Blotechnol. Prog. lW2, 8, 527-532
527
Dextran Retention in the Rat Brain Following Release from a Polymer Implant Wenbin Dang and W.Mark Saltzman' Departments of Chemical Engineering and Biomedical Engineering, The Johns Hopkins University, Baltimore, Maryland 21218
Intracranial controlled release polymers may improve drug administration to the brain, where therapy is frequently limited due to the low permeability of brain capillaries t o therapeutic agents. On the basis of drug transport and elimination rates, we proposed that high molecular weight, water-soluble molecules would be retained in the brain space following release from an intracranial implant. T o test this hypothesis, solid particles of different molecular weight fractions of fluorescein isothiocyanate labeled dextran (FITC-dextran; 4 X lo3Da (4kDa) < weight-averaged molecular weight (M,) < 150 kDa) or fluorescein were uniformly dispersed in matrices of a polyanhydride copolymer synthesized from a fatty acid dimer and sebacic acid in a 5050 ratio, P(FAD: SA). When incubated in buffered saline, FITC-dextran fractions of 70 kDa M, were released from the polymer within 48 h; 4 kDa M, FITC-dextran and fluorescein were released more slowly. Following implantation of P(FAD:SA) matrices containing either 70 kDa M, FITC-dextran, 4 kDa M, FITC-dextran, or fluorescein into the brains of normal rats, fluorescent tracers were continuously released into the brain tissue for 30 days. Tracer concentrations within the brain were significantly higher for large molecular weight tracers (70kDa M, FITC-dextran >> 4 kDa M, FITC-dextran > fluorescein). The rate of elimination, kapp, of each tracer from the brain was determined by comparing experimental data with a model describing tracer diffusion/elimination in the brain extracellular space; kapp decreased with increasing molecular weight (fluorescein > 4 kDa M, FITC-dextran > 70 kDa M, FITC-dextran).
Introduction Controlled release polymers, implanted directly into the intracranial space, have been used to deliver bioactive agents to the brain. While the first intracranial controlled release polymers were used to address research questions (Moskowitz et al., 19811,systemsfor treating brain disease have now been developed using anticancer agents (Chasin et al., 19901,neurotransmitters (During et al., 1989;Freese et al., 1989),steroids (Reinhard et al., 1991),and growth factors (Camarata et al., 1992;Hoffman et al., 1990;Powell et al., 1990). A biodegradable polymer system delivering l,&bis(2-chloroethyl)-l-nitrosourea (BCNU) has been tested in humans at medical centers throughout the United States (Brem et al., 1991). In spite of this progress and the likelihood that new clinical implant systems will be available soon, little is known about the fate of agents released from polymers directly into the brain space. In a previous study, we developed quantitative criteria for predicting the effectiveness of drug delivery to the brain on the basis of rates of drug diffusion and elimination within the central nervous system (CNS) (Saltzman and Radomsky, 1991). These models confirmed that controlled release to the brain is an effective method for increasing local drug concentrations, but they also suggested that certain classes of compounds should be better candidates for intracranial delivery than others. In particular, agents that are water-soluble, slowly eliminated, and diffusible were predicted to be the best candidates for direct delivery
* Address correspondence to this author at the Department of ChemicalEngineering,The Johns HopkinsUniversity, Room 42 New Engineering Building, 3401 N. Charles St.,Baltimore, MD 21218. Phone: (410)516-8480. FAX: (410)516-5510. 8756-7938/92/300&0527$03.00/ 0
to the brain. In the present report, we tested these predictions by designing polymer implants for the controlled release of fluorescent tracers with different molecular weights and implanting them into the brains of normal rats.
Experimental Procedure Materials. Fluorescein and fractions of fluorescein isothiocyanate labeled dextran (FITC-dextran, 4,40,70, and 150X lo3Da (kDa)weight-averaged molecular weight (M,); polydispersity < 1.25-1.5)were obtained from Sigma Chemical Company (St.Louis, MO) and used as received. Polyanhydridecopolymers composed of a fatty acid dimer and sebacic acid were kindly provided by Nova Pharmaceutical Corporation (Baltimore,MD;M, = 25 OOO). These copolymers, designated P(FAD:SA), were composed of dimers of stearic acid, CH~(CHZ)&OOH,and sebacic acid, HOOC(CHz)&OOH, in a 1:l ratio as indicated
r
1
Preparation of Controlled Release Polymers. Tracer particles were encapsulated in a polymer matrix at 105%
0 1992 American Chemical Society and American Institute of Chemical Engineers
Biotechnol. Prog., lQ92,Vol. 8, No. 6
528
(w/w) loading. To attain a homogeneous distribution of tracer particles within the polymer matrix, the tracer particles were crushed and sieved so that each particle was not larger than 117 pm. P(FAD:SA) was melted at 80 OC in a Teflon-coated mold, tracer particles were added, and the suspension was mixed by vigorous stirring for 5 min or until the suspension was homogeneous. The mixture was cooled to room temperature, and the solidified mixture was removed from the mold. Small cylindrical pellets (2 mm thick, 3 mm in diameter, weighing between 15 and 17 mg) were cut from the resulting slabs with a cork bore. For calculation purposes, we assumed that the pellets were approximately spherical with a 1.5-mmradius. Release from Polymers into a Well-Stirred Solution, To characterizethe release of drug from the polymer devices, the polymer pellets were submerged into 1mL of phosphate buffered saline (PBS, pH 7.4, 120 mM NaC1, 2.7 mM KC1, 10 mM phosphate with 0.02% gentamicin, all from Sigma) and incubated at 37 "C with constant shaking (90 rpm). The buffered saline was removed and replaced with new PBS at specific time intervals. The tracer concentration in the old solution was determined by measuring the absorbance at 490 nm. The cumulative mass released from the polymer implants as a function of time following immersion, Mt, was calculated from these measured concentrations. Release from Polymers into the Rat Brain. Male Fisher 344 rats (6-7 weeks old, Harlan Sprague Dawley, Indianapolis, IN) were used in this study. The polymer pellets were sterilized under ultraviolet light for 1h prior to surgery. In each experimental animal, an intracranial polymer implant was placed 5 mm posterior to the bregma and 3 mm lateral to the sagittal suture. Surgery was performed under aseptic conditions. Animals were anesthetized by an intraperitoneal injection of 3.0-3.5 mL/kg of freshly prepared anesthetic solution containing 25 mL of ketamine hydrochloride (100 mg/mL, Parke-Davis, Morris Plains, NJ), 2.5 mL of xylazine (100 mg/mL, Rompun, Mobay Company, Shawnee, KS), 14.2 mL of ethanol (loo%),and 58 mL of sterile saline (0.9% NaC1). To insert the polymer implant, a small burr hole was drilled in the cranium at the desired site, a 3 mm deep cortical incision was made, and the implant was inserted. For comparison, instead of inserting an implant, aqueous solutions of the tracer compounds (12 pL of 83 mg/mL) were directly injected at the same tissue site in some animals. Animals were anesthetized, as described above, and euthanized by exsanguination at 1, 3, 7, 14, or 30 days followingpolymer implantation or bolus injection. Blood samples were collected into heparinized syringes by direct cardiac puncture and plasma was obtained by centrifugation. To determine tracer concentration in the brain tissue, the brain was removed from the animal, the polymer implant was carefully removed from the tissue, and the cerebrum was divided into two hemispheres. Each hemisphere was weighed and placed in a plastic tube. PBS was added to the tube (2X the hemisphere weight) and the tissue was homogenized under sonication in an ice bath for 5 min. The tissue sample was centrifuged, and the supernatant was analyzed for the presence of the tracer by spectrofluorimetry (excitation at 490 nm and emission at 515 nm). Data Analysis. Assuming diffusion is the predominant mechanism for drug release from the nearly spherical polymer pellets, release of tracer molecules into a wellstirred reservoir of buffered saline can be described by the
familiar Fickian equation for desorption from a sphere:
where C, is the local concentration of solute in the polymer as a function of time and position, t is the time following immersion into the reservoir, r is the radial position within the polymer matrix (r = 0 a t center of sphere; r = R at outer surface), and D, is the diffusion coefficientfor solute transport within the polymer matrix. We further assume that (i) tracer particle distribution is initially uniform throughout the pellet (C, = C,O a t t = 0,O 5 r 5 R); (ii) tracer concentration at the pellet surface is zero throughout release (C, = 0 at r = R, t > 0); and (iii) there is no flux of tracer through the center of the sphere (aC,lar = 0 at r = 0, t > 0). With these initial and boundary conditions, eq 2 can be solved to yield an expression for C, as a function of r and t (Crank, 1975). When integrated over radial position, an expression for the mass of tracer released from the sphere as a function of time, Mt, is obtained. For short times following initiation of release, the solution for Mt can be reduced to (Vergnaud, 1991)
(3) where Mt is the total amount of tracer released from the polymer pellet (a function oft) and M, is the totalamount of tracer initially dispersed in the pellet. We used this description to characterize the release of tracers from polymer pellets into well-stirred reservoirs by comparing the initial data (Mt versus t 1 / 2for short times following immersion) to eq 3 and, by linear regression, determining the D, that best represents the data. In some experiments, tracer compounds were directly injected into the brain tissue. In these cases, we assumed that the tracer was eliminated by an unspecified, firstorder process with a characteristic rate constant, kei, which was simply related to the half-lifefollowingdirect injection, ti$ction.
C = C, exp(-keit)
(4)
where tyection = In (2)lkei and Co is the initial concentration of tracer in the brain, t is the time following injection, and C is the concentration as a function of time. To aid in the interpretation of the experimentsinvolving implantation of polymers into the rat brain, we compared our results to a model for diffusionlelimination of drugs in the brain space following release from a polymer matrix (Saltzman and Radomsky, 1991). The model has been completely developed and analyzed previously, but is briefly summarized here. We assume that drugs are released from the polymer by diffusion, and therefore drug concentration in the polymer is described by eq 2. Once released from the polymer pellet, tracers diffuse through the brain extracellular space with a characteristic diffusion coefficient, Db, and are eliminated from the brain by a first-order process, characterized by a rate constant kapp. While kapp is a lumped parameter and may account for solute elimination by metabolism, internalization, or uptake by the ventricular system, we expect that partitioning into the systemic circulation will be the dominant mode of tracer elimination from the brain. With these assumptions,tracer transport within the brain is described
WEotechnol. Rog., 1992, Vol. 8, No. 8 100,
100,
0
24
48
72
96
120
144
168
0
2
4
6
8
10
12
14
Time (hr) Square Root Time (hr’”) Figure 1. Release of fluorescent tracers from P(FAD:SA) pellets. (A) The cumulative percent of fluorescent tracer released from the polymer is plotted versus time for polymer pellets containing 70 kDa M, FITC-dextran (a),4 kDa M, FITC-dextran (O), and fluorescein(A). Each symbol indicates the mean value obtained for three identical polymer pellets. In all cases, the standard deviation was smaller than the symbol used to indicate the mean. (B)The cumulative percent of fluorescent tracer released from the polymer is dotted versus the sauare root of time: data from panel (A) is replotted with the same symbols. Dashed lines indicate the best fit ofeq 3 to the initial dits points.
by a Fickian expression similar to eq 2:
(5) For simplicity, we assume that the tracer molecules do not bind to any fixed tissue components. To determine tracer concentrations in the brain as a function of time, eqs 2 and 5 must be solved simultaneously. Initially, we assume that tracer is uniformly distributed within the polymer pellet and not present in the brain:
C = 0 for r > R C, = , C for 0 < r < R
1
for t = 0
(ea) (6b)
Equations 2 and 5 are coupled through two boundary conditions at the polymer/tissue interface: KC = C, for r = R, t
(”)
>0
(7a)
(7b) -D ar = - D p ( 2 ) } where K is the equilibrium partition coefficient between the polymer and tissue phases. For a hydrophobic polymer releasing a water-soluble drug into the brain tissue, we expect that the drug will be present predominantly within the aqueous pore space of the polymer and the extracellular space in the brain. Therefore, we assume that K is -1 since the pore space in the polymer is approximately equal to the drug loading, 105% , and the extracellular space constitutes -18% of the brain tissue (Saltzman and Radomsky, 1991). To complete the description, we assume that tracer concentration is zero far from the polymer surface and that there is no flux of tracer through the center of the polymer pellet (the polymer implant is symmetrical): C=O -ac, =0 ar
asr-w atr=O
t>O
(8)
t>O
Equations 2 and 5-9 were solved numerically with two adjustable parameters, Db and = R(kapp/Db)1/2, yielding tracer concentrations in the brain as a function of time and position. These concentration profileswere integrated
to obtain the total mass of tracer present in the brain a t any time following implantation,Mpredickd. This quantity was compared to the experimentally determined mass of tracer present in the brain, Merperbend. Values of the two parameters were adjusted to fit the model predictions to the experimentaldata, by minimizing the root mean square (rms) error, as described previously (Saltzman and Radomsky, 1991).
Results When incubated in buffered saline a t 37 “C, P(FAD: SA) matrices containing tracer particles remained macroscopically intact for at least 20 days. After the first 2 days, the pellets became noticeably softer, although they could still be handled without much difficulty. Following several months in saline, a rim of partially degraded material was apparent around the pellet; this material was easily removed by shaking. Following removal of the degraded rim, a large core of apparently intact polymer remained. The release rate of different molecular weight fractions of FITC-dextran and fluoresceinfrom P(FAD:SA) matrices was determined by incubating small polymer pelleta in buffered saline a t 37 “C (Figure 1A). FITC-dextran, 4 kDa M,, and fluorescein were released more slowly than the high molecular weight, 70 kDa, FITC-dextran fraction. When plotted versus the square root of time, all of the curves were initially linear, suggesting that diffusion through the polymer controls the rate of drug release (Figure 1B). The effective diffusion coefficients for tracer transport in the polymer pellet, obtained by comparing the curves in Figure 1B to eq 3, are shown in Table I. Polymer/fluorescent tracer pellets, identical to those characterized in Figure 1, were implanted into the brains of normal rata. The total concentration of tracer was determined in brain tissue homogenates and blood plasma as a function of time followingimplantation;Figure 2 shows results for P(FAD:SA) implants containing 70 kDa M, FITC-dextran. FITC-dextran was continuously released from the polymer for 30days following implantation. Over the experimental period, the concentration of FITC-
Eiotechnoi. Prog., 1992, Vol. 8, No. 8
530
Table I. Characteristics of Tracer Molecules Used for Direct Delivery to the Brain Tissue.
fluorescein FITC-dextran
74b
4X1O3MW 70 X 1@M,
O.OOO7
loooj
27
0.087
0.7
