Novel sustained-release dosage forms of proteins using polyglycerol esters of fatty acids

Journal of Controlled Release 63 (2000) 319–329 www.elsevier.com / locate / jconrel

Novel sustained-release dosage forms of proteins using polyglycerol esters of fatty acids Yutaka Yamagata*, Katsumi Iga, Yasuaki Ogawa DDS Research Laboratories, Pharmaceutical Research Division, Takeda Chemical Industries, Ltd., 17 -85 Jusohonmachi 2 -chome, Yodogawa-ku, Osaka 532 -8686, Japan Received 16 July 1999; accepted 16 September 1999

Abstract In order to develop a novel delivery system for proteins based on polyglycerol esters of fatty acids (PGEFs), we studied a model system using interferon-a (IFN-a) as the test protein. A cylindrical matrix was prepared by a heat extrusion technique using a lyophilized powder of the protein and 11 different types of synthetic PGEFs, which varied in degree of glycerol polymerization (di- and tetra-), chain length of fatty acids (myristate, palmitate and stearate) and degree of fatty acid esterification (mono-, di- and tri-). In an in-vitro release study using an enzyme-linked immunosorbent assay (ELISA) as a detection method, the matrices prepared from a monoglyceride (used for comparison) and from diglycerol esters exhibited a biphasic release pattern with a large initial burst followed by slow release. In contrast, the matrices prepared from tetraglycerol esters showed a steady rate of release without a large initial burst. In an in vivo release study, initial bursts of IFN-a release were, also, dramatically reduced when the matrices were prepared from the tetraglycerol esters of palmitate and stearate, and the mean residence time (MRT) of IFN-a was prolonged, whereas the matrices prepared from monoglyceride and from diglycerol esters showed large initial bursts of IFN-a release. Since the release rates from the matrices prepared from the tetraglycerol esters of palmitate and stearate were governed by Jander’s equation modified for a cylindrical matrix, the release from those matrices was concluded to be a diffusion-controlled process. The bioavailability of IFN-a after implantation of the matrix formulation prepared using all types of PGEFs, except for tetraglycerol triesters, was almost equivalent to that after injection of IFN-a solution; consequently, IFN-a in these matrices appears to remain stable during the release period.  2000 Elsevier Science B.V. All rights reserved. Keywords: Sustained delivery of proteins; Polyglycerol esters of fatty acids (PGEF); Interferon-a (IFN-a); Enzyme-linked immunosorbent assay (ELISA); Jander’s equation; Hydrophile-lipophile balance (HLB); Stability in a matrix

1. Introduction Biologically active peptides and proteins are increasingly becoming a very important class of thera*Corresponding author. Tel.: 181-6-6300-6173; fax: 181-66300-6582. E-mail address: Yamagata [email protected] (Y. Yamagata) ]

peutic agents because of their extremely specific activity and high tolerability by the body. However, these peptides and proteins are poorly absorbed by the gastrointestinal tract after oral administration. In addition, because of their rapid clearance from the body, repeated injections are required if therapy is to be continued for a long period. To avoid the trauma and inconvenience of injections, a number of con-

0168-3659 / 00 / $ – see front matter  2000 Elsevier Science B.V. All rights reserved. PII: S0168-3659( 99 )00206-0

320

Y. Yamagata et al. / Journal of Controlled Release 63 (2000) 319 – 329

trolled delivery systems have been studied, and some successes were reported recently [1,2]. Those systems achieved sustained release of peptides over 1 month by using synthetic biodegradable polymers. For proteins, however, denaturation tends to occur in synthetic biodegradable polymers because of their higher susceptibility [3,4]. In other approaches, natural biomaterials, such as alginate [5], collagen [6–8], lipids and glycerides [9,10], have been used as the basis of matrices to achieve continuous delivery of proteins. A disadvantage of using natural materials as a matrix is that few variations in physico-chemical characteristics are available. Polyglycerol esters of fatty acids (PGEFs), which resemble glycerides in the structure, are expected to be biocompatible materials. Release of low-molecularweight drugs from microspheres can be regulated by selecting a PGEF with an appropriate hydrophilelipophile balance (HLB) value, which is determined by degree of glycerol polymerization, chain length of fatty acid and degree of fatty acid esterification [11]. Therefore, it would be of interest to investigate if the release of proteins from a PGEF-based matrix could be controlled by selecting the type of PGEF. Drug release from a wax-based matrix has been shown to be expressed by Higuchi’s equation [12,13]. The mechanism of release of low-molecularweight drugs from PGEF-based microspheres has been investigated and can be expressed by Jander’s equation [14]. However, it is difficult to evaluate the release of proteins because of their extreme susceptibility. For example, Lu et al. reported that proteins incorporated in biodegradable polymers degenerated during the release period in vitro [3,4]. In this study, we used an enzyme-linked immunosorbent assay (ELISA), which detects proteins having an intact conformation, to study the release of protein from the PGEF-based matrix. Fujioka et al. prepared a cylindrical matrix containing interferon-a (IFN-a) from collagen and demonstrated that IFN-a release from the matrix was accelerated by adding human serum albumin, which is a hydrophilic excipient [7]. Various kinds of PGEFs with different HLB values can be obtained [11], so it was of interest to investigate if IFN-a release from a PGEF-based matrix can be controlled by selecting a PGEF with an appropriate HLB value. We prepared a cylindrical matrix with various

PGEFs and a lyophilized powder of IFN-a, evaluated the release kinetics of IFN-a from these matrices both in vitro and in vivo, and elucidated the factors influencing the rate and mechanism of protein release from a PGEF-based cylindrical matrix.

2. Materials and methods

2.1. Materials Interferon-a 2a, synthesized at Takeda Chemical Industries, Ltd. (Osaka, Japan) [15], was dialyzed against distilled water and lyophilized prior to use. PGEFs (DGMM, DGMP, TGMM, TGMP, TGMS, TGDM, TGDP, TUDS, TGTP and TGTS) were kindly given by Sakamoto Yakuhin Kogyo Co., Ltd. (Osaka, Japan). The monoglycerides (MGML, MGMP and MGMS) used for comparison and DGMS were generously given by Riken Vitamin Co., Ltd. (Tokyo, Japan). Canferon  A300 (Takeda Chemical Industries) was used as a standard in the enzyme- linked immunosorbent assay (ELISA). The water used during the study was ultra pure grade from a Milli-Q plus  system (Millipore, Paris, France). Other chemicals were of reagent grade.

2.2. Characterization of PGEFS and monoglycerides The melting temperature (T m ) of each PGEF and monoglyceride was determined by differential scanning calorimetry (DSC) with a Perkin Elmer DSC 7. The sample was placed in an aluminum pan which was heated to 108C at a rate of 1008C / min. The HLB value of each sample was calculated by Griffin’s equation [16].

2.3. Thermo-stabiity of a lyophilized IFN-a powder during preparation Lyophilized IFN-a power (1 mg) was added to melted TGMS (43 mg), heated to 608C and mixed for 15 s. The mixtures were incubated at 608C for 0.1, 0.5 or 5 h. After cooling, each mixture was suspended in 200 ml of phosphate-buffered saline (PBS) containing 10% Block ace  (Snow Brand Milk Products Co. Ltd., Tokyo, Japan). The bio-

Y. Yamagata et al. / Journal of Controlled Release 63 (2000) 319 – 329

logical activity of IFN-a in the suspensions was determined by anti-viral assays employing in-vitro infection of Madin-Darby bovine kidney (MDBK) cells with vesicular stomatitis virus (VSV) [17].

321

ether anesthesia. After drug administrations, about 0.5 ml of blood was periodically withdrawn through the tail vein. The serum samples were stored at 2408C until assay of their IFN-a content by ELISA. Three rats received each dose level.

2.4. Matrix preparation Each PGEF or monoglyceride was heated to above its T m , and then lyophilized IFN-a powder was mixed in. The melted mixture was loaded into a 14 G stainless steel needle and extruded to form a rod with a diameter of 1.2 mm. The finished cylindrical matrix was produced by cuffing a 10 mm length from the extruded rod. The dose in each matrix was calculated from the actual weight of the implant, taking 1 mg of the lyophilized IFN-a powder as 1.74310 8 IU, as determined by ELISA.

2.5. In-vitro release study The in-vitro release study was conducted using PBS containing 0.02% bovine serum albumin (BSA; Cohn fraction V, Seikagaku Corporation, Tokyo, Japan) and 0.05% sodium azide as the dissolution medium as follows. The matrix was placed in a glass tube with a screw cap containing 50 ml of the dissolution medium, and the tube was incubated at 378C on a roller (STOVALL  Low Profile Roller ROL115; Life Science, Inc., Greensboro, NC) rotating at 15 rpm. At predetermined times, 0.2 ml of the medium was withdrawn and 0.2 ml of fresh medium was added. Each experiment consisted of 3–4 samples.

2.6. In vivo release study Male Sprague Dawley rats were purchased from Clea Japan, Inc. (Shizuoka, Japan) at 4 weeks of age and were used for the pharmacokinetic studies after a further 2 weeks of in-house rearing. An aqueous solution of IFN-a was prepared in PBS containing 0.5% BSA. The IFN-a solution was administered to rats by intravenous (iv) injection into a femoral vein or subcutaneous (sc) injection into the nape of the neck at a dose of 1.86310 8 IU / kg under ether anesthesia. The matrix implants were implanted through an implanting needle into the nape of the neck under

2.7. Determination of IFN-a concentration by ELISA Concentrations of IFN-a in the injection solution and serum samples were determined by ELISA in the following way. Each well of a 96-well immunoplates (Nunc A / S, Roskilde, Denmark) was coated with 100 ml of horse anti-human IFN-a polyclonal antibody (Japan Chemical Research Co., Tokyo, Japan) diluted with 50 mM carbonate buffer (pH 9.8) containing 0.05% sodium azide as a preservative. The plates were maintained at 48C until use. An aqueous solution containing 137 mM NaCl, 2.7 mM KCl, 1.5 mM KH 2 PO 4 , 8.1 mM Na 2 HPO 4 and 0.05% Tween 20 was used to wash the plate. The samples were added to the wells, followed by incubation at room temperature (RT) for 2 h. Mouse anti-human monoclonal antibody (Hayasibara Chemical Co., Okayama, Japan) was added and incubated at RT for 2 h. Any mouse antibody bound to the plates was detected using goat anti-mouse immunoglobulin G labeled with horseradish peroxidase (Chemicon International Inc., Temicula, CA). The plates were developed with TMblue  (TSICenter for Diagnostic Products, Milford, MA), and the reaction was terminated by adding 1 N H 2 SO 4 solution. The absorbance at 450 nm was measured using a MPR-32 platereader (Corona Electric Co., Ltd., Ibaragi, Japan), and the concentration in the samples was calculated using Canferon  300 as a standard. The lower limit of detection was 1.0 IU / ml.

2.8. Pharmacokinetic analysis Serum data following iv injection of the IFN-a solution were fitted to a two-compartment model, and serum data following sc injection of the solution were fitted to a one-compartment model. The maximum concentration (Cmax ) and the time taken to reach the maximum concentration (t max ) were read directly from the serum concentration–time data.

Y. Yamagata et al. / Journal of Controlled Release 63 (2000) 319 – 329

322

The area under the curve of serum concentration versus time (AUC) and the area under the curve of serum concentration3time versus time (AUMC) were determined by the trapezoidal method. The MRT was obtained using Eq. (1): MRT 5 AUMC /AUC

(1)

The bioavailable fraction (F ) of the matrix was calculated using Eq. (2): F 5 Dose (iv) / Dose (matrix) 3 AUC (matrix) /AUC (iv) 3 100

(2)

where Dose (iv) represents the dose of IFN-a in the iv injected solution, Dose (matrix) represents the dose in the matrix, AUC (iv) represents the AUC after iv injections of the solution and AUC (matrix) represents the AUC after implantation of the matrix formulation.

3. Results

degree of glycerol polymerization increases, and as the chain length of the fatty acid decreases; the degree of fatty acid esterification does not affect T m . The T m values of monoglycerides are 10–208C higher than those of PGEFs containing the same fatty acid. The HLB values of monoglycerides range from 4.0 to 5.4, indicating their hydrophobicity. Those of PGEFs, however, range widely from 4.5 to 9.6. The HLB values of PGEFs decrease as the degree of glycerol polymerization decreases, the chain length of the fatty acid increases and the degree of fatty acid esterification increases.

3.2. Thermo-stability of the lyophilized IFN-a powder The antiviral activity of the lyophilized IFN-a powder in a melted solution of TGMS was measured during incubation at 608C. The activity decreased about 95% at 6 min compared with the initial activity, but then did not change up to 5 h. There were no statistically significant differences between them.

3.1. Characteristics of PGEFs and monoglycerides 3.3. In-vitro release profiles The T m and HLB values of PGEFs and monoglycerides used in the present study are summarized in Table 1. The Tm values of PGEFs decrease as the Table 1 T m s and HLB values of various types of monoglycerides and PGEFs Materials

T m (8C)a

HLB b

MGML MGMP MGMS DGMM DGMP DGMS TGMM TGMP TGMS TGDM TGDP TGDS TGTP TGTS

59.0 71.8 73.0 40.2 52.3 61.3 35.5 48.2 50.9 35.7 48.4 50.9 48.4 50.7

5.4 4.8 4.0 6.7 6.1 5.8 9.6 8.9 8.4 7.4 6.8 6.4 5.1 4.5

a

T m of each PGEF was determined by DSC. b HLB value of each monoglyceride and PGEF was calculated from Griffin’s equation.

In-vitro release profiles of the matrices prepared using MGMP, DGMP and TGMP (the MGMP-, DGMP-, and TGMP-matrices, respectively) are shown in Fig. 1. When fatty acid esterification was restricted to a palmitate monoester, drug release from

Fig. 1. In-vitro release profiles of IFN-a from matrices prepared using esters of palmitate (open triangles, MGMP; solid triangles, DGMP; solid circles, TGMP). Each point represents the mean (n53–4). Bars represent SE.

Y. Yamagata et al. / Journal of Controlled Release 63 (2000) 319 – 329

323

the matrix was dramatically reduced when the degree of glycerol polymerization was four. About 60% of the IFN-a was released from the MGMP-matrix, which is based on monoglyceride and was used for comparison, and from the DGMP-matrix within 24 h. On the contrary, the initial drug release from the TGMP-matrix was significantly reduced and it was followed by a constant rate of release. The amount of IFN-a released from the MGMP-matrix has been lost during the in-vitro release study. For tetraglycerol ester of palmitate and stearate, drug release was reduced as the degree of fatty acid esterification increased (Fig. 2A, B and C). Drug release from the TGMP- and TGDP-matrix was slower than that from the TGMS- and TGDS-matrix, whereas the TGTPmatrix released IFN-a faster than the TGTS-matrix. The TGMM-matrix melted and dissolved in the dissolution medium within 24 h, so that drug release was quite rapid (Fig. 2A). In contrast, initial drugrelease from the TGDM-matrix was significantly slower than that from the TGDP- and TGDS-matrices (Fig. 2B).

3.4. In-vivo release study The total body clearance (CL total ) and the F value of the IFN-a solution derived from the serum concentration–time profiles after iv and sc injection of the solution are 494.9660.8 ml / h / kg and 41.962.9%, respectively (mean6SE). The MRT of IFN-a after sc injection of the solution is 2.51 h, indicating that IFN-a molecules were immediately eliminated from the body. The serum IFN-a concentration–time profiles after sc implantation of the matrix prepared from the monoglycerides are shown in Fig. 3 A. Serum IFN-a levels were prolonged as the chain length of the fatty acid in the monoglycerides was increased. The MGMS-matrix maintained serum IFN-a levels for 7 days. Serum IFN-a concentration–time profiles after sc implantation of the matrices prepared from PGEFs composed of diglycerol are shown in Fig. 3B. In this case, the DGMP-matrix maintained serum IFN-a levels for 6 days. Serum IFN-a concentration–time profiles after sc implantation of the matrices prepared from PGEFs composed of tetraglycerols are shown in Fig. 4. For the tetraglycerol monoesters (Fig. 4A),

Fig. 2. In-vitro release profiles of IFN-a from matrices prepared using (A) tetraglycerol monoesters (open triangles, TGMM; open circles, TGMP; open squares, TGMS), (B) tetraglycerol diesters (solid triangles, TGDM; solid circles, TGDP; solid squares, TGDS) and (C) tetraglycerol triesters (solid rhombuses, TGTP; x, TGTS). Each point represents the mean (n53–4). Bars represent SE.

the TGMP-matrix caused prolonged high serum IFNa levels as compared with the other matrices. Its release pattern was first order. For the tetraglycerol diesters (Fig. 4B), the TGDP- and TUDS-matrices reduced the initial burst and maintained high serum IFN-a levels for 7 days. Release patterns from these matrices were pseudo zero order. For the tetraglycerol triesters (Fig. 4C), the TGTP- and TGTSmatrices significantly reduced the initial burst, but

324

Y. Yamagata et al. / Journal of Controlled Release 63 (2000) 319 – 329

Fig. 3. Serum IFN-a levels after implantation of a matrix containing IFN-a prepared using (A) monoglycerides (open triangles, MGML; open circles, MGMP; open squares, MGMS) and (B) diglycerol esters (solid triangles, DGMM; solid circles, DGMP; solid squares, DGMS) at a dose of about 2.0310 8 IU / kg in rats. Each point represents the mean (n53). Bars represent SE.

serum IFN-a levels could not be detected after day 9. Both the TGMM- and TGDM-matrices released almost all the IFN-a within 3 days. Pharmacokinetic parameters obtained from the serum IFN-a profiles shown in Figs. 3 and 4 are summarized in Table 2. The value of t max is not significantly different between the matrices, whereas the Cmax values for the TGMP-, TGDP-, TGDS-, TGTP- and TGTS-matrices are decreased by almost one-half when compared with those for the matrices prepared from the monoglycerides and the diglycerol esters. Corresponding to the reduction of Cmax , their MRTs become longer, except for the TGTS-matrix. The bioavailability of IFN-a from the matrices, except for the TGTP- and TGTS-matrices, is almost equivalent to that of the solution.

Fig. 4. Serum IFN-a levels after implantation of a matrix containing IFN-a prepared using (A) tetraglycerol monoesters (open triangles, TGMM; open circles, TGMP; open squares, TGMS), (B) tetraglycerol diesters (solid triangles, TGDM; solid circles, TGDP; solid squares, TGDS) and (C) tetraglycerol triesters (solid rhombuses, TGTP; x, TGTS) at a dose of about 2.0310 8 IU / kg. Each point represents the mean (n53). Bars represent SE.

4. Discussion Since the oral bioavailability of proteins is very low because of their susceptibility to the acidic environment and to proteolytic enzymes, parenteral

Y. Yamagata et al. / Journal of Controlled Release 63 (2000) 319 – 329

325

Table 2 Pharmacokinetic parameters obtained from the serum IFN-a levels after administration of matrices prepared from various triglyceride and PGEFs to rats a Formulation

Dose (310 8 IU / kg)

t max (h)

Cmax (310 4 IU / ml)

MRT (h)

F (%)

MGML-matrix MGMP-matrix MGMS-matrix DGMM-matrix DGMP-matrix DGMS-matrix TGMM-matrix TGMP-matrix TGMS-matrix TGDM-matrix TGDP-matrix TGDS-matrix TGTP-matrix TGTS-matrix

1.96 2.15 2.09 2.19 2.10 1.71 1.70 2.13 2.10 2.53 2.01 2.07 1.90 1.73

2.00 (0.00) 3.33 (1.10) 2.33 (0.72) 3.33 (0.54) 2.00 (0.00) 2.33 (1.89) 2.00 (0.00) 2.00 (0.00) 4.67 (0.54) 2.00 (0.00) 4.67 (0.54) 2.00 (0.00) 1.00 (0.00) 1.00 (0.00)

2.22 1.07 2.04 2.30 1.76 1.70 1.96 0.84 1.50 3.70 0.43 0.44 0.09 0.21

5.40 (0.29) 11.7 (0.79) 16.8 (3.14) 5.93 (0.22) 18.8 (6.53) 15.5 (6.62) 6.97 (0.43) 40.3 (15.4) 9.36 (0.70) 4.37 (0.53) 76.1 (14.0) 93.1 (9.62) 89.2 (4.27) 18.1 (3.48)

45.0 (3.03) 30.1 (2.60) 42.7 (7.53) 47.3 (10.8) 44.5 (8.27) 39.9(5.76) 45.5 (3.85) 38.7 (4.59) 42.1 (5.28) 44.1 (7.10) 45.8 (3.72) 34.8 (6.45) 2.42 (0.61) 3.35 (0.55)

a

(0.17) (0.02) (0.58) (0.39) (0.04) (0.13) (0.22) (0.40) (0.51) (0.03) (0.08) (0.14) (0.04) (0.07)

Each value is the mean (n53), and value in parentheses is S.E.

administration is currently required to achieve therapeutic activity. However, the frequent injections cause trauma and inconvenience to patients. To minimize the drawbacks of frequent parenteral administration, a number of attempts have been made to develop long-acting parenteral preparations of protein drugs. Natural substances, such as alginate [5], collagen [6–8], lipids and glycerides [9,10], are used as a matrix base for sustained release of proteins due to their high biocompatibility. However, the characteristics of excipients derived from natural substances are limited. For instance, glycerides, used as a matrix base for sustained release of insulin have limited variations in the HLB value and T m [9]. On the contrary, a variety of types of PGEFs, which are synthetic substances and resemble glycerides in structure, can be obtained by changing the degree of glycerol polymerization, the chain length of the fatty acid and the degree of fatty acid esterification [11]. Eleven types of PGEFs were used in this study to prepare cylindrical matrices containing IFN-a as a model protein (Table 1). The HLB values increase as the degree of glycerol polymerization increases, the chain length of the fatty acid decreases and fatty acid esterification decreases. The available range of HLB values (4.5–9.6) is wider than that of monoglycerides (4.0–5.4), indicating that a wider variety

of controlled release properties can be achieved with matrices prepared with PGEFs as compared with those prepared with monoglycerides. The T m values decrease as the degree of glycerol polymerization increases and the chain length of the fatty acid decreases. A lower T m favors protein stability because a heat extrusion technique was used to prepare the cylindrical matrix. Zaks et al. reported that a lyophilized powder of porcine pancreatic lipase could withstand heating at 1008C for many hours and exhibited a high catalytic activity at that temperature [18]. We postulated that a lyophilized powder of IFN-a might be stable at the T m of PGEFs. In fact, the antiviral activity of IFN-a was maintained for 5 h in melted TGMS at 608C. Thus, we produced a cylindrical matrix of 10 mm in length and 1.2 mm in diameter by the heat extrusion technique using various PGEFs and the lyophilized powder of IFN-a. The initial release of IFN-a from the matrix prepared with the monoglyceride (MGMP) used for comparison and the diglycerol ester (DGMP) was high in the in-vitro release studies (Fig. 1). On the contrary, the initial drug release from the matrix prepared from the tetraglycerol ester (TGMP) was significantly reduced. In accord with the in vitrorelease study, a high initial release from the matrices prepared from the monoglycerides (MGML, MGMP

326

Y. Yamagata et al. / Journal of Controlled Release 63 (2000) 319 – 329

and MGMS) and the diglycerol esters (DGMM, DGMP and DGMS) was observed in the in-vivo release study (Fig. 3A and B). The values of Cmax after implantation of these matrices were high (Table 2). However, the matrices prepared from the tetraglycerol esters of palmitate and stearate dramatically inhibited the initial burst (Fig. 4A, B and C). The MRT after the implantation of these matrices became longer in inverse proportion to Cmax (Table 2). Sullivan et al. reported that the release of cyclazocine, a narcotic antagonist, from matrices prepared using glycerides became slower as the degree of glycerol esterification increased [10], indicating that the higher the HLB value of the matrix, the faster the drug is released. The HLB values of PGEFs increase as glycerol polymerization increases, meaning that our data conflict with those of Sullivan et al. However, the present study is the first investigation to evaluate the effect of glycerol polymerization on protein release from a matrix prepared using tetraglycerol esters. There might be other factors besides HLB that inhibit protein release from a matrix prepared from tetraglycerol esters. One possible factor is that the hydroxy groups of the polyglycerol may reduce the mobility of macromolecules such as proteins in the matrix after hydration. We investigated protein release from matrices prepared using the tetraglycerol esters in more detail. Drug release in vivo from the matrices prepared from the tetraglycerol esters of myristate (TGMM and TGDM) was rapid (Fig. 4A and B) because their T m values were below 378C, resulting in melting in the body. In the in-vitro release study, the TGMMmatrix melted and dissolved in the dissolution medium within 24 h, so that drug release was quite rapid (Fig. 2A). In contrast, initial drug-release in vitro from the TGDM-matrix was significantly slower than from the TGDP- and TGDS-matrices (Fig. 2B) although it melted at 378C (Table 1). Since TGDM is more lipophilic than TGMM, the melted TGDM-matrix did not dissolve in the dissolution medium. It seems that the diffusion of IFN-a molecules is inhibited in the melted TGDM-matrix. For the tetraglycerol esters of palmitate and stearate, drug release in vitro was reduced as the degree of fatty acid esterification increased (Fig. 2A, B and C). Release rate of low-molecular weight-drugs from

PGEF-based microspheres have been shown to be governed by Jander’s equation [14]. However, that equation refers to a spherical matrix and can not be applied to the cylindrical matrices used in the present study because of the difference in the shape. Thus, we modified Jander’s equation to apply to the drug release rate from a cylindrical matrix. Fig. 5 shows schematically the state of the cylindrical matrix during drug release. A cylindrical, matrix with radius R and length L begins to release the drug at time t50. At time5T, the drug has been released from the matrix to a depth of d, and the part of the depth d is a released part. The drug remains in the part of the matrix with radius (R2d ). If it is assumed that: (a) the matrix maintains its initial shape because of its insolubility, (b) the fraction of drug in diffusive passage across the released part is negligible in comparison with the drug remaining in the unreleased part, and (c) leaching of the drug takes place at the release boundary, then x, the fraction of drug released at time t, is expressed as the following equation: x 5 (p R 2 L 2 p (R 2 d )2 (L 2 2d )) /p R 2 L 5 1 2 (1 2 d /R)2 (1 2 2d /L)

(3)

where d is the depth of the layer released from the matrix surface. If d < L, Eq. (3) approximates to Eq. (4): x 5 1 2 (1 2 d /R)2

(4)

which can be rearranged to:

d 5 R(1 2 (1 2 x)1 / 2 )

(5)

Since the rate of increase of the depth is in inversely

Fig. 5. Schematic description of drug release from a cylindrical matrix.

Y. Yamagata et al. / Journal of Controlled Release 63 (2000) 319 – 329

proportional to the depth, i.e., the square of d is proportional to time, the following equation is obtained:

d 2 5 k9 t

(6)

where k9 is a constant. According to Eqs. (5) and (6), the amount of drug released at time t is expressed as Eq. (7) using k9 or k (5k 9 1 / 2 /R): 1 2 (1 2 x)1 / 2 5 (k 9 1 / 2 /R) t 1 / 2 5 k t 1 / 2

(7)

The exponent of (12x) in Eq. (7) changes to 1 / 2 from 1 / 3 in Jander’s equation developed for the spherical matrix [14] because it is assumed that the release from both plane surfaces is negligible, i.e., release occurs in two dimensions. The release of IFN-a from the matrices prepared with tetraglycerol esters of palmitate and stearate was plotted according to the modified Jander’s equation (Eq. (7)). When 12(12x)1 / 2 was plotted as a function of square root of time, a straight line was obtained (Fig. 6). These results indicate that the release of IFN-a from the matrices prepared using tetraglycerol esters of palmitate and stearate likely to be a diffusion-controlled process within the matrix. Akiyama et al. reported that the release of low-molecular-weight drugs from microspheres prepared with various HLB values was faster as the HLB values increased [11]. Since increasing esterification decreases the HLB value (Table 1), release from the matrix prepared using

Fig. 6. Plot of square root time versus 12(12x)1 / 2 for in-vitro release of IFN-a from matrices prepared using various PGEFs: solid triangles, TGMP; solid circles, TGDP; solid squares, TGTP; open triangles, TGMS; open circles, TGDS; open squares, TGTP. Each line represents the results of least-square linear regression.

327

palmitate and stearate esters with lower HLB values probably becomes slower. For the stearate esters, the release rate increases as the HLB values of the PGEFs used is higher (Fig. 6). Our results confirmed that release of macromolecules from stearate esterbased matrices also becomes faster when stearate esters with higher HLB values are used for the matrix base. On the contrary, drug release from palmitate ester-based matrix was slower while the HLB values of used PGEFs was higher (Fig. 6). These results suggest that some factors may inhibit the diffusion of macromolecules in the palmitate ester-based matrix. In the in-vitro release study, initial drug-release from the TGDM-matrix was significantly slow, although TGDM melted in the dissolution medium. It is postulated that some fractions in palmitate esters might melt at the body temperature. Further investigation of protein release from melted PGEF-based matrices is necessary for complete understanding of these phenomenon. In the in-vivo release study, the MRT of IFN-a after implantation of the matrices prepared from the tetraglycerol esters of palmitate increases in proportion to the degree of esterification (Table 2). Since the increase of esterification makes the HLB value lower (Table 1), release from matrices prepared using palmitate esters with lower HLB values probably becomes slower [10]. For the tetraglycerol esters of stearate, the MRT of TGDS is longer than that of TGMS, but the MRT of TGTS is shorter than that of TGDS (Table 2). Daily injections of a solution of human IFN-a, which is a foreign molecule to animals, were reported to cause antibody generation in several species [19]. Therefore, low serum IFN-a levels could not be detected by ELISA because of neutralization of serum IFN-a by the antibodies produced from 7 days after implantation. The bioavailabilities obtained after sc implantation of all matrices prepared using PGEFs, except for TGTP and TGTS, are almost equivalent to that obtained after sc injection of IFN-a solution (Table 2). Lu et al. recently reported that proteins entrapped in microspheres prepared using copoly (dl-lactic / glycolic acid) (PLGA) were denatured during the in-vitro release test [3,4], indicating that materials which ensure stability are required for the sustained delivery of proteins. Our results suggest that PGEFs are suitable materials for the sustained release of pro-

Y. Yamagata et al. / Journal of Controlled Release 63 (2000) 319 – 329

328

teins because of the limited protein denaturation observed during the release periods. Serum IFN-a levels after implantation of the TGDP-matrix are almost constant for 7 days. To evaluate the extent of in-vivo release from the TGDP-matrix, we calculated the serum levels of the drug at steady state at a dose of 2.01310 8 IU / kg / week using the kinetic parameters obtained for the IFN-a solution. If the system is assumed to obey linear kinetics with this low serum IFN-a level, the following applies: R iv 5 R sc F 5 (2.01 3 10 8 IU / kg / week / 7 days / 24 h) (0.419) 5 5.01 3 10 5 IU / kg / h

(8)

followed by a constant rate of release. It was suggested that some factors prevent release from matrices prepared using the tetraglycerol esters of palmitate and stearate. Since the release rate from these matrices fitted well to Jander’s equation modified for the cylindrical matrix, drug release from these matrices is a diffusion-controlled process. The bioavailability of IFN-a after implantation of matrices prepared using all PGEFs, except for TGTP and TGTS, was high, indicating that IFN-a in these matrices was stable during the release period. Since the TGDP-matrix maintained serum IFN-a levels almost equivalent to the theoretically calculated Css for 1 week, it may be a suitable dosage form of IFN-a for 1-week sustained delivery.

Css 5 R iv / CL total 5 (5.01 3 10 5 IU / kg / h) /(494.9 ml / h / kg) 5 1012 IU / ml

6. Notations (9)

where Css is the steady-state drug level for 2.01310 8 IU / kg / week, R sc is the subcutaneous constant infusion rate, R iv is the intravenously available constant infusion rate and, CL total represents the total body clearance of the drug after iv injection. R sc is calculated as if IFN-a is constantly infused for a week at a dose of 2.01310 8 IU / kg. The serum IFN-a level is almost equivalent to the calculated Css , indicating that the TGDP-matrix is suitable for a 1-week sustained release formulation of IFN-a. Infused delivery of IFN-a has been shown to have an equivalent effect to daily injections and to reduce the incidence of adverse effects such as fever [20]. The TGDP-matrix developed in this study might allow continuous delivery of IFN-a.

PGEF; MGML; MGMP; MGMS; DGMM; DGMP; DGMS; TGMM; TGMP; TGMS; TGDM; TGDP; TGDS; TGTP; TGTS;

polyglycerol ester of fatty acid, monoglycerol monolaurate, monoglycerol monopalmitate, monocglycerol monostearate, diglycerol monomyristate, diglycerol monopalmitate, diglycerol monostearate, tetraglycerol monomyristate, tetraglycerol monopalmitate, tetraglycerol monostearate, tetraglycerol dimyristate, tetraglycerol dipalmitate, tetraglycerol distearate, tetraglycerol tripalmitate, tetraglycerol tristearate

Acknowledgements 5. Conclusions In the present study, a sustained-release dosage form of a protein was produced using a lyophilized powder of IFN-a as a model protein and 11 types of PGEFs. In both in-vitro and in-vivo release studies, IFN-a release from matrices prepared from monoglycerides and diglycerol esters was initially high. On the contrary, the initial release from matrices prepared using the tetraglycerol esters of palmitate and stearate was significantly reduced and was

The authors greatly appreciate Dr. Osamu Nishimura in our company for supplying IFN-a. The authors gratefully acknowledge the technical assistance of Ms. Sakae Yonezawa for ELISA.

References [1] Y. Ogawa, M. Yamamoto, H. Okada, T. Yashiki, T. Toguchi, A new technique to efficiently entrap leuprolide acetate into

Y. Yamagata et al. / Journal of Controlled Release 63 (2000) 319 – 329

[2]

[3]

[4]

[5] [6]

[7]

[8]

[9] [10]

[11]

microcapsules of polylactic acid or copoly (lactic / glycolic) acid, Chem. Pharm. Bull. 36 (3) (1988) 1095–1103. T. Heya, H. Okada, Y. Tanigawara, Y. Ogawa, H. Toguchi, Effects of counteranion of TRH and loading amount on control of TRH release from copoly (dl-lactic / glycolic acid) microspheres prepared by an in-water drying method, Int. J. Pharm. 69 (1991) 69–75. W. Lu, T.G. Park, Protein release from poly (lactic-coglycolic acid) microspheres: protein stability problems, PDA, J. Pharm. Sci. Tech. 49 (1) (1995) 13–19. W. Lu, T.G. Park, In vitro release profiles of eristostatin from biodegradable polymeric microspheres: protein aggregation problem, Biotechnol. Prog. 11 (2) (1995) 224–227. W.R. Gombotz, S.F. Wee, Protein release from alginate matrices, Adv. Drug Delivery Rev. 31 (1998) 267–285. K. Fujioka, Y. Takada, S. Sato, T. Miyata, Novel delivery system for proteins using collagen as a carrier materials: the minipellet, J. Control. Release 33 (1995) 307–315. K. Fujioka, Y. Takada, S. Sato, T. Miyata, Long-acting delivery system of interferon: IFN minipellet, J. Control. Release 33 (1995) 317–323. K. Fujioka, M. Maeda, T. Hojo, A. Sano, Protein release from collagen matrices, Adv. Drug Delivery Rev. 31 (1998) 247–266. P.Y. Wang, Lipid as excipient in sustained release insulin implants, Int. J. Pharm. 54 (1989) 223–230. M.F. Sullivan, P.S. Ruemmler, D.R. Kalkwarf, Sustained administration of cyclazocine for antagonism of morphine, Drug Alc. Depend. 1 (1975 / 76) 415–428. Y. Akiyama, M. Yoshioka, H. Horibe, S. Hirai, N. Kitamori, H. Toguchi, Novel oral controlled-release microspheres

[12]

[13]

[14]

[15]

[16] [17]

[18] [19] [20]

329

using polyglycerol esters of fatty acids, J. Control. Release 26 (1993) 1–10. J.B. Scbwartz, A.P. Simonelli, W.I. Higuchi, Drug release from wax matrices I: analysis of data with first-order kinetics and with the diffusion-controlled model, J. Pharm. Sci. 57 (1968) 274–277. J.B. Schwartz, A.P. Simonelli, W.I. Higuchi, Drug release from wax matrices II: application of a mixture theory to the sulfanilamide-wax system, J. Pharm. Sci. 57 (1968) 278– 282. Y. Akiyama, M. Yoshioka, H. Horibe, S. Hirai, N. Kitamori, H. Toguchi, Mechanism of drug release from polyglycerol esters of fatty acid-based microspheres, J. Control. Release 27 (1993) 37–45. S. Honda, H. Sugino, K. Nishi, A. Kakinuma, Purification of human leukocyte interferon A derived from Eschericha coli: an aging process to prevent the formation of its oligomers, J. Biotechnol. 5 (1987) 39–51. W.C. Griffin, Classification of surface-active agents by HLB, J. Soc. Cosmet. Chem. 1 (1949) 311–327. M.J. Kramer, R. Dennin, C. Kramer et al., Cell and virus sensitivity studies with recombinant human leukocyte interferon, J. Interferon Res. 3 (1983) 425–435. A. Zaks, A.J. Russell, Enzymes in organic solvents: properties and applications, J. Biotech. 8 (1988) 259–270. P.W. Trown, R.J. Wills, J.J. Kamm, The preclinical development of Roferon-A, Cancer 5 (8) (1986) 1648–1656. T. Poynard, V. Leroy, M. Cohard et al., Activity and tolerance of a continuous subcutaneous infusion of interferon-a2b in patients with chronic hepatitis C, J. Interferon Cytokine Res. 17 (1997) 665–670.