Posterior Segment Ocular Pharmacokinetics Using Microdialysis in a Conscious Rabbit Model

Abstract

purpose. To develop and establish a conscious rabbit model for ocular pharmacokinetic studies and to delineate the effects of anesthesia and probe implantation on the ocular disposition of ganciclovir.

methods. A conscious rabbit model was developed for microdialysis of the posterior ocular segment. Rabbits were divided into three groups. Group I consisted of rabbits with no recovery period after probe implantation and were anesthetized throughout the experiment. Group II consisted of rabbits that had a more than 5-day recovery period and were conscious during the experiment. Group III consisted of rabbits that had a more than 5-day recovery period and were anesthetized during the experiment. ³[H] ganciclovir was administered (50 μL) intravitreously in all groups, and ocular levels were determined for 10 hours at appropriate time intervals. Data obtained were subjected to noncompartmental modeling.

results. Probe calibration studies indicated that the probes were functional for at least 14 days. The anesthetized groups, regardless of the period of recovery from probe implantation, exhibited higher areas under the curve than did the conscious group. The vitreous half-life of ganciclovir was significantly shorter in the groups with a recovery period of more than 5 days compared with the group with no recovery period.

conclusions. The conscious rabbit model was developed and can be used for a period of at least 14 days. Anesthesia increased ocular bioavailability of intravitreously administered ganciclovir, whereas probe implantation led to increased protein efflux into the vitreous, which may be the reason for the increased half -life of ganciclovir in group I.

Microdialysis is a valuable sampling technique that has been applied to ocular drug research. Ocular microdialysis is a relatively recent event first reported in the late 1980s. ¹ The application of this technique in ocular drug kinetics has had a major impact on the study of drug disposition in the eye. It has been used in both the aqueous and vitreous chambers in several laboratories. ^{2 3 4} In our laboratory, we have successfully used this technique to study the disposition of drugs in the aqueous and vitreous chambers simultaneously. ⁵ However, thus far, the animal model used has involved anesthetized rabbits. Waga et al. ^{2 6 7} used a conscious rabbit model in studies involving the posterior segment of the eye. Rittenhouse et al. ^{8 9} used an awake rabbit model for anterior segment studies. To improve the existing model in our laboratory, it was necessary to develop a rabbit model in which the animal could remain conscious during the entire study period. Although Waga et al. ⁶ developed their model in 1991, very little has been done to replicate the model for use in ocular studies, largely because of the detailed surgical techniques involved. Our initial goal was to establish a conscious animal model for posterior segment disposition studies. Although the development of our model was based on procedures reported by Waga et al., ⁶ our goal was to establish a simpler and less tedious surgical procedure for developing the model. 

To induce anesthesia in rabbits, a combination of ketamine and xylazine is commonly used. This combination produces a suppressive effect on the heart rate and respiratory rate. ^{10 11} In addition, both ketamine and xylazine exhibit effects on the intraocular pressure (IOP). ^{12 13 14} These side effects may have an impact on the disposition of drugs in the posterior segment. Ganciclovir (GCV) is an acycloguanosine analogue and is currently the drug of choice for the treatment of cytomegalovirus (CMV) retinitis. ^{15 16} GCV is administered by various routes, including intravitreous injection. To delineate the effect of anesthesia on the disposition of ocular drug using microdialysis, we chose GCV as the drug to be studied. In addition to studying the effects of anesthesia, it was necessary to investigate the effect of probe implantation on drug disposition. Because the anesthetized model may not have been allowed a sufficient period of recovery, it was important to investigate whether any alterations in the protein content of the vitreous occurred after implantation of the probe. 

Materials and Methods

Materials

³[H] GCV and ¹⁴C mannitol were purchased from Moravek Biochemicals, Inc. (Brea, CA). Linear microdialysis probes (polyacrylonitrile membrane with 10 mm length) used for the sampling of vitreous humor were purchased from Bioanalytical Systems (West Lafayette, IN). The microinjection pump (CMA/100) necessary for perfusing isotonic phosphate-buffered saline (IPBS) through the probes, was obtained from CMA/Microdialysis (Solna, Sweden). Biodegradable sutures were purchased from Ethicon Inc. (Somerville, NJ); ketamine HCl and xylazine from Fort Dodge Laboratories (Fort Dodge, IA) and Bayer Animal Health (Shawnee Mission, KS), respectively; pentobarbital sodium from Abbott Laboratories (Abbott Park, IL); and tropicamide and 0.1% dexamethasone phosphate from Bausch & Lomb (Tampa, FL). New Zealand male albino rabbits weighing between 2 and 2.5 kg were purchased from Myrtle’s Rabbitry (Thompson Station, TN). All experimental protocols followed the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Probe Calibration

In vivo retrodialysis with an internal standard method was used for calibration of the probe. In this case ³[H] GCV was the compound of interest. Ideally, the internal standard should be similar to GCV in its diffusion characteristics. In our studies ¹⁴C mannitol was used as the internal standard, first, because it was a ¹⁴C-labeled compound and could be monitored simultaneously along with ³[H] GCV. Second, it would not interfere with the ocular disposition of GCV in case there were any facilitative processes involved. In addition, mannitol behaves very similarly to GCV in terms of probe recovery. Briefly, the calibration procedure involved perfusion of both the compound of interest and the internal standard through the probe at 2 μL/min. The probe was placed in IPBS and the dialysate was collected at 20-minute intervals (Fig. 1) . The samples were analyzed for radioactivity, and the loss of both compounds was measured. Recovery was calculated as  

\[\mathrm{Recovery}\ {=}\ (C_{\mathrm{in}}{-}C_{\mathrm{out}})/C_{\mathrm{in}}\]

where C _in is the concentration of analytes entering the probe and C _out is the concentration of analytes leaving the probe. The recoveries of the two compounds were compared using the recovery ratio calculated by  

\[\mathrm{Recovery}_{\mathrm{Mannitol}}/\mathrm{Recovery}_{\mathrm{GCV}}\]

Once the probe was calibrated for both compounds, it was then used in the conscious-animal surgery. After surgery, the probe was perfused with IPBS containing a known concentration of the internal standard (mannitol). 

The in vivo recovery from implantation of the probe for mannitol was determined and, using the predetermined recovery ratio, the recovery factor for GCV was determined. This procedure was performed before each pharmacokinetic study. To determine the in vivo viability of the probe, mannitol recoveries were determined daily for at least 14 days. If there was any bacterial or protein build-up or disintegration of the probe, recovery would change dramatically, indicating that the probe was no longer viable. 

Surgical Procedure

Animals were anesthetized by intramuscular administration of ketamine (50 mg/kg) and xylazine (5 mg/kg). One to two drops of tropicamide were applied to dilate the pupil. The eye was proptosed, a 25-gauge needle was inserted in the eye approximately 8 mm below the sclera–limbus junction and exited at the opposite side. A linear probe (MD-2000; Bioanalytical Systems) was inserted into the needle, and the probe was pulled through the eye by withdrawing the needle. The probe was positioned in such a way that the entire membrane area lay in the vitreous chamber and was slightly angled so as to avoid contact with the lens. After the probe’s position had been visually inspected under a microscope, the probe was fixed to the conjunctiva with a 6-0 surgical chromic gut suture. Subsequent to implantation of the probe in the eye, the area between the ears was shaved and a 4-mm incision was made with a sterile disposable scalpel. A subcutaneous tunnel was then created from the area of the incision to the upper eyelid with forceps. Once the tunnel was created, a small incision was made in the upper eyelid close to the eyeball. The two ends of the forceps were introduced through the subcutaneous tunnel and externalized through the incision area in the upper eyelid. The two ends of the probe were then pulled through the tunnel with the forceps. The outlet and inlet ends exited completely through the area between the ears. The incision was then closed, and the probe was secured to the skin with a 3-0 suture. 

Throughout the procedure, saline was flushed through the probe by a microdialysis pump, to ensure that the probe had not cracked or leaked. Once the skin sutures were in place, the outlet and inlet ends were cut to the appropriate size to prevent the rabbit from accessing it. After the surgical procedure, the animals were observed at least once daily for signs of irritation and inflammation. One to 2 drops of 0.1% dexamethasone phosphate was applied topically to counter any inflammation or redness of the sclera or conjunctiva. The probes were also flushed daily (20 μL/min) with IPBS containing a cocktail of penicillin and streptomycin to prevent any bacterial growth in the eye and around the probe membrane. Only animals that showed no signs of infection or blockage of the probe were used in the studies. 

Ocular Pharmacokinetic Experiments

Because there were two variables (anesthesia and probe implantation) to be studied, rabbits were divided into three groups (Table 1) . In all, 24 rabbits were included in the study. Of these, three rabbits showed development of infection in the eye, whereas four rabbits managed to dislodge the probe during the recovery period of 5 days. These rabbits were excluded from the study, and the remaining were grouped according to treatment. 

Group I (n = 6) consisted of animals that were not allowed a recovery period after implantation of the probe and were kept under anesthesia throughout the experiment. Group II (n = 7) consisted of animals that were given at least a 5-day recovery period after implantation of the probe and were kept conscious throughout the pharmacokinetic study. Group III (n = 4) consisted of animals that were given at least a 5-day recovery period, but were kept anesthetized during the experiment. Rabbits were anesthetized with intramuscular injection of ketamine and xylazine. An appropriate dose of ³[H] GCV (50 μL, 2 μCi) was injected into the midvitreous region with a tuberculin syringe and a 30-gauge needle. Subsequent to administration of the drug, microdialysis samples were collected at 2 μL/min every 20 minutes. The concentrations obtained were plotted against the midpoint of each 20-minute interval. Group II animals were allowed to recover from anesthesia (which usually took 45 minutes) and were placed in steel restrainers during the sampling period. Samples were collected at appropriate time intervals with periodic breaks in sampling to provide access to food and water. To replicate conditions, similar sampling time intervals were followed in other experimental groups. All samples were assayed with a multipurpose scintillation counter (LS 6500; Beckman, Carlsbad, CA). 

Protein Estimation Studies

The purpose of this study was to investigate the effect of implantation of a probe on protein release into the vitreous chamber. Rabbits were anesthetized as described previously. The linear probe was implanted in the vitreous chamber, and animals were killed with an overdose of pentobarbital sodium injected into the marginal ear vein at different time points (n = 3 for each time point). Both eyes were enucleated, and approximately 0.5 mL of vitreous was aspirated from each eye. The vitreous was then centrifuged at 5000 rpm for 15 minutes to clear it of any cellular debris. The supernatant was collected and assayed for protein content with the Bradford assay. The protein content of the experimental eye (eye in which probe was implanted) was compared with the contralateral eye, which served as the control. 

Results

Probe Function

To check the viability of the probe in the conscious rabbit, in vivo calibration of the probe was performed by retrodialysis. These studies were conducted for a period of 14 days and calibration was performed daily with ¹⁴C mannitol as the marker. Generally, the probe remained viable for the period under study (Fig. 2) . Although probe recovery decreased after the surgical procedure, the values by the fifth day after surgery were almost identical (90%–95% of the initial probe recovery value) with those obtained before surgery. A decrease in probe recovery value immediately after the surgery may have been due to release of certain inflammatory proteins, which hindered the diffusion of the marker. A topical anti-inflammatory agent such as dexamethasone was used to counter any inflammation of the sclera or conjunctiva such as redness of the tissue. The animals responded very well to the dexamethasone treatment and showed no signs of inflammation after 2 days. These data indicate that the probes remained functional for at least 14 days. 

Ocular Pharmacokinetics of GCV

The intravitreous pharmacokinetics of GCV are illustrated in Figure 3 . Figure 3 , top, represents the vitreous pharmacokinetic profile of GCV in the animals in group I after an intravitreous injection. When a compound is injected in the midvitreous region, it may exhibit a diffusion-based equilibration–distribution phase followed by steady state elimination once equilibration is achieved. The profiles in Figure 3 , top indicate an initial increase in concentrations followed by a steady state monoexponential decline. The increase in concentration was merely because the injection position (midvitreous) was away from the microdialysis probe (8 mm below the sclera–limbus junction), which was positioned close to the back of the eye. The compound exhibited a distribution phase and had to travel a path from the center of the vitreous chamber to the point where the probe was positioned. Therefore, the initial phase of the profiles demonstrated an increase in vitreous concentrations. Once the compound had equilibrated in the vitreous, it underwent steady state elimination. Figure 3 , middle, represents profiles obtained from animals that had a 5-day recovery period and were conscious during sampling. All the profiles exhibit a biexponential behavior, with the initial distribution lasting for approximately 2 hours However, in this group of animals there was no concentration build-up in the distribution phase, because the probe’s position was close to the point of injection near the midvitreous. Such positioning was possible in the conscious animals because the probes were fixed to the eye with conjunctival biodegradable sutures. Hence, after 5 days it is likely that the probe’s position had shifted toward the center of the vitreous chamber because of the activity of the animal and also because of degradation of the sutures to some extent. 

In group III animals (Fig. 3 , bottom), similar results were recorded in some animals, whereas other animals exhibited a small increases in concentration followed by steady state elimination. All the elimination phases were monoexponential, and to make comparisons, the distributional phases (120 minutes) were not considered. Only the terminal elimination phases were subjected to a noncompartmental modeling approach (WinNonlin; Pharsight, Mountain View, CA). The terminal elimination phases (Fig. 4) provided pharmacokinetic parameters, such as half-life (t _1/2) and area under the curve (AUC) for comparison (Table 2) . In general it was noted that anesthesia increased steady state levels compared with the conscious animal group II. AUC values in group II were significantly lower than those obtained in the other groups (a sixfold difference), in which the animals were kept under anesthesia during the experiment. The intravitreous t _1/2 in group II was determined to be approximately 3.5 hours with similar values obtained in animals that were allowed to recover from surgery but were kept under anesthesia. Therefore, it appears that anesthesia did not affect the rate of elimination at steady state. However, the vitreous t _1/2 for GCV in group I animals was elevated (6 hours). This increase in t _1/2 of GCV suggests that the probe implantation led to changes in vitreous dynamics that decreased the rate of elimination of GCV. The possible effects after surgery may have been trauma to the tissue leading to inflammatory reactions and release of various inflammatory proteins. There may also have been a temporary breakdown in the blood–retinal barrier leading to increased protein levels in the vitreous humor. The increased protein levels could have altered drug disposition, leading to slower elimination rates due to protein binding. 

Effect of Probe Implantation on Protein Content

To investigate the effect of probe implantation on vitreous humor physiology, protein content was determined. Protein content of the experimental eye was compared with that in the contralateral eye, which served as the control. The protein content ratios were found to be higher than unity in all cases, with this ratio generally increasing with time (Fig. 5) . Although there was an increase in protein content with probe implantation, the levels did not indicate a breakdown in the blood–retinal barrier wherein at least a 20- to 30-fold difference would have been observed. The highest ratio obtained was approximately fourfold at 6 hours after probe implantation. Although this ratio is not considered to be extremely elevated, it may be high enough to alter the kinetics of a drug in the vitreous, especially if the compound exhibits protein-binding properties. Increased protein levels in group I animals in which there was no recovery period after surgery could be the reason why GCV was eliminated at a slower rate in this study group. 

Discussion

Microdialysis has been a valuable sampling technique in drug disposition and pharmacokinetics. The main feature of this sampling procedure is that it does not induce any changes in the physiological volume of the sampling compartment. In the eye, it is a tool that enables one to obtain a complete pharmacokinetic profile from a single subject, thus reducing the number of animals involved. However, implantation of the probe can lead to trauma and tissue injury, and this could lead to alteration in drug disposition, especially if the injury is severe or if the animal is not allowed to recover from the trauma. The eye is a specialized organ that provides a very small physiological sampling compartment for pharmacokinetic studies. Therefore, any alterations in the composition of the sampling compartment could lead to a change in drug disposition. Rittenhouse et al. ¹⁷ have demonstrated that the microdialysis probe increased the protein content in the aqueous humor by almost 30-fold, which led to alterations in the disposition of propranolol. In our study, we observed a similar, though smaller, effect on the protein content in the vitreous. Increased protein levels could lead to increased drug binding and hence form a depot in the vitreous leading to slower elimination of free drug, assuming the binding process is reversible. As the free drug is eliminated from the vitreous, there may be a shift in the binding equilibrium, and drug molecules bound to protein are released at the moment the vitreous is no longer saturated. The protein levels were elevated approximately three- to fourfold with time in rabbits in which no recovery period was allowed after the surgery. These levels may not affect disposition of compounds that do not exhibit sufficient protein binding. The protein concentration at 8 hours after probe implantation was 2.5-fold higher in comparison with that in the control eye. All experiments were performed well after the completion of 5 days of recovery, and hence the protein concentration would be close to baseline. As a result, protein concentrations were not determined after 5 days. GCV exhibits 1% to 2% plasma protein binding and thus the effect of the increased protein content on its vitreous elimination is not strong. However, if a highly protein bound molecule is the subject of investigation, appropriate recovery time should be allowed to prevent any artifactual alterations in its vitreous disposition. 

Anesthesia has a suppressive effect on many physiological processes, including heart rate and respiratory rate. These effects can lead to slower blood flow and fluid exchange. The distribution of a compound in the vitreous is mainly governed by diffusion. However, convective forces also play an important role in this process. Convective forces develop due to the pressure differences between the anterior part of the eye and the retinal surface. ¹⁸ A change in the pressure gradient can affect these convective forces, in turn affecting the distribution of a compound. The anesthetic combination used in these studies has been shown to alter the IOP and thus may affect the distribution of the drug. In addition, any tissue uptake processes involved in the distribution of the drug may be slowed, leading to higher concentrations remaining in the vitreous chamber. These physiologic changes are probably the reasons why anesthesia leads to higher steady state levels. In addition, the intravitreous injection was made through the sclera and not through an extraocular muscle. The 50-μL injection may have transiently increased the IOP, leading to reflux of some of the injected drug through the needle hole in the sclera. This may have contributed in part to the lower AUCs obtained in the conscious group of animals. In our studies, anesthesia increased the exposure levels but did not have a noticeable effect on the elimination rate with the vitreous t _1/2 comparable to that in the conscious group II. The vitreous t _1/2 of GCV was short (3–6 hours), indicating a major transretinal route of elimination in addition to any contribution from the aqueous outflow pathways. Mannitol, a paracellular marker, exhibits an intravitreous t _1/2 of 4 hours (Atluri and Mitra, unpublished results, 2001). The intravitreous t _1/2 for GCV in group I animals was 6 to 7 hours, which is longer than that obtained for mannitol. This result indicates that GCV, like mannitol, may follow a similar passive paracellular route of elimination. However, the possibility of a facilitative mechanism contributing to the rapid elimination of GCV cannot be ruled out and is an area that needs further investigation. 

GCV has been previously studied in the eye. ¹⁹ The reported intravitreous t _1/2 was 7 hours for a 196-μg dose. In addition, Macha and Mitra ²⁰ have reported the GCV t _1/2 to be 7 hours, by using ocular microdialysis. The intravitreous t _1/2 in the animal group that was kept anesthetized and not given a sufficient recovery period after probe implantation (group I) was determined to be 6 hours (±1.5 hours), which is similar to values reported in the literature. However, an important point to be noted is that the animal models in the case of Lopez et al. ¹⁹ were pigmented. It is well known that ocular pigmentation can affect the elimination of a drug from the vitreous. The t _1/2 of GCV obtained in groups that were given a sufficient recovery period (groups II and III) is in the range of 4 hours, which is slightly lower. The rabbits used in the current study were albino rabbits, and hence the difference in the intravitreous t _1/2 compared with that obtained by Lopez et al. could be attributed to the effect of ocular pigmentation. Determinations in the model without a sufficient recovery period (group I) generated a t _1/2 for GCV which is almost equal to the one obtained by Lopez et al. The release of small amounts of protein somehow makes up for the absence of ocular pigmentation in this model. Anesthesia is not expected to alter drug diffusion across the retinal barrier, which may signal a similar rate of elimination. 

The studies presented in this article indicate that the choice of the pharmacokinetic model for studying disposition in the eye has to be made carefully. This decision should be based on several factors, such as the purpose of the study, the molecule under investigation, and the various possible interactions between the analyte and ocular tissues and fluids. For example, if the study is a comparative study between two compounds, the anesthetized model may be suitable for short-term studies to generate important pharmacokinetic parameters provided that the protein-binding characteristics of the two compounds are similar. In general, if there are no time constraints, the conscious model presented herein would be most suitable for preclinical research because of the absence of variables such as anesthesia and elevated protein levels. Parameters obtained from this model would provide true pharmacokinetic and disposition data necessary for preclinical studies involving ophthalmic drugs. 

Conclusions

The surgical procedure was well tolerated by most animals. Microdialysis probes remained functional for at least 14 days. Ganciclovir was rapidly eliminated in all animal groups. Anesthesia did not affect the rate of elimination of drug; however, it may have affected drug distribution in the vitreous, leading to higher exposure levels. The elevated protein levels after probe implantation necessitates a suitable recovery period after surgery. The conscious model in conjunction with the microdialysis technique is a valuable tool in assessing the intravitreous disposition of a compound, especially in cases in which the compound exhibits a long t _1/2. The proposed model may stimulate research in the area of drug delivery to the posterior segment of the eye. 

 Supported by National Eye Institute Grants RO1EY09171-08 and RO1EY10659-07.

 Submitted for publication June 11, 2002; revised July 16 and August 15, 2002; accepted August 20, 2002.

 Disclosure: C.S. Dias, None; A.K. Mitra, None

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Retrodialysis method used in the study.

Table 1.

 

Experimental Protocol for Study of Ocular Pharmacokinetics

Experimental Variables	Group I	Group II	Group III
Anesthesia	Yes	No	Yes
Recovery period	No	Yes	Yes

Figure 2.

 

Probe recovery values for mannitol in the conscious animal expressed as a percentage of value on day of surgery (Day 0).

Figure 3.

 

Ocular pharmacokinetics after an intravitreous bolus dose of GCV in (top, group I) anesthetized rabbits (n = 6) that were allowed no time for recovery from probe implantation, (middle, group II) conscious rabbits (n = 7) that had a 5-day or longer recovery from probe implantation, and (bottom, group III) anesthetized rabbits (n = 4) that had a 5-day or longer recovery from probe implantation.

Figure 4.

 

A comparison of the terminal elimination phases of GCV kinetics in the different study groups.

Table 2.

 

Ocular Pharmacokinetic Parameters of Ganciclovir after Intravitreous Administration

Parameters	Group I	Group II	Group III
AUC (μCi/mL · min)	* 650.71 (±264.70)	95.16 (±78.58)	* 564.34 (±227.67)
Half-life (min)	360.39 (±91.16)	, † 210.63 (±56.77)	, † 239.58 (±30.72)
AUC ratio compared with Group II	6.84	—	5.93

Figure 5.

 

Protein content of rabbit vitreous humor after probe implantation compared with the contralateral eye (control).

The authors thank David K. Peters from the Laboratory Animal Center, University of Missouri-Kansas City for valuable suggestions and advice in developing the surgical procedure; and Banmeet S. Anand for assistance with the conscious-animal surgery and animal management during the experiments.