Preparation of thermo-responsive membranes. I1 Iwao Nozawa,* Yosuke Suzuki, and Shuji Sat0 Applied Research Laboratory IlI, Lion Co., Ltd., 7-13-12 Hirai, Edogawa-ku, Tokyo 132, Japan Kenji Sugibayashi and Yasunori Morimoto Faculty of Pharmaceutical Sciences, Iosai University, 1-1 Keyakidai, Sakado, Saitama 350-02, Japan Two types of liquid crystal (LC)-immobilized membranes were prepared by a soaking method and sandwich method to control the permeation of indomethacin, as a model drug, in response to local and systemic fever. Monooxyethylene trimethylolpropane tristearate (MTTS) was used as a model LC because it has a gel-liquid crystal phase transition temperature near the body temperature, 39-40°C in phosphate buffered saline (pH 7.4). Two porous polypropylene (PP) membranes were soaked into 20% MTTS chloroform solution in the soaking method, and two PP membranes were poured with the melted MTTS and pressed in the
sandwich method. Thermo-response efficacy of the soaked membrane was dependent upon the content of MTTS in MTTS membrane, and the MTTS content above the void volume of PP membrane (38%) was needed for high efficacy. On the other hand, the sandwich membrane exhibited higher thermo-response efficacy than the soaked membrane, because more LC was embedded in the pores of sandwich membrane than that of the soaked membrane. The sandwich membrane permeation of indomethacin was sharply controlled by temperature changes between 32 and 38°C.
INTRODUCTION
For developing transdermal therapeutic systems (TTS), achievement of controlled release is one of the most important subjects. For this purpose several polymers can be used as membranes and/or matrices for sustaining drug release.’ For example, Transderm-Scop (Alza Co. and Ciba-Geigy L td.)2-4and Transderm-Nitro (Alza Co. and Ciba-Geigy Ltd.)56have been fabricated with an ethylenehinylacetate copolymer (EVA) porous membrane for sustained release and Nitrodisc (Searle Ltd.) has a silicon elastomer matrix for diffusion contr01.”~However, these TTS are not able to release drugs according to the condition of the patients or a preprogrammed schedule. Recently, intelligent materials which change their physical properties in response to external stimuli, such as temperature, photoirradiation, and pH, have been studied and they have been attractively applied to drug delivery system. *To whom correspondence should be addressed at Faculty of Pharmaceutical Sciences, Josai University, 1-1Keyakidai, Sakado, Saitama 350-02, Japan. Journal of Biomedical Materials Research, Vol. 25, 577-588 (1991) 0 1991 John Wiley & Sons, Inc. CCC 002l-9304/91/050577-12$4.00
NOZAWA ET AL.
578
In our previous paper: liquid crystal (LC)-immobilized membranes were studied to fabricate drug delivery systems with thermo-sensitivity. LCabsorbed membranes obtained by absorbing LC into porous hydrophobic polymer membranes could control the permeation of drug in a pulsatile manner in response to minute temperature changes. In the present article, we describe the influence of the preparation methods for the LC-immobilized membranes on the thermo-response efficacy. EXPERIMENTAL
Materials Monooxyethylene trimethylolpropane tristearate (MTTS) was purchased from Nihon Emulsion Co., Ltd. (Tokyo, Japan). The structural formula is shown in Figure 1. Polytetraf liioroethylene (PTFE) membrane filter (PT-020, Toyo Roshi Co., Tokyo, Japan) and porous polypropylene (PP) membrane (Celgard 2400, Hoechst-Celanese Co., USA) were used as MTTS supporting membranes. The properties of these membranes are shown in Table I. Indomethacin (IND) was a commercial product of Japanese Pharmacopoeia grade. All other chemicals were of reagent-grade products. Preparation alf thermo-responsive membranes Thermo-responsive membranes were prepared by a soaking method and a sandwich method. Two PI' membranes or one PTFE membrane filter were soaked in given concentrations of MTTS chloroform solution for 1 h at room temperature and then dried in uacuo for 24 h in the soaking method. Melted MTTS was ploured between two Celgard 2400 sheets, and the sandwiched sheets were rolled out to squeeze excess MTTS and was pressed at 25 x lo3 kg/cm2 for 10 min in the sandwich method. MTTS contents in the soaked membrane prepared by using 20% MTTS chloroform solution and in the sandwich membrane were 48 and 38% per membrane weight, respectively. The thicknesses of these membranes were 62 and 55 pm, respectively. MTTS contents were calculated from the difference of membrane weights before and after 3 times washing of MTTS-immobilized membranes in EtOH at 40°C to complletely remove MTTS from the membrane.
CH3CH2C
t
CH20CH2CH20COC17H35 CH20CH2CH20COC17H35 CH20CH2CH20COC,7H35
Monooxyethylene T r i m e t h y l o l p r o p a n e T r i s t e a r a t e ( M T T S )
Figure 1. Structural formula of liquid crystal.
THERMO-RESPONSIVE MEMBRANES. I1
579
TABLE I The Properties of Two Types of LC-Supporting Polymer Membranes
(w-4
Thickness (Pm)
Void Volume"
Membrane
Pore Size PT-020 CelgardR2400
0.2 0.2 x 0.02
70 25
74
(%I 38
"[Void Volume (%)I = (actual density of polymer) - (apparent density of polymer membrane) x 100 (actual density of polymer)
Characterization of LC-immobilized membranes
Differential scanning calorimeter (DSC, SSC/560, Daini Seikosha, Tokyo) was used to evaluate the phase transition of MTTS in the MTTS-soaked membranes in PBS. The thermo-responsive membrane was gold-coated by an Ion Coater and observed by a scanning electron microscope (SEM, Type S520, Hitachi Ltd., Tokyo) to study surface morphology. Measurements of the permeation coefficients of IND through LC-immobilized membranes A two-chamber glass diffusion cell was used for the measurement of permeation coefficients of IND through test membranes.' Two half-cells (volume, 100 mL) were separated by a LC-immobilized membrane (effective area, 7.7 cm2).The donor chamber was filled with 100 mL of 0.05% IND in 40% ethanol (EtOH) or 50% N-methyl-2-pyrrolidone aqueous solution, and the receiver with 100 mL of the same solution without IND. Permeation of IND through the membranes was measured with magnetic stirring at 32 or 38°C. The permeation of drug was pursued by sampling 50 pl aliquots from the receiver chamber and assaying with HPLC. The permeation coefficients were calculated from the steady state period obtained after 10 h with the following Eq. (1):
ln(1 - 2Ct/Co) = -2PAt/V
(1)
where Co is the initial drug concentration in the donor chamber, Ct is the drug concentration in the receiver chamber at time t, P is the permeation coefficient, A is the effective area for permeation, and V is the volume of each chamber. In vitro permeation experiments through hairless rat skin Full-thickness skin (approximately 0.8 mm of thickness) was excised from the abdominal region of male hairless rats (WBN/kob strain, weighing about 150 g) under anesthesia. The small two-chamber glass diffusion cells (volume of donor and receiver chamber, 2 mL each; effective area, 0.93 cm') were
NOZAWA ET AL.
580
used for the in vitro skin permeation experiment." The excised skin alone or LC-immobilized membrane piled on the stratum corneum of the skin were supported between two half-cells. The donor chamber was filled with 2 mL of 2.4% IND suspension in 50% N-methyl-2-pyrrolidone aqueous solution and the receiver was filled with phosphate buffered saline (PBS) (pH 7.4). At appropriate intervals, 50 pl aliquots were withdrawn from the receiver chamber and then assayed with HPLC. RESULTS A N D DISCUSSION
Thermo-response efficacy of the soaked membranes The thermo-controlled permeation of IND at 32 and 38°C through the soaked membrane was studied. The thermo-response efficacy for IND permeation expressed as a ratio of permeation coefficients at 38 and 32°C (P3,/P3,) of the soaked PP-membrane (62 pm of thickness) was 123 times. This value was larger than that through the soaked PTFE-membrane filter (75 pm of thickness) (54 times). The P3,/P3, value for the PTFE membrane prior to introlducing MTTS (70 pm of thickness) was 1.4. Because little IND permeated through PP membrane without MTTS at 32 or 38"C, the thermoresponse efficacy for PP membrane could not be calculated. MTTS contents in MTTS-soaked PP and PTFE membranes were identical, 48% per membrane weight. The reason why there is a significant difference in thermoresponse efficacy between two membranes in spite of the same MTTS content, can be explained by a difference in porosities of these membranes. As shown in Table I, the PTFE membrane filter is more porous (void volume, 74%)than the PP sheet (void volume, 38%),and hence the 48% MTTS content was too low to achieve high efficacy with the PTFE membrane. It was found from the results that the soaked PP-membrane would be more profitable than PTFE as a material for high thermo-control property. Figure 2 shows the results of the in vitro IND permeation experiments through the hairless rat skin with the soaked membrane from 50% N-methyl-2-pyrrolidone aqueous solution containing 2.4% IND suspension. Organic solvents were often composed in TTS as solubilizers for drugs and/or penetration enhancers. We had studied the effect of N-methyl-2-pyrrolidone on the skin permeation of drug in our previous papers.","! Based on the previous experiments, N-methyl-2-pyrrolidone was used as a model organic solvent in the present study. The cumulative amount of INlD permeated through the soaked PP-membrane for 28 h at 38°C was 120 times higher than that at 32°C (Fig. 2b), while the ratio of the soaked PTFE-membrane filter was only 13 times (Fig. 2a). The permeation coefficient of IND through the skin withbut the MTTS-soaked membrane at 38°C was only 1.2 times compared with that at 32°C. The ratio of the permeation coefficients at 38 to 32°C through the skin piled with the MTTS-soaked PP membrane was about 40 times. From these results, the high thermo-response efficacy of IND permeation through skin must be achieved by application of the MTTS-soaked PP membrane to the skin. MTTS-soaked PP membranes
THERMO-RESPONSIVE MEMBRANES. I1 N--
400
E
u
u
\
\
D 3
0,
-
1
300
7L)
W
+
U
m
m
; 200
W
2oc
W
W
a
a
n z
v1 U
n 4
D
z
100
0
z
30C
V
W
+
40(
N -
~
-
581
A o
10
20
1oc
’c 0 v1 22
30
z
0
10
20
30
a Time ( h )
Time ( h )
Figure 2. Control of IND permeation through the excised hairless rat skin by (a) MTTS-soaked PTFE membrane and (b) MTTS-soaked PP membrane. Results are expressed as the mean 2 SD of three experiments.
seem to be more suitable as materials for TTS, not only due to a good cost/benefit ratio but also due to the thermo-response efficacy. The desorption of MTTS from the MTTS-soaked membrane was evaluated by weight loss of the membrane after immersing it in 40% EtOH aqueous solution or 50% N-methyl-2-pyrrolidone aqueous solution, as model organic solvent-water systems, at 25 and 40°C (Table 11). Although these data are somewhat scattered, the MTTS-soaked PP membrane may be stable for more than 6 months at 40°C which is above the Tc of MTTS. The IND permeation experiments were also performed by using the MTTS-soaked PP membranes after immersing in 40% EtOH aqueous solution at 40°C for 6 months to examine the stability in terms of thermo-response efficacy. The value of P3,/P3,, as an index of thermo-response efficacy, after immersing was almost the same as the value before immersing. Therefore, it can be concluded that there was no effect of immersing on the thermo-response efficacy, and that the MTTSsoaked membrane was stable in these organic solvent-water systems. TABLE I1 The Weight Loss of LC-Immobilized PP Membrane after Incubation in Organic Solution for Given Periods Percentage of Remaining LC (76)” ~
40% EtOH
50% N-methyl-2-pyrrolidone
Time (month)
40°C
25°C
40°C
25°C
1 3 6
95.5 f 4.5 86.4 f 7.0 92.3 & 5.2
96.1 f 2.1 95.6 f 2.0 95.7 f 3.2
87.5 C 6.7 87.1 f 9.2 93.0 f 6.7
95.3 f 3.7 98.2 f 1.5 98.5 f 1.6
aMeans 2 SD (n = 3).
NOZAWA ET AL.
582
Preparation conditions of thermo-responsive membranes
In order to study the effect of degree of orientation and arrangement of MTTS on the thermo-response efficacy, IND permeation experiments were performed by using the MTTS-soaked PP membranes which were prepared by three different ways; (a) by immersing the membrane into warm water near the gel-liquid crystal phase transition temperature (Tc) (40°C) to promote the arrangement of MTTS, (b) by wiping excess MTTS with looser and more unoriented structure from the membrane surface with a doctor-blade, (c) by immobilizing MTTS by soaking PP membrane into MTTS chloroform solution at 40°C to promote the arrangement of MTTS. MTTS-soaked membranes were also prepared by combining these treatments. The influence of these treatments on the thickness of these membranes was not found. The thickness of the membrane treated as described above was 62 pm and the value was the same before the treatments. Measurement of MTTS content, DSC analysis, and X-ray diffraction study of these membranes were carried out. Figure 3 shows DSC thermogram of the MTTS-ssaked membranes obtained in PBS. New peaks probably due to rearrangement of MTTS were observed. In Figure 4, two examples of X-ray diffraction patterns of MTTS-soaked membranes after these treatments (i.e., immersing into warm water, wiping excess MTTS, soaking at 40°C) are shown. From these figures, it is evident that these treatments can change and rearrange the state of MTTS molecular packing. IND permeation experiments through the MTTS-soaked membranes with different MTTS contents were carried out at 32 and 38°C to examine the effects of these treatments on the thermo-response efficacy. These experiments were done by using 50% N-methyl-2-pyrrolidoneas a donor and receiver solution. The addition of organic solvents such as N-methyl-2-pyrrolidone and EtOH reduceld the Tc of MTTS from 40 to 38°C due to their plasticizing effect as described in our previous paper.' The relationship between thermo-
34
38
42
Temperature ( " C )
Figure 3. DSC charts for several treated membranes. (a) Standard membrane, (b) soaked MTTS at 40°C, wiped and immersed at 40°C, (c) soaked MTTS at 25°C and immersed at 40°C, (d) soaked MTTS at 25"C, immersed at 40°C and wiped.
THERMO-RESPONSIVE MEMBRANES. I1
0.2 0.5 2
583
1.0
1.5
e
Figure 4. X-ray diffraction patterns of the MTTS-soaked membrane (a) without any treatment, (b) soaked MTTS at 2 5 T , wiped and immersed at 40°C, and (c) soaked MTTS at 40°C and wiped.
response efficacy, P38/P32,for the IND permeation, and MTTS content is shown in Figure 5. MTTS content in the MTTS-soaked membranes was controlled by MTTS concentration of MTTS chloroform solution at membrane preparation and by treatments for soaked membranes as described above. Although some scatter was noted, in this figure, all data points from different ..c
m
m
150
m a
-
h
:1 0 0 U .r
+ 'c w
$
50
S
0
a
. _
0 al I I-
10
20
30
MTTS C o n t e n t
40
50
( % )
Figure 5. The relationship between the thermo-response efficacy and MTTS contents in the MTTS-soaked membrane. The dotted line indicates the void volume of PP-membrane.
584
NOZAWA ET AL.
membranes are on one curve. The value of P3,/P3, steeply increases above the MTTS content of 38% which is equal to the void volume of PP membrane. A 38% MTTS content must be necessary to obtain a MTTS-immobilized membrane which MTTS completely fills the void space of the PP membrane (38%). Because MTTS exists not only in the pores but also on the surface, an MTTS content above the void volume of PP membrane (38%)should be needed to obtain compllete pouring of MTTS into the pores. It is evident from Figures 3, 4, and 5 that the MTTS content is one of the most important factors for high thermo-response efficacy with MTTS-soaked membranes. There is no influence of MTTS reorganization after treatments on the membrane permeation of IND. These results suggested that the endothermic behavior below a peak of a DSC thermogram may not relate to the IND permeation through the MTTS-immobilized membranes. In other words, membrane permeation of IND may occur easily at the foot region below the Tc, and there is no significant difference between membranes before and after the treatments in microviscosity surrounding alkyl chains of MTTS at 38°C. Surface mlorphology of the soaked membranes was studied using SEM (Fig. 6). The extent of surface coating by MTTS increased with increasing MTTS content and the pores on the surfaces were completely covered with MTTS above 38% of MTTS content, where the thermo-response efficacy became large (Fig. 5). From the results mentioned above, it became clear that
Figure 6. Scanning electron micrographs of MTTS-soaked membrane. MTTS content is (a) 0%, (b) 27%, (c) 3076, (d) 38%, and (e) 48%.
THERMO-RESPONSIVE MEMBRANES. I1
585
high MTTS content and embedding of the pores of polymer membranes with MTTS so as to completely cover the membrane surfaces with MTTS were important to obtain enough thermo-response efficacy. It is clear that most IND is permeated through MTTS-portion and little IND through inert polymer matrix, because little IND was permeated through the MTTS-immobilized membrane at 32°C (Table 111). Therefore, the state of the MTTS-immobilization must be one of the most important factors for an understanding of the thermo-response efficacy of the MTTS-immobilized membranes. Although the pores of the membrane might be completely embedded by MTTS theoretically at 38% MTTS content, because the void volume of PP membrane is at that value, it was clear from Figure 6d that the MTTS-immobilized membrane, prepared by the soaking-method, had the MTTS on the membrane surface and was not effectively embedded with MTTS. In order to clarify the roles of membrane surface-coated MTTS and pore-embedded MTTS in the thermo-response efficacy of the MTTS-immobilized membranes, it should be necessary to evaluate the IND permeation behavior through the membrane with a high MTTS-embedding efficiency. A sandwich membrane may be a better candidate as a MTTS-immobilized membrane, instead of the soaked membrane.
Preparation of the sandwich membrane A sandwich membrane was prepared according to the method described in the experimental part of this article, and its thermo-response efficacy was compared with that of the MTTS-soaked membrane (Table 111). The sandwich membrane with a 38% MTTS content showed markedly higher thermoresponse efficacy than the soaked membrane with the same MTTS content (38%) and had a better thermo-response efficacy than that by the soaked membrane with 48% of MTTS content. The fact that the thermo-response efficacy of the soaked membrane (MTTS content, 48%) completely coated with excess MTTS was lower than that of the sandwich-membrane with a low MTTS content (38%)probably indicates that the role of coated MTTS on the membrane surface may be not so important in the thermo-response efficacy for the IND permeation. From this result, it can be concluded that the sandwich method is more suitable to prepare MTTS-immobilized membranes with a high MTTS density in the pores and a high thermo-response efficacy. TABLE 111 Thermo-Response Efficacies of MTTS-Immobilized Membranes
Soaked membrane Sandwich membrane
48.6 38.4 38.1
aMean It_ SD for three experiments.
3.12 & 0.37 2.82 & 0.17 2.61 f 0.04
2.65 f 1.05 13.00 f 4.90 1.46 f 0.67
120 22 186
NOZAWA ET AL.
586
A nomograph for evaluation of thermo-response efficacy
An overall function of these thermo-responsive membranes cannot be evaluated only by the values of P38/P32 as an index of the thermo-response efficacy.Even with the membranes having a high thermo-response efficacy, the total efficacy must be low when the absolute amount of the drug permeated through skin with the membranes is so low that the effective blood (body) concentration of the drug is not obtained at 38°C. On the other hand, in the case that the permeability of the membranes is too high compared to the skin permeability, an overall permeability through the skin cannot be controlled by the thermo-responsive membranes at 32°C. Therefore, the function of the thermo-responsive membranes should not be evaluated only by the thermo-response efficacy, but also by the absolute permeability through the membranes. When the drug absorption rate is controlled by the endo- or exogenous temperature changes (e.g., antiinflammatory drugs and antipyretic drugs), the skin Permeability at 38°C must be over a minimum boarder value of total skin permeability ( Pt38)required for a minimum effective concentration of the drug and also be under a maximum boarder value of the total skin permeability at 32°C (Pt32)to maintain the drug concentration under a ineffective body level. a and p were defined as coefficients between each boarder value (Pt38,Pt32)and the intact skin permeability (lJs38,Ps32) as described in the following two Eqs. (2) and (3): Pt38 = a . PS38
(2)
Pt32 = p . Ps32
(3)
The miniscules, t and s, show total (skin with the membrane) and skin, respectively. a and /3 should be determined by the values of Pt38and Pt32 required phamacodynamically. A nomograph was established for the total evaluation of the thermoresponsive membranes (Fig. 7). Since it is a model nomograph, a and p are assumed to be 0.5 for simplification. The thermo-response efficacy of the MTTS-immobilized membrane alone between 38 and 32°C (Pm38/Pm32,miniscule m means membrane) is plotted against the total thermo-response efficacy of skin with the MTTS-immobilized membrane (Pt38/Pt32). The solid lines in the figure show the theoretical values calculated from two border values ((2) and (3)) and a following Eq. (4).13 1/PtS8= 1/Ps38+ 1/Pm38
(4)
1/Pt32= 1/Ps32+ 1/Pm32 Parameter a can be determined to ensure the minimum skin permeation rate at 38"C, whereas p can be determined to ensure maximum skin permeation at 32°C. This nomograph is useful to select and evaluate the thermo-responsive membranes. For instance, in the case that the permeation coefficient of the MTTSimmobilized membrane at 38°C is very low and the permeation of drug through skin is reduced lower than one half by application of the MTTS-
THERMO-RESPONSIVE MEMBRANES. I1
587
h u 5
u .r
Lc %
w
N
m
100
-
L e u
+
a
0 . . I
s
t
r
0
n
0 c
5 Y
0 I-
1
10
100
1000
Pm3a/Pm3z On-off Efficacy o f M T T S - i m m o b i l i z e d Membrane
Figure 7. Nomograph for evaluation of thermo-response efficacy.
immobilized membrane to the skin (i.e., Pt3*< 0.5PS38), the total thermoresponse efficacy of skin piled with the MTTS-membrane (Pt38/Pt32)is located in region 1. In this case, it is true that the total thermo-response efficacy of skin with the MTTS-immobilized membrane becomes fairly high, but the total amount of drug permeated at 38°C becomes less than the minimum effective dose of drug. On the other hand, in the case that the permeation coefficient of the MTTS-immobilized membrane at 32°C is not so low (i.e., Pt32> 0 . 5 P ~ ~the ~ ) ,total thermo-response efficacy of skin with the MTTSimmobilized membrane (Pt38/Pt32) is located in region 3. In this case, the thermo-response efficacy, Pt38/Pt32,becomes very low. The optimum region should be therefore located in region 2. The MTTS-immobilized membranes used in this study, the soaked membrane and the sandwich membrane, are represented as + and respectively, in Figure 7. From this result it can be seen that the sandwich membrane is more suitable as a thermo-responsive membrane than the soaked membrane.
*,
CONCLUSION
The sandwich membrane which is easily prepared by pouring the melted LC into two PP porous membranes achieved high thermo-response efficacy. It can be considered that this membrane may enable us to develop a TTS which attains ”on-off switching release” in response to thermal stimuli. References 1. Y. W. Chien, “Logics of transdermal controlled drug administration,” Drug Dev.Ind. Pharrn., 9,497-520 (1983). 2. A. Zaffaroni, “Microporous Bandages,” U.S. Patent 3,797,494 (March 19, 1974). 3. S. K. Chandrasekaran, “Controlled release of scoporamine for prophylaxis of motion sickness,” Drug Dev. Ind. Pharrn., 9, 627-646 (1983).
NOZAWA ET AL.
588 4.
5. 6.
7. 8. 9. 10.
11.
12, 13..
J. E. Shaw and S. K. Chandrasekaran, "Controlled topical delivery of drugs for systemic action," Drug Metab. Rev., 8, 223-233 (1978). J. F. Desta and D. R. Geraets, "Topical nitroglycerin: A new twist to an old standby," Am. Pharm., NS22(2), 29-35 (1982). W. R. Good, "Transderm-Nitro controlled delivery of nitroglycerin via the transdermal route," Drug Dev. Ind. Pkarm., 9, 647-670 (1983). A. Karim, "Transdermal absorption; A unique opportunity for constant delivery of nitroglycerin," Drug Rev. Ind. Pkarm., 9, 671-689 (1983). I. Nozawa, Y. Suzuki, S. Sato, K. Sugibayashi, and Y. Morimoto, "Preparation of thermo-responsive polymer membranes I," J. Biomed. Muter. Res., to appear. S. J. Lee, T. Kurihara-Bergstrom, and S.W. Kim, "Nonaqueous drug permeation through synthetic membranes," J. Controlled Rel., 5, 253262 (1988). Y.Morimoto, K. Sugibayashi, K. Hosoya, and W. I. Higuchi, "Penetration enhancing effect of Azone on the transport of 5-fluorouracil across the hairless rat skin," Int. J. Pharmnceut., 32, 31-38 (1986). K. Sugibayashi, M. Nemoto, and Y. Morimoto, "Effect of several permeation enhancers on the percutaneous absorption of indomethacin in hairless rats," Ckem. Pkarm. Bull., 36, 1519-1528 (1988). K . Sugibayashi, C. Sakanoue, and Y. Morimoto, "Utility of topical formulation of morphine hydrochloride containing Azone and N-methyl2-pyrrolidone," Selective Cancer Tker., 5, 119-128 (1989). T. Okano, S.W. Kim, F. Komada, G. Imanidis, S. Nishiyama, and W.I. Higuchi, "Control of drug concentration-time profiles in vivo by zeroorder transdermal delivery systems," J. Controlled Rel., 6, 99-106 (1987).
Received November 1,1990 Accepted November 15, 1990
sandwich method. Thermo-response efficacy of the soaked membrane was dependent upon the content of MTTS in MTTS membrane, and the MTTS content above the void volume of PP membrane (38%) was needed for high efficacy. On the other hand, the sandwich membrane exhibited higher thermo-response efficacy than the soaked membrane, because more LC was embedded in the pores of sandwich membrane than that of the soaked membrane. The sandwich membrane permeation of indomethacin was sharply controlled by temperature changes between 32 and 38°C.
INTRODUCTION
For developing transdermal therapeutic systems (TTS), achievement of controlled release is one of the most important subjects. For this purpose several polymers can be used as membranes and/or matrices for sustaining drug release.’ For example, Transderm-Scop (Alza Co. and Ciba-Geigy L td.)2-4and Transderm-Nitro (Alza Co. and Ciba-Geigy Ltd.)56have been fabricated with an ethylenehinylacetate copolymer (EVA) porous membrane for sustained release and Nitrodisc (Searle Ltd.) has a silicon elastomer matrix for diffusion contr01.”~However, these TTS are not able to release drugs according to the condition of the patients or a preprogrammed schedule. Recently, intelligent materials which change their physical properties in response to external stimuli, such as temperature, photoirradiation, and pH, have been studied and they have been attractively applied to drug delivery system. *To whom correspondence should be addressed at Faculty of Pharmaceutical Sciences, Josai University, 1-1Keyakidai, Sakado, Saitama 350-02, Japan. Journal of Biomedical Materials Research, Vol. 25, 577-588 (1991) 0 1991 John Wiley & Sons, Inc. CCC 002l-9304/91/050577-12$4.00
NOZAWA ET AL.
578
In our previous paper: liquid crystal (LC)-immobilized membranes were studied to fabricate drug delivery systems with thermo-sensitivity. LCabsorbed membranes obtained by absorbing LC into porous hydrophobic polymer membranes could control the permeation of drug in a pulsatile manner in response to minute temperature changes. In the present article, we describe the influence of the preparation methods for the LC-immobilized membranes on the thermo-response efficacy. EXPERIMENTAL
Materials Monooxyethylene trimethylolpropane tristearate (MTTS) was purchased from Nihon Emulsion Co., Ltd. (Tokyo, Japan). The structural formula is shown in Figure 1. Polytetraf liioroethylene (PTFE) membrane filter (PT-020, Toyo Roshi Co., Tokyo, Japan) and porous polypropylene (PP) membrane (Celgard 2400, Hoechst-Celanese Co., USA) were used as MTTS supporting membranes. The properties of these membranes are shown in Table I. Indomethacin (IND) was a commercial product of Japanese Pharmacopoeia grade. All other chemicals were of reagent-grade products. Preparation alf thermo-responsive membranes Thermo-responsive membranes were prepared by a soaking method and a sandwich method. Two PI' membranes or one PTFE membrane filter were soaked in given concentrations of MTTS chloroform solution for 1 h at room temperature and then dried in uacuo for 24 h in the soaking method. Melted MTTS was ploured between two Celgard 2400 sheets, and the sandwiched sheets were rolled out to squeeze excess MTTS and was pressed at 25 x lo3 kg/cm2 for 10 min in the sandwich method. MTTS contents in the soaked membrane prepared by using 20% MTTS chloroform solution and in the sandwich membrane were 48 and 38% per membrane weight, respectively. The thicknesses of these membranes were 62 and 55 pm, respectively. MTTS contents were calculated from the difference of membrane weights before and after 3 times washing of MTTS-immobilized membranes in EtOH at 40°C to complletely remove MTTS from the membrane.
CH3CH2C
t
CH20CH2CH20COC17H35 CH20CH2CH20COC17H35 CH20CH2CH20COC,7H35
Monooxyethylene T r i m e t h y l o l p r o p a n e T r i s t e a r a t e ( M T T S )
Figure 1. Structural formula of liquid crystal.
THERMO-RESPONSIVE MEMBRANES. I1
579
TABLE I The Properties of Two Types of LC-Supporting Polymer Membranes
(w-4
Thickness (Pm)
Void Volume"
Membrane
Pore Size PT-020 CelgardR2400
0.2 0.2 x 0.02
70 25
74
(%I 38
"[Void Volume (%)I = (actual density of polymer) - (apparent density of polymer membrane) x 100 (actual density of polymer)
Characterization of LC-immobilized membranes
Differential scanning calorimeter (DSC, SSC/560, Daini Seikosha, Tokyo) was used to evaluate the phase transition of MTTS in the MTTS-soaked membranes in PBS. The thermo-responsive membrane was gold-coated by an Ion Coater and observed by a scanning electron microscope (SEM, Type S520, Hitachi Ltd., Tokyo) to study surface morphology. Measurements of the permeation coefficients of IND through LC-immobilized membranes A two-chamber glass diffusion cell was used for the measurement of permeation coefficients of IND through test membranes.' Two half-cells (volume, 100 mL) were separated by a LC-immobilized membrane (effective area, 7.7 cm2).The donor chamber was filled with 100 mL of 0.05% IND in 40% ethanol (EtOH) or 50% N-methyl-2-pyrrolidone aqueous solution, and the receiver with 100 mL of the same solution without IND. Permeation of IND through the membranes was measured with magnetic stirring at 32 or 38°C. The permeation of drug was pursued by sampling 50 pl aliquots from the receiver chamber and assaying with HPLC. The permeation coefficients were calculated from the steady state period obtained after 10 h with the following Eq. (1):
ln(1 - 2Ct/Co) = -2PAt/V
(1)
where Co is the initial drug concentration in the donor chamber, Ct is the drug concentration in the receiver chamber at time t, P is the permeation coefficient, A is the effective area for permeation, and V is the volume of each chamber. In vitro permeation experiments through hairless rat skin Full-thickness skin (approximately 0.8 mm of thickness) was excised from the abdominal region of male hairless rats (WBN/kob strain, weighing about 150 g) under anesthesia. The small two-chamber glass diffusion cells (volume of donor and receiver chamber, 2 mL each; effective area, 0.93 cm') were
NOZAWA ET AL.
580
used for the in vitro skin permeation experiment." The excised skin alone or LC-immobilized membrane piled on the stratum corneum of the skin were supported between two half-cells. The donor chamber was filled with 2 mL of 2.4% IND suspension in 50% N-methyl-2-pyrrolidone aqueous solution and the receiver was filled with phosphate buffered saline (PBS) (pH 7.4). At appropriate intervals, 50 pl aliquots were withdrawn from the receiver chamber and then assayed with HPLC. RESULTS A N D DISCUSSION
Thermo-response efficacy of the soaked membranes The thermo-controlled permeation of IND at 32 and 38°C through the soaked membrane was studied. The thermo-response efficacy for IND permeation expressed as a ratio of permeation coefficients at 38 and 32°C (P3,/P3,) of the soaked PP-membrane (62 pm of thickness) was 123 times. This value was larger than that through the soaked PTFE-membrane filter (75 pm of thickness) (54 times). The P3,/P3, value for the PTFE membrane prior to introlducing MTTS (70 pm of thickness) was 1.4. Because little IND permeated through PP membrane without MTTS at 32 or 38"C, the thermoresponse efficacy for PP membrane could not be calculated. MTTS contents in MTTS-soaked PP and PTFE membranes were identical, 48% per membrane weight. The reason why there is a significant difference in thermoresponse efficacy between two membranes in spite of the same MTTS content, can be explained by a difference in porosities of these membranes. As shown in Table I, the PTFE membrane filter is more porous (void volume, 74%)than the PP sheet (void volume, 38%),and hence the 48% MTTS content was too low to achieve high efficacy with the PTFE membrane. It was found from the results that the soaked PP-membrane would be more profitable than PTFE as a material for high thermo-control property. Figure 2 shows the results of the in vitro IND permeation experiments through the hairless rat skin with the soaked membrane from 50% N-methyl-2-pyrrolidone aqueous solution containing 2.4% IND suspension. Organic solvents were often composed in TTS as solubilizers for drugs and/or penetration enhancers. We had studied the effect of N-methyl-2-pyrrolidone on the skin permeation of drug in our previous papers.","! Based on the previous experiments, N-methyl-2-pyrrolidone was used as a model organic solvent in the present study. The cumulative amount of INlD permeated through the soaked PP-membrane for 28 h at 38°C was 120 times higher than that at 32°C (Fig. 2b), while the ratio of the soaked PTFE-membrane filter was only 13 times (Fig. 2a). The permeation coefficient of IND through the skin withbut the MTTS-soaked membrane at 38°C was only 1.2 times compared with that at 32°C. The ratio of the permeation coefficients at 38 to 32°C through the skin piled with the MTTS-soaked PP membrane was about 40 times. From these results, the high thermo-response efficacy of IND permeation through skin must be achieved by application of the MTTS-soaked PP membrane to the skin. MTTS-soaked PP membranes
THERMO-RESPONSIVE MEMBRANES. I1 N--
400
E
u
u
\
\
D 3
0,
-
1
300
7L)
W
+
U
m
m
; 200
W
2oc
W
W
a
a
n z
v1 U
n 4
D
z
100
0
z
30C
V
W
+
40(
N -
~
-
581
A o
10
20
1oc
’c 0 v1 22
30
z
0
10
20
30
a Time ( h )
Time ( h )
Figure 2. Control of IND permeation through the excised hairless rat skin by (a) MTTS-soaked PTFE membrane and (b) MTTS-soaked PP membrane. Results are expressed as the mean 2 SD of three experiments.
seem to be more suitable as materials for TTS, not only due to a good cost/benefit ratio but also due to the thermo-response efficacy. The desorption of MTTS from the MTTS-soaked membrane was evaluated by weight loss of the membrane after immersing it in 40% EtOH aqueous solution or 50% N-methyl-2-pyrrolidone aqueous solution, as model organic solvent-water systems, at 25 and 40°C (Table 11). Although these data are somewhat scattered, the MTTS-soaked PP membrane may be stable for more than 6 months at 40°C which is above the Tc of MTTS. The IND permeation experiments were also performed by using the MTTS-soaked PP membranes after immersing in 40% EtOH aqueous solution at 40°C for 6 months to examine the stability in terms of thermo-response efficacy. The value of P3,/P3,, as an index of thermo-response efficacy, after immersing was almost the same as the value before immersing. Therefore, it can be concluded that there was no effect of immersing on the thermo-response efficacy, and that the MTTSsoaked membrane was stable in these organic solvent-water systems. TABLE I1 The Weight Loss of LC-Immobilized PP Membrane after Incubation in Organic Solution for Given Periods Percentage of Remaining LC (76)” ~
40% EtOH
50% N-methyl-2-pyrrolidone
Time (month)
40°C
25°C
40°C
25°C
1 3 6
95.5 f 4.5 86.4 f 7.0 92.3 & 5.2
96.1 f 2.1 95.6 f 2.0 95.7 f 3.2
87.5 C 6.7 87.1 f 9.2 93.0 f 6.7
95.3 f 3.7 98.2 f 1.5 98.5 f 1.6
aMeans 2 SD (n = 3).
NOZAWA ET AL.
582
Preparation conditions of thermo-responsive membranes
In order to study the effect of degree of orientation and arrangement of MTTS on the thermo-response efficacy, IND permeation experiments were performed by using the MTTS-soaked PP membranes which were prepared by three different ways; (a) by immersing the membrane into warm water near the gel-liquid crystal phase transition temperature (Tc) (40°C) to promote the arrangement of MTTS, (b) by wiping excess MTTS with looser and more unoriented structure from the membrane surface with a doctor-blade, (c) by immobilizing MTTS by soaking PP membrane into MTTS chloroform solution at 40°C to promote the arrangement of MTTS. MTTS-soaked membranes were also prepared by combining these treatments. The influence of these treatments on the thickness of these membranes was not found. The thickness of the membrane treated as described above was 62 pm and the value was the same before the treatments. Measurement of MTTS content, DSC analysis, and X-ray diffraction study of these membranes were carried out. Figure 3 shows DSC thermogram of the MTTS-ssaked membranes obtained in PBS. New peaks probably due to rearrangement of MTTS were observed. In Figure 4, two examples of X-ray diffraction patterns of MTTS-soaked membranes after these treatments (i.e., immersing into warm water, wiping excess MTTS, soaking at 40°C) are shown. From these figures, it is evident that these treatments can change and rearrange the state of MTTS molecular packing. IND permeation experiments through the MTTS-soaked membranes with different MTTS contents were carried out at 32 and 38°C to examine the effects of these treatments on the thermo-response efficacy. These experiments were done by using 50% N-methyl-2-pyrrolidoneas a donor and receiver solution. The addition of organic solvents such as N-methyl-2-pyrrolidone and EtOH reduceld the Tc of MTTS from 40 to 38°C due to their plasticizing effect as described in our previous paper.' The relationship between thermo-
34
38
42
Temperature ( " C )
Figure 3. DSC charts for several treated membranes. (a) Standard membrane, (b) soaked MTTS at 40°C, wiped and immersed at 40°C, (c) soaked MTTS at 25°C and immersed at 40°C, (d) soaked MTTS at 25"C, immersed at 40°C and wiped.
THERMO-RESPONSIVE MEMBRANES. I1
0.2 0.5 2
583
1.0
1.5
e
Figure 4. X-ray diffraction patterns of the MTTS-soaked membrane (a) without any treatment, (b) soaked MTTS at 2 5 T , wiped and immersed at 40°C, and (c) soaked MTTS at 40°C and wiped.
response efficacy, P38/P32,for the IND permeation, and MTTS content is shown in Figure 5. MTTS content in the MTTS-soaked membranes was controlled by MTTS concentration of MTTS chloroform solution at membrane preparation and by treatments for soaked membranes as described above. Although some scatter was noted, in this figure, all data points from different ..c
m
m
150
m a
-
h
:1 0 0 U .r
+ 'c w
$
50
S
0
a
. _
0 al I I-
10
20
30
MTTS C o n t e n t
40
50
( % )
Figure 5. The relationship between the thermo-response efficacy and MTTS contents in the MTTS-soaked membrane. The dotted line indicates the void volume of PP-membrane.
584
NOZAWA ET AL.
membranes are on one curve. The value of P3,/P3, steeply increases above the MTTS content of 38% which is equal to the void volume of PP membrane. A 38% MTTS content must be necessary to obtain a MTTS-immobilized membrane which MTTS completely fills the void space of the PP membrane (38%). Because MTTS exists not only in the pores but also on the surface, an MTTS content above the void volume of PP membrane (38%)should be needed to obtain compllete pouring of MTTS into the pores. It is evident from Figures 3, 4, and 5 that the MTTS content is one of the most important factors for high thermo-response efficacy with MTTS-soaked membranes. There is no influence of MTTS reorganization after treatments on the membrane permeation of IND. These results suggested that the endothermic behavior below a peak of a DSC thermogram may not relate to the IND permeation through the MTTS-immobilized membranes. In other words, membrane permeation of IND may occur easily at the foot region below the Tc, and there is no significant difference between membranes before and after the treatments in microviscosity surrounding alkyl chains of MTTS at 38°C. Surface mlorphology of the soaked membranes was studied using SEM (Fig. 6). The extent of surface coating by MTTS increased with increasing MTTS content and the pores on the surfaces were completely covered with MTTS above 38% of MTTS content, where the thermo-response efficacy became large (Fig. 5). From the results mentioned above, it became clear that
Figure 6. Scanning electron micrographs of MTTS-soaked membrane. MTTS content is (a) 0%, (b) 27%, (c) 3076, (d) 38%, and (e) 48%.
THERMO-RESPONSIVE MEMBRANES. I1
585
high MTTS content and embedding of the pores of polymer membranes with MTTS so as to completely cover the membrane surfaces with MTTS were important to obtain enough thermo-response efficacy. It is clear that most IND is permeated through MTTS-portion and little IND through inert polymer matrix, because little IND was permeated through the MTTS-immobilized membrane at 32°C (Table 111). Therefore, the state of the MTTS-immobilization must be one of the most important factors for an understanding of the thermo-response efficacy of the MTTS-immobilized membranes. Although the pores of the membrane might be completely embedded by MTTS theoretically at 38% MTTS content, because the void volume of PP membrane is at that value, it was clear from Figure 6d that the MTTS-immobilized membrane, prepared by the soaking-method, had the MTTS on the membrane surface and was not effectively embedded with MTTS. In order to clarify the roles of membrane surface-coated MTTS and pore-embedded MTTS in the thermo-response efficacy of the MTTS-immobilized membranes, it should be necessary to evaluate the IND permeation behavior through the membrane with a high MTTS-embedding efficiency. A sandwich membrane may be a better candidate as a MTTS-immobilized membrane, instead of the soaked membrane.
Preparation of the sandwich membrane A sandwich membrane was prepared according to the method described in the experimental part of this article, and its thermo-response efficacy was compared with that of the MTTS-soaked membrane (Table 111). The sandwich membrane with a 38% MTTS content showed markedly higher thermoresponse efficacy than the soaked membrane with the same MTTS content (38%) and had a better thermo-response efficacy than that by the soaked membrane with 48% of MTTS content. The fact that the thermo-response efficacy of the soaked membrane (MTTS content, 48%) completely coated with excess MTTS was lower than that of the sandwich-membrane with a low MTTS content (38%)probably indicates that the role of coated MTTS on the membrane surface may be not so important in the thermo-response efficacy for the IND permeation. From this result, it can be concluded that the sandwich method is more suitable to prepare MTTS-immobilized membranes with a high MTTS density in the pores and a high thermo-response efficacy. TABLE 111 Thermo-Response Efficacies of MTTS-Immobilized Membranes
Soaked membrane Sandwich membrane
48.6 38.4 38.1
aMean It_ SD for three experiments.
3.12 & 0.37 2.82 & 0.17 2.61 f 0.04
2.65 f 1.05 13.00 f 4.90 1.46 f 0.67
120 22 186
NOZAWA ET AL.
586
A nomograph for evaluation of thermo-response efficacy
An overall function of these thermo-responsive membranes cannot be evaluated only by the values of P38/P32 as an index of the thermo-response efficacy.Even with the membranes having a high thermo-response efficacy, the total efficacy must be low when the absolute amount of the drug permeated through skin with the membranes is so low that the effective blood (body) concentration of the drug is not obtained at 38°C. On the other hand, in the case that the permeability of the membranes is too high compared to the skin permeability, an overall permeability through the skin cannot be controlled by the thermo-responsive membranes at 32°C. Therefore, the function of the thermo-responsive membranes should not be evaluated only by the thermo-response efficacy, but also by the absolute permeability through the membranes. When the drug absorption rate is controlled by the endo- or exogenous temperature changes (e.g., antiinflammatory drugs and antipyretic drugs), the skin Permeability at 38°C must be over a minimum boarder value of total skin permeability ( Pt38)required for a minimum effective concentration of the drug and also be under a maximum boarder value of the total skin permeability at 32°C (Pt32)to maintain the drug concentration under a ineffective body level. a and p were defined as coefficients between each boarder value (Pt38,Pt32)and the intact skin permeability (lJs38,Ps32) as described in the following two Eqs. (2) and (3): Pt38 = a . PS38
(2)
Pt32 = p . Ps32
(3)
The miniscules, t and s, show total (skin with the membrane) and skin, respectively. a and /3 should be determined by the values of Pt38and Pt32 required phamacodynamically. A nomograph was established for the total evaluation of the thermoresponsive membranes (Fig. 7). Since it is a model nomograph, a and p are assumed to be 0.5 for simplification. The thermo-response efficacy of the MTTS-immobilized membrane alone between 38 and 32°C (Pm38/Pm32,miniscule m means membrane) is plotted against the total thermo-response efficacy of skin with the MTTS-immobilized membrane (Pt38/Pt32). The solid lines in the figure show the theoretical values calculated from two border values ((2) and (3)) and a following Eq. (4).13 1/PtS8= 1/Ps38+ 1/Pm38
(4)
1/Pt32= 1/Ps32+ 1/Pm32 Parameter a can be determined to ensure the minimum skin permeation rate at 38"C, whereas p can be determined to ensure maximum skin permeation at 32°C. This nomograph is useful to select and evaluate the thermo-responsive membranes. For instance, in the case that the permeation coefficient of the MTTSimmobilized membrane at 38°C is very low and the permeation of drug through skin is reduced lower than one half by application of the MTTS-
THERMO-RESPONSIVE MEMBRANES. I1
587
h u 5
u .r
Lc %
w
N
m
100
-
L e u
+
a
0 . . I
s
t
r
0
n
0 c
5 Y
0 I-
1
10
100
1000
Pm3a/Pm3z On-off Efficacy o f M T T S - i m m o b i l i z e d Membrane
Figure 7. Nomograph for evaluation of thermo-response efficacy.
immobilized membrane to the skin (i.e., Pt3*< 0.5PS38), the total thermoresponse efficacy of skin piled with the MTTS-membrane (Pt38/Pt32)is located in region 1. In this case, it is true that the total thermo-response efficacy of skin with the MTTS-immobilized membrane becomes fairly high, but the total amount of drug permeated at 38°C becomes less than the minimum effective dose of drug. On the other hand, in the case that the permeation coefficient of the MTTS-immobilized membrane at 32°C is not so low (i.e., Pt32> 0 . 5 P ~ ~the ~ ) ,total thermo-response efficacy of skin with the MTTSimmobilized membrane (Pt38/Pt32) is located in region 3. In this case, the thermo-response efficacy, Pt38/Pt32,becomes very low. The optimum region should be therefore located in region 2. The MTTS-immobilized membranes used in this study, the soaked membrane and the sandwich membrane, are represented as + and respectively, in Figure 7. From this result it can be seen that the sandwich membrane is more suitable as a thermo-responsive membrane than the soaked membrane.
*,
CONCLUSION
The sandwich membrane which is easily prepared by pouring the melted LC into two PP porous membranes achieved high thermo-response efficacy. It can be considered that this membrane may enable us to develop a TTS which attains ”on-off switching release” in response to thermal stimuli. References 1. Y. W. Chien, “Logics of transdermal controlled drug administration,” Drug Dev.Ind. Pharrn., 9,497-520 (1983). 2. A. Zaffaroni, “Microporous Bandages,” U.S. Patent 3,797,494 (March 19, 1974). 3. S. K. Chandrasekaran, “Controlled release of scoporamine for prophylaxis of motion sickness,” Drug Dev. Ind. Pharrn., 9, 627-646 (1983).
NOZAWA ET AL.
588 4.
5. 6.
7. 8. 9. 10.
11.
12, 13..
J. E. Shaw and S. K. Chandrasekaran, "Controlled topical delivery of drugs for systemic action," Drug Metab. Rev., 8, 223-233 (1978). J. F. Desta and D. R. Geraets, "Topical nitroglycerin: A new twist to an old standby," Am. Pharm., NS22(2), 29-35 (1982). W. R. Good, "Transderm-Nitro controlled delivery of nitroglycerin via the transdermal route," Drug Dev. Ind. Pkarm., 9, 647-670 (1983). A. Karim, "Transdermal absorption; A unique opportunity for constant delivery of nitroglycerin," Drug Rev. Ind. Pkarm., 9, 671-689 (1983). I. Nozawa, Y. Suzuki, S. Sato, K. Sugibayashi, and Y. Morimoto, "Preparation of thermo-responsive polymer membranes I," J. Biomed. Muter. Res., to appear. S. J. Lee, T. Kurihara-Bergstrom, and S.W. Kim, "Nonaqueous drug permeation through synthetic membranes," J. Controlled Rel., 5, 253262 (1988). Y.Morimoto, K. Sugibayashi, K. Hosoya, and W. I. Higuchi, "Penetration enhancing effect of Azone on the transport of 5-fluorouracil across the hairless rat skin," Int. J. Pharmnceut., 32, 31-38 (1986). K. Sugibayashi, M. Nemoto, and Y. Morimoto, "Effect of several permeation enhancers on the percutaneous absorption of indomethacin in hairless rats," Ckem. Pkarm. Bull., 36, 1519-1528 (1988). K . Sugibayashi, C. Sakanoue, and Y. Morimoto, "Utility of topical formulation of morphine hydrochloride containing Azone and N-methyl2-pyrrolidone," Selective Cancer Tker., 5, 119-128 (1989). T. Okano, S.W. Kim, F. Komada, G. Imanidis, S. Nishiyama, and W.I. Higuchi, "Control of drug concentration-time profiles in vivo by zeroorder transdermal delivery systems," J. Controlled Rel., 6, 99-106 (1987).
Received November 1,1990 Accepted November 15, 1990
