Cha 1990

Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Thebiodegradabilityofpolyesterblends

Y. Cha and C.G. Pitt


Research Triangle Institute, Research Triangle Park NC 27709, USA
(Received 10 July 1989; revised 14 September 1989; accepted 19 September 1989)

Blends of poly(s-caprolactone) (PCL) and poly(L-lactic acid) (PM) with polyglycolic acid-co-l-lactic acid
(PGLA) were prepared by three methods: compression moulding, coprecipitation, and solvent evaporation
of a methylene chloride-in-water emulsion of the polymers. The rates of hydrolytic chain scission of each
component of the blends were determined by deconvolution of GPC traces of samples maintained in
phosphate buffer, pH 7.4,37%, for up to 3000 h. The observed rates were dependent on the method of
blending. For compression moulded blends, the rate of chain scission of PGLA was decreased and that of
PC1 and PLlA increased. A corresponding delay in the onset of weight loss was also observed. There was
no evidence of blend miscibility.

Keywords: Biodegradation, poly(e-caprolactone), poly(L-lactic acid), polyglycolic acid-co-L-lactic acid. polymer blends. hydrolysis,
gel permeation chromatography

Aliphatic polyesters derived from glycolic acid, DL- and 140°C for 18 h in an evacuated glass vessel in the presence
L-lactic acid, and c-caprolactone have found frequent appli- of stannous octoate4, 5. The polymers were purified by
cation as biodegradable matrices for prosthetics’ and precipitation from methylene chloride with methanol and
controlled drug delivery’. 3. The optimization of the key dried in vacua. The sample of PGLA was determined to
properties of these biomaterials, i.e. the permeability, the contain 83 mol% of lactide by ’ H-n.m.r. spectroscopy (CDCI,
rate of biodegradation, and the tensile properties, has solvent) using a Bruker Model Wm-250 Supercon spectro-
generally been achieved by copolymerization4-6. Blending of meter. Blends were prepared by casting common solutions
the homo- and copolymers represents an alternative but less of the polymers in methylene chloride, then compression
exploited means of tailoring the material properties. The moulding the cast films to a thickness of 0.6 mm in a Carver
changes in the permeability of polyesters effected by press for 2 min at 2000 p.s.i. then 1.5 min at 20 000 p.s.i.
blending are predictable and have already been used to tailor at elevated temperatures: PGLA 1 OO”C, PGLA-PCL and PCL
the rate and duration of drug delivery systems. For example, 125-l 3O”C, PGLA-PLLA 190-200°C. Discs, diameter
blending poly(s-caprolactone) (PCL) or its copolymers with 0.64 cm, approx. 20 mg, were punched from the moulded
increasing amounts of poly(L-lactic acid) (PLLA)‘, cellu- films. Precipitated blends of PCL and PGLA were prepared
loses*,’ or other glassy polymers permits a systematic by dissolution of a 1 g mixture of the two polymers in
reduction in the permeability of the polymer matrix. In methylene chloride (6 ml), concentration of the continuously
contrast to these diffusion studies, the biodegradation of mixed solutionin vacua at 35”C, and addition of an excess of
blends is largely unexplored. Several recent publications7. lo methanol. The precipitated mass was dried in vacua, and cut
suggest that blending may be used to manipulate the rates of into 3 mm3 pieces (about 20 mg). Microspheres were
biodegradation-controlled delivery of drugs, but the prepared by the solvent evaporation method” at ambient
mechanism of the process is not known. We report here the pressures. Typically, a solution of the polyesters (300 mg) in
effect of blending on the rates of hydrolytic degradation of CH,CI, (4 ml) was poured rapidly into 20 ml of water
the above polyesters and show that, despite the incompati- containing 0.4 wt%of poly(vinyl alcohol) (/VI, 25 000; 88%
bility of the blends, the rate of hydrolytic chain scission of hydrolysed). The mixture was stirred with a magnetic stirrer
each component is significantly modified by the presence of to form an emulsion; stirring was continued at 37°C until the
the second polymer. CH,CI, had evaporated. The microspheres were separated,
washed with deionized water, and dried in vacua.
Polymer hydrolysis was measured by immersing
METHODS
samples in 40 ml of 0.1 M phosphate buffer at pH 7.4 and
37°C. Duplicate samples were withdrawn at different time
PCL, PLLA and polyglycolic acid-co-L-lactic acid (PGLA)
intervals and weight loss was determined gravimetrically
were prepared by bulk polymerization of redistilled
after drying for 24 h in vacua. Molecular weights were
c-caprolactone or recrystallized L-dilactide and diglycolide at
determined by gel permeation chromatography (GPC) in
Correspondence to Dr C.G. Pitt. chloroform using a set of five p-Styragel columns (Waters
The authors are presently at Amgen Inc.. Thousand Oaks, CA 9 1320. USA Assoc.) with nominal pore sizes of 1 05, 1 04, 1 03, 1 02, and

o 1990 Butterworth Et Co (Publishers) Ltd. 0142-9612/90/020108-05

108 Biomaterials 1990, Vol 11 March


Biodegradability of polymer blends: K Cha and C.G. Pitt

5 X 10’ nm, and an eluent flow rate of 1 ml/min. The GPC 600 -7

traces were evaluated by the universal calibration method’”


using polystyrene standards (Waters Assoc., Ventron) and
published Mark-Houwink constants for PGLA13, PCL14, and 500
PLLA15. Curve fitting of the GPC traces of blends to two
overlapping Gaussian functions was carried out using a
least-squares computer program (RS?) on an IBM PC. 400
Polymer crystallinity, or absence thereof, was determined by
differential scanning calorimetry using a Perkin-Elmer
Model DSC-2 instrument, a heating rate of 10 or 20”C/min,
and an indium standard. The percentage crystallinity was
derived from the measured heat of fusion using the reported
heat of fusion of 139.5 J/g of 100% crystalline PCL”.
Thermomechanical analysis (TMA) was carried out using a
Perkin-Elmer Model DSC-2 instrument and a heating rate of
5”C/min. Dynamic mechanical analysis of blends was 100 -

conducted with a Rheovibron and a temperature range of


- 120 to 100°C.
-
0

RESULTS AND DtSCUSSlON


a
-100 -_

The rates of hydrolytic chain scission of blends of PCL with I I I I I 313 I I I I I I I I


2j 29 31 35 37 39 41
PGLA and PLLA with PGLA were first measured by
immersion of compression moulded films in phosphate ELUTION VOLUME (ML)

buffer at pH 7.4 and 37°C. At various time intervals,


samples were retrieved and the changes in the molecular 600
weights of the component polymers determined by GPC.
Overlapping GPC peaks were deconvoluted analytically by
the least-squares fit of two Gaussian curves to the data, after 500
verifying that the unblended polymers peaks could be
described by this function. Deconvolution using other
functions, such as the Tung equation’? and the Wesslau or 400
log-normal distribution’*, was less successful. An example
of deconvoluted GPC peaks is shown in Figure 1. The results
of the GPC measurements of PGLA-PCL and PGLA-PLLA K
i?j 300
blends are shown as semilog plots of /M, versus time in
5
Figures 2 and 3.
2
It has previously been established that the in v&-o 200 -
degradation of PCL, PLLA, PGLA, and other polyesters
occurs at the same rate in vitro, with no significant enzymatic
contribution to the process4. Rather, biodegradation of these 100 -
polyesters proceeds by random hydrolytic chain scission of
ester links, until the molecular weight has decreased to the
point that continued cleavage produces fragments small 0 -4
enough to diffuse from the polymer bulk; weight loss then
ensues. The kinetic laws governing chain cleavage are
derived from the assumption that cleavage is autocalysed by -100 -- Ib
1111111111111111111111
the carboxy end groups generated and is also proportional to 27 29 31 33 35 37 39 41 43 45 47 49
the water and ester concentrations (Equation 7). While the
ELUTiON VOLUME (ML)
number of cleavages is small, [HzO] and [Ester] may be
considered constant and integration of Equation 7, coupled Figure 1 Gel permeation chromatography traces (outersolidlines] from [a)
with the relationship [COOH] = M,-‘, leads to Equation 2. PCL and (bj a I:2 PCL-PGLA blend, and the least squares fit of Gaussian
functions (a, + symbols) to the experimental data.
[COOW], = k’[COOHJO[H20][Ester] (1)
M,’ = M,’ expf -kt) (2)
rapidly degraded polymer and to increase the rate of
Thus, a semilog plot of M, versus time is linear during this cleavage of the less reactive polymer, in approximate
initial phase of degradation. proportion to the weight fraction of each component. For the
It is evident from the semilog plots in Figures 2 and 3 1: 1 blends of PGLA with either PLLA or PCL, the rates of
that the linear relationship between ln(n/l,) and time is also cleavage of PCL and PLLA are increased and that of PGLA
observed with the compression moulded polyester blends, decreased. Increasing the proportion of one component of
although curvature is observed at later times as degradation the blend produces a weighted change in the rate constants,
progresses. The rate constants (k) of Equation 2 estimated evident from the results with I:2 and 1 :3 blends. The rates
from the initial slopes of semilog plots of the experimental of cleavage of the two components of the 1: 1 blends of both
data in Figures 2 and 3 are listed in Tab/e 1. The effect of PCL:PGLA and PLLA:PGLA are nearly identical.
blending is to slow the rate of chain scission of the more Not unexpectedly, the retardation of the rate of
Biodegradability of polymer blends: Y. Cha and C.G. Pitt

Table 2 Weight loss (%) of PGLA and its blends with PCL as a function of
time in phosphate buffer, pH 7.4, 37°C

Polymer 500h lOOOh 1500h 2010h 3020h 4000h

PGLAa 4 9 41 53 69 -
PCL:PGLA (1:3) 0 0 0 0 28 -
PCL:PGLA (1:2) 0 0 0 0 4.5 -
PCL:PGLA (1:l) 0 0 0 0 4 17
PCLb 0 0 0 0 0 0

alnitial M, 10 000; blnitial M, 76 000.

weight loss. Unblended PCL (n/r, 76 000) lost no weight in


this time period, consistent with earlier observations’g.
Dynamic mechanical analysis measurements were
102 I I I I undertaken to assess the miscibility of the compression
0 500 1000 1500 2000 2500
moulded blends. The Ts of PCL at -60°C in 1: 1 blends with
TIME (HRS)
PGLA and PLLAwas unchanged, indicating lack of miscibility.
Figure 2 The change in the molecular weights (M,) of unblended (0) PCL The closeness of the Tgof PGLA and the T,,, of PCL prevented
and PGLA and 13 (+). 12 (X), 1: 1 (m) compression moulded blends of the
observation of the Ts of PGLA in the blend; however the
same polymers, in phosphate buffer, pH 7.4, 37°C.
failure to observe a new averaged Ts in the vicinity of 0°C also
suggests lack of miscibility. The heat of fusion of PCL,
measured by DSC, was proportional to the weight fraction of
PCL incorporated in the blend, demonstrating no change in
the crystallinity of PCL on blending.
The lack of miscibility of the blends was not unexpected,
because of the large difference in solubility parameters of the
constituent polymers calculated from Fedor’s constants”,” :
PCL 20.8; PLLA 22.7; PGLA 25.4 J”’ cmm3”. This lack of
miscibility would appear to rule out a mechanistic explanation
whereby the hydrolysis rates were determined by the
weighted concentrations of the carboxy end groups of PGLA
and PCL (cf. Equation l), the more acidic PGLA carboxy end
group being the more effective catalyst. Similarly an increase
in the water content of the blend associated with the more
.- hydrophilic PGLA cannot be invoked if separate polymer
0 1000 20’00 30bo 40’00

TIME(HRS)
domains exist. However, despite the apparent lack of
miscibility, the large perturbation of the rates of chain
Figure 3 The change in the molecular weights (M,) of unblended and
scission suggested that there must be a strong interaction of
compression moulded It1 blends of PGLA and PLLA in phosphate buffer,
pH 7.4, 37°C. The initial M, of the unblended sample of PLLA was slightly
the polymers or a significant change in their morphology on
smaller than that used to prepare the blend. Symbols: unblended PLLA (U), blending.
PGLA:PCL blend (0). precipitation (0); and solvent evaporation of an The importance of the morphology of the polymers
emulsion (X). was examined by comparing the rates of hydrolysis of
PCL:PGLA blends prepared by two additional methods.
hydrolytic chain cleavage of PGLA in PCL-PGLA blends was These were: coprecipitation from a common solution with a
accompanied by an increase in the induction period before non-solvent and solvent evaporation of an emulsion in water.
weight loss. The weight losses of 1: 1, I:2 and 1:3 blends The latter procedure is a common method of preparation of
during a period of 4000 h are listed in Table 2. The greater microspheres for drug delivery. The result of these changes
the proportion of PCL, the slower the onset and extent of in sample preparation was a reduction in the rate of
hydrolysis of unblended PGLA (Figure 4). as well as a change
Table 1 Rate constants (X 1O4 h-‘) derived from initial slopes (first in the behaviour of the blends (figures5 6).The initial rateof
1000 h) of semilog plots of the molecular weights (M,) of polyesters and hydrolysis of PGLA microspheres was approximately one-
their blends versus time in phosphate buffer, pH 7.4, 37°C
third of the rate of compression moulded PGLA discs.
Blend ratio Moulded Precipitated Solvent evaporated
However, the rate increased with time and after 1500 h was
of polymers comparable to that of compression moulded PGLA. The rate
PCL PGLA PCL PGLA PCL PGLA of hydrolysis of precipitated PGLAwas intermediate in value.
The hydrolysis of PCL was less affected by the method of
1a 1.7 - 2.1 - 1.4 -
preparation, and the changes in molecular weight were
1:l 3.8 4.4
1:2 6.2 8.2 2.5 4.9 1.7 7.0 small. The rates of hydrolysis of PCL in coprecipitated 1 :2
1:3 11 8.8 2.7 3.3 2.1 6.4 and 1 :3 blends with PGLA were the same and significantly
0:l 31 - 24 - 9.7 lower than those observed with compression moulded
PLLA PGLA blends (Table 7). The rates of hydrolysis of PGLA in
microsphere blends with PCL were similar to the rates of
1 :o 2.8 - compression moulded blends and slightly greater than the
1:l 5.0 9.5
rates of coprecipitated blends.
0:l 31
No correlation between hydrolysis rates and the

110 Biomaterials 1990, Vol 11 March


Biodegradability of polymer blends. Y Cha and C-G. Pm

v_ . . _~
+- PCL

5
G
0 PGLA
Lu 108
g

0 500 1000
TIME(HAS)
1500 2000 lo2w- TIME(HAS)
2000

Figure 4 The change in the molacolar weights (M,j of PGlA in phosphate figtire 6 The change in the molecular werghts (NI,j oioobleoded (0) PCL
buffer. pH 7.4,3 7°C Samples were prepared& compression moolding (Oj, and PGL/L and lt2 (X) and 1:3 (0) blends of the same polymers, in
compression moulding as a 3: 1 PGLAPCL blend (Oj, precipitation (0). and phosphate buffer, pH 7.4, 37°C. Samples were prepared by solvent
solvent evaporation of an emulsion IXJ. evaporation of a methylene chlorrde solution of the polymers emulsified rn
aqueous po/y(vinyl akohol).

mechanical strength4. It is also evident that the method of


fabrication and the resulting morphology of the polymers
and their blends play a critical role in determining their
relative rates of hydrolytic degradation.

ACKNOWLEDGEMENTS

This work was supported in part by the National Institute on


Drug Abuse, Grant No. DABB 5-ROI -DA-36 16-03.

REFERENCES

d2 0 500 1000 1500 2000 2500


Hoffman, AS., Medical appltcations of polymeric fibers, /. Appi.
Polym. Sci.. Appl. Poiym. Symp. 1977, 33, 3 t 3-334
Pttt, C.G. and Schindler. A., Btodegradatton of polymers, tn Controlled
Drug Delivery (Ed. SD. Bruck), CRC Press, USA, 1983, pp 53-80
TIME(HRS) Heller. J., Biodegradable polymers in controlled drug delivery, CNt.
Rev. Ther. Drug Carrier Syst. 1984, 1 I 39-90
Figure 5 The change in the molecular weights (M,j of unblended (UJ PCL Pitt, C.G., Jeffcoat, R.A., Schindler, A. and Zweidinger, R.A., Sustained
and PGLA. and 1.2 (ij and 13 (OJ blends of the same polymers, in drug deltveiy systems. I. The permeabrlity of polyja-caprolactcne).
phosphate buffer, pH 7.4, 37°C. Samples were prepared by coprecipitation polyfot-lacttc acid), and thetr copolymers, J. Biomed. Mater. Res.
from methylene chloride with methanol. 1979,13,497-507
5 Pitt, C.G., Gratzl, M.M., Ktmmel, G.L., Surles, J. and Schlndler. A.,
Altphatlc polyesters. Il. The degradation of poly(oL-lacttde), poly(e-
caprolactone), and thew copolymers in viva, Biomateenals 198 1I 2,
or
AHfusion Ta of samples was observed. The initial crystal-
215-220
linities of unblended PCL, calculated from the heat of fusion, 6 Reed, A.M. and Gilding, D.K.. Biodegradable polymers for use tn
were 49% (moulded), 43% (microspheres) and 50% surgery: poly(glycolic)/poly(lactic actd) home- and copolymers. 2.
(precipitated). The Tgs of PGLA determined from the ln wtro degradation, Polymer 198 1, 22. 494-498
softening point by TMA were (n = 2): 31.8 + 1.8”C 7 Cha, Y. and Pitt, C.G., A one-week subdermai delivery system for
i-methadone based on biodegradable mkxocapsules. J Controlled
(moulded), 36.3 i 22°C (precipitated), and 34.2 Ifr 7.8”C
R&ease 1988. 7. 69-78
(mi~rospheres}. 8 Chang, R.K., Prtce. J.C. and Whttworth, C.W., Control of drug release
While the molecular basisforthe observed differences rates through the use of mtxtures of polycaprolactone and cellulose
in hydrolysis rates is not clear, the practical value of these acetate butyrate polymers, Pharm. Technol. 1986. 10, 24-33
9 Chang, A.K., Price, J.C. and Whltworth, C.W., Control of drug release
results is the identification of a new means by which the
rate by use of mixtures of polycaprolactone and cellulose acetate
properties of the polyesters may betailored for drug delivery. butyrate polymers Drug Dev. Ind. Pharm. 1987, 13. 1 1 19-l 135
For example, PCL is much more permeable than PGLA but 10 Hutchinson, F.G. and Fun. B.J.A., Brodegradable polymers for the
degrades very slowly. Blending of PCL with PGLA provides a sustatned release of peptides. Trans. 609th Biochem. Sac. Meetrng,
means of retaining the permeability and form stability of PCL Leeds, UK. 1985, 13. 520-524
11 Deasy, P.B., Microencapsulation and Related Drug Processes,
while increasing the rate of degradation. This is less easily
M. Dekker, New York, NY, 1984
accomplished by copolymerization; random copolymers of 12 Grubistc, Z., Remmp. P. and Benoit. Ii., A unwefsal calibratton for gel
&-caprolactone with glycolide or lactide have comparable permeation chromatography, J. Polym. Sci, Part ff 1967. 5. 753
permeabilities to PCL, and degrade rapidly, but have little 13 Pitt. C.G. and Gu. Z.W., Modification of the rates of charn cleavage of

Biomaterials 1990, Vol If March 111


poly(s-caprolactone) and related polyesters in the solid state, 17 Tung, L.H., Fractionation of polyethylene. J. Polym. Sci. 1956, 20.
J. Contraffed Reiease 1987, 4, 283-292 495-506
14 Koteske, J.V. and Lundberg, RD., Lactone polymers. Ii. Hydrodynamic 28 We&au, l-f., Die molekuiarg~ichtsuerte~lung einiger niederdruck-
properties and unperturbed dimensions of poly-s-caprotactone, pofyathytene. Macromol. &hem. 1956.20. 1 1 l- 142
J. Polym. Sci.. Part A2 1969. 7, 897-907 19 Pitt, C.G., Chasalow, F.I.. Hibionada, Y.M., Klimas, D.M. and
15 Schindler, A., Hibionada, Y. and Pitt, C.G., Aliphatic polyesters. III. Schindler, A. Aliphatic polyesters. I, The degradation of poly(s-
Molecular weight and molecular weight distribution in atcohol caprolactone) in v&o. ./. &pi. PO&?. Sci. 196 1~ 26, 3779-3787
initiated polymerization of s-caprofactone, & Poiym. Sci, &hem. Ed. 20 Fedors, RF., Method for estimating both solubility parameters and
198X,20,3 19-326 molar valumes of liquids, Polym. Eng. Sci. 1974, 14, 147-
16 Crescenzi, V.. Manzini. G., Calzolari, G. and Borri, C., Thermodynamics 154
of Fusionof poly-~-p~p~lacton~ and poly-c-caprolactone. Comparative 21 Van Krevelen, D.W., Proper&es of Polymers. Their Estimstion and
analysis of the melting of polylactone and polyester chains, Etrr. Correlation w&h Cbemicaf S%nztwe, Elsevier Scientific Publishing
Polym~ J. 1972. 8. 449-463 Co., New York, 1976. Ch 7

11.2 Biomaterials 1990, !/ol 11 March

You might also like