Journal of Controlled Release, 12 (1990) 133-141 Elsevier Science Publishers B.V.

, Amsterdam - Printed in The Netherlands

133

INFLUENCE OF MOLECULAR SIZE AND WATER SOLUBILITY OF THE SOLUTE ON ITS RELEASE FROM SWELLING AND EROSION CONTROLLED POLYMERIC MATRICES* K.V. Ranga Rao, K. Padmalatha Devi and P. Buri
School of Pharmacy, University of Geneva, 30 quai Ernest-Ansermet, (Received November 3,1988; Geneva 4 (Switzerland)

accepted in revised form August 14,1989)

Key words: oral release systems; hydrophylic matrix tablets; drug solubility; drug molecular size; matrix swelling; matrix erosion

The release of23 drugs of vario~ sol~~itks and mokc~r we~h~ thro~h males of hydroxypropylmethykellulose (HPMC; matrix I) and a mixture (1: 1) of HPMC and sodium carboxymethylcellulose (matrix II) was studied. Very soluble and soluble beta blockers and a freely soluble vitamin ~th~rni~ hydro~hkride) were rek~ed more skwly from matrix II than from ~tr~ I due to complex formation between the eationic drug and the anionic polymer. For 15 of the drugs, the release rates from matrix II were much higher than from matrix I owing to the rapid erosion of the swollen gel of ~trix II com~red to ~tr~ I, Increase in ~kcu~r we~ht (a granter that may be indirectly related to the mokcular size) decreased the rekase rate in a nonlinear manner from both the matrices for soluble (L-ephedrine and propranolol hydrochloride) and certain sl~htly soluble drugs ~~~pi~ and quini~ s~fate, as well as ~et~o~rn~e, p~~barbit~ and erythromycin), and from matrix I alone for very soluble beta blochers. For methyltestosterone, erythromycin and diazepam, the release rates from matrix I are nearly the same, although their ~k~u~r weights and sol~ilitks differ widely. When the release data (< 60%) was fitted to the simple power law equation, the mode of drug rekase from matrix I and matrix II was non-Fickian and super case II type, respectively, for sparingly soluble, slightly solubk and very slightly soluble drugs. The study reveakd that drugs falling into this category can be released at nearly zero-order rate thro~h HPMC matrices. Since these matrices are prepared by direct compression, cellulose matrices may be preferred for oral controlled release formulations. INTRODUCTION In recent years, much a~ntion has been devoted to the formulation of swelling controlled

*Presented at the 3rd European Congress of Biopharmaceutics and Pharmacokinetics, April 21-24,1987, Freiburg, F.R.G.

release matrix formulations. Various types of swellable polymers used as matrices [ 1,21,the mechanism of release of drug from such matrices and their modelling aspects have been reviewed [ 3-71. Among the various hydrophilic polymers, cellulose ethers may be preferred for making matrix formulations intended for oral administration because they are directly compressible, they can a~cornrno~~ a relatively

0168-3659/~/$03.50

0 1990 Elsevier Science Publishers B.V.

134

high percentage of the drug, they hydrate rapidly at body temperature and the processing variables have minimum influence on the release of drug. The release of freely soluble drugs through hydroxypropylmethylcellulose (HPMC) matrices is known to follow the square-root of time relationship due to an increase in diffusional pathlength for the drug with time [8-111. By optimizing the swelling and erosion rates of the matrices, i.e., by mixing HPMC, hydroxypropylcellulose (HPC ) or methylcellulose (MC) with sodium carboxymethylcellulose (NaCMC) in an optimum ratio, certain very soluble drugs were released at a nearly zero-order rate [ 12-161. Various factors influencing the release of drugs through cellulose matrices have been reviewed [4,5]. Diffusion coefficients of drugs through membranes were found to decrease linearly as the square of solute radius increased [ 17,181. Release rates of some bronchodilators through HPMC matrices were reported to be inversely proportional to the logarithm of their accessible surface area [lo]. Release of three drugs differing widely in their molecular weights, namely potassium chloride (mol. wt. 74.55), phenylpropanolamine hydrochloride (mol. wt. 187.67) and bovine serum albumin (mol. wt. 69,000), through poly(viny1 alcohol) matrices was studied, and the results revealed that, as the molecular size of the solute increases, the release rate decreases [ 191. This phenomenon was attributed to decreased diffusivity. Since most of the drugs used in formulations have a narrow molecular size range, the objective of this study is to know the influence of aqueous solubility and molecular size of drugs on their release through cellulose matrices. For this purpose, the release of 23 drugs of various solubilities (1 in < 1 to 1 in 10,000) and molecular weights (122-734) through matrices of HPMC (matrix I ) and a mixture ( 1: 1) of HPMC and NaCMC (matrix II) was studied, and the results are summarized in this paper.

EXPERIMENTAL Materials

Hydroxypropylmethylcellulose (HPMC, Methocel K4M Premium, Colorcon, U.K.) having 22% methoxy and 8% hydroxypropyl content was used. Its gelation temperature and degree of methoxy substitution were 85 C and 1.4, respectively. A 2% aqueous solution has a viscosity of ca. 4000 cP at 20C. Sodium carboxymethylcellulose (NaCMC, Blanose 7H4FD, Aqualon, France) has 0.65-0.90 as substitution range and 7.0-8.9% sodium content. A 1% aqueous solution has a viscosity in the range 2500-4000 CP at 25C. Solutes were alprenolol hydrochloride and metoprolol tartrate (Hassle, Sweden); oxprenolol hydrochloride (Ciba-Geigy, Switzerland); timolol maleate (Torrent Laboratories, India); saccharin sodium, thiamine hydrochloride L-ephedrine, aspirin, benzoic acid, chloramphenicol, phenobarbital and pentobarbital (Siegfried, Switzerland) ; propranolol hydrochloride (Indian Explosives, India); quinidine sulfate, quinine sulfate, acetazolamide, sulfamethizole, 17crmethyltestosterone and methylprednisolone (Sigma, USA); atropine and erythromycin (Fluka, Switzerland); phenacemide (Aldrich, F.R.G. ); diazepam (Hoffman la Roche, Switzerland). All the materials were fractioned and the particles of 63-125 ,um size range were used. The list of drugs, their molecular weights and approximate aqueous solubilities (the number of parts of water required for dissolving 1 part of the drug) according to the U.S.P. XXI are given in Table 1.
Preparation studies of matrices and in vitro release

Drug was mixed with HPMC ( 1: 1) or HPMC and NaCMC ( 1: 0.5 : 0.5) and compressed manually into flat-faced tablets (200 mg, 8.0 mm

135 TABLE 1 Molecular weights and approximate aqueous solubilities of drugs and their release parameters (intercept and slope) through matrices of HPMC (matrix I) and HPMC + NaCMC (matrix II) Solute Mol.wt. Approx. solubilityj Release parameters matrix I intercept, % Alprenolol HCl Oxprenolol HCl Metoprolol tartrate Timolol maleate Saccharin sodium Thiamine HCl L-Ephedrine Propranolol HCl Quinidine sulfate Benzoic acid Aspirin Diazepam Chloramphenicol Atropine Quinine sulfate Acetazolamide Phenobarbital Erythromycin Phenacemide Pentobarbitai Sulfamethizole 17a-Methyltestosterone Methylprednisolone 249.34 265.34 267.38 316.42 205.16 337.28 165.23 259.34 324.41 122.12 180.15 284.76 323.14 289.38 373.50 222.25 232.23 733.92 178.19 226.30 270.33 302.44 374.46 <l <l <l tl 1.2 1 20 20 100 300 300 333 400 460 500 1000 1000 1000 2000 2000 2000 10000 10000 9.17 14.07 12.91 17.97 14.88 10.47 23.68 10.15 1.53 6.75 2.88 1.81 1.56 5.31 -0.15 - 1.02 0.95 -0.13 7.35 -0.24 -0.42 -0.57 -0.86 slopeb, % h- 16.86 14.72 12.76 11.60 12.86 31.35 9.12 8.08 4.97 3.49 3.56 2.44 3.92 7.57 4.63 4.51 3.63 2.66 9.06 5.78 3.94 2.81 5.23 matrix II intercept, % 12.81 9.81 3.65 7.25 11.71 13.11 3.25 29.71 -2.42 - 6.59 2.22 - 1.76 - 4.98 -9.64 - 1.42 -4.45 -3.66 - 4.80 - 13.17 -6.52 - 1.33 10.70 -5.58 slopeb, % h- 6.41 8.62 9.38 9.84 24.85 24.86 20.78 5.35 5.64 16.64 8.14 10.688 15.12 23.40 7.74 17.49 13.67 9.88 21.15 20.71 16.478 3.01d 14.99

mol. wt. of base: br > 0.992; r = 0.978; do= 0.933;*r= 0.989;'r= 0.984;V= 0.990;'r= 0.986;obtained by regressing the dissolution data < 60%; r= coefficient of correlation; number of parts of water required to dissolve 1 part of the drug according to U.S.P. XXI.

diameter) using a Specac hydraulic press (Specat, U.K. ) at constant pressure (9.35 kN). The amount of drug present in each tablet was lOO? 2 mg. Three tablets of each formulation were subjected to dissolution using a U.S.P. dissolution apparatus 2 (Sotax AT6, Sotax, Switzerland) in 1000 ml of distilled water maintained at 375 0.2C and with the paddle rotating at 100 rpm. A flow-through apparatus connected to a spectrophotometer (Beckman model 35, Beckman Instruments, U.S.A.)

through a piston pump (Dissotest CY, Sotax, Switzerland) was used throughout the study. The pump was operated at a delivery rate of 15 ml min-. The absorbance of the dissolution medium at the corresponding wavelength maximum of the drug was recorded automatically at regular intervals. When the tablets containing barbiturates, ephedrine and erythromycin were subjected to dissolution, samples (5 ml) were withdrawn at regular intervals and assayed for the drug. An equivalent volume of dis-

136

tilled water maintained at 37 4 0.2 C was added immediately, and the dilution caused by this was corrected for when calculating the amount of drug released into the medium. Barbiturates were assayed by adjusting the pH of the sample to 13 with sodium hydroxide solution. Ephedrine was estimated as its hydrochloride by adding 0.05 ml of cont. HCl to 5 ml of the sample. E~hromycin was determined calorimetrically

TABLE 2 Values of kinetic constant (K) , release exponent (n ) and correlation coefficient (r*) following linear regression of log fraction released (Q0.6) versus log time of HPMC {matrix I) and HPMC+NaCMC (matrix II) containing formuiations

Dw

Matrix I K n r*

Matrix II K n r2

I201.

Alprenolol HCI Oxprenolol HCI Metoprolol t&rate Timolol maleate

0.2819 0.6198 0.9756 0.1784 0.5862 0.9995 0.2961 0.5890 0.9995 0.1805 0.6591 0.9979 0.2672 0.6118 0.9956 0.1254 0.8834 0.9917 0.2840 0.5635 0.9995 0.1643 0.7496 0.9945

RESULTS AND DISCUSSION Plots of percentage released versus time (mean of three tablets) for all the drugs from both the matrices are given in Figs l-6. Based on aqueous solubility, the drugs are classified as in the U.S.P. into very soluble, freely soluble, soluble, sparingly soluble, slightly soluble and very slightly soluble categories. According to U.S.P. XXI, the number of parts of water required for dissolving 1 part of the drug are: < 1 for very soluble, l-10for freely soluble, 10-30 for soluble, 30-100 for sparingly soluble, lOO1000 for slightly soluble, and lOOO-10,000 for very slightly soluble drugs. The release rates (% h-l) were calculated by subjecting the dissolution data <60% to least squares linear fitting. The values obtained for all the drugs are given in Table 1. In order to understand the mode of release of drug from these matrices, the data ( < 60% ) was fitted to the following power law equation [ 211:

Saccharin sodium 0.2865 0.5863 0.9995 0.3749 0.6972 0.9997 Thiamine HCI 0.4323 0.6845 0.9983 0.1042 0.6488 0.9922 L-Ephedrine Propranolol HCl Quinidine sulfate Benzoic acid Aspirin Diazepam Chloramphenicol Atropine Quinine sulfate Acetasolamide Phenobarbital Erythromycin Phenacemide Pentobarbital S~famethi~le 17crMethyltestosmrone Methylprednisolone 0.3309 0.4059 0.9960 0.2493 0.8280 0.9969 0.1833 0.6314 0.9979 0.3658 0.2522 0.9844 0.0541 0.9836 0.9948 0.0357 1.1987 0.9948 0.0829 0.7310 0.8805 0.1040 1.2643 0.9925 0.0545 0.8502 0.9990 0.0936 0.9545 0.9988 0.0311 0.9342 0.9974 0.0743 1.2461 0.9854 0.0519 0.8892 0.9969 0.0975 1.2904 0.9989 0.0940 0.8805 0.9933 0.0716 1.3699 0.9942 0.0481 0.9962 0.9580 0.0626 1.1072 0.9986 0.2355 1.0006 0.9968 0.0811 1.3122 0.9993 0.1953 0.8284 0.9840 0.0984 1.1962 0.9988 0.0265 0.9861 0.9922 0.0512 1.4682 0.9499 0.1505 0.7844 0.9929 0.0713 1.8360 0.9951 0.1597 0.9270 0.9894 0.0740 1.2923 0.9994 0.0362 1.0312 0.9989 0.1121 1.2771 0.9983 0.0203 1.1756 0.9762 0.1021 0.5992 0.9233

0.2036 1.0147 0.9984 0.1167 1.4175 0.9971

where A&/M, is the fraction of drug released up to time t, K is a constant incorporating the structural and geometric characteristics of the release device, and n is the release exponent indicative of the mechanism of release. The value of n for a cylinder is GO.45 for Fickian release, >0.45 and co.89 for nonFickian release, 0.89 for case II release and > 0.89 for super case II type of release. The val-

ues of K, n and coefficient of correlation (r) obtained are given in Table 2. A close look at the data given in Table 1 and Figs. l-3 reveals that, for certain very soluble ( 1 in c 1) beta adrenergic blockers, namely alprenolol hydrochloride, metoprolol tartrate, oxprenolol hydrochloride and timolol maleate, for a soluble (1 in 20) beta blocker, propranolol hydrochloride and a freely soluble (1 in 1) vi-

Very soluble drugs

60

80

0 2 3 E $ = $ P

60

60

40

40

20

20

Matrix I
0 0 2 4 6 8 10 12 14 0 0 2 4 6 8

Matrix II
10 12 14

Time, h

Time, h

Fig. 1. Release profile8 of drug8 through HPMC (matrix I) and HPMC + NaCMC (matrix II). ( A ) Oxprenolol HCl, ( 0 )metoprolol tartrate, (0)alprenolol HCl and (o )timolol maleate.
Freely soluble drugs
106

80

0 $ t t

60

E 40 p: it P 20

Matrix I
2 4 Time,h 6 8 10 12

Matrix II

1
14

a
2 4 Time,h 6 8 10 12

4
14

Fig. 2. Release profile8 of drugs through HPMC (matrix I) and HPMC+NaCMC (0 )saccharin sodium.

(matrix II). (A)Thiamine

HCl and

tamin, thiamine hydrochloride, the release rates from matrix II are much slower than those from matrix I. These cationic drugs might be forming a complex with the anionic NaCMC and the complex might be diffusing from the swollen gel at a much slower rate. A similar theory was proposed and proved for the release of chlorpheniramine maleate from matrices containing

NaCMC [22]. For a sparingly soluble drug, quinidine sulfate ( 1 in 100)) and a very slightly soluble (1 in 10,000) drug, 17a-methyltestosterone, the release rates are nearly the same from both the matrices. For all the remaining 15 drugs, interestingly, the release rates from matrix II are much higher than those from matrix I. It was observed during the experiment

138

Soluble, sparingly soluble and slightly soluble drugs

10

12

14

10

12

14

lime,

Time,

Fig. 3. Release profiles of drugs through HPMC (matrix I) and HPCM + NaCMC (matrix II). ( A )L-Ephedrine, pranolol HCl, (0)quinidine sulfate and (o )chloramphenicol.

( 0 )pro-

that, for the last-named category of 15 drugs, the tablets of matrix II swell and erode at a relatively faster rate compared with those of matrix I. This was confirmed by subjecting the tablets (200 mg) of matrices I and II (without the drug) to dissolution in the usual manner. At the end of 4 h, the matrices were taken out from the beaker and dried to constant weight in a hot. air oven (SO'C ). The percentages remaining (w/w) of matrices I and II were 86 and 47, respectively, compared with their initial weight. Similar results were reported earlier with cellulose matrices containing 50% of drug, i.e., the erosion rate of the matrices containing 50% of drug 50% of anionic and nonionicethers (HPMC + NaCMC or cellulose HPC + NaCMC ) was much higher compared to matrices containing only a nonionic polymer, such as HPMC or HPC [ 15,231. Therefore, at any particular time, due to rapid erosion of the gel of matrix II, the diffusional pathlength for the drug in matrix II was smaller than that in matrix I and hence the release rate was faster. Disintegration of tablets matrix II containing 17cw-methyltestosterone, methylprednisolone and diazepam occurred after ca. 3 h. An increase in molecular weight decreased the

release rate in a nonlinear manner from both the matrices for soluble (1 in 20) drugs, namely L-ephedrine and propranolol hydrochloride, slightly soluble (1 in 1000) drugs, namely, acetazolamide, phenobarbital and erythromycin, and other slightly soluble (1 in 460 and 1 in 500) drugs, namely atropine and quinine sulfate. Such a phenomenon is seen from matrix I alone for very soluble ( 1 in < 1) beta blockers (taking into account the mol. wt. of the base). However, this trend could not be extrapolated to a soluble (1 in 20) beta blocker, propranolol, although its molecular weight is closer to those of very soluble beta blockers. Release rates from matrix I for 17cu-methyltestosterone (mol.wt. 302.5), erythromycin (mol. wt. 733.9) and diazepam (mol. wt. 284.7) are 2.81,2.66 and 2.44% h-l, respectively, while their approximate solubilities are 1 in 10,000, 1 in 1000 and 1 in 333. Methylprednisolone, which has a higher molecular weight than methyltestosterone, has a higher release rate through both the matrices, although their solubilities are same. This suggest that aqueous solubility and molecular size are not the only factors governing the release of solutes through cellulose matrices. It is also clear from this study that the

139

Slightly soluble drugs

100

80

a0

$ 60 i f
E
0) r 40 p 20

60

2 Time, h

a
Time, h

10

12

14

Fig. 4. Release profiles of drugs through HPMC (matrix I) and HPMC +NaCMC (matrix II). (A )Atropine, (0 )quinine sulfate, (O)benzoic acid, (o )aspirin and ( + )diazepam.

Slightly soluble and very slightly soluble drugs

100 100

a0

a0
60

20

Matrix II

4 Time,

6 h

10

12

14

4 Time,

6 h

10

12

14

Fig. 5. Release profiles of drugs through HPMC (matrix I) and HPMC + NaCMC (matrix II). ( A ) Methylprednisolone, (0 )acet.azolamide, (O)phenobarbital, (o ) 17wmethyltestosterone and ( + )erythromycin.

release phenomenon that applies with matrix I is not always applied with matrix II for the same solutes. It may be concluded that, in addition to solubility and molecular size of solute, certain other factors, like interaction between the polymer (s) and the drug, penetration of the solvent into the matrix, erosion of the swollen gel, influence of the nature of the drug on the erosion

of the gel, solubilization of the drug by the polymer( s), etc., might also be governing the release. Experiments to quantify the influence of these factors are in progress in our laboratory. It is also clear from Figs. 3-6 that several sparingly soluble,slightly soluble and very slightly soluble drugs are released from cellulose matrices at a nearly zero-order rate. It is

140

Very slightly soluble drugs

6 Time, h

10

12

14

6 Time, h

10

12

14

Fig. 6. Release profiles of drugs through HPMC (matrix I) and HPMC +NaCMC tobarbital and (0)sulfamethizole.

(matrix II). (A ) Phenacemide,

(0 )pen-

only for the very soluble and freely soluble drugs that the release rate is rapid initially and later decreases with time. The values of the release exponent n given in Table 2 for matrix II are higher than those for matrix I for most of the drugs. The mode of drug release from matrices I and II is non-Fickian and super case II type, respectively, for sparingly soluble, slightly, and very slightly soluble drugs, which indicates that the drugs falling into this category can be released at a nearly zero-order rate through HPMC matrices. As the method involved in the preparation of these matrices is very simple, cellulose matrices may be preferred for oral controlled release formulations.

REFERENCES
P. Buri and E. Doelker, Formulation des cornprimes a liberation prolong&. II. Matrices hydrophiles, Pharm. Acta Helv., 55 (1980) 189-197. D.A. Alderman, A. review of cellulose ethers in hydrophilic matrices for oral controlled-release dosage forms, Int. J. Pharm. Technol. Prod. Manuf., 5 (1984) l-9. N.A. Peppas and R.W. Korsmeyer, Dynamically swelling hydrogels in controlled release applications, in: N.A. Peppas (Ed.), Hydrogels in Medicine and Pharmacy, Vol. III, Properties and Applications, CRC Press, Boca Raton, FL, 1987, pp. 109-135. E. Doelker, Water-swollen cellulose derivatives in pharmacy, in: N.A. Peppas (Ed.), Hydrogels in Medicine and Pharmacy, Vol. II, Polymers, CRC Press, Boca Raton, FL, 1987, pp. 115-160. K. Padmalatha Devi, Studies on formulation of sustained release dosage forms of some beta adrenergic blockers, Ph.D. Thesis, Panjab University, Chandigarh, India, 1987. P.I. Lee, Interpretation of drug-release kinetics from hydrogel matrices in terms of time-dependent diffusion coefficients, in: P.I. Lee and W.R. Good (Eds.), ACS Symp. Ser., No. 348, American Chemical Society, Washington, DC, 1987, pp. 71-83. K.V. Ranga Rao and K. Padmalatha Devi, Swelling

ACKNOWLEDGEMENTS

We are grateful to Hlissle (Sweden), CibaGeigy (Switzerland), Hoffmann la Roche (Switzerland), Indian Explosives (India) and Torrent Laboratories (India), for supplying the samples of drugs.

141

10

11

12

13

14

15

16

controlled-release systems: recent developments and applications, Int. J. Pharm., 48 (1988) 1-13. J.L. Ford, M.H. Rubinstein and J.E. Hogan, Formulation of sustained release promethazine hydrochloride tablets using hydroxypropylmethylcellulose matrices, Int. J. Pharm., 24 (1985) 327-338. J.L. Ford, M.H. Rubinatein and J.E. Hogan, Propran0101 hy~~hloride and aminophy~ine release from matrix tablets containing hy~ox~ropylmethylcellulose matrices, Int. J. Pharm., 24 (1985) 339-350. S.K. Baveja, K.V. Ranga Rao, A. Singh and V.K. Gombar, Release characteristics of some bronchodilators from compressed hy~ph~i~ polymeric matrices and their correlation with molecular geometry, Int. J. Pharm., 41 (1988) 55-62. J.L. Salomon, E. Doelker and P. Buri, Importance de la technologie et de la formulation pour le m&an&me de liberation du chlorure de potassium contenu dans des matrices hydrophiles. I. Influence de la viscositi et du pourcentage de gelifiant, Pharm. Acta Helv., 54 ( 1979) 82-85. K.V. Ranga Rao, Studies on fo~u~tion of sustained release dosage forms of centperazine, Ph.D. Thesis, Panjab University, Chandigarh, India, 1984. S.K. Baveja and K.V. Rangao Rae, Sustained release tablet formulation of centperazine, Int. J. Pharm., 31 (1986) 169-174. S.K. Baveja, K.V. Ranga lXaoand K. Padmalatha Devi, Zero-order release hydrophilic matrix tablets of beta adrenergic blockers, Int. J. Pharm., 39 (1987) 39-45. K.V. Ranga Rae, K. Padmalatha Devi and P. Buri, Cellulose matrices for zero-order release of soluble drugs, Drug Dev. Ind. Pharm., 14 (1988) 2299-2320. K. Padmalatha Devi, K.V. Ranga Rae, SK. Baveja,

M. Fathi and M. Roth, Zero-order release formulation of oxprenolol hydrochloride with swelling and erosion control, Pharm. Res., 6 (1989) 313-317. 17 S. Wisniewski and S.W. Kim, Permeation of watersoluble solutes through poly(2-hydroxyethyl methacrylate ) and poly (2-hydroxyethyl methacrylate ) crosslinked with ethylene glycol dimethacrylate, J. Membrane Sci., 6 (1980) 299-308. 18 S.W. Kim, J.R. Cardinal, S. Wisniewski and G.M. Zentner, Solute permeation through hydrogel membranes: hydrophilic vs. hydrophobic solutes, in: S.P. Rowland (Ed.), ACS Symp. Ser., No. 127, American Chemical Society, Washington, DC, 1980, pp. 347-359. 19 R.W. Konmeyer, R. Gurny, E. Doelker, P. Buri and N.A. Peppas, Mechanisms of solute release from porous hydrophilic polymers, Int. J. Pharm., 15 (1983) 25-35. 20 J.H. Ford, G.C. Prescott, J.W. Hinman and E.L. Caron, Calorimetric estimation of erythromycin, Anal. Chem., 25 (1953) 1195-1197. 21 P.L. Ritger and N.A. Peppas, A simple equation for description of solute release. II. Fickian and anomalous release from swellable devices, J. Controlled Release, 5 (1987) 37-42. 22 L.C. Feely and S.S. Davis, The influence of polymeric excipienta on drug release from hydroxypropylmethylcellulose matrices, Int. J. Pharm., 44 (1988) 131139. 23 K.V. Ranga Rao, K. Pa~alatha Devi, F. Kubel and P. Buri, Studies on factors affecting the release of drugs through cellulose matrices, Proc. Int. Symp. Controlled Release Bioact. Mater, 15 (1988) 101-102.