Compressed polymeric mini-matrices for drug release control

Journal of Controlled Release, 1(1985) 283-289 Elsevier Science Publishers B.V., Amsterdam -Printed

COMPRESSED POLYMERIC

283 in The Netherlands

MINI-MATRICES

FOR DRUG RELEASE CONTROL

P. Colombo, U. Conte, C. Caramella, A. Gazzaniga and A. La Manna Dipartimento (Received

Chimica

Farmaceutica,

June 19, 1984; accepted

Universite

di Pavia,

Via Taramelli

in revised form February

12, 27100

Pavia (Italy)

25, 1985)

Matrices of reduced dimensions of either inert or swellable polymers for multiple-unit administration, containing diprophylline as the model drug, were made by compression. Some swellable mini-matrices prepared from poly(viny1 alcohol) were externally crosslinked by soaking in an acidic formalin solution, and subsequent drying and exposure to UV or thermal energy sources. The release kinetics of the model drug from the mini-matrices was investigated by analyzing the early time portion of the release curves according to a simplified exponential equation which allows the comprehension of the release mechanism. The results showed that the release kinetics may be Fickian or anomalous nonFickian depending on the solubility of the polymer employed. Zero-order release kinetics were obtained by using mini-matrices, which were crosslinked at the surface.

INTRODUCTION

The optimization of drug activity can often be achieved by an appropriate formulation. Drug delivery systems represent the practical realization of this concept; in many cases the “best fit” between pharmacological activity and drug administration can be satisfactorily achieved with a constant drug input. Of those oral drug delivery systems which are obtained by compression, matrices (monoliths) are the easiest to prepare; drug release from matrices is diffusion-controlled and the fraction released is usually proportional to the square root of time. Zero-order release from a matrix can be obtained by using either the appropriate device geometry [ 11, hydrophobic porous materials [2], or hydrophilic soluble polymers capable of modifying the effective diffusivity of the active principle 131. The use of a multiple-unit instead of a

0168-3659/85/$03.30

0

1985

single-unit dosage form permits the reduction of the inherently large inter- and intrasubject variations linked to gastrointestinal transit time, thus improving the safety of use [4]. Following these principles and with the aim of investigating the influence of the dimensions and composition of the matrices on the release kinetics we have prepared matrices of small dimensions (mini-matrices for multiple-unit dosage form), by compression, using either inert or swellable polymers. The objective of zero-order release kinetics could be achieved by changing the formula composition and also by controlling the solute transfer rate at the fluid/matrix interface by directly crosslinking the polymer at the surface of the mini-matrices. To investigate the relationship between formulation parameters and release kinetics the semiempirical equation recently used by Korsmeyer et al. [ 31 was applied. We have selected polymers which have been studied previously and which have

Elsevier Science Publishers B.V.

284

good compac~bi~ty characteristics [ 51. Diprophylline was chosen as a model drug because of its well-known solubility charaeteristics and pharmacokinetics.

MATERIALS

The drug and polymer mixtures were granulated by wetting with the binding solutions in a planetary mixer (Table l), passing through a 60-mesh ASTM sieve and drying overnight at 40’C. The mini-matrices were prepared by compression at a force level of 2.5 MN and checked for weight (28 mg), height (2.4 mm), diameter (3.3 mm) and release rates. Tablets with a diameter of 11.4 mm and height of 1.64 mm were also prepared from the same mixtures. Cross~nking in the he~rogeneo~ phase [S] was effected in two steps: (i) 20 minimatrices were soaked in 30 ml of a 1% HCl solution of formalin for five- or six-minute periods. The variation in soaking time leads to different penetrating solvent depths; (ii) the mini-matrices were dried in a stream of warm air and exposed either to a UV energy source (by rotating the pan containing the mini-matrices (5 rpm) under a UV lamp (254 nm) of 8 W placed at the distance of 4 cm), or to a thermal energy sourc!e (by heating them in an oven at a controlled temperature for a fixed time, see also Table 3). The fraction of crosslinked polymer was measured by dissolving a few mini-matrices in 250 ml of hot water (80°C) and salvaging the insoluble crosslinked shells. These were then dried in vacua at 60°C and weighed.

AND METHODS

Two grades of poly(viny1 alcohol) (PVA) were used for the preparation of swellable matrices : Mowiol 40-88 (MW 130,000; degree of hydrolysis 87.7%; 4% aqueous solution at 2O”C, viscosity 40 mPa s; Hoechst, Frankfurt a.M.) and Mowiol 28-99 (MW 88,000; degree of hydrolysis 99.4%; 4% aqueous solution at 20°C, viscosity 28 mPa s; Hoechst, Frankfurt a.M.). For the preparation of inert matrices, we have employed the following materials: cellulose acetate propionate (CAP 482-20; T, 14?C; hydroxyl content 2.1%; EastmanKodak, U.S.A.), a mixture of 91% poly(methyl methac~late~ and 9% poly( ethyl rnet~c~~te) (ParaIoid D 120, MW 1,000,000, Rohm and Haas, U.S.A.) and Ethocei HV (ethylcellulose USP XX grade, Dow Chemical, U.S.A.). All the polymers were used without further purification. The model drug used was diprophylline (MW 254; mp 158°C; solubility in water 10% w/v; Rhone-Poulenc Chimica, Milan, Italy).

TABLE 1 Composition of mixtures (%)

Diprophylline CAP 482120 Paraloid K120 Mowiol40-88 Mowiol28-99 Ethocel HV Talc Binding agentb

IN1

IN2

IN3

IN4

IN5

SW1

SW2

SW3

SW4

SW5

50 40 -

37.5 25 37.5 (a)

37.5 37.5 25 (b)

37.5 62.5 -

50 50

50 -

60 -

-

30 -

50 -

50 30 20 -

60 30 1oa

60 10a 30 -

(b)

Ce)

z

(c)

Ce)

Gl)

(d)

ta9

aDiiolved in solvent (d). bThe binding solutions were: (a) = acetone; 1 :l; (e) = methocel 2% w/v in water.

(b) = water; (c) = water-ethanol

1:l;

(d) ehloroform~thanol

285

The ratio between the weight of the shells and the total amount of polymer in the mini-matrices was taken as the fraction of crosslinked polymer. Scanning electron microscopy photographs were taken on the intact, crosslinked and exhausted mini-matrices using a JSM 35C apparatus (Japan Electron Optical Laboratory, Tokyo). The release tests (6 replicates) were performed in water (1000 ml, 37oC) at 100 rpm using the U.S.P. XX paddle apparatus. Diprophylhne was assayed continuously spectrophotomehically (at 273 nm) using a Perkin-Elmer Coleman 111 model UVVIS spectrophotometer and a recorder. The release data were analyzed according to eqn. (1) [3,7]

Mt

.-..v=ktn

M,

TABLE 2 Inert and swellable mini-matrices release parameters according to eqn. (1); values f 95% confidence limits Formula

n

k

Inert IN1 IN2 IN3 IN4 IN5

0.49 0.47 0.52 0.55 0.48

f f + f f

0.03 0.01 0.02 0.05 0.01

0.09 0.09 0.08 0.10 0.12

f 0.01 +_0.02 * 0.006 * 0.01 i 0.004

0.63 0.62 0.59 0.57 0.57

+ * f f f

0.01 0.004 0.01 0.01 0.009

0.05 0.05 0.06 0.07 0.07

* * f + f

Swellable SW1 SW2 SW3 SW4 SW5

0,003 0,001 0.002 0.003 0.01

(1)

where ~~/~~ is the drug fraction, &, released at time t and k and n are the const~ts characteristic of the matrix-eluent system. The exponent n indicates the release kinetics: for n = 0.5 the equation describes Fickian release, for n > 0.5 non-Fickian, anomalous cases, and for n = 1 the limiting non-Fickian case, i.e., the constant release of drug. Data analysis was performed with a program run on a Mint 11 computer, which is capable of effecting the best data fit, chosen on the basis of variance analysis, between eqn. (l), the first-order and the zero-order equations. The fitting of the data was accomplished on the early portion of the curve (M&f, < 0.70; degrees of freedom > 10; confidence limits calculated at P = 0.95) [7,81* RESULTS AND DISCUSSION

The parameters of eqn. (1) obtained from the release data of mini-matrices are given in Table 2. Values of the exponent n near 0.5 were obtained for the inert minimatrices, thus indicating a Fickian release.

Swellable m~i-rna~ic~ show n values si~~ic~tly higher than 0.5, thus ~dica~ng a non-Fickian anomalous release; this suggests that these systems are in moving boundary conditions, since the swelling and dissolution of the polymer continuously modify the effective diffusivity of the drug. This is particularly evident for those mini-matrices containing the more soluble PVA grade (Mowiol 40-88) (SW1 and SW2 formulas). By partially substituting Mowiol 40-88 with the less soluble grade (Mowiol 28-99) or with insoluble polymers (CAP, Ethocel) the n values tend to approach 0.5 (SW4 and SW5 formulas). The II values obtained for those tablets (diameter = 11.4 mm; height = 1.64 mm) which had the same formulas as the minimatrices were not significantly different, thus indicating that the release mechanism is not influenced by the decreased dimensions of the monoliths. In the case of SW3 formula mini-matrices which are crosslinked under different conditions, the values of the exponent n gradually approach one, and therefore a zero-order release is reached (Table 3).

286 TABLE 3 SW3 crosslinked mini-matrices release parameters according to eqn. (1); values ?: 95% confidence limits Formulaa

II

k

SWl(not crosslinked) SW3(5’,UV, ZO”, 420’) SW3( 6’,UV, 20”) 420’)

0.59 f 0.01 0.72 f 0.02 0.98 + 0.04

0.060 i 0,002 0.017 f 0.002 0.003 f 0.001

SW3(6’,TH, 40”) 300’) SW3(6’,TH, 40”, 600’) SW3(6’,TH, 40”,1440’) SW3( 6’,TH,lOO”, 60’)

0.77 0.88 0.90 0.93

0.010 0.007 0.006 0.003

f f f f

0.02 0.04 0.04 0.04

f * f *

0.002 0.001 0.001 0.001

BThe crosslinking reaction conditions are indicated in brackets in the following order: soaking time (min), UV or thermal (TH) energy source, temperature (“C), exposure time to energy source (min).

0

1.5

Time

3.0

4.5

6.0

(hours)

Fig. 1. Computer-simulated release curves obtained by increasing the value n in eqn. (I) from 0.6 to 1. All graphs for constant MJM- = 0.6.

The release rate of the drug from minimatrices treated with thermal energy was lower than that from UV-irradiated systems. The evolution of the systems towards zero-order kinetics is illustrated by computer-simulated release curves obtained by increasing the n values from 0.5 to 1 (Fig. 1). When the release data are fitted according to the zero-order kinetic equation, the portion of the release curve which can be fitted to a straight line increases from B-56% for the non-crosslinked mini-matrices to 5-70% for the SW3(6’,TH,lOO” ,60’) minimatrices (Table 4). For some systems, such

Fig. 2. Microphot~ap~ (SEM) of mini-matrices: (a) before cro~linking (16.7X), (b) after crosslinking (16.7X), (c) exhausted system (16.7X), (d) surface conditions before crosslinking (125X), (e) surface conditions after crosslinking (125X).

287

288

TABLE

4

Comparison between kinetics equation

fits of release

data treated

according

to eqn. (1) and to the zero-order

Formula SW3(not

crosslinked)

20” , 420’)

SW3( 5’,UV,

SW3(6’,UV,

SW3(6’,TH,

SWS(G’,TH,

SW3(6’,TH,

20”) 420’)

40”) 300’)

40”, 600’)

40” ,144O’)

SW3(6’,TH,100°,

aThe

t91.

linear

portion

60’)

Q = 0.06 PJ9 Q = 0.0090t + 0.15

(uy = 9.5 (4 = 2.9

x x0--5) x lo-‘)

0.29-0.568

Q = 0.017P Q = 0.0036t

(“Y = 1.3 (4 = 4.2

x 10-4) x 10-4)

0.12-0.60a

Q = 0.0034Pa Q = 0.0031t - 0.004

(4 (4

x lo-‘) x lO-q)

0.03-0.63a

Q = O.OlOP*” Q = 0.0034t + 0.063

@Y = 3.79 x 10-y (uy = 4.84 x 10-l)

0.10-0.57a

Q = 0.007Ps8 Q = 0.0034t + 0.029

(I+ = 1.32 x 1O-3) (uy = 1.45 x lo-$)

0.09-0.65a

Q = 0.005P’ Q = 0.0028t

:L$ = 1.55 x 10-s) ,, = 1.63 x lo-“)

0.06-0.66a

(4 = 3.5 (uy = 6.7

0.05-0.70a

+ 0.023

Q = 0.003P3 Q = 0.0018t - 0.007

of the release

curves

derived

from

= 6.9 = 5.5

x 10-a) x lo-‘)

a Durbin-Watson

serial correlation

-

SW3(6’,UV,20° ,420’) and SW3(6’,TH, 100“,60’), the release data are even better fitted by the zero-order equation than by eqn. (l), as shown by the y variance values, uy. The SEM photomicrographs taken of SW3 mini-matrices at different stages (before and after crosslinking and after dissolution; Fig. 2) show that the surface is modified by crosslinking, which at those stages has the appearance of a continuo~ film (Figs. 2c, d) ; the exhausted system, salvaged at the end of the release test (Fig. 2e) looks like an empty capsule, the shell of which appears to be continuous with a thickness of about 300 nm. In the case of the more deeply crosslinked mini-matrices the shell weighed as much as 35% of the total amount of polymer in the rn~i-rna~c~. The results obtained show that during the crosslinking procedure the soaking produces a swollen outer layer (around a glassy core) which is then transformed, by UV or thermal as

+ 0.08

energy, into an insoluble layer of crosslinked polymer. It is suggested that the crosslinked outer polymer acts as a more or less porous polymeric barrier, thus affecting the drug diffusivity, and that, upon increasing the crosslinking intensity, this crosslinked polymer barrier becomes a ratecontrolling membrane. CONCLUSIONS

The results obtained show that for diprophylline-containing inert mini-matrices, a Fickian diffusion mechanism predominates, while for swellable mini-matrices a nonFickian mechanism occurs; surface crosslinking of the mini-matrices forms a barrier which limits drug release. Release rates were progressively reduced by changing the crosslinking conditions, i.e., the soaking time, the energy source and the energy exposure time. The release

289

kinetics can be modulated from the Fickian cases case (n = 0.5) through non-Fickian (n > 0.5) to a zero-order release (n = 1). It is therefore concluded that the crosslinking reaction creates a crosslinked polymer barrier which acts as a membrane and transforms the matrix into a quasi-reservoir (hybrid) sys tern. The use of eqn. (I), which is considered as an extension of the Higuchi treatment equation, permits the comprehension of the physical evolution of the systems. From the practical point of view the use of matrices with reduced dimensions and variation of the number of units used would facilitate control of the amount of drug released.

ACKNOWLEDGEMENTS

The authors gratefully thank Prof. N.A. Peppas (School of Chemical Engineering, Purdue University) for helpful comments. In addition, the authors wish to acknowledge Dr E. Pisoni (postgraduate researcher at Recordati S.p.A., Milan) for experimental cooperation, Mrs. M.C. Sacchi for assistance in text preparation and Mr. R. Bonecchi (Centro “Gino Bozza”, Politecnico, Milan) for SEM microphotographs. This work was supported by the “‘Progetto Finalizzato Chimica Fine e Secondaria” (Grant N.

82.00619.95 of Research

of the Italian C.N.R.).

National

Council

REFERENCES D.S.T. Hsieh, W.D. Rhine and R. Langer, Zeroorder controlled-release polymer matrices for micro- and macromolecules, J. Pharm. Sci., 72 (1983) 17-22. E.A. Swan and N.A. Peppas, Drug release kinetics from hydrophobic porous monolithic devices, Proc. Symp. Contr. Rel. Bioact. Mater., 8 (1981) 18-23. R.W. Korsmeyer, 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. H. Bechgaard and F.N. Christensen, Gastrointestinal transit time of pharmaceutical dosage form, Proc. Int. Symp. Contr. Rel. Bioact. Mat., 10 (1983) 240. U. Conte, C. CaramelIa, P. Colombo and A. La Manna, The compactability of polymers for drug delivery systems, 11 Farmaco Ed. Prat., 38 (1983) 361-368. E.S. Lee, SW. Kim, S.H. Kim, J.R. Cardinal and H. Jacobs, Drug release from hydrogel devices with rate-controlling barrier, J. Membrane Sci., 7 (1980) 293-303. N.A. Peppas, Analysis of Fickian and non-Fickian drug release from polymers, Pharm, Acta Helv., in press. G.W. Sinclair and N.A. Peppas, Analysis of nonFickian transport in polymers using simplified exponential expressions, J. Membrane Sci., 17 (1984) 329-331. H. van der Voet, P. de Haan and D.A. Doornbos, The use of the Durbin-Watson statistic for testing the validity of kinetic models for dissolution, Int. J. Pharm., 14 (1983) 291-298.