September/October 2004
Published by the American Chemical Society
Volume 5, Number 5
© Copyright 2004 by the American Chemical Society
Reviews Poly(ortho esters)sFrom Concept to Reality† Jorge Heller* and John Barr AP Pharma, 123 Saginaw Drive, Redwood City, California 94063 Received May 19, 2004
The development of poly(ortho esters) dates back to the early 1970s, and during that time, four distinct families were developed. These polymers can be prepared by a transesterification reaction or by the addition of polyols to diketene acetals, and it is the latter method that has proven to be preferred one. The latest polymer, now under intense development, incorporates a latent acid segment in the polymer backbone that takes advantage of the acid-labile nature of the ortho ester linkages and allows control over erosion rates. By use of diols having selected chain flexibility, polymers that range from hard, brittle materials to materials that have a gel-like consistency at room temperature can be obtained. Drug release from solid materials will be illustrated with 5-fluorouacil and bovine serum albumin, and drug release from gel-like materials will be illustrated with mepivacaine, now in Phase II clinical trials as a delivery system to treat post-operative pain. A brief summary of preclinical toxicology studies is also presented. Introduction Poly(ortho esters) were developed in the very early 1970s, and at that time, the only bioerodible polymers known were poly(DL- and L-lactic acid), poly(glycolic acid), and their copolymers. However, these materials were designed as surgical repair materials and sutures, were not designed for drug delivery, and as a consequence had less than optimal properties for such applications. For this reason, the decision was made to prepare a bioerodible polymer, specifically designed for drug delivery. In designing such a polymer, our intent was to develop a material where the erosion process would be confined predominantly to the surface layers because in the absence of significant diffusional release, a drug uniformly dispersed in such a polymer should exhibit good constant release kinetics. This is in contrast to lactide and glycolide polymers, where hydrolysis occurs throughout the bulk of the materials and where drug release is a combination of diffusion and erosion. † This paper was presented at the ICMAT 2003 conference, held in Singapore June 29th through July 4th, 2003. * To whom correspondence should be addressed. Dr. Jorge Heller, P.O. Box 3519, Ashland, OR 97520. Phone: (541) 552-0941. Fax: (541) 5520941. E-mail:
[email protected].
Table 1. Relative Reactivities of of an Acetal, a Ketal, and an Ortho Estera
a Adapted from Cordes, E. H.; Bull, H. G. Chem. Rev. 1974, 74, 581605.
It was our hypothesis that surface erosion should occur with polymers that are highly hydrophobic to minimize water penetration into the bulk material and that the linkages that connect segments of the polymer backbone should undergo a rapid hydrolysis. After surveying various linkages, we narrowed the selection to acetals, ketals, and ortho esters. On the basis of the relative reactivities shown in Table 1, ortho ester linkages were selected.
10.1021/bm040049n CCC: $27.50 © 2004 American Chemical Society Published on Web 08/17/2004
1626
Biomacromolecules, Vol. 5, No. 5, 2004
Heller and Barr Scheme 1
Scheme 2
Development of Poly(ortho esters) Poly(ortho esters) have been under development since the early 1970s and during that time, four families of poly(ortho esters) have been described. They have recently been comprehensively reviewed.1 The four families are shown in Scheme 1. The polymers shown in Scheme 1 have been prepared by two general reaction processes. In one process, the polymers are prepared by a transesterification reaction, illustrated for POE I and POE III in Scheme 2. However, this type of reaction involves long reaction times to drive the equilibrium toward polymer and is very difficult to scale up, and accurate control of the molecular weight is virtually impossible. For this reason, despite interesting properties, particularly for POE III which has shown excellent ocular biocompatibility,2 these materials have not been commercialized. An alternate means of preparing poly(ortho esters) is by the addition of diols to diketene acetals. Unfortunately, very few diketene acetals have been described in the literature
and our first attempt to prepare poly(ortho esters) by the addition of diols to diketene acetals was based on the known 1,1,4,4-tetramethoxy-1,3-butadiene.3 This attempt is shown in Scheme 3. While the reaction with 1,1,4,4-tetramethoxy-1,3-butadiene proceeded readily, and as expected, a pendant methoxy group on an ortho ester linkage is highly reactive and reacts with unreacted diol to yield a cross-linked structure, as shown. There were no reaction conditions that we could find that would eliminate this side reaction, so this approach was abandoned. Clearly, such a synthetic procedure can only be used if the formation of a pendant methoxy group on the ortho ester linkages is prevented, and to do that, a cyclic diketene acetal must be used. One such cyclic diketene acetal described in the literature is 3,9-dimethylene-2,4,8,10-tetraoxaspiro[5.5]undecane,4 and the addition of a diol to this diketene acetal is shown in Scheme 4. However, in this diketene acetal, there are two electron donor groups on each double bond so that even minute traces of acid will induce a cationic polymerization. And because this diketene acetal has two double bonds per molecule, a cross-linked product will result. Because the addition of diols to diketene acetals requires the use of an acidic catalyst, attempts to prepare linear poly(ortho esters) only produced cross-linked materials as a result of the competing cationic polymerization. We have finally found that the competing cationic polymerization could be suppressed by using I2 in pyridine, and a linear polymer could be obtained.5 To inhibit the competing cationic polymerization, a new diketene acetal was synthesized where a methyl group was
Scheme 3
Biomacromolecules, Vol. 5, No. 5, 2004 1627
Poly(ortho esters)
Figure 1. Weight loss as a function of time for a polymer prepared from 3,9-dimethylene-2,4,8,10-tetraoxaspiro[5.5]undecane and 1,6hexanediol. 0.05 M PBS, pH 7.4, 37 °C. Scheme 4
Figure 2. Relationship between lactic acid release (b) and weight loss (9) for a poly(ortho ester) prepared from 3,9-diethylidene-2,4,8,10-tetraoxaspiro[5.5]undecane and a 100:70:30 mixture of 1,10decanediol and 1,10-decanediol lactide. 0.13 M, pH 7.4, sodium phosphate buffer at 37 °C (from ref 10). Scheme 6
Scheme 5
attached to the double bond to sterically hinder the cationic polymerization. Polymer synthesis using the new diketene acetal, 3,9-diethylidene-2,4,8,10-tetraoxaspiro[5.5]undecane, is shown in Scheme 5.6 Even though ortho ester linkages are very reactive, the polymer is highly hydrophobic so that the amount of water that can penetrate the polymer is very limited. For this reason, when a poly(ortho ester) device such as a compression molded disk is placed in buffer at pH 7.4 and 37 °C, the erosion rate is extremely slow, as shown in Figure 1.7 As would be expected, increasing the polymer hydrophilicity increases erosion rates, and in the extreme case when a soluble poly(ortho ester) is synthesized by the use of poly(ethylene glycol) as the diol, the polymer completely hydrolyzes to monomers in a matter of a few hours.8 Clearly, the synthesis of drug delivery systems that have realistic delivery times requires the development of a means of accelerating the erosion rate in a controllable and reproducible manner. To do that, we have taken advantage of the acid-labile nature of the ortho ester linkage and built into the polymer a latent acid catalyst. The incorporation of a latent acid into the polymer is shown in Scheme 6, and the polymer has been designated as POE IV.9 Polymer Properties Hydrolysis. Polymer hydrolysis occurs as shown in Scheme 7.10 As depicted, the hydrolysis proceeds in three consecutive steps. In the first step, the lactic acid or glycolic acid segment
in the polymer backbone hydrolyzes to generate a polymer fragment containing a carboxylic acid end group that will lower the pH and catalyze ortho ester hydrolysis. A second cleavage produces free R-hydroxy acid that provides the major catalytic effect for the hydrolysis of the ortho ester linkages. Further hydrolysis then proceeds in two steps to first generate the diol or mixture of diols used in the synthesis and pentaerythritol dipropionate, followed by ester hydrolysis to produce pentaerythritol and propionic acid. Figure 2 shows polymer weight loss and release of lactic acid.10 The concomitant weight loss and release of lactic acid argues convincingly for an erosion process confined predominantly to the surface layers of the polymer matrix. However, the process is not pure surface erosion because no polymer is so hydrophobic that no water can penetrate the bulk, and there is a significant drop in molecular weight of the remaining polymer, indicating some bulk erosion. However, because the polymer is very hydrophobic, the water concentration in the bulk is very low and, hence, the rate of hydrolysis is limited by the amount of available water. Because the water concentration in the outer eroding layers is high, the rate of hydrolysis is also high. Glass Transition Temperature. An important attribute of POE IV is the ability to vary independently thermal/ mechanical properties and erosion rates. Mechanical and
1628
Biomacromolecules, Vol. 5, No. 5, 2004
Heller and Barr Scheme 7
thermal properties can be varied by choosing the appropriate R group in the diol and in the latent acid. The use of rigid diols produces materials having high glass transition temperatures and, hence, high fabrication temperatures, while the use of flexible diols produces materials having low glass transition temperatures and hence, low fabrication temperatures The effect of the diol structure on the glass transition temperature is shown in Figures 3 and 4. Figure 3 shows the effect of varying the ratio of a rigid diol, transcyclohexanedimethanol, and a flexible diol, 1,6-hexanediol, on the glass transition temperature.11 Clearly, by an appropriate choice of diol ratios, any glass transition temperature between 110 and 20 °C can be selected. Figure 4 shows the effect of the chain length of a diol on the glass transition temperature.12 The ability to vary glass transition temperatures within very wide limits allows the preparation of two major types
Figure 3. Glass transition temperature of 3,9-diethylidene-2,4,8,10tetraoxaspiro[5.5]undecane, trans-cyclohexanedimethanol, and 1,6hexanediol polymer as a function of mole percent 1,6-hexanediol (from ref 11).
of materials, solid materials and gel-like materials. If the latter materials are properly constructed, they are directly injectable, and such materials will be covered later in this manuscript. Solid Materials Control of Erosion Rates. The effect of latent acid content on the polymer erosion rate is shown in Figure 5, where the latent acid content was varied from 5 to 0.1 mol %. Clearly, variation of the latent acid content translates into excellent control over the polymer erosion rate. Fabrication. Poly(ortho esters) are excellent thermoplastic materials and can be easily fabricated by conventional thermoplastic fabrication methods such as compression molding, extrusion, or injection molding. Because glass transition temperatures are adjustable within wide limits, fabrication temperatures can be customized for specific drugs taking into account their thermal stability. A detailed study
Figure 4. Effect of diol chain length on the glass transition temperature of polymers prepared from 3,9-diethylidene-2,4,8,10-tetraoxaspiro[5.5]undecane and R,ω-diols (from ref 12).
Poly(ortho esters)
Figure 5. Effect of latent acid content on erosion rates for a polymer prepared from 3,9-diethylidene-2,4,8,10-tetraoxaspiro[5.5]undecane, cyclohexanedimethanol, decanediol, triethylene glycol, and triethylene glycol glycolide. (9) 40:45:10:5; (b) 40:49:10:1; (2) 40:49.9:10:0.1.
Biomacromolecules, Vol. 5, No. 5, 2004 1629
Figure 7. Polymer weight loss (b) and 5-FU release (9) from a polymer prepared from 3,9-diethylidene-2,4,8,10-tetraoxaspiro[5.5]undecane and 1,3-propanediol/triethylene glycol diglycolide (90:10). Compression-molded films, cut to 5 × 5 × 0.1 mm. Drug loading 20 wt %, 0.05 M PBS, pH 7.4, 37 °C (from ref 14).
particular polymer used in this stability study is a hydrophilic polymer containing 40 mol % latent acid that has been ground to produce microparticles, thus, greatly increasing surface area. This is a very rapidly eroding polymer that would completely erode in a matter of a few days if placed in an aqueous buffer.1 Drug Delivery
Figure 6. Stability of a polymer prepared from 3,9-diethylidene-2,4,8,10-tetraoxaspiro[5.5]undecane, trans-cyclohexanedimethanol, triethylene glycol, and triethylene glycol glycolide (35:25:40) stored at room temperature and under anhydrous conditions. The four polymer samples represent four different preparations.
of polymer extrusion and drug incorporation has been carried out.13 The polymer is soluble in solvents such as methylene chloride, tetrahydrofuran, and ethyl acetate so that that films prepared by solution casting or coating of devices such as cardiovascular stents can be easily carried out. Further, the vast array of microencapsulation methods is also available. Polymer Stability. As shown in Figure 6, poly(ortho esters) have excellent stability and, when stored under anhydrous conditions, are stable at room temperature. The Table 2. Focus Diols
In discussing drug delivery, it is of interest to consider two types of drugs, low-molecular-weight water-soluble drugs and macromolecular drugs. Clearly, in constructing drug delivery devices, it is possible to use a great variety of different diols. However, to reduce the toxicology studies necessary for regulatory approval, the number of diols used needs to be limited. For this reason, we have limited, whenever possible, the diols to those shown in Table 2 and, based on various diols, have created a number of “focus polymer” groups. Low-Molecular-Weight Water-Soluble Drugs. The release of a low-molecular-weight water-soluble drug can be illustrated with 5-fluorouracil (5-FU), shown in Figure 7.14 Although some diffusional release cannot be ruled out, the concomitant drug release and weight-loss of the device provide strong evidence for a predominantly erosioncontrolled process. In the data shown, the 5-FU material balance is a little low, but there is little doubt that the predominant mechanism controlling release of 5-FU is erosion and not diffusion.
1630
Biomacromolecules, Vol. 5, No. 5, 2004
Figure 8. Effect of loading on amount of 5-FU released from a poly(ortho ester) prepared from 15 mol % trans-cyclohexanedimethanol, 40 mol % 1,6-hexanediol, 40 mol % triethylene glycol, and 5 mol % triethylene glycol glycolide. (9) 5.5 wt % 5-FU (14 mg), (0) 11.6 wt % 5-FU (28 mg), (b) 23.6 wt % 5-FU (56 mg). 0.01 M PBS, pH 7.4, 37 °C (from ref 14).
Further, as shown in Figure 8, the rate of drug release is proportional to drug loading.15 This is also consistent with a surface erosion process and provides additional evidence for the hydrolysis and erosion study already described. Macromolecular Drugs. Solventless incorporation of peptides and proteins into thin, extruded strands is a particularly attractive method, if the extrusion can be carried out at temperatures lower than the temperature at which the protein begins to denature. Such a study has been carried out with a model protein fluorescein isothiocyanate-bovine serum albumin (FITCBSA) that was mixed with polymer and extruded into 1-mm strands cut to 10-mm lengths. The extrusion was carried out at 75 °C. The strands were then placed in a pH 7.4 phosphate buffered saline (PBS) at 37 °C, and weight loss and FITCBSA release determinations were carried out. Results of this study are shown in Figure 9.16 To date, no protein stability studies have been carried out. Because the long lag time is not optimal in most applications, we have investigated a means of reducing the lag time and found that the addition of very small amounts of poly(ethylene glycol) to the matrix prior to extrusion was effective in reducing lag time. On the basis of these data, we have used an AB-block copolymer of poly(ortho ester) and poly(ethylene glycol) as the matrix and achieved the FTIC-BSA release shown in Figure 10.16 Clearly, the use of such block copolymers yields much improved release kinetics. The block copolymer was prepared as shown in Scheme 8.17 Gel-like Materials Gel-like materials have a number of significant advantages. Dominant among these is the ability to incorporate a therapeutic agent by a simple mixing procedure at room temperature, without the need to use solvents, and, when properly constructed, direct injectability. However, this latter
Heller and Barr
Figure 9. Release of FITC-BSA (2) and weight loss (9) from a poly(ortho ester) prepared from 3,9-diethylidene-2,4,8,10-tetraoxaspiro[5.5]undecane, 1,4-pentanediol, and 1,6-hexanediol glycolide (100: 95:5). Strands, 1 × 10 mm, extruded at 70 °C. 0.01 M PBS, pH 7.4, 37 °C. FITC-BSA loading 15 wt % (from ref 16).
Figure 10. Release of FITC-BSA from an AB-block copolymer containing 6 wt % 2 kDa poly(ethylene glycol). Poly(ortho ester) was prepared from 3,9-diethylidene-2,4,8,10-tetraoxaspiro[5.5]undecane and a 85:15 mol % mixture of 1,3-propanediol and buffered saline, pH 7.4, 37 °C. FITC-BSA loading 15 wt % (from ref 16).
application requires a limitation in molecular weights to about 5 kDa to reduce viscosity and, in some cases, the addition of excipients to further reduce viscosity. Molecular weights can be controlled by either skewing stoichiometry where an excess of diol is used or by using a monofunctional alcohol that acts as a chain stopper.18 Either procedure is very satisfactory, and materials having reproducible molecular weights can be readily prepared. Polymer Stability Gel-like materials are also stable at room temperature provided that moisture is rigorously excluded. The long-term stability of a low-molecular-weight gel-like material is shown in Figure 11. Erosion Rate Control As with solid polymers, the erosion rate can be controlled by the amount of latent acid incorporated into the polymer
Biomacromolecules, Vol. 5, No. 5, 2004 1631
Poly(ortho esters) Scheme 8
backbone. However, the erosion rate of these low-molecularweight polymers can also be controlled within fairly wide limits by the hydrophilicity of the diol(s) used. This is illustrated in Figure 12, which shows the lifetime of gellike materials as determined by a nuclear magnetic resonance (NMR) scan of such materials implanted subcutaneously in mice. With this polymer, the hydrophilic material prepared using triethylene glycol has a lifetime of only about 2 days, whereas the hydrophobic material prepared using decanediol has a lifetime in excess of 3 weeks. Both materials contained identical amounts of latent acid.
In this work, a gel-like material based on triethylene glycol and triethylene glycol glycolide with a molecular weight of about 6 kDa was used. To further reduce viscosity and facilitate administration, a methoxy poly(ethylene glycol) having a molecular weight of 500 Da was added. The specific composition, designated APF 112, contained 77.6 wt % polymer, 19.4 wt % methoxy poly(ethylene glycol), and 3 wt % mepivacaine. This polymer has a lifetime of about 3 days.
Development of an Injectable Formulation To Control Post-Operative Pain
Only a summary of these studies will be presented here, and a detailed description will be the subject of a separate publication. Two types of studies were carried out, (a) using a polymer hydrolysate and (b) using the APF 112 formulation. Polymer Hydrolysate. Hydrolyzing the polymer into its hydrolysis products simulates the instantaneous erosion of an implant and, thus, represents a worse case scenario.
In this specific application, the intent is to instil the gellike material containing mepivacaine into the surgical incision prior to closure. The rationale is to provide a high local concentration of mepivacaine within the incision while at the same time maintaining a low systemic concentration. And if the analgesic effect at the surgical site can be maintained, the patient’s dependence on orally administered opiates with their well-known side effects is greatly reduced.
Figure 11. Post-production stability of a gel-like poly(ortho ester) prepared from 3,9-diethylidene-2,4,8,10-tetraoxaspiro[5.5]undecane, triethylene glycol, and triethylene glycol glycolide.
Preclinical Toxicology Studies Preparatory to Investigational New Drug (IND) Filing
Figure 12. Depot volume remaining as a function of time as determined by nuclear magnetic resonance (NMR). Material injected subcutaneously in rats. Gel-like poly(ortho ester) prepared from 3,9diethylidene-2,4,8,10-tetraoxaspiro[5.5]undecane, (9) triethylene glycol, and triethylene glycol glycolide and (b) decanediol and triethylene glycol glycolide.
1632
Biomacromolecules, Vol. 5, No. 5, 2004
The hydrolysate was prepared by hydrolyzing the polymer in PBS at 80 °C for 24 h, adjusting the pH to 7.4 with NaOH, adding the methoxy poly(ethylene glycol), mixing thoroughly, adding deionized water to adjust osmolarity, and finally filtering through a 0.45-µm filter. The solution was then injected subcutaneously into male and female SpragueDowley rats and into male and female beagle dogs. In the rat study, the doses used were 0, 1, 3, and 10 mL/kg, and in beagle dogs, the doses were 0, 0.05, 0.1, and 0.2 mL/kg. Both animal species were observed for 14 days, and no adverse effects by clinical observation and gross necropsies were found at any time, up to 14 days. Studies with APF 112. The following incisional wound instillation study was carried out. A 1-cm full thickness incision was made, a subcutaneous pocket was thus created by blunt dissection, the APF 112 formulation was administered into the subcutaneous pocket, and the skin was closed with 4-0 nylon sutures which were removed after 7 days. The study was carried out using Sprague-Dowley male and female rats using a 500 and 1000 µL single doses, and the rats were sacrificed at day 8. Both doses were well tolerated, but the 1000 µL dose was too large and resulted in leakage and wound distension due to the excessive amount of formulation used. IND Filing. On the basis of these preclinical data, an IND was filed on August 26, 2003, and a Phase II clinical trial using treatment of pain following repair of an inguinal hernia using human volunteers is currently ongoing. References and Notes (1) Heller, J.; Barr, J.; Ng, S. Y.; Schwach-Abdellauoi, K.; Gurny, R. AdV. Drug DeliVery ReV. 2002, 54, 1015-1039.
Heller and Barr (2) Zignani, M.; Bernatchez, S. F.; LeMinh, T.; Tabatabay, C.; Anderson, J. M.; Gurny, R. J. Biomed. Mater. Res. 1998, 39, 277-285. (3) Scheeren, J. W.; Aben, R. W. Tetrahedron Lett. 1974, 12, 10191020. (4) Yasnitskii, B. G.; Sarkisyants, S. A.; Ivanyua, E. G. Zh. Obshch. Khim. 1974, 34, 1019-1020. (5) Heller, J.; Penhale, D. W. H.; Helwing, R. F. J. Polym. Sci., Polym. Lett. Ed. 1980, 18, 619-624. (6) Ng, S. Y.; Penhale, D. W. H.; Heller, J. Macromol. Synth. 1992, 11, 23-26. (7) Heller, J.; Penhale, D. W. H.; Helwing, R. F.; Fritzinger, B. K. Polym. Eng. Sci. 1981, 21, 727-731. (8) Heller, J.; Vandamme, T. Unpublished results. (9) Ng, S. Y.; Vandamme, T.; Taylor, M. S.; Heller, J. Macromolecules 1997, 30, 770-772. (10) Schwach-Abdellaoui, K.; Heller, J.; Gurny, R. Macromolecules 1999, 32, 301-307. (11) Heller, J.; Penhale, D. W. H.; Fritzinger, B. K.; Rose, J. E.; Helwing, R. F. Contracept. DeliVery Syst. 1983, 4, 43-53. (12) Heller, J.; Rime, A. F.; Rao, S. S.; Fritzinger, B. K. In Trends and Future PerspectiVes in Peptide and Protein DeliVery; Lee, V. H. L., Hashida, M., Mitsushima, Y., Eds.; Harwood Academic Publishers: 7000 Chur, Switzerland, 1995; pp 39-56. (13) Heller, J.; Barr, J.; Shah, D. T.; Ng, S. Y.; Shen, H.-R.; Baxter, B. Poly(ortho esters). In Scaffolding in Tissue Engineering; Ma, P. X., Elisseeff, J., Eds.; Marcel Dekker: New York, in press (14) Heller, J.; Barr, J.; Ng, S. Y.; Shen, H.-R.; Schwach-Abdellaoui, K.; Einmahl, S.; Rothen-Weinhold, A.; Gurny, R. Eur. J. Pharm. Biopharm. 2000, 50, 121-128. (15) Ng, S. Y.; Shen, H.-R.; Lopez, E.; Zherebin, Y.; Barr, J.; Schacht, E.; Heller, J. J. Controlled Release 2000, 65, 367-374. (16) Rothen-Weinhold, A.; Schwach-Abdellaoui, K.; Barr, J.; Ng, S. Y.; Shen, H.-R.; Gurny, R.; Heller, J. J. Controlled Release 2001, 71, 31-37. (17) Toncheva, V.; Schacht, E.; Ng, S. Y.; Barr, J.; Heller, J. J. Drug Targeting 2003, 11, 345-353. (18) Schwach-Abdellaoui, K.; Heller, J.; Barr, J.; Gurny, R. Int. J. Polym. Anal. Charact. 2002, 7, 145-161.
BM040049N
