Liquid, Injectable, Hydrophobic and ... - Wiley Online Library

Feature Article

Liquid, Injectable, Hydrophobic and Biodegradable Polymers as Drug Delivery Vehicles Brian G. Amsden

New delivery approaches to achieve minimally invasive, sustained and local release of drugs are needed for more effective treatment of conditions such as cancer and ischemia. Hydrophobic, biodegradable, liquid injectable polymers possess a number of potential advantages for this purpose. This review examines various approaches that have been explored for the preparation of these types of polymers, their ability to control the release of various drugs ranging from low-molecular-weight hydrophobic compounds to protein therapeutics, and finally their degradation rates and the tissue response to them upon implantation.

Introduction Advances in biotechnology and drug development have resulted in an increase in the number of active pharmaceutical agents that are difficult to administer by conventional means due to their low aqueous solubility or degradation and instability in the aqueous environment. Moreover, there is an increased demand for loco-regional administration combined with sustained and controlled release in the treatment of conditions such as ischemia, chronic pain, and cancer. To overcome these disadvantages, a number of different formulation approaches have been investigated that can be injected directly into the required site.[1,2] Among these, viscous, amorphous, liquid or low melting point, hydrophobic and biodegradable polymers possess distinct advantages, including: facile incorporation of thermally sensitive drugs such as proteins and peptides by simple mixing, injectability through standard gage B. G. Amsden Department of Chemical Engineering, Queen’s University, Kingston, Ontario, Canada E-mail: [email protected] Macromol. Biosci. 2010, 10, 825–835 ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

needles and thus administration via minimally invasive means, no need for device retrieval, restricted water penetration, which may provide enhanced stability for drugs incorporated as solid particles, and the viscous, amorphous nature of the polymer may limit irritation when implanted in soft tissue. In the past, oily vehicle parenterals have been used, such as: olive oil, sesame oil, and castor oil. These vehicles are slowly cleared from the injection site, through absorption into the capillaries or lymphatic system, by phagocytosis, and by in situ metabolism. Disadvantages of these formulations are possible reactions to the vehicle itself such as allergic reactions, cyst formation, pulmonary embolism following translocation of oil droplets, and localized lymphadenopathy.[3] A possible solution to these disadvantages is the use of a biodegradable, synthetic, liquid polymer. In this review, the various biodegradable, liquid polymer compositions that have been explored will be outlined and the various advantages and disadvantages of the approaches described. The review is restricted to those polymers that

DOI: 10.1002/mabi.200900465

825

B. G. Amsden

remain liquid upon injection. Other liquid polymer approaches, such as in situ setting through crystallization,[4] gel formation,[5] or solute leaching,[6,7] are not considered and have been discussed in recent reviews.[1,2] General Concepts To ensure that the polymer used is a liquid at temperatures reasonable for injection into patients, it must have a low molecular weight. At sufficiently low molecular weights, the polymer chains in the melt are not entangled with each other. As molecular weight increases, there exists a particular molecular weight for each polymer at which chain entanglement occurs, called the critical entanglement length (Me). Below Me, the polymer melt viscosity is typically Newtonian, and given by[8] 1

h ¼ KL Mw

(1)

where KL is a constant. Hence, as the polymer molecular weight increases, the viscosity increases. Above Me, the polymer melt viscosity is given by h¼

3:4 KH M w

(2)

in which KH is a constant. Above Me, chain entanglements result in marked increases in melt viscosity as molecular weight increases by acting as temporary crosslink points providing additional resistance to chain movement. Thus, it is important to ensure that the molecular weight used is below Me to ensure a viscosity low enough to permit injection through a needle. Molecular weight also influences the melting point of the polymer, with the melting point decreasing as molecular weight increases. If the polymer chosen, for reasons of degradability perhaps, is semi-crystalline, molecular weight can also be manipulated to ensure that it is amorphous at its operating temperature of 37 8C. To ensure that it can flow readily through a needle, the polymer must also possess a low glass transition temperature. The glass transition temperature dependence of the melt viscosity can be described using the Williams-LandelFerry equation,[8] h log hTg

!

C1 T Tg ¼ C1 þ T Tg

(3)

in which C1 and C2 are constants for a given polymer, and hTg is the melt viscosity of the polymer at its glass transition temperature. For many linear amorphous polymers, C1 and C2 can be approximated as 17.44 and 51.6, respectively.[8] As the glass transition temperature decreases, for example, by designing the polymer so as to possess flexible linkages in

826

Macromol. Biosci. 2010, 10, 825–835 ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

Brian Amsden obtained his PhD in Chemical Engineering from Queen’s University in 1996, in the area of therapeutic protein delivery from hydrogels and polymer microspheres. He worked for Angiotech Pharmaceuticals from 1996-97 as a Research Associate, leading projects involving the formulation of paclitaxel for localized delivery to treat post-operative adhesions and psoriasis, and participating in projects developing degradable microsphere and micellar formulations of paclitaxel for intra-articular delivery and systemic delivery, respectively. He left Angiotech to join the Faculty of Pharmacy at the University of Alberta and is currently a Professor in the Department of Chemical Engineering at Queen’s University where he has been since July 2000. His current research interests include the development of biodegradable elastomers, hydrogels, and low viscosity polymers for the local delivery of small molecules, peptides, and proteins, and as scaffolds for soft and connective tissue regeneration.

the backbone or through elimination of polymer chain interactions by inclusion of a plasticizer, the viscosity decreases. Furthermore, hTg is also molecular weight dependent, decreasing as molecular weight increases. For biodegradability, most polymer designs include hydrolyzable linkages such as esters or anhydrides. For these polymers, degradation rate is then dependent on the relative rates of water diffusion into the bulk of the polymer and the kinetics of the hydrolysis reaction.[9] The rate of water penetration is determined by the diffusivity of water into the polymer and the relative hydrophobicity of the polymer. Water diffusivity into polymers is dependent on polymer chain flexibility, increasing as chain flexibility increase, and hence as glass transition temperature decreases.[10] Degradation rates are also influenced by the nature of the drug incorporated; hydrophilic drugs will act in an osmotic fashion to draw water into the polymer bulk, whereas hydrophobic drugs can act as plasticizers or can retard water penetration into the bulk.[11] The hydrolysis kinetics are determined by the susceptibility of the linkage to hydrolysis; for example, anhydrides undergo hydrolysis much faster than do esters.[12] Moreover, esters and anhydrides are susceptible to auto-catalyzed hydrolysis, as they generate acidic degradation products that catalyze the hydrolysis, and hence the internal pH may decrease more rapidly than the pH at the surface of the implanted polymer. Poly(ortho ester)s The first descriptions of biodegradable, liquid polymers to be used for drug delivery were from Heller et al. concerning poly(ortho ester)s (POE).[13,14] Much of the early work was

DOI: 10.1002/mabi.200900465

Liquid, Injectable, Hydrophobic and Biodegradable Polymers as . . .

Scheme 1. Structures of poly(ortho esters) developed by Heller et al.[16] Top: poly(ortho ester) class III, prepared through the transesterification reaction of 1,2,6-hexanetriol and trimethyl orthoacetate. Bottom: poly(ortho ester) class IV.[18] R represents the chemical group that arises from the choice of diol used (e.g., 1,10-decanediol or triethylene glycol), while R0 represents the diol initiator used during the ringopening polymerization to yield either oligo(lactide) (shown above), or oligo(glycolide).

done using the poly(ortho ester) family prepared through the transesterification reaction of 1,2,6-hexanetriol and trimethyl orthoacetate, designated POE III (Scheme 1). These polymers degraded by hydrolysis of the ester linkages yielding acetic acid and the original triol as degradation products. The degradation rate of the polymer was controllable by incorporating acidic or basic compounds. Moreover, they were injectable, sterilizable, and possessed good biocompatibility when injected into the ocular subconjunctival space.[15] However, this polymerization approach required long reaction times, control of the molecular weight was unachievable, and it was very difficult to scale up and so further development of POE III was abandoned.[16] To overcome these problems, a new class of poly(ortho ester)s was developed, designated POE IV. These polymers were prepared through polyaddition of polyols with a diketene acetal, 3,9-diethylidene-2,4,8,10-tetraoxaspiro[5.5]undecane (DETOSU). Examples of polyols used included 1,10-decanediol or triethylene glycol to provide backbone flexibility, and oligo(lactic acid) or oligo(glycolic acid) (termed latent acid diols) to provide controllable degradation rates (Table 1). The structure of these polymers is provided in Scheme 1. It can be appreciated that a large number of different poly(ortho ester)s can be prepared in this fashion through utilization of different diols. The

nature of the diol used influenced the physical properties of the final polymer. For example, the use of 1,10-decanediol resulted in a hydrophobic polymer while triethylene glycol resulted in more hydrophilic polymers. The glass transition temperature of the polymers was also determined by choice of the diol used. For example, inclusion of trans-cyclohexanedimethanol, which imparts rigidity to the polymer backbone, increased the polymer glass transition temperature, whereas inclusion of longer chain alkane diols decreased the polymer glass transition temperature;[16,17] glass transition temperatures as low as 46 8C have been reported.[17] Control over the molecular weight of POE IV polymers was achieved by using an excess of diol relative to the diketene acetal or through the use of a chain-stopper monofunctional alcohol, such as, for example, 1-decanol.[17] The number-average molecular weight for injectable purposes was maintained below 5 000 Da, and the polydispersities ranged from 1.6 to 1.9.[17] The degradation of POE IV is believed to occur in the following sequence. First, the lactide or glycolide segment is hydrolyzed yielding a shorter polymer chain possessing a carboxylic acid end group. Further cleavage results in the liberation of free lactic or glycolic acid. These acidic groups then catalyze hydrolysis of the ortho ester groups. The final products of the degradation of this poly(ortho ester) class are the original diol used (e.g., triethylene glycol or 1,10-

Table 1. Physical properties of POE IV polymers developed as injectable delivery systems.

Polymer

DETOSU/1,10decanediol/1,10 decanediol dilactate

DETOSU/triethylene glycol/triethylene glycol monoglycolide a)

Mn;

M w;

Tg

h at 37 -C

Da

Da

-C

Pa s

5 200

9 700

25

148a) a)

4 500

7 400

33

70.8

3 800

6 100

46

60.9a)

6 410

8 850

25

1 200–1 300b)

Ref.

[17]

[24]

Complex viscosity; b)Zero-shear viscosity.


www.mbs-journal.de

827

B. G. Amsden

decandiol), pentaerythritol, propionic acid, and lactic or glycolic acid.[18] This process of degradation is supported by in vitro degradation data showing that the mass loss of high molecular weight poly(ortho ester)s prepared using 1,10decanediol, 1,10-decanediol dilactate, and DETOSU polymer was closely matched by the release of lactic and propionic acid.[19] The resulting degradation also appears to proceed predominantly via surface erosion, as the surface contains more water; however some bulk erosion also occurs. The degradation rates of the POE IV polymers depend on their monomer composition. Increasing the amount of oligolactide diol in the polymer increased the in vitro degradation rate of high molecular weight POE IV polymers prepared using 1,10-decanediol and 1,10-decanediol dilactate.[19] The same effect would be expected for low-molecular-weight polymers. Moreover, the use of an oligoglycolide diol would be expected to further increase the degradation rate given that polyglycolide is more hydrophilic and degrades faster than polylactide.[20] Similarly, the hydrophilicity of the diol can be utilized to control the degradation rate. For example, low-molecular-weight poly(ortho ester)s composed of DETOSU, triethylene glycol, and triethylene glycol glycolide degraded much more rapidly than those composed of DETOSU, decanediol and triethylene glycol glycolide when implanted subcutaneously in mice; the former degraded within 2 days while the latter degraded completely between 23 and 25 days.[16] The in vivo biocompatibility of these injectable polymers has been examined and reported after ocular implantation in rabbits,[21] and implantation into the periodontal pocket in humans.[22] In both cases the polymer was composed of DETOSU, 1,10-decanediol, and 1,10-decanediol dilactate, and had a weight-average molecular weight of 5 000 to 6 000 Da. The implants were well tolerated, with only mild inflammatory responses reported. Interestingly, the degradation rate of the implants depended on the ocular implantation site, with polymer lifetimes ranging from 5 weeks after subconjunctival implantation to 6 months following suprachoroidal or intracameral implantation. The differences in degradation rate were not discussed, but may be due to differences in local tissue movement. In the subconjunctical space, the polymer may have been exposed to mechanical stress as a result of eye movement, resulting in a physical erosion of the polymer, whereas in the suprachoroidal or intracameral space, the extent of mechanical stresses would likely have been much smaller. A number of different drugs or drug analogs have been incorporated into injectable POE IV polymers, and their release rates monitored. For example, in vitro and in vivo release studies have been undertaken with the local anaesthetic bupivacaine,[23] while in vitro release has been demonstrated using protein drug analogs such as bovine serum albumin,[23,24] lysozyme and lactalbumin,[24] and the angiogenic protein vascular endothelial growth factor

828


Figure 1. A) Representative release profiles of various drugs loaded into POE IV polymers composed of DETOSU, triethylene glycol, and triethylene glycol diglycolide.[23,24] B) A closer comparison of the release of positively charge drugs from the POE IV polymer showing that their release rates are approximately linear and essentially identical beyond an initial period. The solid lines represent a linear regression to the data. (LYS ¼ lysozyme, BSA ¼ bovine serum albumin, and VEGF ¼ vascular endothelial growth factor).

(VEGF)[24] from poly(ortho ester)s composed of DETOSU, triethylene glycol, and triethylene glycol diglycolide. Their release rates are shown in Figure 1. For all the drugs but VEGF, release began with an initial faster, burst phase, followed by a nearly linear release period. The release profiles of the low-molecular-weight bupivacaine (288.43 Da) and the higher molecular weight lysozyme (LYS, 14.7 kDa) are essentially the same, up to about 30 h, after which time the lysozyme release is much slower (Figure 1A). Despite exhibiting an initial lag period, VEGF (VEGF, 45 kDa) was released at essentially the same rate as well (Figure 1B). By contrast, bovine serum albumin (BSA, 67 kDa) was released much more slowly (Figure 1A). These release characteristics cannot be explained by the different size and

DOI: 10.1002/mabi.200900465


therefore diffusivity of the molecules within the polymer matrix, but appear to be dependent on the overall charge of the molecule. Bupivacaine (pKa ¼ 8 [25]), LYS (pHi ¼ 11 [25]), and VEGF (pHi ¼ 8.9 [26]) are all positively charged, whereas BSA (pHi ¼ 4.7 [27]) is negatively charged, at the pH conditions existing within the degrading polymer. This influence of molecular charge was noted by Weert et al. when examining protein release from the polymers.[24] They postulated that the more basic proteins effectively catalyze the degradation of the poly(ortho ester), enhancing their release, which is controlled by the polymer degradation rate. This explanation is supported by the fact that the bupivacaine is also released at the same rate as the positively charged proteins. In vitro release experiments have also been undertaken with poly(ortho ester)s composed of DETOSU, 1,10-decanediol, and 1,10-decanediol dilactate (molar ratios of 100:50:50) using tetracycline.[28] Tetracycline release from this more hydrophobic polymer lasted longer (approximately 15 days), but exhibited a more sigmoidal release pattern. This formulation was tested in humans for treatment of infection in the periodontal pocket.[22] Despite the formulation being apparently well tolerated by the patients, the polymer was not retained at the injection site, with only 2 out of 24 sites having the formulation present by day 11. A drawback of the use of these polymers with respect to drug delivery is that the accumulation of acidic degradation products within the polymer resulted in a drop in pH to less than 5.2 within a day within triethylene glycol-based poly(ortho ester)s.[24] This drop in pH has been cited as the cause of both protein aggregation and VEGF denaturation through hydrolysis when released from these polymers.[24] Attempts to overcome this denaturation could include the incorporation of basic compounds, as has been done previously with earlier POE polymers to control their degradation rate.[15] Nevertheless, such attempts to control the acidic microenvironment within poly(lactide-co-glycolide) microspheres has been somewhat unsuccessful as pH varies throughout the interior of the polymer,[29] and so the use of these injectable polymers may need to be restricted to drugs that are not acid sensitive.

Poly(ester)s A series of patents assigned to Ethicon Inc., describe the preparation and use of liquid or low melting point, biodegradable poly(a-hydroxy acid)s as injectable delivery systems.[30–33] All the patents describe the preparation of low-molecular-weight co- or terpolymers of e-caprolactone with either lactide, para-dioxanone, or trimethylene carbonate through ring-opening polymerization. The ecaprolactone is considered a necessary comonomer because Macromol. Biosci. 2010, 10, 825–835 ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

it imparts a low glass transition temperature and thus a low viscosity to the final polymer (the glass transition temperature of poly(e-caprolactone) is 60 8C [34]). Properties of examples of these polymers are given in Table 2. The data in Table 2 illustrate the degree of control over the final viscosity of the polymers achieved through copolymerization and by manipulation of the molecular weight. Polymer viscosity can also be controlled by the choice of initiator used. For low-molecular-weight polymers, the initiator comprises a significant portion of the overall macromolecule. In the examples in Table 2 taken from Bezwada et al.[32] and Scopelianos et al.,[33] the initiator used was propylene glycol, whereas the initiator used by Sharifpoor et al. was 1-octanol.[35] It has been demonstrated that the length of alkanol initiators used, and the presence of carbon-carbon double bonds in the middle region of the initiator used, influences the melt viscosity of the final lowmolecular-weight polymer, all other parameters being kept constant. For primary alcohol initiators, the longer the chain length of the initiator the lower the melt viscosity of the polymer, until a chain length of eight carbons, after which there was no noticeable effect. The use of secondary alcohols resulted in higher viscosities while the use of an unsaturated alcohol, also reduced the melt viscosity of the polymer.[36,37] The influence of the initiator is explained in terms of the flexibility of the overall polymer chain; the more flexible the polymer chain, the more readily it orients in the direction of the applied force, thus exhibiting a lower viscosity. Increasing the number of carbons in an alkane portion of the polymer increases its chain flexibility, while bulky pendant groups increase steric hindrance for bond rotation about the chain and decrease chain flexibility. The presence of unsaturated groups along the backbone results in eased rotation of the neighboring carbon groups on the backbone in the melt.[38] The architecture of the polymer also has been used to manipulate its viscosity. The use of polyol initiators such as glycerol or pentaerythritol in the ring-opening polymerization leads to star polymers, and low-molecular-weight starshaped polymers possess a lower melt viscosity than their linear counterparts of equal molecular weight.[38] This approach was taken by Sokolsky-Papkov and Domb, who used castor oil, a triglyceride of ricinoleic acid, as an initiator in a polycondensation reaction of L- and D,L-lactic acid to form a star polymer with three arms (Scheme 2); the polymer has a castor oil core with radiating poly(lactic acid) blocks and the length of the poly(lactic acid) blocks is controlled by the feed ratio of lactic acid to castor oil.[39,40] The viscosity of the polymers formed was determined primarily by the molecular weight and were viscous liquids due to the unsaturated group along the ricinoleic acid backbone and the low molecular weight (Table 2). There was also an influence of the stereoregularity of the polymer on viscosity; the poly(L-lactic acid) had a higher viscosity at a

www.mbs-journal.de

829

B. G. Amsden

Table 2. Physical properties of poly(a-hydroxy acid)s developed as liquid injectable delivery systems (CL ¼ e-caprolactone, PDO ¼ paradioxanone, LLA ¼ L-lactide, DLLA ¼ D,L-lactide, CO ¼ castor oil, TMC ¼ trimethylene carbonate, HLA ¼ monohexyl-substituted lactide, diHLA ¼ dihexyl-substituted lactide).

Monomers

a)

Composition

Mn

Mw

[h]a)

Tg

mol-%

Da

Da

dL g1

-C

CL:PDO

60:40

1 990

3 230

50:50

1 850

3 290

CL:PDO

50:50

Ref.

ha)

T

Pa s

-C

7.62

25

0.22

11.2

25

0.08

1.6

37

0.14

3.2

37

0.22

9.2

37

CL:LLA

50:50

0.14

CL:TMC

50:50

0.14

CL:TMC

88:12

1 550

2 810

67

50:50

1 400

2 525

62

LLA:CO

55:45

3 300

4 800

12.7

37

62:38

2 600

3 670

4.3

37

52:48

52:48

3 250

4 650

5.5

37

56:44

2 800

3 800

2.0

37

29:71

2 100

2 800

0.9

37

mHLA

100

4 800

6 000

17

715

25

diHLA

100

4 500

5 620

42

40

25

37

[32] [33]

37

[33]

37

[33]

1.1

37

[35]

2.7

37

3.4

[39,40]

[41]

[h] ¼ inherent viscosity in a 0.1 g dL1 hexafluorisopropyl alcohol solution at 25 8C,h ¼ zero-shear viscosity.

given molecular weight than did the poly(D,L-lactic acid), likely due to enhanced polymer-polymer chain interactions. While bulky pendant groups decrease polymer chain flexibility, resulting in increased viscosity, flexible pendant groups increase polymer chain flexibility. This fact has been exploited by Trimaille et al., who have synthesized alkylsubstituted lactide (Scheme 3) and used these monomers to prepare viscous liquid polymers through ring-opening polymerization.[41] The viscosity of the poly-(hexyl-substituted lactide) thus prepared decreased as the degree of hexyl substitution increased (Table 2). Examples of the in vitro degradation rate of poly(ahydroxy acid) approaches examined in the literature are given in Figure 2. These polymers degraded through a bulk degradation process, producing the characteristic weight loss profile; an initial period in which little weight loss is observed, followed by a period in which weight loss increases markedly.[11] In the data of Tomkins et al. (Figure 2A),[42] the 2 800 Da star poly[(e-caprolactone)-co(D,L-lactide)] (SCP) degraded at a faster rate than the 1 400 Da poly(e-caprolactone) (PCL) because of the incorporation of

830

0.19

Viscosity


the hydrophilic lactide monomer in the SCP. Moreover, the SCP was amorphous while the PCL was semi-crystalline, and crystalline regions resist hydrolysis and slow down the rate at which water can penetrate the polymer.[20] In the approach of Trimaille et al. [41], the grafting of a single hexyl pendant group to the lactide monomer reduced the glass transition temperature of the resulting polymer to the point where the 4 500 Da polymer was a viscous liquid (Table 2); however, the increased hydrophobicity that was generated reduced the overall degradation rate from that of poly(D,Llactide) of the same molecular weight (Figure 2B). The increase in degradation rate that resulted from grafting two hexyl groups to the lactide was due to the significant (25 8C) drop in glass transition temperature. This degree of decrease in glass transition temperature allowed for an increase in water diffusion through the polymer, leading to increased hydrolysis. Despite possessing low-molecularweight, the hydrophobic castor oil central core resulted in a slow degradation rate of the star-poly(lactic acid) approach of Sokolsky-Papkov (Figure 2C). Moreover, there was no influence of lactic acid stereoregularity on the degradation rate, likely because each polymer was amorphous as the

DOI: 10.1002/mabi.200900465


Scheme 2. Polycondensation reaction of castor oil and lactic poly(lactic acid) according to Sokolsky-Papkov and Domb.[39]

poly(L-lactic acid) blocks were too short to promote crystallization. Examples of the release of low-molecular-weight drug salts from viscous liquid poly(a-hydroxy acids) are shown in Figure 3A, while the release of vitamin B12 is shown in Figure 3B. The drug salt release rates vary from 1 week to 3 weeks, and are incomplete for the higher initial drug salt loadings of 10%. The incomplete release could be due to interaction between the positively charged tetracycline or tamsulosin, and the negatively charged polymer degradation products, although this is a tentative explanation, as the degradation rates appear to be slow (Figure 2C). Another possibility is that the dissolved drug salt within the polymer depots eventually precipitated from solution as the free base form of the drug and the limited solubility of the free base form in the aqueous surrounding medium produced a much slower release rate. Furthermore, there is not much influence of initial tamsulosin HCl loading on the release rate, until day 10. Beyond day 10, the tamsulosin HCl is released at a greater rate normalized to the initial drug loading from the depots containing 5% drug than from the Macromol. Biosci. 2010, 10, 825–835 ß 2010 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim

depots containing 10% drug. This result is in agreement with the lack of influence of loading on ofloxacin release from linear poly[(e-caprolactone)-co-lactide] of Bezwada et al.[32] By comparison, vitamin B12, a neutral, highly water soluble compound, was completely released from poly[(trimethylene carbonate)-co-(e-caprolactone)] depots (Figure 3B).[35] The release of vitamin B12 increased as the polymer vehicle viscosity decreased; the viscosity at 37 8C of the 1 500 Da, 50:50 poly[(trimethylene carbonate)-co-(e-caprolactone)] was 2.7 Pa s (Table 2), while that of the poly(trimethylene carbonate) was 16.8 Pa s.[35] The vitamin B12 was mixed into the polymer at only 1 wt.-%. Its release thus has been explained in terms of an osmotic pressure driven mechanism. Upon immersion into water, water penetrates the polymer as a diffusional front, until it reaches a particle. The water dissolves the particle, creating a saturated solution. As a result of the increase in the water activity gradient, water begins to be drawn toward the dissolving particle, generating a pressure equal to the osmotic pressure of the drug solution. The pressure begins to push on the acid yielding a star polymer, generating a pore that ultimately reaches the surface. This release mechanism may also be occurring for the drug salt loaded depots in Figure 3A. To date, none of these polymers have been used in the delivery of protein therapeutics. This is likely because these hydrolyzable polyesters will degrade to generate an acidic microenvironment, detrimental to the stability of the entrapped protein. Indeed, Trimaille et al. reported that tetracycline HCl was degraded within the poly(monohexylsubstituted lactide) vehicle, and attributed this degradation to the acidic microenvironment that is generated during polymer hydrolysis.[41] There has been little data published on the biocompatibility of these polyester liquid vehicles. Ekholm et al. have reported that low-molecular-weight amorphous pastes of poly[(e-caprolactone)-co-(D,L-lactide)] induced a severe inflammatory response when implanted in the abdominal muscle of rats.[43] Although not discussed by these authors, the severe inflammatory response is likely due to the rapid degradation and concomitant release of a high local concentration of acidic degradation products. SokolskyPapkov et al. have reported on the in vivo delivery of bupivacaine, a local anesthetic, from their star poly(lactic

www.mbs-journal.de

831

B. G. Amsden

Scheme 3. Structures of monohexyl-substituted lactide (mHLA) (left), dihexyl-substituted lactide (diHLA), and the poly(hexylsubstituted lactide)s formed through ring-opening polymerization (ROP) of these monomers.[41]

acid) depots;[40] however, the tissue response to the implant was not assessed. A different approach to obtaining liquid, hydrophobic polymers containing hydrolyzable ester linkages was taken by Nathan et al. Scheme 4.[44,45] This approach comprised a polycondensation reaction between a polybasic acid, such as succinic anhydride, with a monosubstituted fatty acid. A representative approach is illustrated in Scheme 3, showing the formation of a branched, ester linked, fatty acid polymer from the reaction of glyceryl monolinoleate with succinic anhydride. The resulting polymer was a viscous liquid with a numberaverage molecular weight of 2 260 Da and a weight-average molecular weight of 3 955 Da; the viscosity of the polymer was not disclosed. Examples of other fatty acids that were used in the polymerization process included glyceryl monostearate, glyceryl monooleate, glyceryl monodecanoate, and glyceryl monolaurate. These fatty acids were also copolymerized with each other. Furthermore, the hydrophilicity of the polymer could be increased by copolymerization of poly(ethylene glycol) diol.[44] In an in vivo experiment, 2 115 Da poly[(monostearoyl glycerol)-co(glyceryl monolinoleate/succinate)] was implanted into bone defects in rabbits by injection. In two of the four implants, bone regeneration was observed at 8 weeks.[45] No other description of tissue response was provided. Poly(carbonate)s Recognizing that the acidic degradation products of injectable polyesters adversely affects incorporated acid sensitive drugs, such as proteins and peptides, we have recently explored the potential of low-molecular-weight

832


Figure 2. Weight loss during in vitro degradation of low-molecularweight, viscous liquid poly(a-hydroxy acid)s. A) Data from Tomkins et al., comparing the degradation rate of 1 400 Da 1-octanolinitiated PCL to 2 800 Da star poly[(e-caprolactone)-co-(D,L-lactide)] (50:50 mol:mol) (SCP) and blends of the SCP and PCL of different wt.-% SCP.[42] B) Data from Trimaille et al. comparing the degradation rate of 4 500 Da poly(D,L-lactide) (PLA) to poly(monohexyl-substituted lactide) (PmHLA) and poly(dihexylsubstituted lactide) (PdiHLA).[41] C) Data from Sokolsky-Papkov et al., for three-armed star poly(lactic acid) initiated with castor oil, illustrating the influence of lactic acid stereoregularity.[39]

DOI: 10.1002/mabi.200900465


Figure 4. In vitro release of VEGF from 1 600 Da poly(trimethylene carbonate). The VEGF was mixed in as