
Figure 2-Solubility curve for sulfamethoxazolein methylpyrrolidonehater mixtures. Error bars represent standard deviation (n = 3). Estimated log PCOlwCalculation-Estimated log PCo,wvalues were calculated1'J2 a s an index of polarity. The Nys-Rekker method involves a contribution scheme where each substituent group on a molecule is assigned an "f" number. The sum of all *f" numbers plus correction factors due to proximity effects of substituent groups is an estimate of log PC0/,. Statistical Analysis-Best fit plots and curves were determined with SigmaPlot version 5.0(Jandel Scientific, Corte Madera, CA).
Results and Discussion Solubility curves for the compounds under investigation appear in Figures 1-6. The solubility increase of each compound when methylpyrrolidone is added to water a t a volume fraction of 0.2 is shown in Table 1 along with other relevant data. 1214 / Journal of Pharmaceutical Sciences Vol. 83, No. 9, September 1994
I
I
0.4
0.6
I
0.8
I 1 .o
Volume Fraction Methylpyrrolidone Figure 4-Solubility curve for theophylline in rnethylpyrrolidoneiwater mixtures. Error bars represent standard deviation (n = 3).
As shown in Table 1, methylpyrrolidoneis a useful cosolvent for all the drugs tested. Increase in solubility ranged from approximately 2- to 500-fold a t a methylpyrrolidone volume fraction of 0.2 in water. Equation 2 fit well (correlation coefficient = 0.99) for each solubility curve except digoxin in which it was valid for only the monotonically increasing portion of the curve CV,less than or equal to 0.4). Except for phenytoin, there appears to be a relationship between the estimated log PC and the efficiency of methylpyrrolidone as a cosolvent. The bl parameter from eq 4 as well as the increase in solubility observed a t a methylpyrrolidone volume fraction of 0.2 indicate that, in general, the cosolvency effect of methylpyrrolidone increased as solute polarity increased. This is not unexpected since methylpyrrolidone itself is an extremely polar molecule as evidenced by a dipole moment of 4.09 Debye.] The ring structure of
1000
T
0.01
0.2 0.4 0.6 0.8 1 .0 Volume Fraction Methylpyrrolidone Figure 5-Solubility curve for phenytoin in rnethylpyrrolidone/water mixtures. Error bars represent standard deviation ( n = 3). 0.0
’Oo0
7
0.0
0.2
0.4
0.6
0.8
1 .o
solubilities in methylpyrrolidone/water mixtures. The solutes displaying the smallest interactions (hydrocortisone acetate, theophylline, and reserpine) were the least polar solutes and all had positive estimated log PC values. Reserpine, the least polar compound, exhibited a negative deviation. Phenytoin clearly did not follow this trend with a positive estimated log PC but also a significant positive deviation. Rubino and Yalkowsky13observed deviation from log-linear behavior in solubility plots for diazepam, phenytoin, and benzocaine in the polar solvents dimethyl sulfoxide, dimethylacetamide, and dimethylformamide. In these experiments the less polar diazepam (estimated log PC = 3.62) displayed a negative deviation from linearity in the polar solvents while benzocaine (estimated log PC = -0.38) and phenytoin displayed positive deviations. This phenomenon was attributed to the absence of proton-donating groups on diazepam to interact with dimethyl sulfoxide, dimethylacetamide, and dimethylformamide which all contain carbonyl groups and are thus good proton acceptors. Both benzocaine and phenytoin have primary and secondary amino groups, respectively, and thus are predicted to interact with dimethyl sulfoxide, dimethylacetamide, and dimethylformamide. Methylpyrrolidone, also containing a carbonyl, would be expected to behave similarly to dimethyl sulfoxide, dimethylacetamide, and dimethylformamide and in fact it did. As shown in Table 1,the compounds with multiple protondonating sites were more readily solubilized with methylpyrrolidone than those containing only one site for interaction. This is even more clearly demonstrated if the molecular weight differences are taken into account by calculating the ratio of proton-donating groups per milligram of each molecule (also shown in Table 1). Phenytoin on a weight basis has a relatively high number of proton-donating groups. This could account for its extremely high solubility in methylpyrrolidone/ water systems and positive deviation from linearity exhibited in its solubility plot.
Volume Fraction Methylpyrrolidone
Figure GSolubility curve for reserpine in methylpyrrolidoneiwatermixtures. Error bars represent standard deviation (n = 3). methylpyrrolidone facilitates the delocalization of electrons in the NC=O function resulting in a molecule more polar than water.14 This has been supported by kinetic studies of alkylation of malonic esters by alkyl halides in which methylpyrrolidone was used as the reaction medium.15 Examination of Figures 1-6 shows that there are significant deviations from linearity in the solubility curves of all compounds except hydrocortisone acetate and theophylline. These deviations may be attributed to changes in the solutewater-cosolvent mixture that do not occur in mixtures of the solute and the cosolvent or the solute and water. Many solutes reach a maximum solubility in cosolvent-solvent mixtures.I6-l8 It has also been demonstrated that some solutes exhibit peak solubilities at specific dielectric cons t a n t ~ In . ~fact, ~ multiple peaks have also been demonstrated for some solute-cosolvent-solvent mixtures.20 Such observations may be due to solvate formation or change in the crystalline form of the solute.20 The magnitude of deviation from linearity was also related to the estimated log PC, again with the exception of phenytoin. The ratio of the observed solubility (So)to the solubility predicted by eq 1 was calculated for each drug at Vf = 0.2 appears in Table 1. Digoxin, sulfamethoxazole, and phenytoin all displayed significant positive deviations in solubilization indicative of changes that would lead to solubility in excess of the predicted values. Digoxin and sulfamethoxazole were the most polar molecules in the study; both compounds have negative estimated log PC values and exhibited maxima
Conclusions Low solubility polar molecules or those with proton-donating groups are perhaps the best candidates for solubilization with methylpyrrolidone. The enormous increase in solubility observed for digoxin and phenytoin suggests that methylpyrrolidone may be used in low concentrations to effect solubilization of some drug compounds. In light of this, methylpyrrolidone may be more desirable than the more commonly used cosolvents propylene glycol and poly(ethylene glycol) for selected drug molecules. The consistent observed deviation from log-linear behavior also suggests that methylpyrrolidone participates in some yet to be defined solute-water-cosolvent interaction with many drug molecules.
References and Notes 1. GAF. M-PYROL Handbook; GAF, New York, 1972, p 5. 2. Sasaki, H.; Kojima, M.; Masaki, M.; Mori, Y.; Makamura, J.; Shibasaki. J. Int. J . Pharm. 1988. 44. 15-18. 3. Hoelgard,’A.; Mollgard, B.; Baker; E.’lnt. J . Pharm. 1988,43, 233-240. 4. Morgan, J. P.; Malook, S. U.; Boon, P. F. G. Belgium Patent 905 528 A l . 1987. 5. Reference 1, p 175. 6. Bartsch, W..;-Spooner, G.; Dietmann, K.; Fuchs, A. Arzneim. Forsch. 1976,26,1581-1591. 7. Lee, K. P.; Chromey, K. C.; Culik, R.; Barnes, J. R.; Schneider, P. W. Fundam. A p p l . Toxicol. 1987.9, 222-235. 8. Cohen, J. L. Theophylline. In Analytical Profiles of Drug Substances; Vol. 4; Florey, K., Ed.; Academic Press: New York, 1975; pp 466-493.
Journal of Pharmaceutical Sciences / 1215 Vol. 83, No. 9, September 1994
9. Yalkowsky, S. H.; Roseman, T. J. In Techniques of Solubilization of Drugs; Marcel Dekker: New York, 1981; Chapter 3. 10. Martin, A.; Newberger, J.; Adjei, A. J . Pharm. Sci. 1980, 69, 487 __ . -49 . - 1-. 11. Nys, C. G.; Rekker, R. F. Eur. J . Med. Chern. 1974,9,361-365. 12. Rekker, R. F.; de Kort, H. M. Eur. J . Med. Chern. 1979,9,479488. 13. Rubino, J. T.; Yalkowsky, S. H. Pharm. Res. 1987,4, 231-236. 14. CRC Handbook of Chemistry and Physics, 68th ed.; Weast, R. C., Ed.; CRC Press Inc.: Boca Raton, 1987; p 61. 15. Zaugg, H. E.; Horrom, B. W.; Borgwardt, S. J . Am. Chem. SOC. 1960;82, 2895-2903. 16. Krause, G. M.; Cross, J. M. J . Pharm. Sci.1951,40, 137.
1216 / Journal of Pharmaceutical Sciences Vol. 83, No. 9, September 1994
17. Peterson, C.; Hopponen, R. E. J . Pharm. Sci. 1953, 42, 541. 18. Ng, W. F.; Poe, C. F. J . Pharm. Sci. 1956, 45, 531. 19. Semenchenko, V. K. Vestnik Moskou Uniu. 1947, 5, 49; Chem. Abstr. 1948, 42, 3240. 20. Paruta, A. N.; Sciarrone, B. J.; Lordi, N. G. J . Pharm. Sci. 1965, 54.838-841.
Acknowledgments The authors sincerely wish t o thank Dr. Rodolfo Pinal for his comments and suggestions.
AND
A. WASEEM MALICK
Received March 1, 1993, from the Pharmaceutical Research and Development, Hofhann-La Roche, Inc., Nutley, NJ 07110. for publication May 4, 1994@. *Current address: R. W. Johnson Pharmaceutical Research Institute, Raritan, NJ 08876. Abstract 0 KMethyl-2-pyrrolidone(methylpyrrolidone), a cosolvent which has been used in veterinary medicine and in transdermal delivery devices, was investigated as a cosolvent for model drug compounds of widely varying polarity. These compounds were digoxin, sulfamethoxazole, hydrocortisone acetate, theophylline, phenytoin, and reserpine. Methylpyrrolidonewas found to be an extremely efficient cosolvent for low solubility polar drugs such as digoxin or drugs containing multiple proton-donating groups such as phenytoin. The increase in solubility observed in aqueous solutions of digoxin and phenytoin to which 0.2 volume fraction of methylpyrrolidone was added was 500x and 65x, respectively. Significant deviations from log-linear solubilization were observed with digoxin, sulfamethoxazole, phenytoin, and reserpine, indicating significant water-solute-cosolvent interactions.
Introduction Methylpyrrolidone is a water-miscible, relatively polar, aprotic solvent that is thermally stable and has a high boiling point (202 "C).' Due to these properties, its major commercial use has been as a chemical reaction medium. It has been used in pharmaceuticals to enhance the percutaneous absorption of drugs2v3and in veterinary injections of tetracycline derivat i v e ~ Methylpyrrolidone .~ has an acute oral LD50 in rats of 7 cm3/kg.5 The intraperitoneal LDsOin the mouse and rat have been reported as 2.4 and 4.3 gkg, respectively.6 Lee et al. has provided a comprehensive summary of the toxicological studies of methylpyrr~lidone.~It is apparent from these studies that 1 mL of a solution with 0.2 volume fraction methylpyrrolidone administered to a rat of average weight (250 g) is within the guideline proposed by Bartsch and so may also be suitable for use as a vehicle for pharmacokinetic or toxicological evaluation of drugs in animals. A methylpyrrolidone volume fraction of 0.2 is therefore considered to be a "benchmark" concentration for evaluating the usefulness of the cosolvent in this investigation. The use of methylpyrrolidone as a cosolvent for several model drugs of widely varying polarity was investigated. These drugs were digoxin, sulfamethoxazole, hydrocortisone acetate, theophylline, phenytoin, and reserpine. All of the compounds except theophylline are practically insoluble in water. The solubility of theophylline in water is approximately 10 mg/mL.8 It has previously been showngthat for many solutes a log-linear increase in solubility occurs with increasing volume fraction of cosolvent. Equation 1describes this log-linear relationship: log
s, = fllog S,) + (1 - fllog s,
(1)
In eq 1,S , is the solubility of drug in a water-cosolvent mixture where the volume fraction of the cosolvent is f. S, and S, are the solubilities of the drug in the cosolvent and ~~~
@Abstractpublished in Advance ACS Abstracts, August 15,1994.
0 1994, American Chemical Society and American Pharmaceutical Association
Accepted
water respectively. Yalkowskygshowed that by assuming the properties of a solute-water-cosolvent system can be determined by a linear combination of the calculated solubilities of a solute in the cosolvent and water separately, eq 1takes an the form
log S, = log S,
+ af
(2)
where al is the solubilizing power of the cosolvent. Furthermore, he found that for many drugs exhibiting nonlinear solubility curves, the empirical relationship log
s, = log b, + b f +
(3)
where bl and b2 are constants, appears to be valid. Solubility data for the model drug compounds in this study were fit to eqs 2 and 3. log PCd, values have been successfully used for a number of compounds to estimate aqueous solubility1° and the slopes of solubilization curves have been shown to be related t o PC,,,,.9 Estimated log PC,,w values were calculated by the method of Nys and Rekkerl1J2and used as a polarity index.
Experimental Section Materials-N-Methyl-2-pyrrolidone(99.5%)(M-PYROL) was supplied by GAF (Wayne, NJ). Reserpine (percent compound purity, 99%; maximum absorbance wavelength, 216 nm), digoxin (97%,220 nm), and phenytoin (99%,226 nm) were obtained from Aldrich Chemical Co. (Milwaukee, WI). Theophylline (99%,270 nm), hydrocortisone acetate (98%,244 nm), and sulfamethoxazole (99%, 256 nm) were obtained from Sigma Chemical Co. (St. Louis, MO). All materials were reagent grade and used as received. Solubility Determinations-Methylpyrrolidone and water components of the mixtures were measured as separate volumes using volumetric glass pipettes and dispensed into borosilicate glass vials. The volume fraction of methylpyrrolidone in the test solutions were 0,0.2,0.4,0.6,0.8,or 1.0. Each drug was added to methylpyrrolidone/ water mixtures in 15-mL vials until complete saturation occurred and there was an excess of undissolved material. Saturated equilibrium was found to occur a t varying times of 2-10 h depending on the drug used and the volume fraction of methylpyrrolidondwater mixture. To assure that all drugs were at saturated equilibria and handled in a consistent manner, the saturated solutions were equilibrated for 24 h at 25 4z 0.5 "C in a shaking water bath (Tecator 1024, Fisher Scientific, Fairlawn, NJ). Each sample was then filtered through a 0.45-pm Millipore Durapore filter (Millipore Products, Inc., Bedford, MA). Ultraviolet absorbance (Hewlett-Packard Diode Array Spectrophotometer 8425A, Hewlett-Packard Co., San Diego, CA) was used to determine drug concentration in the filtrate after dilution in a miscible solvent that did not compete with the ultraviolet maximum absorbance range for the drug of interest. These diluted samples were compared against linear standard solutions of the drugs dissolved at appropriate concentrations in the miscible solvents to be within the W detectable range. Methylpyrrolidone/water mixtures used (corresponding to the proper volume fractions) without drug and diluted into solvent were used as blanking controls. Samples were run in triplicate.
0022-3549/94/1200-1213$04.50/0
Journal of Pharmaceutical Sciences / 1213 Vol. 83, No. 9, September 1994
PC and Relevant Solubility Data for Model Drug Compounds in MethylpyrrolidoneMTater Mixtures
Table 1-log
estimated log PC
drug
-3.94 -1.42 1.31 1.85 2.14 5.25
Digoxin Sulfamethoxazole Hydrocortisoneacetate Theophylline Phenytoin Reserpine a
log SdS, at vt = 0.2
SdSaat
vt = 0.2
Proton-Donating Groups per Molecule
Proton-DonatingGroups per mg x 10-l8
bi
1.85 0.63 -0.08 0.097 1.03 -0.36
500 14 8 2 65 2
6 3 3 1 2 1
4.62 7.14 4.47 3.34 4.76 0.90
17.2 5.5 1.9 1.o 7.4 0.1
S, = predicted solubility, So = observed solubility, S, = aqueous solubility. 1000
T
loooo
T
100
10
E r L 1
0.1
-I
0.01
0.0
I
,
I
1
0.2
0.4
0.6
0.8
I 1 .o
Volume Fraction Methylpyrrolidone Figure 1-Solubility curve for digoxin in methylpyrrolidone/water mixtures. Error bars represent standard deviation (n = 3).
'Oo0
0.0
c-
0.2
0.0
0.2
I I I I 0.4 0.6 0.8 1 .o Volume Fraction Methylpyrrolidone Figure 3-Solubility curve for hydrocortisone acetate in methylpyrrolidone/water mixtures. Error bars represent standard deviation (n = 3).
0.01
I I
T
0.1
0.0
0.2
0.4 0.6 0.8 Volume Fraction Methylpyrrolidone
1 .o
Figure 2-Solubility curve for sulfamethoxazolein methylpyrrolidonehater mixtures. Error bars represent standard deviation (n = 3). Estimated log PCOlwCalculation-Estimated log PCo,wvalues were calculated1'J2 a s an index of polarity. The Nys-Rekker method involves a contribution scheme where each substituent group on a molecule is assigned an "f" number. The sum of all *f" numbers plus correction factors due to proximity effects of substituent groups is an estimate of log PC0/,. Statistical Analysis-Best fit plots and curves were determined with SigmaPlot version 5.0(Jandel Scientific, Corte Madera, CA).
Results and Discussion Solubility curves for the compounds under investigation appear in Figures 1-6. The solubility increase of each compound when methylpyrrolidone is added to water a t a volume fraction of 0.2 is shown in Table 1 along with other relevant data. 1214 / Journal of Pharmaceutical Sciences Vol. 83, No. 9, September 1994
I
I
0.4
0.6
I
0.8
I 1 .o
Volume Fraction Methylpyrrolidone Figure 4-Solubility curve for theophylline in rnethylpyrrolidoneiwater mixtures. Error bars represent standard deviation (n = 3).
As shown in Table 1, methylpyrrolidoneis a useful cosolvent for all the drugs tested. Increase in solubility ranged from approximately 2- to 500-fold a t a methylpyrrolidone volume fraction of 0.2 in water. Equation 2 fit well (correlation coefficient = 0.99) for each solubility curve except digoxin in which it was valid for only the monotonically increasing portion of the curve CV,less than or equal to 0.4). Except for phenytoin, there appears to be a relationship between the estimated log PC and the efficiency of methylpyrrolidone as a cosolvent. The bl parameter from eq 4 as well as the increase in solubility observed a t a methylpyrrolidone volume fraction of 0.2 indicate that, in general, the cosolvency effect of methylpyrrolidone increased as solute polarity increased. This is not unexpected since methylpyrrolidone itself is an extremely polar molecule as evidenced by a dipole moment of 4.09 Debye.] The ring structure of
1000
T
0.01
0.2 0.4 0.6 0.8 1 .0 Volume Fraction Methylpyrrolidone Figure 5-Solubility curve for phenytoin in rnethylpyrrolidone/water mixtures. Error bars represent standard deviation ( n = 3). 0.0
’Oo0
7
0.0
0.2
0.4
0.6
0.8
1 .o
solubilities in methylpyrrolidone/water mixtures. The solutes displaying the smallest interactions (hydrocortisone acetate, theophylline, and reserpine) were the least polar solutes and all had positive estimated log PC values. Reserpine, the least polar compound, exhibited a negative deviation. Phenytoin clearly did not follow this trend with a positive estimated log PC but also a significant positive deviation. Rubino and Yalkowsky13observed deviation from log-linear behavior in solubility plots for diazepam, phenytoin, and benzocaine in the polar solvents dimethyl sulfoxide, dimethylacetamide, and dimethylformamide. In these experiments the less polar diazepam (estimated log PC = 3.62) displayed a negative deviation from linearity in the polar solvents while benzocaine (estimated log PC = -0.38) and phenytoin displayed positive deviations. This phenomenon was attributed to the absence of proton-donating groups on diazepam to interact with dimethyl sulfoxide, dimethylacetamide, and dimethylformamide which all contain carbonyl groups and are thus good proton acceptors. Both benzocaine and phenytoin have primary and secondary amino groups, respectively, and thus are predicted to interact with dimethyl sulfoxide, dimethylacetamide, and dimethylformamide. Methylpyrrolidone, also containing a carbonyl, would be expected to behave similarly to dimethyl sulfoxide, dimethylacetamide, and dimethylformamide and in fact it did. As shown in Table 1,the compounds with multiple protondonating sites were more readily solubilized with methylpyrrolidone than those containing only one site for interaction. This is even more clearly demonstrated if the molecular weight differences are taken into account by calculating the ratio of proton-donating groups per milligram of each molecule (also shown in Table 1). Phenytoin on a weight basis has a relatively high number of proton-donating groups. This could account for its extremely high solubility in methylpyrrolidone/ water systems and positive deviation from linearity exhibited in its solubility plot.
Volume Fraction Methylpyrrolidone
Figure GSolubility curve for reserpine in methylpyrrolidoneiwatermixtures. Error bars represent standard deviation (n = 3). methylpyrrolidone facilitates the delocalization of electrons in the NC=O function resulting in a molecule more polar than water.14 This has been supported by kinetic studies of alkylation of malonic esters by alkyl halides in which methylpyrrolidone was used as the reaction medium.15 Examination of Figures 1-6 shows that there are significant deviations from linearity in the solubility curves of all compounds except hydrocortisone acetate and theophylline. These deviations may be attributed to changes in the solutewater-cosolvent mixture that do not occur in mixtures of the solute and the cosolvent or the solute and water. Many solutes reach a maximum solubility in cosolvent-solvent mixtures.I6-l8 It has also been demonstrated that some solutes exhibit peak solubilities at specific dielectric cons t a n t ~ In . ~fact, ~ multiple peaks have also been demonstrated for some solute-cosolvent-solvent mixtures.20 Such observations may be due to solvate formation or change in the crystalline form of the solute.20 The magnitude of deviation from linearity was also related to the estimated log PC, again with the exception of phenytoin. The ratio of the observed solubility (So)to the solubility predicted by eq 1 was calculated for each drug at Vf = 0.2 appears in Table 1. Digoxin, sulfamethoxazole, and phenytoin all displayed significant positive deviations in solubilization indicative of changes that would lead to solubility in excess of the predicted values. Digoxin and sulfamethoxazole were the most polar molecules in the study; both compounds have negative estimated log PC values and exhibited maxima
Conclusions Low solubility polar molecules or those with proton-donating groups are perhaps the best candidates for solubilization with methylpyrrolidone. The enormous increase in solubility observed for digoxin and phenytoin suggests that methylpyrrolidone may be used in low concentrations to effect solubilization of some drug compounds. In light of this, methylpyrrolidone may be more desirable than the more commonly used cosolvents propylene glycol and poly(ethylene glycol) for selected drug molecules. The consistent observed deviation from log-linear behavior also suggests that methylpyrrolidone participates in some yet to be defined solute-water-cosolvent interaction with many drug molecules.
References and Notes 1. GAF. M-PYROL Handbook; GAF, New York, 1972, p 5. 2. Sasaki, H.; Kojima, M.; Masaki, M.; Mori, Y.; Makamura, J.; Shibasaki. J. Int. J . Pharm. 1988. 44. 15-18. 3. Hoelgard,’A.; Mollgard, B.; Baker; E.’lnt. J . Pharm. 1988,43, 233-240. 4. Morgan, J. P.; Malook, S. U.; Boon, P. F. G. Belgium Patent 905 528 A l . 1987. 5. Reference 1, p 175. 6. Bartsch, W..;-Spooner, G.; Dietmann, K.; Fuchs, A. Arzneim. Forsch. 1976,26,1581-1591. 7. Lee, K. P.; Chromey, K. C.; Culik, R.; Barnes, J. R.; Schneider, P. W. Fundam. A p p l . Toxicol. 1987.9, 222-235. 8. Cohen, J. L. Theophylline. In Analytical Profiles of Drug Substances; Vol. 4; Florey, K., Ed.; Academic Press: New York, 1975; pp 466-493.
Journal of Pharmaceutical Sciences / 1215 Vol. 83, No. 9, September 1994
9. Yalkowsky, S. H.; Roseman, T. J. In Techniques of Solubilization of Drugs; Marcel Dekker: New York, 1981; Chapter 3. 10. Martin, A.; Newberger, J.; Adjei, A. J . Pharm. Sci. 1980, 69, 487 __ . -49 . - 1-. 11. Nys, C. G.; Rekker, R. F. Eur. J . Med. Chern. 1974,9,361-365. 12. Rekker, R. F.; de Kort, H. M. Eur. J . Med. Chern. 1979,9,479488. 13. Rubino, J. T.; Yalkowsky, S. H. Pharm. Res. 1987,4, 231-236. 14. CRC Handbook of Chemistry and Physics, 68th ed.; Weast, R. C., Ed.; CRC Press Inc.: Boca Raton, 1987; p 61. 15. Zaugg, H. E.; Horrom, B. W.; Borgwardt, S. J . Am. Chem. SOC. 1960;82, 2895-2903. 16. Krause, G. M.; Cross, J. M. J . Pharm. Sci.1951,40, 137.
1216 / Journal of Pharmaceutical Sciences Vol. 83, No. 9, September 1994
17. Peterson, C.; Hopponen, R. E. J . Pharm. Sci. 1953, 42, 541. 18. Ng, W. F.; Poe, C. F. J . Pharm. Sci. 1956, 45, 531. 19. Semenchenko, V. K. Vestnik Moskou Uniu. 1947, 5, 49; Chem. Abstr. 1948, 42, 3240. 20. Paruta, A. N.; Sciarrone, B. J.; Lordi, N. G. J . Pharm. Sci. 1965, 54.838-841.
Acknowledgments The authors sincerely wish t o thank Dr. Rodolfo Pinal for his comments and suggestions.