Controlled release of U-86983 from double-layer biodegradable matrices: effect of additives on release mechanism and kinetics

journal of

ELSEVIER

Journal of Controlled Release 45 (1997) 177-192

controlled release

Controlled release of U-86983 from double-layer biodegradable matrices: effect of additives on release mechanism and kinetics a ab C . X . S o n g , V. L a b h a s e t w a r " , R.J. L e v y a ' b ' * aThe University of Michigan Medical Center, Division of Pediatric Cardiology, Department of Pediatrics and Communicable Diseases, R 5014, Kresge II, Box 0576, Ann Arbor, MI 48109-0576, USA bCollege of Pharmacy, The University of Michigan, Ann Arbor, MI 48109-0576, USA

Received 14 May 1996; revised 25 August 1996; accepted 9 October 1996

Abstract

Effects of additives on the drug release kinetics from biodegradable matrices is an important determinant in designing a drug delivery system. These experiments introduced the influence of an array of additives on the drug release from double-layered poly(lactic-co-glycolic acid) (PLGA) matrices. Various additives such as L-tartaric acid dimethyl ester (DMT), Pluronic® F127 (F127); 2-hydroxypropyl derivative of/3-cyclodextrin (HPB), methyl derivative of/3-cyclodextrin (MMB) and Beeswax (Wax), differing in molecular size, hydrophilicity and steric configuration were selected for this study. An antiproliferative 2-aminochromone, U-86983 (U-86, Pharmacia and Upjohn), was used as a model agent because of our interest in investigating local drug delivery systems for the inhibition of restenosis. The in vitro release of U-86 from PLGA matrices without additive showed a typical biphasic release kinetics, i.e. a slow diffusion release (Phase I) followed by a fast erosion-mediated release (Phase II). The water-soluble additives in PLGA matrices changed the biphasic release pattern to a near monophasic profile by increasing the release rate of the Phase I. Increasing the ratio of additives to PLGA in matrices causes a significant increase in the U-86 release rates. The high molecular weight water-soluble additive, Pluronic® F127, resulted in a matrix showing perfect zero-order release kinetics. The water-soluble cyclodextrin derivative, HPB, gave the highest release rate among all the matrices formulated. A hydrophobic additive, Beeswax, however showed biphasic release kinetics comparable to PLGA control matrices, but delayed the onset of the Phase II by 4 days. The U-86 release profiles were in good agreement with the mass loss profiles of these matrices except for the matrices with F127 and HPB additives. The morphologic evaluation of matrices using scanning electron microscopy indicates that the water-soluble additives are leachable and thus generate a highly porous structure in the matrices. The matrix pore configuration (e.g. interconnected or closed) created with different additives determined the mechanism of drug release kinetics from the various matrix formulations. In conclusion, the feasibility of modulating release rates and kinetics of an agent from PLGA monolithic matrices by utilizing various types of additives is demonstrated. Water-solubility, molecular size and steric configuration of the additives are the important determinants in generating various types of pore structures in polymer matrix which in turn affect the release mechanism and release kinetics. Keywords: Poly(lactic-co-gtycolic acid) (PLGA); Additives; Release kinetics; Morphology; Restenosis

Corresponding author. Tel.: + 1 3t3 9362850; fax: +1 313 7473270; e-mail: [email protected]. 0168-3659/97/$17.00 Copyright © 1997 Elsevier Science Ireland Ltd. All rights reserved PII $0168-3659(96)0155 1~9

178

C.X. Song et al. / Journal of Controlled Release 45 (1997) 177-192

1. Introduction Lactide/glycolide co-polymers (PLGA) are a widely investigated biodegradable material for controlled drug delivery applications. Drug release from the PLGA matrix is a combined result of two processes: drug diffusion and matrix erosion [1]. Erosion-mediated drug release does not become significant until the molecular weight of the polymer decreases to a point where there is an onset of polymer mass loss [1-3]. Therefore, drug release from the PLGA film-type monolithic matrix follows a typical biphasic release kinetics: an initial slow release period (lag time or Phase I) due to the low permeability of drug through PLGA followed by a sudden increase in release rate at the onset of polymer mass loss (Phase II) [4,5]. Restenosis is a pathologic process of reobstruction of an artery following an angioplasty induced injury response [6]. Direct implantation of a drug-loaded matrix at the site of an injured artery is useful for local specific therapy to prevent restenosis. Periadventitial delivery of dexamethasone through a silicone matrix has shown to be effective in decreasing neointimal formation in a rat carotid model [7]. Biodegradable materials are in some ways more suitable for in vivo drug delivery implants for a disorder such as restenosis because a biodegradable matrix does not require removal after depletion of the active agents. Only a limited number of studies have been done to alter the inherent biphasic drug release kinetics of PLGA matrices, especially to eliminate the lag time [8,9]. Most of the attempts were related to changing the degradation rate of PLGA [8,9]. Some improvements in release rate and kinetics have been achieved by varying the composition ratio of the two monomers in the co-polymer [8,9]; by selecting an optimal starting molecular weight of PLGA [10], and by using co-polymers of PLGA with hydrophilic segments [11]. However, these approaches tend to significantly change the mechanical and chemical features of the original polymers. Drug release from polymer matrices can also be modulated by inclusion of different additives. The potential of using additives as a mechanism for regulating drug release from polymer matrix has not been fully explored. One of the main advantages of

the additive approach is that it could be potent enough so that a small incremental amount in the matrices could cause an effective change in the drug release without altering the mechanical and chemical properties of the predominant material. The effect of each additive depends on its lipophilic nature, solubility, volatility, location and interaction with the polymer [12]. Due to the complexity of the factors involved during the action of additives, criteria for selecting a specific additive to achieve a desired release rate profile is not clearly understood. The aim of this study was to explore the potential of utilizing a small portion of hydrophilic additives in the PLGA (50:50) matrix to enhance U-86 release. Four hydrophilic materials differing in molecular size, hydrophilicity and steric features were evaluated as release-enhancing additives. A hydrophobic compound, Beeswax, was also studied for comparison to establish a correlation between the matrix composition and drug release.

2. Chemical rationale of the additives used in this study All five additives utilized in this study are nontoxic and well established pharmaceutical excipients but differ in molecular size, hydrophilicity and steric features as described below. However, it should be noted that derivatives of cyclodextrin and Pluronic® F-127 at this time are not authorized or approved by the U.S. Food and Drug Administration as injectable o r implantable formulation. Regulatory considerations of these additives thus far are limited to the oral or percutaneous route.

1. L-Tartaric acid dimethyl ester (DMT) is a low molecular weight compound (MW= 178.1) made from naturally occurring e-tartaric acid. It is both organic and water soluble, and thus is compatible with PLGA and is hypothesized to impart hydrophilicity to PLGA. The DMT molecules are leachable in aqueous solution. 2. Pluronic® F127 (F127) is a block co-polymer of poly(ethylene oxide) (PEO) and poly(propylene oxide) (POP) possessing the following characteristics: high molecular weight (MW=12 600

C.X. Song et al. / Journal of Controlled Release 45 (1997) 177-192

Da), both water and organic solvent solubility and leachability in water. 3. The 2-hydroxypropyl derivative of/3-cyclodextrin (HPB) and the methyl derivative of /3-cyclodextrin (MMB) are chemically modified cyclodextrins (CDs). These are water soluble, hydrophilic, cavity-forming cyclic oligosaccharides. They can accommodate drugs to form watersoluble inclusion complexes. HPB has comparatively higher water solubility than MMB (75 g/ 100 ml water for HPB and 25 g/100 ml for MMB). The average MW of these CDs is in the range of 3000-8000. 4. Beeswax consists of esters of long-chain monohydric alcohols with even-numbered carbon chains (C24-C36), esterified with long-chain acids, also having even numbers of carbon atoms (up to 36). Beeswax is water insoluble and hydrophobic, and was used as a hydrophobic additive for the purpose of comparison. The goals of this study were:

1. To formulate and characterize double-layered PLGA matrices with and without different additives using U-86983 as a model compound. 2. To assess the in vitro release kinetics of the model agents from these differently constructed matrices with respect to the nature and the amount of additives utilized. 3. To arrive at a general understanding of the mechanism of this approach for modulating release kinetics of U-86983.

3. Materials and methods 3.1. Materials

Poly (d,l-lactic-co-glycolic acid) (PLGA) 50/50 of inherent viscosity of 0.67 dl/g measured in hexafluoroisopropanol (estimated MW=52 000 Da) was purchased from Birmingham Polymers, Inc., Birmingham, AL. Pluronic® F127 was purchased from the BASF Co., Parsippany, NJ. 2-Hydroxypropyl derivative of /3-cyclodextrin (HPB) and methyl derivative of /3-cyclodextrin (MMB) were

179

Molecusol® cyclodextrin derivatives from the Pharmatec, Inc., Alachua, FL. L-Tartaric acid dimethyl ester (DMT) and Beeswax were Sigma reagents (Sigma Chemical Co., St. Louis, MO). U-86983 (2-(4-morpholinyl)- 8 -(3-pyridinylmethoxy)- 4H -1benzopyran-4-one) was kindly provided by Pharmacia and Upjohn Laboratories, Kalamazoo, MI. All organic solvents were either HPLC grade or American Chemical Society analytical grade reagents. 3.2. Preparation o f U-86983-loaded double-layer .films

The double-layer films were formulated with a U-86 containing layer and a drug-free protecting layer. The drug-containing layer consisted of U-86/ additive/PLGA in the ratio of 10:10:80 by weight, whereas the protecting layer consisted of PLGA only. The films were prepared by a combined solvent casting and melt-compressing method. For the drugcontaining layer, 320 mg PLGA, 40 mg of the additive compound and 40 mg of U-86 were dissolved in 10 ml of methylene chloride, except for HPB and MMB which were dispersed in the PLGA/ U-86 methylene chloride solution due to their insolubility in the organic solvent. Various types of matrices were solvent casted into Petri dishes precovered by a Teflon® film. The solvent was evaporated at room temperature in a hood and then at 37°C overnight. The PLGA matrices without additive and U-86 were prepared in the same way. The solvent casted films were first compressed at 110°C under 8000 to 10 000 psi to a thickness of - 8 0 / x m using a CARVER® Laboratory Press (Fred S. Carver, Inc., Menomonee Falls, WI). The two layers, one containing an additive and U-86, and the other layer with only PLGA, were overlaid on each other, and again compressed, but at lower pressure (2000 psi) to form a double-layer matrix with a final thickness of about 150 #m. Samples 1.0× 1.0 cm square were used in subsequent studies. 3.3. In vitro drug release

Double-chamber diffusion cells were used to measure the release rates of U-86 from the two layers of the matrix. The double-layer film was

180

C.X. Song et al. / Journal of Controlled Release 45 (1997) 177-192

placed between the two cells as a barrier. The cells on both sides of the film were filled with phosphate buffer solution (0.05 M, pH 7.4, PBS) and placed on a rotary shaker at 37°C. The buffer in both cells were replaced with fresh buffer at predetermined time intervals. The drug amounts released to both sides of the cell were assayed at 312 nm wavelength using a Perkin-Elmer (Norwalk, CT) Lambda 3B UV/VIS Spectrophotometer. Four parallel samples were tested for each type of matrix.

3.4. Mass loss, water uptake and morphology o f the matrix

Matrices with 10% additive and 90% PLGA without U-86 and control PLGA matrices were prepared by the procedure described above were used for mass loss, water uptake and morphologic evaluation. Pre-weighed (recorded as Wo) 1× 1 cm squares ( 1 5 0 / x m thickness) were used as of matrix samples were placed in 15 ml of PBS at 37°C. Three parallel samples were tested for each type of film. The medium was replaced with fresh PBS every day to eliminate the factor of autocatalysis of the matrix degradation by the polymer degradation products. At each time point, the samples were recovered and rinsed thoroughly with water. The surface water was removed by blotting and the wet weight of the matrices was recorded. The matrix samples were then dried by lyophilization to a constant weight and the dry weight was recorded. Water uptake was calculated according to the difference between the wet weight (Ww) and the dry weight (Wd) and expressed as: Water uptake (%) = ( % - Wd)IWd

(1)

The mass loss was expressed as: Mass loss (%) = ( Wo - Wa)IWo

(2)

The dried matrix samples were further used for morphological characterization. The morphological study of both the surface and the cross section of the matrices was carried out by scanning electron microscopy ($570 Hitachi, Cherry Hill, NJ). To obtain a cross fracture, the films were frozen in liquid nitrogen and sliced at - 5 6 ° C with a razor blade.

3.5. U-86 Solubility in water and in cyclodextrin aqueous solution

A mass of 3.5 g of HPB or MMB was dissolved in 10 ml of de-gassed distilled water to form a clear solution. To establish the solubility of U-86 in these solutions, an excess of amount of U-86 was stirred at 37°C. An aliquot of 1 ml of the solution was taken every 4 h and centrifuged at 14 000 rpm for 6 min. The supernatant was assayed for U-86 concentration by HPLC. The constant highest value was considered as the solubility of U-86 in these solutions. For the HPLC method of U-86 assay refer to our previous publication [13]. U-86 solubility in double distilled water without cyclodextrin was established in the same way.

4. Results and discussion 4. I. Theoretical considerations

Drug delivery rates from a film-type matrix can be characterized in terms of their kinetics and physical processes. In principle, the most common kinetics from a nondegradable monolithic film results in a release rate proportional to t 1~2. This square-root-oftime dependency was characterized by a declined rate with time reflecting a mass transport by simply diffusion [14]. However, the biodegradable monolithic film-type device is more complicated. Drug release from this type of matrix can be viewed as a result of two concurrent processes: (i) the diffusion of drug molecules from the polymer matrix, and (ii) the hydrolytic erosion or degradation of the polymers. Therefore, a biodegradable system may yield zero-order constant release with a simple monolithic device, if somehow the matrix degradation can compensate for the rate decline caused by a simple diffusion mechanism [15]. The degradation-mediated release is mainly through two processes: exposing drugs directly to the solvent at the releasing front for dissolution, and creating porous channels due to drug particle dissolution along with the loss of the degradation product. Pitt et al. [3] have described the first phase in the degradation of polyesters as only a decrease in molecular weight caused by random cleavage of the

C.X. Song et al. / Journal of Controlled Release 45 (1997) 177-192

ester linkage without detectable change in polymer mass. The second phase is the onset of the matrix weight loss. Kenley et al. [16] further found that no matrix mass loss is observed until the number-average molecular weight (Mn) of the polymer drops to 2800 Da. For a PLGA matrix with a molecular weight higher than 2800 Da, the drug release is mainly due to a diffusion mechanism resulting in a slow release period (first phase of release), as was observed in our studies. When the molecular weight of the matrix degraded to the point of the onset of mass loss, the erosion-mediated release mechanism becomes significant, resulting in a rapidly increased release rate (second phase of release). This is the classical biphasic release kinetics which is typically observed from the PLGA monolithic slab-type matrix [4,5]. In order to overcome such inherent drawbacks with a PLGA matrix, especially to achieve an overall constant release kinetics, our design considerations for the PLGA film system were concentrated on enhancing the initial drug release rate, and therefore, to eliminate the lag-time of drug release by using additives in the matrix. It has been demonstrated that hydrophilic molecules, especially if leachable, would enhance drug release from a hydrophobic matrix [17]. On the other hand, some lipophilic molecules, such as waxes, fatty esters, etc. will act as a sealant when added into the porous matrix, thus slowing down the drug release rate [18]. If porous channels are present, these may prove to be a predominant route for the drug release. Although all hydrophilic additives could hypothetically enhance drug release from a hydrophobic matrix, different mechanisms can be hypothesized due to the differences in molecular size, solubility and steric configurations of the additives utilized which will inevitably result in an overall different release kinetics~

4.2. Release rates of U-86 from the matrices with different additives The percent cumulative release of U-86 from the drug-containing layer of the double-layer matrix with different additives was plotted against time. As shown in Fig. 1, a typical biphasic release kinetics profile was observed for the PLGA matrix without any additives. It demonstrated a slow release phase

181

100

80

60

.=~ 40

20

0 ff.,,,,~-r~- T 0 4

HPB; .L

.

~ , 8 12 T i m e (day) •

F127;

M M B ; ---.-ii----- Wax

-

I 16

~

,

U 20

DMT; •

PLGA only;

Fig. 1. In vitro release of U-86983 from the drug-containing layers of the double-layer PLGA/U-86/additive matrices, showing the different release profiles with different additives.

up to 13 days followed by a rapid releasing phase. The additive-containing matrices showed different release profiles. The water-soluble additives (HPB, F127 and DMT) showed much higher drug release rates at an early with an apparent reversal of the control biphasic pattern. For instance, the matrices with HPB showed a burst release for the first 3 days followed by a relatively stable rate with 100% of the U-86 released within 13 days, whereas during the same time period, only 10% drug was released from the PLGA control matrices. The matrices with Pluronic® F127 resulted in a perfect constant release for the entire experimental period without any lag time. The DMT, a water-soluble but smaller molecule than F127, diminished the lag-time and smoothed the biphasic pattern. On the other hand, Beeswax, a more hydrophobic molecule, caused a prolongation of lag-time and delayed the onset of the second phase by 4 days. MMB, hydrophilic but less water-soluble than HPB, showed the same biphasic kinetics comparable to PLGA control matrices but advanced the onset of the second phase by 3 days. The understanding of the observed effects of additives on kinetics would be improved if information were available concerning the miscibility of the various additives with PLGA, the effect of additives

182

C.X. Song et al. / Journal of Controlled Release 45 (1997) 177-192

on water uptake, and the rate of chain cleavage of PLGA depending on the various additives. Further valuable studies would involve the effect of the additives on the glass transition temperature of the PLGA. Unfortunately, these experiments have not been done by others, and were beyond the scope of these initial studies.

4.3. The influence of the additive ratio in the PLGA matrix The effect of the amount of additive in the matrix was investigated utilizing DMT as a model. Fig. 2 shows that the percent of DMT in PLGA matrices significantly affects the release rate and the overall pattern of release kinetics. With an increase in the ratio of DMT to PLGA in the matrix, the drug release rate increased and the lag-time proportionally diminished. With 20% DMT, a burst release was observed for the first 24 h. These results suggest that the effect of additives on the release rates is positively proportional to the amount used in the matrix. Thus, it is possible to change the biphasic kinetics into a relatively monophasic profile by properly adjusting the ratio of additive to PLGA.

100

s0

~6 o ~ 40,

10% DMT 5% DMT PLGA only

v

,

3

,

6

,

9

,

t

,

i

12 15 18 Time (day)

t

,

21

I

24

,

t

27

•

i

30

Fig. 2. In vitro release of U-86983 from the double-layer PLGA! U-86 matrices with different amount of dimethyl tartarate (DMT) as additive: the effect of the ratio of DMT to PLGA in the matrix on U-86 release rate and pattern.

4.4. Kinetics considerations Of the six various types of matrices tested, the observed drug delivery profiles can be classified into the following three patterns: (1) a biphasic pattern starting with a slow release period (lag time), followed by a suddenly increased release rate (as the case of PLGA, Beeswax and MMB); (2) a biphasic pattern with an initial burst followed by a period of constant release (the HPB matrix); and (3) an overall constant release without phase transition (the F127 matrix). It was noted that there was a period of steady-state of release within each stage. A straight line can approximate the steady-state for each stage enabling a quantitative comparison among these matrices as illustrated in Fig. 3. The straight line representing more than 90% of the data points of each curve can be expressed by following simple linear equation: M~/M~ × 100(%) = Kt

(3)

where Mt/M ~ is the cumulative fraction released; and K, the slope of the straight portion of the curve, is the steady-state release rate (% release/day per 0.1 /xm thickness) for each stage. The constants and parameters delivered from equation Eq. (3) are listed in Table 1. The K value for the additive-containing matrix was compared with that of the PLGA control and expressed by 'ratio to PLGA,' which is the measure of the effect of additives on the release rate for each phase. For the same matrix, the ratio of the K values of the two phases (K2/K1) was calculated. This reflects how much the release rate varied from Phase I to Phase II. The transition time is the turning point of the two phases. As shown in Table 1, the PLGA control matrix showed the lowest release rate in the first phase (K1) and the higher rate in the second phase (K2), which was characterized by the highest K2/K1 value, indicating that PLGA demonstrated the largest variation in release rate during the entire release period. The water soluble hydrophilic additives, HPB, F127 and DMT, increased the U-86 release rate of Phase I by factors of 36.6, 7.7 and 4.4, respectively compared to PLGA control matrices (Table 1). Concerning the water-soluble additives (HPB, F127 and DMT), although each enhanced U-86 release rates, their effect on the release kinetics was

C.X. Song et al. / Journal of Controlled Release 45 (1997) 177-192

100 ['A: PLGA

B: MMB

183

C: Wax

s0

"~

40 20

o 80

60

Y,

•.=

40 20

0

0

5

10

15

20 0

5

Time (day)

10

15

20 0

4

8

12

16

20

Time (day)

Time (day)

Fig. 3. Release kinetics of U-86983 from the double-layer PLGA/U-86/additive matrices based on the biphasic concept often observed with PLGA matrix, showing that the linear slope and duration in each phase varied remarkably with different additives. The biphasic pattern was smoothed somewhat to a monophasic profile with the water-soluble additives (F127, DMT and HPB).

varied. As shown in Table 1, the differences in release rates were more significant in the first phase of release kinetics. The K1 values were 19.4, 4.09

and 2.31 for HPB, F127 and DMT, respectively. For the HPB containing matrix, 60% drug was released within the first 3 days followed by a reduced rate

Table 1 The parameters from the plots of cumulative release against time (Fig. 3): a comparison of the effect of additives on release rate and transition of the two phases Additive type

None (PLGA) HPB F127 DMT MMB Wax

Slope of the 1st Phase

Slope of the 2nd Phase

KI**

Ratio to PLGA

K2**

Ratio to PLGA

0.53 19.4 4.09 2.31 0.61 0.95

1.00 36.60 7.72 4.36 1.15 1.79

7.85 4.1 4.09 5.23 8.32 7.23

1.00 0.52 0.52 0.67 1.06 0.92

*The transition time from the first phase to the second phase. **K, Percent release/day/0.1 /zm thickness of matrix.

K2/K1

Transition* time (day)

14.8 0.2 1.0 2.3 13.6 7.6

13 3 None 10 10 17

184

c.x. Song et al. / Journal of Controlled Release 45 (1997) 177-192

during the second phase ( K 2 / K 1 = 0 . 2 ) . As a contrast, the D M T matrix had only about 30% of the drug released before the phase transition at the 10th day with a higher rate during the second phase ( K 2 / K l = 2 . 3 ) . The matrices with F127 demonstrated an overall constant release rate without phase transition characterized by the K 2 / K 1 value equal to 1.0 and with the regression coefficient equal to 1.0, thus indicating a perfect zero-order kinetics. It is important to note that during the second stage of release, matrices with HPB, F127 and D M T showed almost the same U-86 rates: K 2 = 4 . 1 0 , 4.09 and 5.23 for HPB, F127 and DMT, respectively. These results thus suggest that the release rate during the early stage is a key factor in determining the overall kinetics. Information from the mass loss and morphological studies revealed that different mecha-

100

A: PLGA

nisms apply to these various formulations of matrices. 4.5. M e c h a n i s m considerations

The mass loss profile of all six matrices followed almost the same biphasic pattern as shown in Fig. 4, with a plateau first phase and an onset of rapid mass loss (second phase). The P L G A control showed less than 3% mass loss for the first 13 days and followed by a rapid loss of weight. The same pattern was found for the M M B and Wax matrices. The matrices with F127, HPB and D M T had an immediate 1 0 15% weight loss within the first 3 days suggesting the quick leaching of these water soluble additives from the matrices. The drug release from the P L G A control, M M B and Wax matrices showed a good

"B: MMB

c:w i -

80

I

Drug Release

O

60 ~D

40 20 0 100

"D: DMT

F: HPB

"E: F127

80 O

,.A

60 "~

40

,o

20 i

0

I

10

i

I

,

I

20 30 Time (day)

i

I

40

I

,

0

I

10

r

I

I

20 30 Time (day)

I

I

40

10

0

I

i

I

20 30 Time (day)

i

I

i

40

Fig. 4. The profiles of mass loss of PLGA matrices with different additives compared with their drug release profiles. The mass loss accordance of drug release was seen for PLGA control, and PLGA with MMB and Wax matrices but not for those with water soluble additives (F127, DMT and HPB).

C.X. Song et al. / Journal of Controlled Release 45 (1997) 177-192

accordance with their mass loss profile as shown in Fig. 4A, Fig. 4B and Fig. 4C, i.e. the same biphasic pattern with almost the same transition time, indicating that drug release from these matrices was dominated by an erosion-mediated mechanism. However, the matrices containing F127 or HPB did not show the mass-loss accordance of release rate. Drug depletion from these types of matrices was much faster than their mass loss rate (Fig. 4E and Fig. 4F). There may be other mechanisms that are more significant than matrix erosion in governing drug release kinetics from the matrices with water-soluble additives. The matrices recovered after 1-day, 6-day and 18-day incubations in pH 7.4 PBS at 37°C were examined with scanning electron microscopy. The PLGA control matrices were intact during Phase I (Fig. 5A, Fig. 5B and Fig. 5C). A highly porous structure was observed only for the samples recovered at the 18th day (Fig. 5D). On the other hand, the matrices with water-soluble additives formed porous structure at a very early stage (Fig. 6, Fig. 7 and Fig. 8). The F127 matrices formed relatively large interconnected open pores throughout the matrix as shown in Fig. 6. Furthermore, the surface had almost the same porous structure as that of the cross section. This particular structure could be attributed to the chemical and physical features of the F127 molecules. As mentioned above, F127 has a high molecular weight and is both organic and water soluble. Obviously, such a molecular feature was highly favorable for generating the interconnected open pore structure when leached out from the matrix. According to Kuu et al. [19], these interconnected pores would form solvent-filled channels. Since the pore size in the matrix was much larger than that of the drug molecules, the restriction of the pores on diffusion was not significant [19]. Thus, it is hypothesized that the interconnected water channels in the F127 matrices were the main route for drug release, although further studies are needed to substantiate this. The DMT matrix also was noted to have an inner porous structure as seen from the cross-section micrograph, but with a much less porous surface (Fig. 7A, Fig. 7B and Fig. 7C). The pore size on its surface was much smaller than that of the F127

185

matrices, and the inner pores are not interconnected. These are separated by non-porous layers with a thickness of 3 - 5 # m . This particular pore structure was more likely caused by the leaching of the small molecules of DMT. Since there is no interconnected channel, drug release from this kind of matrix is still diffusion-controlled [16]. However, the inner porous structure greatly reduced the actual drug diffusion distance, thus resulting in a higher release rate than the PLGA control matrix. During the second phase, the DMT matrix showed a more porous structure (Fig. 7D) which was approximately the same as the PLGA control matrices (Fig. 5D) suggesting that the highly porous structure produced during Phase II was more likely due to the PLGA mass loss, that was responsible for the observed rate increase of U-86 release from the DMT matrices during Phase li. The micrographs of HPB matrices are shown in Fig. 8. The pore size in the HPB matrices was larger than that in the DMT matrices, but much smaller than that in the F127 matrices. One would expect a slower drug release rate from this kind of structure (HPB matrices) compared to F127. In fact, HPB caused an initial burst release and the overall rate was much higher than the F127 matrices. This phenomenon cannot be explained solely by the porous structure. Solubility studies showed that HPB increased the water solubility of U-86 by a factor of 23 compared to the U-86 in water. It is more likely that, in the case of HPB/PLGA matrix, most of the U-86 molecules are included in the cavity of HPB molecules forming a water-soluble cyclodextrin complex. However, there is no direct evidence of drug inclusion in the cyclodextrin molecules. The appropriate differential scanning calorimetry, and infrared spectroscopy studies were beyond the scope of the present experiments. Since HPB is highly water soluble, with the quick leaching of the HPB-U-86 complex, a large amount of U-86 included in the complex was thus released resulting in the observed burst effect. The water uptake profiles of the three releaseenhancing matrices are shown in Fig. 9. The results were in good accordance with the pore structures found from the morphological observations. With the inner-closed pores, the DMT matrix retained a large amount of water. On the other hand, with the interconnected open pores, the F127 matrices

186

C.X. Song et al. / Journal of Controlled Release 45 (1997) 177-192

Fig. 5. Scanning electron micrographs of the PLGA matrices without any additives after 1-day (A: surface), 6-day (B: surface; C: cross section) and 18-day (D) incubations in pH 7.4 phosphate buffer.

C.X. Song et al. / Journal of Controlled Release 45 (1997) 177-192

187

Fig. 6. Scanning electron micrographs of the PLGA/FI27 (90:10, w/w) matrices after 1-day (A: surface), 6-day (B: surface; C: cross section) and 18-day (D) incubations in pH 7.4 phosphate buffer, showing the interconnected open pore structure.

188

C.X. Song et al. / Journal of Controlled Release 45 (1997) 177-192

Fig. 7. Scanning electron micrographs of the PLGA/DMT (90:10, w/w) matrices after 1-day (A: surface), 6-day (B: surface; C: cross section) and 18-day (D) incubations in pH 7.4 phosphate buffer, showing the inner porous structure with a less porous surface. The style of the inner pores are closed.

C.X. Song et al. / Journal of Controlled Release 45 (1997) 177-192

189

Fig. 8. Scanning electron micro graphs of the PLGA/HPB (90:10, w/w) films after 1-day (A: surface), 6-day (B: surface; C: cross section) and 18-day (D) incubations in pH 7.4 phosphate buffer, showing the inner porous structure with semi-interconnected pores.

C.g. Song et al. / Journal of Controlled Release 45 (19.97) 177-192

190

180

DMT

~

ttPB

¢ 150

120

°°

0

i 0

I 4

,

I 8

i

I 12

,

I 16

,

I 20

Time (day) Fig. 9. Water uptake profiles of the PLGA matrices with different water soluble additives.

showed much less water uptake compared to the DMT matrix. HPB, with semi-interconnected pores, had a moderate water uptake level. It is noteworthy that in this study, the water uptake capacity was not directly related to drug release rate. Variations in the actual block copolymer composition were not explored in these studies. However, Ruiz and collaborators [20,21] have extensively investigated this area by examining the phenomena resulting changing the proportion of residual oligomers and the starting PLGA. This would be an interesting approach to consider in light of the findings of our studies with the additives investigated. The interest in the approach of Ruiz and colleagues would focus on the fact that their changes in residual oligomer content should not result in changes in the chemical features of the original polymer, merely the molecular weight distribution [20,21].

4.6. Drug-containing layer vs. drug-free layer In the double-layer matrix, one surface of the drug-containing layer (drug side) was covered by a drug-free PLGA film (polymer layer). Ideally, U-86 will release only from the uncovered surface of the drug layer so that when it is applied to the arterial

segment adventitially, the non-drug PLGA sealant layer could protect drug from loss into the surrounding area. U-86 release rate from both sides of the double-layer matrix was monitored separately and the results are plotted in Fig. 10. For every doublelayer matrix, the release rate of U-86 from the polymer side was significantly lower than that from the drug side. The amount of U-86 released from the polymer side stayed close to zero for the first 13 days. The ratio of drug release from the two sides was varied significantly with different types of matrices. As shown in Fig. 10A, for the HPB matrices 99% of the drug was released from the drug side and only 1% was released from the polymer side. Fig. 10B shows that for the Pluronic® F127 matrix, 90% U-86 was released from the drug side and 10% was released from the polymer side. As a contrast, for the PLGA control matrices, 65% U-86 was released from the drug side and 35% was released from the polymer side (Fig. 10C). These results indicate that the drug-free PLGA sealant layer can effectively prevent drug loss for a period of about 2 weeks.

5. Conclusions The feasibility of using water-soluble substances, such as F127, HPB and DMT, as additives for enhancing U-86 release rates and for modulating the release kinetics from PLGA monolithic matrices was demonstrated. These additives changed the biphasic release profile typically observed with PLGA matrix to a monophasic profile by increasing the release rate of the first phase. It is clear that these hydrophilic water-soluble additives are water-leachable and able to generate a highly porous structure. The pore characteristics, namely interconnected or closed, determine the mechanisms by which these release kinetics occurred. Besides water-solubility, such factors as molecular size, and steric configuration of the additives were among the most important features in generating pores of different dimensions and characteristics. F127 caused an interconnected openporous structure resulting in a dissolution-controlled zero order release kinetics. In contrast, the low molecular and water-soluble DMT produced internally closed pores in the matrix which were responsible

191

C.X. Song et al. / Journal of Controlled Release 45 (1997) 177-192

100 r A: F127 / ~ ~ g layer

~

D: PLGA

20 0 ~

0

0

4

120

8

12

16

20

0

B: HPB

1~

4

8

12

16

20

C: DMT

4o 20 j 0

~-6,0,o~..o~.~-~'

0

4

8

i

12

.

16

20 10 0

20

0

4

Time (day)

8

12

16

20

Time (day)

Fig. 10. In vitro release of U-86983 from the drug-free PLGA protection layer of the double-layer PLGA/U-86/additive matrices compared to that from the drug-containing layers.

for the combined dissolution/diffusion matrix erosion mechanism. HPB and medium size, with a reasonable semi-interconnected porous structure, resulted in the highest release rate~ that could be attributed to its cavity-forming configuration. It is hypothesized that HPB forms a drug-inclusion cyclodextrin complex. The dissolution of this complex could be the main route of U-86 release. Increasing the ratio of additives to P L G A in the matrix caused a more significant influence on U-86 release. In regard to the hydrophobic and less water soluble additives, no significant effect on drug release kinetics was seen in this study. In conclusion, using water-soluble additives in a P L G A matrix is a versatile practice to enhance drug release rate and to modulate the kinetics.

Acknowledgments This work was supported by Pharmacia and Upjohn, Inc., Kalamazoo, MI. The authors thank Ardith Bates for her assistance in the preparation of the manuscript.

References [1] D.H. Lewis, Controlled release of bioactive agents from lactide/glycolide polymers, in: M. Chasin and R. Langer (Eds.), Biodegradable Polymers as Drug Delivery Systems, Marcel Dekker, Inc. New York, NY, 1990, pp. 1-42. [2] P.E DeLuca, R.C. Mehta, A.G. Hausberger and B.C. Thanoo, Biodegradable polyesters for drug and polypeptide delivery, in: M.A. E1-Nokaly,D.M. Piatt and B.A. Charpentier (Eds.),

192

[3]

[4]

[5]

[6] [7]

[8]

[9]

[10]

[11]

[12]

C.X. Song et al. I Journal of Controlled Release 45 (1997) 177-192

Polymeric Delivery Systems: Properties and Applications, American Chemical Society, Washington, D.C., 1993, pp. 53-79. C.G. Pitt, M.M. Gratzl, G.L. Kimmel, J. Surles and A. Schindler, Aliphatic polyesters: II. The degradation of poly (d,Mactide), poly (E-caprolactone) and their co-polymers in vivo, Biomaterials 2 (1981) 215-220. D.H. Lewis and T.R. Tice, Polymeric considerations in the design of microencapsulated contraceptive steroids, in: G.I. Zatuchni (Ed.), Long-Acting Contraceptive Delivery Systems, Harper and Row, Philadelphia, 1984, pp. 77-95. S.S. Shah, Y. Cha and C.G. Pitt, Poly(glycolic acid-co-d,llactic acid): diffusion or degradation controlled drug delivery? J. Control. Release 18 (1992) 261-270. C.C. Haudenscbild, Pathobiology of restenosis after angioplasty, Am. J. Med. 94 (1993) 40S-44S, A. Villa, L. Guzman, W. Chen, G. Golomb, R. Levy and E. Topol, Local delivery of dexamethasone for prevention of neointimal proliferation in rat model of balloon angioplasty. J. Clin. Invest. 93 (1994) 1243-1249. F.G. Hutchinson and B.J.A. Furr, Biodegradable polymer systems for the sustained release of polypeptides, J. Control. Release 13 (1990) 279-294. Miller, J.M. Brady and D.E. Cnlright, Degradation rate of oral resorbable implants (polylactates and polyglycolates): rate modulation with changes in PLA/PGA co-polymer ratios, J. Biomed. Mater. Res. 11 (1977) 711-715. S. Cohen, T. Yshioka, M. Lucarelli, L.H. Hwaug and R. Langer, Controlled delivery systems for proteins based on poly(lactic/glycolic acid) microspheres, Pharm. Res. 8 (1991) 713-720. Z. Brich, F. Nimmerfall, T. Kissel and S. Bantle, Branched ter-polymers: Synthesis, characterization in vitro and in vivo degradation behaviors, Proc. Intern. Symp. Control. Rel. Bioact. Mater. 15 (1988) 95-96. M. Donbrow, Introduction and Overview, in: M. Donbrow (Ed.), Microcapsules and Nanoparticles in Medicine and Pharmacy, CRC Press, Boca Raton, FL, 1992, pp. 1-14.

[13] C.X. Song, V. Labhasetwar, H. Murphy, X. Qu, W.R. Humphrey, RJ. Shebuski and R.J. Levy, Formulation and characterization of biodegradable nanoparticles for intravascular local drug delivery, J. Control. Release, (1996) In press. [14] B. Bemer and S.M. Dinh, Fundamental concepts in controlled release, in: A. Kydonieus (Ed.), Treatise on Controlled Drug Delivery, Fundamentals, Optimization, Applications, Marcel Dekker, Inc. New York, NY, 1992, pp. 1-36. [15] K.W. Leong, Synthetic biodegradable polymer drug delivery systems, in: PJ. Tarcha (Ed.), Polymers For Controlled Drug delivery, CRC Press, Inc., Boca Raton, FL, 1991, pp. 128148. [16] R.A. Kenley, M.O. Lee, T.R. Mahoney II, and L.M. Sanders, Poly(lactide-eo-glycolide) decomposition kinetics in vitro and in vivo, Macromnlecules 20 (1987) 2398-2403. [17] T. Siki and Y. Okahata, Controlled release from a large capsule membrane corked with phosphatidylethanolamine bilayers: effect of temperature, pH and concentration of Ca 2+ ion. J. Microencaps. 2 (1985) 13-19. [18] EB. Deasy, M.R. Brophy, B. Eeanow and M.M. Joy, Effect of ethylcellulose grade and sealant treatments on the production and in vitro release of microencapsulated sodium salicylate. J. Pharm. Pharmacol. 32 (1980) 15-19. [19] W-Y. Kuu, R.W.Wood and T.J. Roseman, Factors influencing the kinetics of solute release, in: A. Kydonieus (Ed,), Treatise on Controlled Drug Delivery, Fundamentals, Optimization, Applications, Marcel Dekker, Inc. New York, NY, 1992, pp. 37-154. [20] J.M. Rniz, J.R Busnel and J.R Benoit, Influence of average molecular weights of poly(d,Mactic acid-co-glycolic acid) copolymers 50/50 on phase separation and in vitro drug release from microsphems. Pharm. Res. 7(9) (1990) 928934. [21] J.M. Ruiz and J.P, Benoit, In vivo peptide release from poly(d,l-lactic acid-co-glycolic acid) copolymer 50/50 microsphems. J. Control. Release 16 (1991) 177-186.