Barrier analysis of periocular drug delivery to the posterior segment

Journal of Controlled Release 148 (2010) 42–48

Contents lists available at ScienceDirect

Journal of Controlled Release j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / j c o n r e l

Barrier analysis of periocular drug delivery to the posterior segment Veli-Pekka Ranta b, Eliisa Mannermaa b, Kirsi Lummepuro d, Astrid Subrizi a, Antti Laukkanen a,1, Maxim Antopolsky a, Lasse Murtomäki a,c, Margit Hornof a, Arto Urtti a,⁎ a

Centre for Drug Research, University of Helsinki, Finland Faculty of Health Sciences, University of Eastern Finland, Finland Department of Chemistry, Aalto University, Finland d Division of Biopharmacy and Pharmacokinetics, University of Helsinki, Finland b c

a r t i c l e

i n f o

Article history: Received 10 July 2010 Accepted 19 August 2010 Available online 7 September 2010 Keywords: Ocular drug delivery Periocular Subconjunctival Trans-scleral drug delivery Retinal drug delivery RPE Pharmacokinetic simulation Posterior segment

a b s t r a c t Periocular administration is a potential way of delivering drugs to their targets in posterior eye segment (vitreous, neural retina, retinal pigment epithelium (RPE), choroid). Purpose of this study was to evaluate the role of the barriers in periocular drug delivery. Permeation of FITC-dextrans and oligonucleotides in the bovine sclera was assessed with and without Pluronic gel in the donor compartment. Computational model for subconjunctival drug delivery to the choroid and neural retina/vitreous was built based on clearance concept. Kinetic parameters for small hydrophilic and lipophilic drug molecules, and a macromolecule were obtained from published ex vivo and in vivo animal experiments. High negative charge field of oligonucleotides slows down their permeation in the sclera. Pluronic does not provide adequate rate control to modify posterior segment drug delivery. Theoretical calculations for subconjunctival drug administration indicated that local clearance by the blood flow and lymphatics removes most of the drug dose which is in accordance with experimental results. Calculations suggested that choroidal blood flow removes most of the drug that has reached the choroid, but this requires experimental verification. Calculations at steady state using the same subconconjunctival input rate showed that the choroidal and vitreal concentrations of the macromolecule is 2–3 orders of magnitude higher than that of small molecules. The evaluation of the roles of the barriers augments to design new drug delivery strategies for posterior segment of the eye. © 2010 Elsevier B.V. All rights reserved.

1. Introduction Posterior segment of the eye is an important target of drug treatment in ophthalmology. Neural retina, choroid, and vitreous are affected in many ophthalmic diseases. For example, age-related macular degeneration (AMD), glaucoma, diabetic retinopathy, and various forms of retinitis pigmentosa are damaging the posterior eye segment, where they may lead to impaired vision and even blindness. AMD is affecting tens of millions inhabitants in Europe, and the overall numbers worldwide are even higher [1]. The clinical and experimental drugs for the posterior segment treatments include both small and large molecules (such as siRNA, antibodies, growth factors, DNA), but their delivery to the posterior segment of the eye is problematic. Topical eye drops do not deliver drug effectively to the retina and choroid: the concentration in those tissues is circa 105 times lower than in the tear fluid [2]. The systemic drug administration is not a viable alternative either. Only very small

⁎ Corresponding author. Centre for Drug Research, University of Helsinki, Viikinkaari 5 E, 00014 University of Helsinki, Finland. Tel.: +358 9191 59636. E-mail address: Arto.urtti@helsinki.fi (A. Urtti). 1 UPM-Kymmene Corporation, UPM Nanocenter, FI-02150 Espoo, Finland. 0168-3659/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.jconrel.2010.08.028

fraction of the systemic dose reaches the eye due to the blood-retina barrier that limits the drug access to the posterior tissues of the eye [3]. Intravitreal injections and implants deliver drugs effectively to the retina and choroid. For example, intravitreal antibodies are widely used in the treatment of the wet form of AMD. Intravitreal drug administration is invasive and potentially risky, and may cause severe adverse effects such as endophthalmitis and retinal detachment. Even though the complications are rare, the number of adverse reactions will inevitably increase when millions of patients will be treated using repeated intravitreal drugs for years or decades. Alternative modes of drug delivery and improved long-acting formulations are needed [4]. Periocular route is less invasive than intravitreal administration and it provides higher retinal and vitreal drug bioavailability (about 0.01–0.1%) compared to the eye drops (about 0.001% or less) [2,5]. Periocular drug delivery can be accomplished as injections or implantation of the drug to the sub-conjunctival, sub-tenon or parabulbar space. The drug must permeate across several barriers to reach the target sites in the choroid, RPE or neural retina [3]. The physical barriers include sclera, choroid and RPE, whereas the lymphatic flow in the conjunctiva and episclera, and the blood flow of the conjunctiva and choroid constitute the physiological barriers. Periocular barriers for drug delivery have been investigated using ocular membranes ex vivo (sclera, RPE) and drug distribution studies

V.-P. Ranta et al. / Journal of Controlled Release 148 (2010) 42–48

in animals in vivo. Conrad and Robinson [6] showed that subconjunctival drug permeation takes place primarily via drug distribution through the sclera into the posterior ocular tissues. Spilling of the drug to the lacrimal fluid and subsequent corneal absorption did not contribute significantly. Ahmed and Patton [7] demonstrated that transscleral permeation of timolol and inulin delivered the drug to the uvea and posterior tissues, but not to the aqueous humour. Injected small molecules are eliminated rapidly from the subconjunctival administration site, presumably via conjunctival and episcleral blood and lymphatic flow [8,9], but the elimination of proteins is 1–2 orders of magnitude slower [10]. The role of blood and lymphatic flow was proven in experiments with post-mortem rabbits showing slower drug elimination from subconjunctival site and significantly improved delivery to the posterior segment [8]. Sclera is permeable to hydrophilic compounds, even macromolecules [11–14], but the permeability in the RPE is 1–2 orders of magnitude lower than in the sclera [15]. Selective block of the choroidal blood flow did not have significant influence on the retinal and subconjunctival drug concentrations suggesting minor barrier role for choroidal blood flow in posterior segment drug delivery [16]. Conjunctival and episcleral blood and lymphatic flows are considered to be the main limiting factors in posterior segment drug distribution after subconjunctival drug administration, whereas anatomic barriers and choroidal blood flow are claimed to be less important [8,16]. The assessment of the relative roles of these barriers is not easy, because they form a complex interacting system. In addition, roles of the barriers are probably dependent on drug properties and this aspect has not been discussed in the literature. Pharmacokinetic models may be helpful in evaluation of the complex interplay between the anatomical and physiological barriers of periocular drug delivery. Lee and Robinson presented simplified three-compartment models with first-order rate constants [17,18]. More complex models have been built and parameters solved by curve fitting [19,20]. Gabhann et al. [21] presented a complex model for periocular administration of green fluorescent protein (GFP), but also this model is partly based on fitted values, mouse data have been used, and only GFP was used as the drug. These models are not entirely physiologically based and the influence of drug on the roles of the barriers has not been taken into account. In this study, we extended our previous work [22], investigated scleral permeability, and built a physiological compartmental model to describe drug administration after periocular drug administration. The model reveals the relative contributions of the anatomical and physiological barriers for small molecules and macromolecules. 2. Materials and methods 2.1. Materials 5(6)-Carboxyfluorescein, FITC-dextrans and poloxamer 407 (Pluronic F127) were obtained from Sigma-Aldrich Chemie, Germany. The FITC-dextrans (mean molecular weight) were 4 kDa (4300), 10 kDa (9500), 20 kDa (20200), 40 kDa (38000), and 70 kDa (77000). Phosphorothioate oligonucleotides were synthesized using ASM 800 DNA/RNA synthesizer (BIOSSET Ltd, Russia) employing the standard coupling protocols of the manufacturer, solid phase support material (US II, Glen Research Corp., USA), fast sulfurizing agent (Glen Research Corp.) and phosphoroamidite monomers (dA–CE-, dC–CE-, dG–CE-, dT–CE- from Sigma-Aldrich). Fluorescein phosphoroamidite (Glen Research Corp.) was used to attach fluorescent label to the 5' terminus of oligonucleotides. Oligonucleotides were cleaved from the solid support with 4 M ammonium in methanol for 40 min., the support was rinsed with 32% water ammonia and final deprotection was carried out for 36 h at room temperature combined water/ methanol/ammonia solution. The solvents were evaporated, the residue dissolved in 0.5 ml of water and then compounds were

43

purified by ion exchange HPLC (Tosoh Bioscience DEAE-2SW; 4.6 mm × 25.0 cm, 5 μm column; from 0 to 50% B in 30 min. linear gradient [A – 0.1 M sodium acetate in 20% acetonitrile , pH = 8; B – 0.1 M sodium acetate and 0.4 M sodium perchlorate in 20% acetonitrile]). Peaks corresponding to the target compounds were collected. After desalting using Sephadex G-25 1 × 30 cm column the structure of compounds were confirmed with mass spectrometry. Three oligonucleotide sequences were synthesized: 12-nt (5'-GTT CCA TTC ATA-3'), 24-nt (5'-ACC TGG GAC ATC GTT CCA TTC ATA-3'), and 36-nt (5'-ACC TGG GAC ATC GTT CCA TTC ATA GTT CCA TTC ATA-3'). Poloxamer solution (20% m/m) was prepared by dissolving the poloxamer in 0.9% saline (4 °C). The solution was autoclaved. The permeating molecules were dissolved at 100 μM to the poloxamer solution that was stored at 4 °C before the experiment. Bovine sclera was obtained from slaughterhouse (Atria, Kuopio, Finland). The sclera was dissected freshly from the eyes, and then frozen at −20 C until used. 2.2. Release and permeation experiments Piece of sclera (1.54 cm2) was attached to the horizontal diffusion chamber apical side facing up. After ensuring that no leaking was taking place, 500 μl of 20% permeant solution (100 μM) in buffer or in poloxamer was pipetted on the sclera. Samples of 200 μl were withdrawn periodically from the donor and receiver compartments. Blank buffer was added as replacement. The experiments were carried out at 37 °C and the receiver chamber solution was mixed at 150 min−1. Concentrations of permeant molecules in the donor and receiver phases were determined by fluorescence (excitation at 485 nm, emission at 535 nm) with multi-label plate reader (Viktor2 1420, Wallac). Apparent permeability coefficients were calculated using pseudo steady state approach from the linear part of the permeated drug quantity vs time plots. 2.3. Compartmental pharmacokinetic model The pharmacokinetic model was an extension of our previous model [22] (Fig. 1). The model was built with Stella software (version 7.0.3, Isee Systems, Lebanon, NH, USA). The drug is administered to the compartment ‘Drug depot’ that is located at the scleral surface in the sub-conjunctival space or more posteriorly at the scleral surface periocularly. Part of the drug dose is eliminated from the depot and enters the capillaries and lymphatics (J10). The remaining drug enters the sclera-choroid border at the rate determined by the scleral permeability (J12). Thereafter, the drug diffuses across the extravascular choroid to the inner choroid (J23) , where the choroidal vasculature clears part of the drug (J30) , and the rest may permeate across the retinal pigment epithelium to the neural retina and vitreous (J34). Finally, the drug is eliminated from the vitreous (J40). In the case of controlled release formulation, the drug is released at constant rate to the ‘subconjunctival drug depot’ compartment

Fig. 1. Compartmental model of periocular drug delivery to the posterior eye segment. ‘Subconjunctival drug depot’ is the drug reservoir in the subconjunctival (or periocular) site after injection or implantation. ‘Retina, vitreous’ compartment includes the vitreous humor and the neural retina. The drug transfer rates are described as fluxes (J).

44

V.-P. Ranta et al. / Journal of Controlled Release 148 (2010) 42–48

(Fig. 1). In the simulation of the suprachoroidal drug delivery, the drug dose is placed to the ‘Sclera-choroid border’ compartment. The simulations were performed using Runge-Kutta 4 algorithm. The parameters in the model are based on the experimental data (Table 1). Cases of constant steady-state delivery (subconjunctival zeroth order drug delivery) and transient drug delivery (subconjunctival injection) were simulated. The simulations were carried out with three different drug properties (small hydrophilic, small lipophilic, macromolecule). Three drug classes were simulated, not specific individual compounds. This approach was adopted for two reasons: 1) not adequate experimental data exists for any individual compound, 2) these three classes of drugs have typical and quite distinct ranges of parameter values (Table 1). It was assumed that the surface area of the drug depot is 1.0 cm2 and the drug permeates to the vitreous in such a way that the exposed surface area of each membrane is always 1.0 cm2. Possible effects of lateral drug diffusion or convective flows in the choriocapillaries were not taken into account in the model. The clearance value for each drug transfer process was calculated as described in Table 1. The total clearance from Drug Depot compartment (compartment 1) is the sum of CL10 and CL12. The total clearance from Inner Choroid compartment (compartment 3) is the sum of CL30 and CL34. Physiological and anatomical barriers of periocular drug delivery were analyzed in series (Fig. 2). Each barrier allows certain fraction (F) of the entering drug quantity to pass through to the next tissue.

Fig. 2. Bioavailability (BA) of periocular drugs. Fbarrier values represent the fraction of the drug in each tissue that is able to pass through the barrier to the next tissue in series. 1 − Fbarrier is the fraction that does not pass through the barrier. In this model vitreal BA is equal to retinal BA, since it is assumed that there is no metabolism in the retina.

The product of individual F values provides the bioavailability (BA) in the vitreous. Fd and Fch were obtained by simulating the fraction of drug flux across the barrier as follows: Fd = J12 = ðJ12 + J10 Þ Fch = J34 = ðJ34 + J30 Þ It is assumed that there is no drug metabolism in sclera, RPE, and retina, and hence the values for Fsc and Frpe are 1 (100%) and vitreal bioavailability is the same as the RPE bioavailability.

Table 1 Parameters of the compartmental model. Factor

Valuesa

Rationale

Drug elimination from the subconjunctival depot

J10 = CL10 C1 CL10 = k10 V1 k10 = 7.6 h−1 (small lipophilic) k10 = 1.7 h−1 (small hydrophilic) k10 = 0.12 h−1 (macromolecule) V1 = 200 μl J12 = CL12 (C1 – C2) CL12 = Psc Ssc Psc = 20 × 10−6 cm/s (small molecules) Psc = 2 × 10−6 cm/s (macromolecule) Ssc = 1 cm2 V2 = 10 μl J23 = CL23 (C2 – C3) CL23 = Dch Sch / hch Dch = 5.3 × 10−6 cm2/s (small molecules) Dch = 2.5 × 10−7 cm2/s (macromolecule) Sch = 1 cm2 hch = 0.01 cm V3 = 10 μl J30 = CL30 C3 CL30 = Ech Qch Ech = 0.35 (small molecules) Ech = 0.000145 (macromolecule) Qch = 9.45 ml/h

k10 values were from experiments with lipophilic prednisolone in rabbits (7.7 h−1) [17] and celecoxib in rats (6.9–8.1 h−1)[19], and hydrophilic Gd-DTPA (1.5–1.9 h−1) [10] and Gd-albumin (0.11–0.13 h−1) in rabbits [10].

Drug permeation across the sclera

Extravascular drug permeation across the choroid

Drug elimination to the choroidal blood flow

Drug permeation across the RPE

Drug elimination from the vitreous

J34 = CL34 (C3 – C4) CL34 = Prpe Srpe Prpe = 16 × 10−6 cm/s (Small lipophilic) Prpe = 2 × 10−6 cm/s (Small hydrophilic) Prpe = 0.03 × 10−6 cm/s (macromolecule); Srpe = 1.0 cm2 J40 = CL40 C4 CL40 = k40 V4 k40 = 0.29 h−1 (small lipophilic) k40 = 0.055 h−1 (small hydrophilic) k40 = 0.0063 h−1 (macromolecule) V4 = 1500 μl

Experimental human Psc values for small and large drugs were used [14]. Lipophilicity does not influence significantly the permeability in the sclera [11].

Diffusivity of fluorescein in the mucus was used as Dch for small molecules [23]. For the macromolecule Dch was the mean D value obtained from eight macromolecular diffusion experiments in matrices resembling sclera [23–26].

Ech for small molecules is the experimental Ech of Cr-EDTA (0.35) in cats [27]. Ech for macromolecule is the mean experimental Ech of albumin (65 kDa; 1.9 × 10−4) and IgG (150 kDa; 1 × 10−4) in rabbits [28,29]. Qch is the experimental value for rabbits scaled to 1 cm2 of choroidal surface [29]. Plasma flow was used because the Ech values were based on plasma flow. Experimental bovine Prpe values for small hydrophilic and lipophilic drugs and macromolecule were obtained from Pitkänen et al. [15].

In vivo rate constants for vitreal drug elimination after intravitreal injection in rabbits were used. The literature values for small lipophilic (0.24–0.39 h−1)[30–32], small hydrophilic (0.02–0.09 h−1) [33,34] and macromolecules (0.0061–0.0066 h−1)[35,36] differed from each other, but they were consistent within each group.

a Abbreviations: J, flux in mass/time; CL, clearance; C, concentration in the compartment; k, first order rate constant; V, volume of the compartment; P, permeability coefficient; S, exposed surface area of tissue; D, diffusion coefficient; h, thickness of tissue; Ech; extraction ratio into choriocapillaris; Qch, plasma flow in choriocapillaries/cm2 of choroid.

V.-P. Ranta et al. / Journal of Controlled Release 148 (2010) 42–48

3. Results and discussion

45

Table 3 Simulated fraction passing from the drug depot (Fd) and the choroid (Fch) to the next layer and vitreal bioavailability (BA(vitreous)) a.

3.1. Permeation studies Scleral permeability of the 6-CF and FITC-dextrans decreased with increasing molecular weight. The scleral permeability (10−6 cm/s± S. D., n = 3) of 6-CF was 1.01± 0.33, and for FITC-dextrans 4 kDa, 10 kDa, 20 kDa, 40 kDa, and 70 kDa the values were 0.71 ± 0.09, 0.38 ± 0.06, 0.38 ± 0.18, 0.26± 0.04, and 0.10 ± 0.06, respectively. Linear relationship was obtained: Log(permeability)(cm/s) = −1.61 × molecular radius (nm) – 5.94 (R2 = 0.972). Molecular weight did not explain the permeability in the sclera as well as the molecular radius confirming the observation of Ambati et al. [12]. The permeability values for thicker bovine sclera were lower than the values of thinner human sclera [14]. Therefore, these values were used only to compare with oligonucleotide permeability, but not in the pharmacokinetic simulations. In the case of oligonucleotides, we did not see clear relationship between the length of the oligonucleotide and permeability. The permeability (10−6 cm/s ± S.D., n = 3) for 12-mer, 24-mer and 36-mer sequences were 0.09 ± 0.02, 0.19 ± 0.01 and 0.17 ± 0.05. The permeability of oligonucleotides was slower than that of FITC-dextran 40 kDa even though the molecular weights of the oligonucleotides are in the range of 4–12 kDa. The molecular conformation and radius of the oligonucleotides may, however, differ from the values of FITCdextrans. Sclera is a negatively charged membrane in the neutral pH [37], and the multiple negative charges of oligonucleotides are expected to impair the scleral permeability by charge repulsion. Previous comparison between negatively charged and neutral small molecules showed opposite results: negative charge increased the rate of permeation. In general, negative charges of the permeant, however, slows down the permeation in negatively charged membranes, which is in accordance with our results. Poloxamer 407 (20%) released all FITC-dextrans (4, 10, 20, 40, 70 kDa) at about the same rate. During 6 h 81–92% of the marker molecule was released from the polymer to the donor phase in the diffusion chamber. This was faster than the transscleral permeation from the gel. During 6 h less than 3% of the FITC-dextrans 10–70 kDa permeated across the sclera, whereas 14.5% of FITC-dextran 4 kDa did permeate through the sclera. These data show that drug delivery to the posterior segment is controlled by the sclera and not by the poloxamer gel. The simulations verified this conclusion (data not shown). In vivo, the issue of rate-control is expected to become even more striking, because the drug permeability in the RPE is usually 1–2 order lower than that of the sclera [15]. Much slower release rate is required for device-controlled kinetics in the periocular drug delivery.

Compound

Fd

Fch

BA (vitreous)

BA (vitreous) (%)

Small lipophilic Small hydrophilic Macromolecule

0.043 0.17 0.048

0.015 0.0019 0.071

0.00065 0.00032 0.0034

0.065 0.032 0.34

a

BA(vitreous) = Fd × Fsc × Fch × Frpe, where Fsc and Frpe are assumed to be 1 (100%).

and RPE (CL12, CL34). Thirdly, the clearance values in the sclera are within a narrow 10-fold range. Subconjunctival elimination (CL10) varies over about 60 times as a function of the molecular properties. However, the clearance from the inner choroid by the blood flow (CL30) and RPE permeation (CL34) are strongly dependent on molecular properties spanning over about 2300 and 500 fold ranges, respectively. 3.3. Bioavailability analysis Bioavailability analysis of the serial periocular barriers was carried out using the simulation model. Choroidal bioavailability (Fd, Fsc) was found to be in the range of 4.3–17% for the three molecular classes, while the vitreal bioavailability is in the range of 0.03–0.3% (Table 3 and Fig. 3). The simulated vitreal bioavailability of small hydrophilic molecule (0.03%) is in the same range with the experimental values for subconjunctivally injected hydrophilic compound (Gd-DTPA 0.06%)[5]. Small hydrophilic drug has the highest bioavailability in the choroid, but the macromolecule shows 5–10 fold higher vitreal bioavailability than the small molecules (Table 3 and Fig. 3). Even though these are low levels, nevertheless, they are much higher than the vitreal bioavailability after topical ocular drug administration (0.001–0.01% for lipophilic small molecules and less for hydrophilic compounds and macromolecules)[2]. 3.4. Roles of the periocular barriers Scleral permeability of the macromolecule is about 10 times slower than the permeation of small molecules [14], but due to their slow elimination from the subconjunctival space [10], the macromolecules have prolonged residence time on the scleral surface. The simulations suggest that, depending on the drug, the loss from the sub-conjunctival depot to the blood and lymphatic vessels is 83–95%. Despite its slow systemic absorption, major fraction of the

3.2. Clearance analysis Periocular route of drug delivery was analyzed using clearance approach. Firstly, Table 2 shows that the clearance values of small molecules are much higher than those of the macromolecule. Secondly, the elimination clearances from the subconjunctival depot (CL10) and choroid (CL30) are substantial relative to the competing clearances from the same compartment by permeation to the sclera

Table 2 Drug clearance from the ocular compartments. Parameter

CL10 CL12 CL23 CL30 CL34 CL40

Clearance (μl/h) Small lipophilic

Small hydrophilic

Macromolecule

1500 72 1900 3300 58 440

340 72 1900 3300 7.2 83

24 7.2 90 1.4 0.11 9.5

Fig. 3. Drug bioavailability in ocular tissues after a sub-conjunctival injection.

46

V.-P. Ranta et al. / Journal of Controlled Release 148 (2010) 42–48

subconjunctival macromolecule is lost to the blood and lymphatic flow. The model supports the conclusion [8] that the major part of the subconjunctival drug is eliminated by the conjunctival and episcleral blood and lymphatic flow processes. Transscleral permeation yields 4.3–17% of the drug dose to the choroid, a highly vascularized tissue. Choroidal extravascular diffusion delivers the drug to the choriocapillaries and the surface of the RPE. Previous publications claim that the choroidal blood flow is an insignificant factor in the periocular drug delivery to the posterior segment [8,16]. Our simulations, based on the quantitative experimental data of Törnquist [27] and Bill [28] suggest that major part of the drug entering the choroid is eliminated by the choroidal blood flow. This is the case for lipophilic (98%), hydrophilic (99%), and macromolecular (93%) drugs. However, these values represent only 4.2–17% of the total drug dose, because only a small fraction of the injected dose eventually enters the choroid (Fig. 3). We simulated the previous animal experiments that involved nearly complete shutdown of the choroidal blood flow [8,16]. As in the experiments [8], this modification had negligible effect on drug concentrations in the subconjunctival space. Nevertheless, in this case, the drug distribution to the vitreous increased several fold. The published study does not rule out the possibility of the increased vitreal drug delivery after the blockade of the choroidal blood flow, because the concentrations in both experiments (with and without choroidal blood flow) were below the detection limit of the analytical method [16]. Also, the similar experimental (0.06%) and simulated (0.03–3%) vitreal bioavailabilities support the role of the choroidal drug elimination. Conjunctival and episcleral elimination alone [10,17,19], without choroidal elimination, should result in vitreal bioavailability that is two orders of magnitude higher than the real experimental values. Choroidal clearance was simulated based on the experimental values from Bill and Törnquist [27–29]. The choroidal elimination is based on the high plasma flow in the choriocapillaries (9.45 ml/h per cm2) and the fenestrated leaky structure of the capillaries. The majority of fenestrations are facing towards Bruch's membrane [38]. The diameter of fenestrations is typically 70–80 nm and the number of fenestrations is 30–50 per μm2 [39,40]. Only a small portion of capillary endothelium in the retinal side is non-fenestrated. Based on these numbers the porosity of choriocapillaris in the retinal side is 10– 20%. However, the actual pore size is reduced by a fenestral diaphragm (diameter 20–30 nm) and filament like connections between the diaphragm and the fenestral rim [41,42]. In addition, the choriocapillary endothelium is surrounded by glycocalyx [43]. These structures restrict the molecular movement from the extravascular choroid into the capillaries. The clearance by free diffusion through the capillary pores obeys the relationship, CL30 = ε Dch A / h, where ε is the porosity of the capillary walls (0.1–0.2), Dch is the diffusivity in the choroid ((0.25–5.3) × 10−6 cm2/s), A is the surface area of the capillaries (1.5 cm2 per 1 cm2 of choroid), and h is the thickness of the capillary wall (2 μm). Completely open fenestrations would result in maximal, completely blood flow limited, clearance of the small molecules (i.e., E = 1.0, CL30 = 9450 μl/h). The experimental CL and E values of small hydrophilic molecule are high (E = 0.35, CL30 = 3300 μl/h), suggesting that the structures of the fenestrations may not significantly hinder the permeation of small molecules into the capillaries. In contrast, the free diffusion of the macromolecule through the open pores (CL30 = 700 μl/h) is 500 times faster than it is in the experiments (CL30 = 1.4 μl/h). It seems, that the ultrastructure of the fenestrations limits significantly the transport of macromolecules leading to slower choroidal clearance. Still, despite those hindrance factors, the choroidal clearance is an important barrier in periocular drug delivery. RPE has an important role in periocular drug delivery to the neural retina and vitreous. Unlike sclera, the RPE has selective permeability properties, and the lipophilic compounds have 10 and 500 times higher permeability than the hydrophilic drugs and macromolecules,

respectively [15]. Likewise, the RPE forms the outer blood retina barrier that effectively hinders the diffusion of xenobiotics that escape from the choroidal blood circulation [3]. The simulations suggest that the RPE is an important limiting membrane in the periocular drug delivery. For maximal retinal and vitreal bioavailability, the drug should have maximal permeability in the RPE, but minimal choroidal clearance. 3.5. Suprachoroidal injection Suprachoroidal injection has been introduced as a potential mode of posterior segment drug delivery [44]. In the simulation of suprachoroidal drug delivery the dose (10 μl) was ‘injected’ into the ‘sclera-choroid border’ compartment. Based on the simulations, suprachoroidal delivery should lead to increased vitreal bioavailability for small lipophilic (1.5%), small hydrophilic (0.19%), and macromolecule 4.2%), thereby showing 6–23 fold improvement in bioavailability as compared to the subconjunctival injection. After suprachoroidal injection, the choroidal blood flow removes most of the small molecular weight drug dose (96–99%), whereas the subconjunctival loss has a minor role (1.1–3.0%). However, the choroidal elimination of macromolecule is slower, and thus the drug is distributed also to the subconjunctival space leading to almost equal elimination by the choriocapillaries (54%) and subconjunctival factors (41%). Injection site seems to have clear impact on the vitreal bioavailability and the role of the elimination processes. 3.6. Steady state drug delivery Chronic retinal and choroidal diseases require continuous treatment with multiple drug doses. Steady state concentrations during the treatment determine the efficacy of the drug in the target tissues. The steady state levels are determined by the rate of drug input and the clearance from the tissue. The steady state pharmacokinetics can be achieved either by using multiple dosing regimens or continuous constant drug input (infusion or implant). We simulated drug delivery using constant zeroeth order drug input to the sub-conjunctival space (100 units per hour). The steadystate drug concentrations in the tissues were simulated for the three model compounds (Fig. 4). The simulated steady-state concentration of the macromolecule in the choroid is three orders of magnitude higher than the steady-state concentrations of the small molecules. In

Fig. 4. Relative steady-state concentrations of the simulated model drugs. Constant release of drug at the rate of 100 units/h was simulated.

V.-P. Ranta et al. / Journal of Controlled Release 148 (2010) 42–48

the vitreous the trend is the same, but the difference is two orders of magnitude. The high concentrations of the macromolecule can be explained by the clearance values, since the steady state concentration is by definition, Css = Jin/CL, where Jin is the influx of the drug and CL is the clearance from the system. The clearance of small molecules from the tissues into the blood circulation (CL10, CL30, CL40) is much greater than that of the macromolecule (Table 2). Compared to the small molecules, the macromolecule has long half-life in the tissues and, therefore, the steady state is reached slowly: vitreal steady-state levels of small lipophilic, small hydrophilic and large molecules are reached approximately at 15 h, 70 h, and after 500 h, respectively. The steady state levels of small molecules are reached rapidly (in less than 4 h) in the subconjunctival space and choroid, whereas the steady state of macromolecule is reached after 40 h. Compared to the small molecules it is possible to reach equivalent steady state concentrations in the choroid and vitreous at 100–1000 times lower rates of macromolecule input, but it takes longer time to achieve the steady state. The simulations suggest the lipophilicity is not a beneficial drug property in periocular delivery. In fact, at steady state the lipophilic drug is expected to have lower concentration than the hydrophilic one (Fig. 4). This is due to the fast clearance of the lipophilic drug to the blood stream (Table 2). Furthermore, the simulation may actually overestimate the levels of the lipophilic drug due to the inaccuracy of two clearance parameters. The subconjunctival clearance of the free dissolved drug may be faster than the experimental values for celecoxib suspension [19]. Also, it is probable that the choroidal clearance of lipophilic drug is higher than the value of hydrophilic CrEDTA (0.35 ml/h), because the lipophilic drugs may permeate also transcellularly across the capillary walls. Finally, lipophilicity is often associated with low water-solubility and, therefore, it is more difficult to achieve high concentration gradient of the drug in aqueous solution. 3.7. Future prospects Periocular delivery of small lipophilic, small hydrophilic and macromolecule drug was modeled. Even though the profiles of individual drugs are different from these model cases, this simple classification is valuable, because many relevant processes (subconjunctival and choroidal elimination, permeability in the sclera and RPE, intravitreal elimination) are strongly dependent on molecular properties, and especially the macromolecule shows distinct profile. The model is based on the data from ex vivo and in vivo experiments, and therefore the simulations coincide with in vivo pharmacokinetics. The model can be further refined to include factors, such as drug metabolism, active transport, binding to the pigment, and lateral drug distribution. These features were not included in this model. This simulation model can be used as a base to build refined models for individual compounds and drug formulations. Acknowledgements Parts of this study were presented in the ARVO Conference in May 2009. The study was supported by the Academy of Finland (A.U.). References [1] D.S. Friedman, B.J. O'Colmain, B. Munoz, S.C. Tomany, C. McCarty, R.T. de Jong, B. Nemesure, P. Mitchell, J. Kempen, Eye diseases prevalence research group, prevalence of agerelated macular degeneration in the United States, Arch. Ophthalmol. 122 (2004) 564–572. [2] A. Urtti, J.D. Pipkin, G. Rork, A.J. Repta, Controlled drug delivery devices for experimental ocular studies with timolol. 1. In vitro release studies, Int. J. Pharm. 61 (1990) 235–240.

47

[3] D.M. Maurice, S. Mishima, Ocular pharmacokinetics, in: M.L. Sears (Ed.), Pharmacology of the Eye, Springler-Verlag, Berlin, 1984, pp. 19–116. [4] E. Del Amo, A. Urtti, Current and future ophthalmic drug delivery systems. A shift to the posterior segment, Drug Discov. Today 13 (2008) 135–143. [5] H. Kim, M.R. Robinson, M. Lizak, G. Tansey, R.J. Lutz, P. Yuan, N.S. Wang, K.G. Csaky, Controlled drug release from an ocular implant: an evaluation using dynamic threedimensional magnetic resonance imaging, Invest. Ophthalmol. Vis. Sci. 45 (2004) 2722–2731. [6] J.M. Conrad, J.R. Robinson, Mechanisms of anterior segment absorption of pilocarpine following subconjunctival injection in albino rabbits, J. Pharm. Sci. 69 (1977) 875–884. [7] I. Ahmed, T.F. Patton, Importance of the noncorneal absorption route in topical ophthalmic drug delivery, Invest. Ophthalmol. Vis. Sci. 26 (1985) 584–587. [8] J.E. Chan, T.A. Pridgen, K.G. Csaky, Episcleral clearance of sodium fluorescein from a bioerodible sub-tenon's implant in the rat, Exp. Eye Res. 90 (2010) 501–506. [9] X. Liu, K.S. Li, E.K. Jeong, Ocular pharmacokinetics study of a corticosteroid by 19F MR, Exp. Eye Res. 91 (2010) 347–352. [10] S.H. Kim, K.G. Csaky, N.S. Wang, R.J. Lutz, Drug elimination kinetics following subconjunctival injection using dynamic contrast-enhanced magnetic resonance imaging, Pharm. Res. 25 (2008) 512–520. [11] M.R. Prausnitz, J.S. Noonan, Permeability of cornea, sclera, and conjunctiva: a literature analysis for drug delivery to the eye, J. Pharm. Sci. 87 (1998) 1479–1488. [12] J. Ambati, C.S. Canakis, J.W. Miller, E.S. Gragoudas, A. Edwards, D.J. Weissgold, I. Kim, F.C. Delori, A.P. Adamis, Diffusion of high molecular weight compounds through sclera, Invest. Ophthalmol. Vis. Sci. 41 (2000) 1181–1185. [13] D.M. Maurice, J. Polgar, Diffusion across the sclera, Exp. Eye Res. 25 (1977) 577–582. [14] T.W. Olsen, H.F. Edelhauser, J.I. Lim, D.H. Geroski, Human scleral permeability: effects of age, cryotherapy, transscleral diode laser, and surgical thinning, Invest. Ophthalmol. Vis. Sci. 36 (1995) 1893–1903. [15] L. Pitkänen, V.P. Ranta, H. Moilanen, A. Urtti, Permeability of retinal pigment epithelium: effect of permeant molecular weight and lipophilicity, Invest. Ophthalmol. Vis. Sci. 46 (2005) 641–646. [16] M.R. Robinson, S.S. Lee, H. Kim, S. Kim, R.J. Lutz, C. Galban, P.M. Bungay, P. Yuan, N.S. Wang, J. Kim, K.G. Csaky, A rabbit model for assessing the ocular barriers to the transscleral delivery of triamcinolone acetonide, Exp. Eye Res. 82 (2006) 479–487. [17] T.W.Y. Lee, J.R. Robinson, Drug delivery to the posterior segment of the eye II: development and validation of a simple pharmacokinetic model for subconjunctival injection, J. Ocul. Pharmacol. Ther. 20 (2004) 43–53. [18] T.W.Y. Lee, J.R. Robinson, Drug delivery to the posterior segment of the eye IV: theoretical formulation of a drug delivery system for subconjunctival injection, J. Ocul. Pharmacol. Therap. 25 (2009) 29–37. [19] A.C. Amrite, H.E. Edelhauser, U.B. Kompella, Modeling of corneal and retinal pharmacokinetics after periocular drug administration, Invest. Ophthalmol. Vis. Sci. 49 (2008) 320–332. [20] R.K. Balachandran, V.H. Barocas, Computer modeling of drug delivery to the posterior eye: effect of active transport and choroidal blood flow, Pharm. Res. 25 (2008) 2685–2695. [21] F.M. Gabhann, A.M. Demetriades, T. Deering, J.D. Packer, S.M. Shah, E. Duh, P.A. Campochiaro, A.S. Popel, Protein transport to choroid and retina following periocular injection: theoretical and experimental study, Ann. Biomed. Engn. 35 (2007) 615–630. [22] V.P. Ranta, A. Urtti, Transscleral drug delivery to the posterior eye: prospects of pharmacokinetic modeling, Adv. Drug Deliv. Rev. 58 (2006) 1164–1181. [23] W.M. Saltzman, M.L. Radomsky, K.J. Whaley, R.A. Cone, Antibody diffusion in human cervical mucus, Biophys. J. 66 (1994) 508–515. [24] R.G. Thorne, A. Lakkaraju, E. Rodriguez-Boulan, C. Nicholson, In vivo diffusion of lactoferrin in brain extracellular space is regulated by interactions with heparan sulfate, Proc. Natl Acad. Sci. USA 105 (2008) 8416–8421. [25] S.R. Chary, R.K. Jain, Direct measurement of interstitial convection and diffusion of albumin in normal and neoplastic tissues by fluorescence photobleaching, Proc. Natl Acad. Sci. USA 86 (1989) 5385–5389. [26] E.B. Brown, Y. Boucher, S. Nasser, R.K. Jain, Measurement of macromolecular diffusion coefficients in human tumors, Microvasc. Res. 67 (2004) 231–236. [27] P. Törnquist, Capillary permeability in cat choroid, studied with the single injection technique (II), Acta Physiol. Scand. 106 (1979) 425–430. [28] A. Bill, Capillary permeability to and extravascular dynamics of myoglobin, albumin and gammaglobulin in the uvea, Acta Physiol. Scand. 73 (1968) 204–219. [29] A. Bill, P. Törnquist, A. Alm, Permeability of the intraocular blood vessels, Trans. Ophthalmol. Soc. UK 100 (1980) 332–336. [30] S. Duvvuri, M.D. Gandhi, A.K. Mitra, Effect of P-glycoprotein on the ocular disposition of a model substrate, quinidine, Curr. Eye Res. 27 (2003) 345–353. [31] C. Durairaj, S.J. Kim, H.F. Edelhauser, J.C. Shah, U.B. Kompella, Influence of dosage form on the intravitreal pharmacokinetics of diclofenac, Invest. Ophthalmol. Vis. Sci. 50 (2009) 4887–4897. [32] S. Majumdar, K. Hippalgaonkar, R. Srirangam, Vitreal kinetics of quinidine in rabbits in the presence of topically coadministered P-glycoprotein substrates/ modulators, Drug Metab. Dispos. 37 (2009) 1718–1725. [33] X. Liu, K.S. Li, E.K. Jeong, Ocular pharmacokinetics study of a corticosteroid by 19F MR, Exp. Eye Res. 91 (2010) 347–352. [34] P. Berthe, C. Baudouin, R. Garraffo, P. Hofmann, A.M. Taburet, P. Lapalus, Toxicologic and pharmacokinetic analysis of intravitreal injections of foscarnet, either alone or in combination with ganciclovir, Invest. Ophthalmol. Vis. Sci. 24 (1994) 1038–1045. [35] M.R. Snyder, J.M. Reid, J.S. Pulido, R.J. Singh, Pharmacokinetics of intravitreal bevacizumab (Avastin), Ophthalmology 114 (2007) 855–859.

48

V.-P. Ranta et al. / Journal of Controlled Release 148 (2010) 42–48

[36] H. Kim, K.G. Csaky, C.C. Chan, P.M. Bungay, R.J. Lutz, R.L. Dedrick, P. Yuan, J. Rosenberg, A.J. Grillo-Lopez, W.H. Wilson, M.R. Robinson, The pharmacokinetics of rituximab following an intravitreal injection, Exp. Eye Res. 82 (2006) 760–766. [37] P.G. Watson, P.D. Young, Scleral structure, organization and disease, Exp. Eye Res. 78 (2004) 609–623. [38] M.H. Bernstein, M.J. Hollenberg, Fine structure of the choriocapillaris and retinal capillaries, Invest. Ophthalmol. 4 (1965) 1016–1025. [39] A. Sugita, M. Hamasaki, R. Higashi, Regional difference in fenestration of choroidal capillaries in Japanese monkey eye, Jpn J. Ophthalmol. 26 (1982) 47–52. [40] R.H. Guymer, A.C. Bird, G.S. Hageman GS, Cytoarchitecture of choroidal capillary endothelial cells, Invest. Ophthalmol. Vis. Sci. 45 (2004) 1660–1666.

[41] S. Melamed, I. Ben-Sira, Y. Ben-Shaul, Ultrastructure of fenestrations in endothelial choriocapillaries of the rabbit–a freeze-fracturing study, Br. J. Ophthalmol. 64 (1980) 537–543. [42] T. Matsuoka, Freeze-fracture studies of the choriocapillary endothelium. 1. Three dimensional visualization of the fenestral diaphragm, Nippon Ganka Gakkai Zasshi 87 (1983) 1065–1076. [43] T. Matsuoka, Freeze-fracture studies on the choriocapillary endothelium. 3. Threedimensional ultrastructure of environment surrounding the endothelium, Nippon Ganka Gakkai Zasshi 89 (1985) 1221–1235. [44] S. Einmahl, M. Savoldelli, F. D'Hermies, C. Tabatabay, R. Gurny, F. Behar-Cohen, Evaluation of a novel biomaterial in the suprachoroidal space of the rabbit eye, Invest. Ophthalmol. Vis. Sci. 43 (2002) 1533–1539.