Evaluation of Clearance Mechanisms with Transscleral Drug Delivery

Abstract

Purpose.: The goal of this study was to examine elimination pathways when delivering subconjunctivally administered hydrophilic agents to the retinas of rat eyes.

Methods.: The distribution of sodium fluorescein released from an episcleral implant was compared in live and postmortem eyes. Elimination of the subconjunctivally administered hydrophilic agent IgG through blood and lymphatic vessels was investigated by immunohistochemistry. Additionally, lymphatic elimination of subconjunctivally injected sodium fluorescein was quantitatively evaluated.

Results.: NaFl released from an episcleral implant was successfully delivered to the subretinal space in the postmortem eye but failed to do so in the live eye. Immunohistochemical visualization of the conjunctival tissue demonstrated dense distribution of blood and lymphatic vessels while also confirming the elimination of subconjunctivally administered IgG through these same vessels. The lymphatic elimination rate after injection of 75.6 μg of a hydrophilic agent, sodium fluorescein, into the subconjunctival space was determined to be 105 ng/min between 30 and 60 minutes.

Conclusions.: Conjunctival blood and lymphatic vessel elimination considerably limit transscleral hydrophilic drug delivery to the retina.

Severe vision loss in age-related macular degeneration (AMD) is caused by pathologic ocular neovascularization. ¹ Although antiangiogenic agents can inhibit choroidal neovascularization, ^2,3 delivering these drugs to the posterior segment of the eye remains challenging. Systemic delivery of these agents often fails to deliver drug to the eye at therapeutic levels because of the blood-retinal barrier. ⁴ Furthermore, systemic delivery of biologically active agents in doses capable of achieving therapeutic levels in the eye may lead to unwanted systemic side effects. Topical administration, such as eye drops, accounts for nearly 90% of the currently accessible ophthalmologic market formulations. ⁵ However, topical eye drops cannot effectively deliver these biologics to the choroid or retina because of a long diffusion distance, counter-directional intraocular convection, lacrimation, and corneal impermeability to large molecules. ⁶ Currently, intravitreal delivery is the most efficient and widely used method for delivering anti–vascular endothelial growth factor (VEGF) agents directly to the posterior segment. Although direct intravitreal injection has been proven to be effective in randomized clinical trials, the short half-life of this method necessitates frequent readministration, which is accompanied by the risk for vitreous hemorrhage, retinal detachment, and endophthalmitis. These complications can be avoided using transscleral drug delivery methods such as subconjunctival or sub-Tenon's injections, which has shown a lower incidence of serious complications than intravitreal injections. ^7,8 In addition, the use of transscleral delivery methods, with their low incidence of adverse effects, could be justified in preventive strategies to reduce the conversion of dry to wet AMD in high-risk eyes. Nevertheless, there are limitations to using this technique in the clinic. 

Transscleral drug delivery may be feasible because of the sclera's large and easily accessible surface area, high degree of hydration (rendering it conducive to water-soluble substances), hypocellularity with an attendant paucity of proteolytic enzymes and protein-binding sites, and permeability that does not appreciably decline with age. ^9–12 Previous in vitro experiments have demonstrated that the sclera is permeable to molecules as large as 120 kDa fluorescein isothiocyanate (FITC)-conjugated dextran. ¹³ However, our previous in vivo studies showed that an episcleral implant did not, in fact, deliver a significant amount of model agents into the vitreous and retina, despite a molecular weight <120 kDa. ^14,15 Based on these previous in vivo studies, ^14,16 we concluded that dynamic or physiological mechanisms of elimination rather than static structural barriers most significantly restrict drug penetration from the episcleral space to the retina and vitreous. Given the limitations of studying transscleral drug delivery in vitro, we investigated transscleral delivery of a hydrophilic agent to the rodent retina in vivo with the hypothesis that conjunctival lymphatics or blood vessels, or both, may be an important barrier to transscleral drug delivery. 

Methods

Animals

Long Evans female rats (weight range, 201–225 g) were purchased from Charles River Laboratories (Wilmington, MA). All procedures adhered to the guidelines of the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were performed with the approval of the Institutional Animal Care and Use Committee at Duke University. 

Episcleral Implant Design

An episcleral implant was manufactured as follows: 15% (wt/vol) hydroxypropyl methyl cellulose (HPMC) solution was formulated by adding 300 mg HPMC into 2 mL of 1× PBS (pH 7.4) in a scintillation glass vial and placed in a water bath at 45°C for 2 hours to dissolve all the HPMC. Six milligrams of fluorescein salt solution (NaFl) in 200 μL water was added and stirred into the HPMC solution to produce a 2% NaFl and 15% HPMC (wt/wt/vol) solution. A portion of the solution was poured onto a glass plate, producing a thin film as it dried at room temperature. Disks were made with a biopsy punch (Acu-Punch; 2.5 mm; Acuderm, Fort Lauderdale, FL) to make 2.5-mm diameter discs. These 2.5-mm diameter disc type implants were suitably sized for insertion into the rat eye subconjunctival space. 

In Vitro Release of the Implant

The in vitro release rate of NaFl from the episcleral implant was determined by placing the implants in 100-mL 1× PBS and stirring with a magnetic bar at 50 rpm at room temperature. One hundred microliters of the solution was assayed every 5 minutes for 30 minutes and every 10 minutes for the next 20 minutes. The NaFl assays were placed in a 96-well plate and performed using a microplate spectrofluorometer (Spectra Max Gemini Microplate Spectrofluorometer; MDS Analytical Technologies Inc., Sunnyvale, CA). A calibration curve was made using a 485-nm excitation wavelength and a 538-nm emission wavelength. The amount of NaFl in the solution at each time point was measured and summated to calculate the cumulative release of NaFl from each implant. 

Episcleral Surgical Implantation

The rats were first anesthetized by intramuscular injection of ketamine (70 mg/kg) and xylazine (30 mg/kg) to maintain a proper level of surgical anesthesia, which was determined by palpebral or hind leg withdrawal reflex. Proparacaine (0.5%) was administered topically for corneal anesthesia. A 2.5-mm superotemporal limbal conjunctival incision was made with scissors to expose the sclera, and blunt dissection parallel to the scleral plane was performed to insert an NaFl implant into the postequatorial subconjunctival space. No suturing was required because of the small incision and the postequatorial position of the implant. CO₂ euthanatization was performed at indicated times after implantation. The eye was enucleated and frozen immediately in OCT compound. The frozen eyes were sectioned into 15-μm thick slices to determine the distribution of NaFl released from the implant in a live rat eye in vivo. 

To determine the distribution of NaFl released from the implant in a postmortem rat eye, an NaFl implant was placed in the postequatorial subconjunctival space immediately after CO₂ euthanatization. At indicated times after implantation, the eye was enucleated and frozen immediately in OCT compound. The frozen eyes were sectioned into 15-μm thick slices. 

Quantitative Analysis of NaFl in the Retina Tissues

The implanted eyes (n = 6 at each time point) were enucleated 1, 2, and 3 hours after episcleral surgical implantation, respectively. The retinal tissues were isolated and incubated in solution (Solvable; PerkinElmer Life Sciences, Boston, MA) in 50°C water bath for 2 hours to solubilize the tissues. The solutions were mixed well by vortex and centrifuged at 13,000 rpm for 5 minutes. Supernatant solutions were placed in a 96-well plate and made ready with a microplate spectrofluorometer (SpectraMax Gemini Dual-Scanning; Molecular Devices, Sunnyvale, CA) at an excitation wavelength of 485 nm and an emission wavelength of 538 nm. All procedures were carried out under reduced ambient light to minimize photobleaching. 

Elimination of Subconjunctivally Administered Hydrophilic Agent through the Conjunctival Lymphatic Vessel

Five microliters of a mixture containing 5% patent blue V dye and fluorescein isothiocyanate-dextran (FD-40; MWt, 40 kDa) at 10 mg/mL were injected into the subconjunctival space of live or postmortem rat eyes in the superotemporal quadrant under general anesthesia. Patent blue V dye is an agent used in humans during lymphangiography to visually identify lymph vessels and nodes. ^17–19 The blue color changes were observed for 5 hours after injection and the pictures were taken. 

In addition, a rat was euthanatized 1 hour after injection, as described, after which an incision was made under the mandible and blunt dissection taken down through the dermis to visualize the blue dye–stained lymphatic vessel and lymph node. The blue-stained lymph node was visualized, isolated, removed, and frozen immediately in OCT compound. The frozen lymph node was sectioned into 12-μm thick slices. The distribution of FD-40 in the sectioned lymph node was examined under an epifluorescence microscope. 

Quantitative Evaluation of Lymphatic Elimination

To quantitatively assess the conjunctival lymphatic elimination, we blocked lymphatic flow distal to the cervical lymph node draining the periocular tissues and administered a fluorescent drug surrogate subconjunctivally. Briefly, rats were injected with 20 μL patent blue V (50 mg/mL) subconjunctivally to visualize the lymph vessels and sentinel nodes draining the eye. The efferent lymphatic vessels from this node were ligated with a 7-0 Vicryl suture before injection of 2 μL of 37.8 mg/mL (75.6 μg) NaFl into the subconjunctival space (see Fig. 7A). The node proximal to the suture was isolated at 30, 45, and 60 minutes after NaFl injection, weighed/diluted to 250 μL with solution (Solvable; PerkinElmer Life Sciences), ²⁰ placed in a 50°C water bath for 2 hours, vortexed, and centrifuged at 13,000 rpm for 5 minutes, and the supernatant was isolated and stored at −80°C. One hundred microliters of sample was added to the well of a 96-well luminescence plate and read at 485 nm excitation/525 nm emission. The standard curve was generated by adding 37.8 mg/mL NaFl to a blue dye positive node and processing it identically to the other samples, as detailed. 

Immunohistochemistry of Lymph Vessels or Blood Vessels in the Conjunctiva

Rabbit anti–mouse polyclonal antibody (LYVE-1; Cell Sciences, Inc., Canton, MA) and mouse anti–human CD31 (PECAM-1) monoclonal antibody (Cell Signaling Technology, Inc., Danvers, MA) were used in immunohistochemistry staining for lymphatic vessels and blood vessels, respectively. For the distribution of lymphatic and blood vessels in the conjunctival tissue, the superficial and deep conjunctival tissues were collected, immediately fixed with 4% paraformaldehyde for 3 hours, and washed in 1× phosphate-buffered saline (PBS) three times. The fixed conjunctival tissues were incubated overnight in 5% goat serum diluted with ICC buffer. Double staining was performed using rabbit anti–mouse LYVE-1 polyclonal antibody (dilution 1:150) or mouse anti–human PECAM-1 polyclonal antibody (dilution 1:150) in 2% goat serum diluted with ICC buffer as a primary antibody solution. The primary antibodies LYVE-1 and anti–PECAM-1 were detected using Alexa Fluor 594 goat anti–rabbit IgG (dilution 1:150) and Alexa Fluor 488 goat anti–mouse IgG (dilution 1:150), respectively, with DAPI (dilution 1:1000). 

Colocalization of Subconjunctivally Administered IgG with Blood and Lymphatic Vessels in the Conjunctiva

Initial efforts to determine the colocalization of NaFl with conjunctival blood and lymphatic vessels after subconjunctival implantation were unsuccessful because NaFl was found to relocalize during the fixation process. Instead, we used Alexa 633–conjugated goat anti–rabbit IgG (H+L) (Invitrogen, Carlsbad, CA) (GAR633) because, as a protein based agent, it would minimally relocalize during the fixation process. Therefore, 5 μL of GAR633 was injected subconjunctivally into the superotemporal quadrant of the right eye with a 32-gauge needle under general anesthesia. The injected eyes were then enucleated 1-hour after injection and frozen immediately in OCT compound. Frozen eyes were sectioned into 15 μm-thick slices. 

Rabbit anti–mouse LYVE-1 polyclonal antibody (Cell Sciences, Inc., Canton, MA) and rabbit anti–human von Willebrand factor (vWF) antibody (DAKO, Carpinteria, CA) were used for immunohistochemical staining of lymphatic vessels and blood vessels, respectively. PECAM-1 antibody worked well for blood vessel immunohistochemistry in a fixed conjunctival tissue without freezing; however, it did not work well in frozen and fixed tissue. Therefore, vWF antibody was used in this colocalization experiment. Frozen 15-μm thick slices showing the distribution of subconjunctivally administered Alexa Fluor 633–conjugated goat anti–rabbit IgG were dried at room temperature for 1 hour, fixed with 4% paraformaldehyde for 30 minutes, and washed in 1× PBS. Sections were blocked with 5% goat serum diluted with 1X PBS for 20 minutes and then incubated for 3 hours in either rabbit anti–mouse LYVE-1 polyclonal antibody (dilution 1:150) or rabbit anti–human von Willebrand factor polyclonal antibody (dilution 1:150) in 1× PBS as a primary antibody solution. The primary antibodies, either anti–LYVE-1 or anti–vWF, were detected using Alexa Fluor 488 goat anti–rabbit IgG (dilution 1:150) with DAPI (dilution 1:1000). The colocalization of subconjunctivally administered antibody with either conjunctival lymphatic or blood vessel was examined under an epifluorescence microscope (Olympus Optical Co., Ltd, Melville, NY) equipped with a charge-coupled device color video camera (Retiga EX; Retiga, Burnaby, BC, Canada). 

Integrity of Anatomic Barriers

To determine the integrity of anatomic barriers throughout the experimental procedure, we examined the structural integrity of retinal pigment epithelial tight junctions by determining the expression of ZO-1, a useful marker of tight junction structure between retinal pigment epithelial. ^21,22 First, we collected eyes 1 hour, 2 hours, and 3 hours after CO₂ euthanatization. The enucleated eyes were immediately sectioned at the equator, and the anterior segment, vitreous, and retina were removed from the eyes. The posterior segment of the eye, including the overlying sclera and choroid, was fixed for 3 hours with 4% paraformaldehyde. After three washings with 1× PBS, the posterior segment of the eye was dissected into quarters by four radial cuts and incubated for 3 hours in 5% goat serum blocking solution diluted with ICC buffer. The posterior segment of the eye was then incubated overnight in rabbit anti–ZO-1 tight junction protein polyclonal antibody (dilution 1:150; Abcam, Cambridge, MA) in the ICC buffer as a primary antibody solution. The primary antibody was detected using Alexa Fluor 488 goat anti–rabbit IgG (dilution 1:150) with DAPI (dilution 1:1000). 

Results

Distribution of NaFl Released from an Episcleral Implant in Live and Postmortem Rat Eyes

Figure 1 shows the distribution of NaFl at various times after implantation after release from an HPMC implant placed in the subconjunctival space of a postmortem eye at 1 hour, 2 hours, and 3 hours and in the eyes of live rat 1 hour, 2 hours, and 3 hours. Qualitatively, we established that NaFl released from the implant easily diffused into the retina of the postmortem eye (Figs. 1A–C) compared with the live eye (Figs. 1D–F), indicating the presence and suggesting the importance of active elimination mechanisms within the conjunctival space of the living eye not present in the postmortem eye. The in vitro release of the episcleral implants (n = 5) demonstrated an initial burst with a declining diffusion phase (Fig. 1G). More than 95% of the release occurred within the 50 minutes of the assay. Quantitatively, the amount of NaFl per unit mass of tissue was measured to 5.8 ± 2.30 ng, 12.0 ± 3.67 ng, and 41.0 ± 26.70 ng in the postmortem eye and 1.0 ± 0.40 ng, 0.7 ± 0.75 ng, and 0.4 ± 0.29 ng in the live eye 1 hour, 2 hours, and 3 hours after episcleral implantation, respectively (Fig. 1H). 

Figure 1.

 

Distribution of sodium fluorescein released from a hydroxypropyl methyl cellulose–based implant placed in the subconjunctival space at various times after implantation in a postmortem eye at (A) 1 hour, (B) 2 hours, and (C) 3 hours and in a live eye at (D) 1 hour, (E) 2 hours, and (F) 3 hours. (G) Cumulative release (%) over time of NaFl from episcleral implants. (H) Amount of sodium fluorescent at unit mass of retina tissue at various time points after implantation in live and postmortem eyes. (C, F, red star) Location of the implant. Scale bars: (A, B, D, E) 100 μm; (C, F) 500 μm. SC, sclera; CH, choroid; RE, retina; AN, anterior segment of the eye; LE, lens; PO, posterior segment of the eye.

Figure 2 demonstrates the amount of patent blue V dye remaining after subconjunctival injection in live or postmortem eyes at various postinjection time points. Given the greater amount of patent blue V remaining within the conjunctiva of the postmortem eye with respect to the live eye at 3 and 5 hours, we concluded that active clearance mechanisms were responsible for increased removal of patent blue V the live eyes and thus play a particularly significant role in the removal of hydrophilic compounds from the conjunctiva. 

Figure 2.

 

Elimination of subconjunctivally injected patent blue V dye from the conjunctival space at various times after injection in a live eye at (A) 1 hour, (B) 3 hours, and (C) 5 hours and in a postmortem eye at (D) 1 hour, (E) 3 hours, and (F) 5 hours.

Integrity of Anatomic Barriers

Given the substantial difference between the penetration of NaFl released from subconjunctivally placed implants in a live versus a postmortem rat eye (Fig. 1), we hypothesized that physiological mechanisms are a greater impediment than structural barriers in delivering subconjunctival therapeutic antibody to the retina. To test this hypothesis, we first wanted to confirm the integrity of the retinal pigment epithelium (RPE), a known structural anatomic barrier, during the postmortem period used in this study. Although the leaky and fuzzy pattern were observed at the small part of the RPE monolayer at Figure 3C, Figure 3 demonstrates that overall the RPE cells showed a clear, lateral membrane staining pattern for ZO-1 protein and outlined the uniform polygonal shape of the RPE cells within the monolayer. This result suggests that the RPE tight junctions remained intact at 3 hours postmortem. 

Figure 3.

 

The integrity of RPE tight junctions at (A) 1 hour, (B) 2 hours, and (C) 3 hours postmortem. A rabbit anti–ZO-1 tight junction protein polyclonal antibody was used for immunohistochemical staining of preserved RPE tight junctions (green). Cell nuclei were stained with DAPI (blue). These data demonstrate the maintenance of RPE tight junction integrity for at least 3 hours after euthanatization.

Lymphatic Vessels as a Rapid Elimination Pathway of Subconjunctivally Administered Drugs

Figure 4A illustrates the primary efferent lymphatic channel and cervical lymph node draining the conjunctival tissue after subconjunctival injection of patent blue V dye into the superotemporal quadrant. This figure illustrates that a remarkable amount of subconjunctivally administered hydrophilic agents is eliminated through lymphatic outflow from the conjunctival space. Figure 4B further supports the notion that active elimination mechanisms, such as lymphatic clearance, represent an important barrier to transscleral drug delivery by demonstrating the existence of FD-40 in the lymph node 1 hour after subconjunctival administration. 

Figure 4.

 

(A) Identification of the conjunctival draining lymphatic channel and primary node after subconjunctival injection of patent blue V dye 1 hour after injection, indicating that subconjunctivally administered hydrophilic agents are at least partially eliminated from the conjunctival space via lymphatic outflow. (B) The presence of FD-40 in the lymph node 1 hour after subconjunctival administration.

Distribution of Blood and Lymphatic Vessels within Conjunctival Tissue and Tenon's Fascia

We hypothesized from Figures 1, 2, and 4 that the active elimination pathways of hydrophilic agents in the conjunctiva were conjunctival lymphatic or blood vessels. To examine this hypothesis, it was essential to first establish the existence and distribution of blood and lymphatic vessels in the conjunctiva with immunohistochemistry. Figure 5 shows the dense distribution of blood vessels (green) and lymphatic vessels (red) in the conjunctival tissue and Tenon's fascia using platelet/endothelial cell adhesion molecule (PECAM-1) and lymphatic vessel endothelial receptor (LYVE)-1, respectively. Blood vessels are denser within Tenon's fascia than they are in conjunctival tissue; on the other hand, lymphatic vessels are denser in the conjunctival tissue than within Tenon's fascia. 

Figure 5.

 

Immunohistochemical staining for blood and lymphatic vessels in rat conjunctival tissue and Tenon's fascia. The first column images are double-stained for blood vessels (green) in superficial conjunctival tissue (A) and Tenon's fascia (D). The second column images are double-stained for lymphatic vessels (red) in superficial conjunctival tissue (B) and Tenon's fascia (E). (C, F) Merged images. Rabbit anti–mouse LYVE-1 polyclonal antibody and mouse anti–human CD31(PECAM-1) monoclonal antibody were used for lymphatic vessel and blood vessel immunohistochemical staining, respectively. Cell nuclei were stained with DAPI (blue).

Colocalization of Subconjunctivally Administered Full-Length IgG with Blood and Lymphatic Vessels

In testing the hypothesis that conjunctival blood and lymphatic vessels are the most significant barrier to delivering subconjunctivally administered hydrophilic agents to the retina, we first assessed the colocalization of subconjunctivally administered full-length IgG with these blood and lymphatic vessels. Figures 6A and 6D show the distribution of subconjunctivally administered full-length IgG in the conjunctival tissue. Figures 6B and 6E show the distribution of conjunctival blood and lymphatic vessels, respectively. Figures 6C and 6F show the colocalization of administered full-length IgG with conjunctival blood and lymphatic vessels. In Figures 6C and 6F, we demonstrated that subconjunctivally administered full-length IgG does in fact colocalize with the conjunctival blood and lymphatic vessels. 

Figure 6.

 

Colocalization of a subconjunctivally administered fluorescence-labeled IgG and conjunctival blood vessels (A–C) or lymphatic vessels (D–F). (A, D) Distribution of subconjunctivally administered fluorescence-labeled full-length IgG in the conjunctival tissues 1 hour after injection. (B, E) Immunohistochemically stained blood vessels (B) and lymphatic vessels (E), respectively. (C, F) Detailed colocalization of subconjunctivally administered full-length IgG and conjunctival blood vessels (C) and lymphatic vessels (F), respectively. Rabbit anti–mouse LYVE-1 polyclonal antibody and rabbit anti–human von Willebrand factor (vWF) antibody were used for immunohistochemical staining of blood vessels (B) and lymphatic vessels (E), respectively (green). Cell nuclei were stained with DAPI (blue).

Quantitative Evaluation of Lymphatic Elimination

The total accumulated amounts (average ± SEM) of NaFl in the lymph node at 30, 45, and 60 minutes after a 75.6 μg of subconjunctival NaFl injection were 1.0 ± 0.27 μg, 2.6 ± 0.38 μg, and 4.1 ± 1.67 μg, representing 1.3%, 3.4%, and 5.4% of the total 75.6 μg injection, respectively (Fig. 7B). We found the amount of NaFl in the lymph node increased linearly at a rate of 0.11 μg/min between 30 and 60 minutes. 

Figure 7.

 

Quantitative evaluation of sodium fluorescein elimination by lymphatic vasculature after subconjunctival administration. (A, left) Preligation image illustrates both the sentinel node draining the eye and the efferent submandibular lymph vessel draining from this lymph node (inset: magnified efferent submandibular lymph vessel). Right: ligation image of the efferent submandibular lymph vessel with a 7-0 Vicryl suture. (B) Total amount of NaFl in the lymph node at 30, 45, and 60 minutes after subconjunctival injection of 75.6 μg NaFl was 1.0 ± 0.27 μg, 2.6 ± 0.38 μg, and 4.1 ± 1.67 μg, representing 1.3%, 3.4%, and 5.4% of the total 75.6-μg injection, respectively.

Discussion

The present study demonstrates that successful transscleral delivery of a hydrophilic agent to the subretinal space did not occur in the live rat eye. However, successful delivery to the subretinal space was achieved in the postmortem eyes. Understanding the barriers involved in transscleral drug transport after a subconjunctival injection is helpful in understanding our results. Kim et al. ²³ established three types of barriers hindering transscleral drug delivery: static, dynamic, and metabolic. However, the bulk of research investigating these barriers to transscleral drug delivery has focused on the role of the sclera as a static, structural barrier to drug penetration. Typically, the sclera is isolated and placed in an Ussing diffusion chamber, and the movement of drugs through the sclera is measured. Multiple investigators have demonstrated that passage through the sclera occurs primarily by diffusion. ^24–27 The interfibrillar aqueous media of scleral proteoglycans results in reduced diffusion through the sclera only for large agents. ¹² In vitro studies suggested that neither lipophilicity nor molecular charge appeared to play a role in this diffusion; however, in vivo there was a suggestion of increased sclera permeability to hydrophilic agents relative to lipophilic agents. ^28,29 Accordingly, many concluded that transscleral drug delivery of hydrophilic macromolecules to the retina was feasible. 

Our previous and current studies in live animals consistently show very limited penetration of hydrophilic drugs (sodium fluorescein) to the retina when drugs are administered in the subconjunctival space of live animals (Fig. 1A). ¹⁴ Interestingly, shortly after an animal is euthanatized, a considerable amount of sodium fluorescein is able to penetrate into the eye (Fig. 1B). One could argue that this hydrophilic drug was able to access the retina after death because static structural barriers, such as the RPE tight junctions, were disrupted. Accordingly, we investigated whether RPE tight junctions were preserved during the time interval used to establish greater drug penetration in postmortem eyes versus live eyes by determining the integrity of ZO-1 protein expression immunohistochemically. Indeed, using this technique of confirming tight junctional protein expression, the results are suggestive that the RPE tight junctions appeared to be functioning for at least 3 hours postmortem (Fig. 3). This confirmed a considerable amount of subconjunctivally administered sodium fluorescein was able to reach the retina despite the maintenance of static barriers. This robust penetration of drug into eyes with structural barriers intact illustrates that static structural barriers play a seemingly minor role in limiting drug penetration from the conjunctival space into the retina. Thus, we hypothesized that dynamic mechanisms of elimination, which are markedly reduced after the death of the animal, might be responsible for restricting transscleral drug delivery in live animals. These dynamic barriers to transscleral drug delivery include conjunctival blood/lymphatic vessels, episcleral veins, uveoscleral outflow, and the choroidal circulation. 

Many investigators feel that the rapid blood flow within the choroid and the subsequent elimination of drug within this space is the principal hurdle in delivering drugs to the retina from outside the eye. Indeed, the blood flow in the choroid is the highest in the body. ³⁰ The in vivo permeability of the choriocapillaris to most molecules is high but is progressively limited with increasing size of the molecule; however, these data were established based on drugs diffusing from the circulation into the extravascular choroidal tissue rather than from the extravascular tissue into the circulation. ³¹ The kinetics of drug movement from the surrounding choroidal tissue into the adjacent choroidal vascular space is unknown. To determine which dynamic mechanisms most extensively limit transscleral drug delivery, experiments were performed in rabbits using cryotherapy to destroy the retinal pigment epithelium and choroidal capillaries. This effectively reduced the dynamic barrier of choroidal circulation while also disrupting retinal pigment epithelial tight junctions. ¹⁶ In this experiment within live animals, there was very little change in the ocular penetration of sub-Tenon's triamcinolone acetonide after cryotherapy. In parallel experiments, an episcleral “window” was produced in which episcleral and conjunctival lymphatics and blood vessels were transected. After this procedure and placement of drug on the episclera, a significant increase in drug penetration to the retina was noted in the live animal. This suggests a high conjunctival clearance through lymphatics and blood outflow may be the critical dynamic barrier that restricts transscleral drug delivery to the retina. However, we did not previously have direct evidence clearly demonstrating clearance of drug through conjunctival blood and lymphatic vessels. In the present study, Figure 6 validates the clearance of subconjunctival IgG through blood and lymphatic vessels with immunostaining. Moreover, Figure 4 provides direct evidence demonstrating the lymphatic clearance of a hydrophilic compound, FD-40, from the conjunctiva. 

The conjunctiva is well vascularized in most mammals. Figure 5 demonstrates a rich supply of both lymphatic and vascular structures in the conjunctiva and Tenon's fascia. Lymphatic vessels in the conjunctiva occur as a superficial plexus immediately beneath the epithelium and as a deeper plexus of larger luminal structures within Tenon's fascia. ^32,33 Indeed, the lymphatic vasculature in the conjunctiva is more extensive than in any other organ system. ³⁴ Singh ³⁵ visualized rapid conjunctival lymphatic elimination of sub-Tenon's trypan blue in humans just seconds after episcleral injection. Figure 2 confirmed this rapid conjunctival elimination of hydrophilic agents by illustrating increased clearance of subconjunctival patent blue V in the live rat eye with respect to the postmortem rat eye at various times after injection. Figure 4A identifies the conjunctival draining lymphatic channel and primary node after the subconjunctival injection of patent blue V dye. In addition to qualitative evidence (Fig. 4), for the first time, to our knowledge, the clearance rate of a subconjunctivally administered hydrophilic agent (NaFl; MWt, 376) has been quantitatively measured through conjunctival lymphatics by ligating the efferent lymphatic vessels from the lymph node (Fig. 7). The ligation of a distal lymph vessel might affect the flow of lymph to the proximal node. However, the amount of NaFl in the lymph node was found to increase linearly from 30 minutes to 1 hour after administration, indicating the limited effect of ligation. The extracted total amount of NaFl in the lymph node at 60 minutes after injection of 75.6 μg NaFl was 4.1 ± 1.67 μg, representing 5.4% of the total 75.6 μg injection (Fig. 7B). A circular lymphatic trunk, termed the pericorneal lymphatic ring, is a centralized collection channel that drains medially and temporally to two regional lymph nodes. Therefore, we can conclude that at least 10% of the small, hydrophilic agent NaFl administered subconjunctivally is eliminated through the lymphatic draining system within the first hour. Subcutaneously administered drugs are transported into the systemic circulation by either blood capillaries or lymphatic vessels. Subcutaneous small molecules (up to 1 kDa) are eliminated predominantly into the systemic circulation by blood capillaries. However, the elimination of subcutaneously administered macromolecules through blood capillaries is limited. Instead, soluble macromolecules were found to enter the systemic circulation indirectly, by way of lymphatic vessels. Supersaxo et al. ³⁶ measured the cumulative recovery (percentage of dose, mean ± SD) of four water-soluble compounds in lymph draining from the site of subcutaneous administration to be 4.0 ± 1.5 (5-fluoro-2′-deoxyuridine; MWt, 246.2), 21.0 ± 7.1 (inulin; MWt, 5200), 38.6 ± 6.7 (cytochrome c, MWt, 12,300), and 59.5 ± 7.7 (human recombinant interferon α-2a; MWt, 19,000), respectively. The cumulative recoveries indicate a linear relationship between the compound's molecular weight and the proportion of administered dose absorbed lymphatically. They concluded that molecules with molecular weights exceeding 16,000 were primarily absorbed and, thus, eliminated from subcutaneous tissue by lymphatic vessels. Based on previous literature and our present study demonstrating subconjunctival NaFl lymphatic clearance of approximately 10%, we conclude that small, hydrophilic drugs administered subconjunctivally are eliminated predominantly through the conjunctival blood vessels; on the other hand, subconjunctivally administered hydrophilic macromolecules are eliminated primarily through conjunctival lymphatic vessels. We aim to further characterize this model in future studies. 

In summary, a hydrophilic agent, NaFl, released by a subconjunctival implant was not successful in delivering the agent to the subretinal space of the rat. However, in postmortem eyes, the agent was delivered successfully to the subretinal space. The dynamic barriers of the conjunctiva, namely the dense network of lymphatics and blood vessels, are likely responsible for our observations. Understanding the barriers in transscleral drug diffusion after subconjunctival injections will be helpful in designing effective transscleral delivery systems. 

Footnotes

 Supported in part by Sogang University Internal Research Grant 200910005.01.

Footnotes

 Disclosure: S.J. Lee, None; W. He, None; S.B. Robinson, None; M.R. Robinson, None; K.G. Csaky, None; H. Kim, None