Intravitreal Sustained-Release Ganciclovir ThomasJ. Smith, MD; P. Andrew Pearson, MD; David L. Blandford, MD; JoelD. Brown, MD; Kenneth A. Goins, MD; Jack L. Hollins, MD; Elmar T. Schmeisser, PhD; Peter Glavinos, PhD; Larry B. Baldwin, MD; Paul Ashton, PhD
cytomegalovipatients with acquired immunodeficiency syndrome involves frequent intravenous administration of sodium ganciclovir that often results in unacceptable side effects. We have developed devices that release ganciclovir at rates of 2 \g=m\g/h and 5 \g=m\g/h in vitro. When implanted into the vitreous of rabbit eyes, mean intravitreal ganciclovir levels of 9 mg/L and 16 mg/L were maintained for more than 80 and 42 days, respectively. Devices were well tolerated, \s=b\ Current
treatment of
retinitis in
rus
with
no
toxic effects attributable to the
polymers used in the devices. This investigation indicates that these devices can maintain therapeutic levels of drug for extended periods and are well tolerated in the rabbit eye. They may prove useful in the clinical management of cytomegalovirus retinitis in patients with acquired
immunodeficiency syndrome. (Arch Ophthalmol. 1992;110:255-258)
jP1 ytomegalovirus (CMV) is the
most
of viral retinitis in with acquired immunodefi¬
common cause
patients ciency syndrome (AIDS), affecting up to 40% of
patients.1"7
If left
untreated,
See also pp 185 and 188.
blindness inevitably results.7,8 Intrave¬ nous sodium ganciclovir is effective in the treatment of CMV retinitis, but requires frequent dosing. This often causes systemic side effects, including
Accepted for publication September 19, 1991. From the Departments of Ophthalmology (Drs
Smith, Pearson, Blandford, Brown, Goins, Hollins, Schmeisser, and Baldwin) and Surgery (Dr Ashton) and the College of Pharmacy (Dr Glavinos), University of Kentucky, Lexington. Reprint requests to the New England Glaucoma Foundation, 100 Charles River Plaza, Boston MA 02114 (Dr Smith).
neutropenia,
that necessitate treat¬
ment withdrawal in
approximately one third of patients.91° Other problems associated with systemic ganciclovir administration include sepsis related to permanent indwelling catheters and
difficulties because of concurrent ther¬ apy with zidovudine. Zidovudine is the only drug shown to prolong life and improve immune function in patients with AIDS." Intravitreal ganciclovir injections provide a higher intraocular drug con¬ centration than systemic therapy and reduce systemic exposure to the drug. The intravitreal half-life of ganciclovir in the human eye (estimated to be approximately 13 hours) necessitates frequent injections (at least one week¬ ly) to maintain therapeutic levels in the eye.12 Intravitreal ganciclovir injec¬ tions of 200 to 400 µg administered weekly have resulted in temporary re¬ mission of CMV retinitis.12"16 Repeated intravitreal injection has an attendant risk of cataract formation, retinal de¬ tachment, cystoid macular edema, and, especially in patients with AIDS,
endophthalmitis.
In vitro studies have indicated the concentration of ganciclovir required for 50% inhibition of CMV replication (ID50) to be between 0.2 and 2.6 mg/L.17"21 This, the short intravitreal half-life of ganciclovir, and its systemic toxicity make the drug a good candi¬ date for a local sustained-release deliv¬ ery system. We have investigated the permeation of ganciclovir through eth¬ ylene vinyl acetate (EVA) and polyvi¬ nyl alcohol (PVA) membranes and have developed implantable devices for the sustained release of ganciclovir. These devices release ganciclovir at rates of 2 or 5 µg/h for an extended period. This study investigates the
Downloaded From: http://archopht.jamanetwork.com/ by a University of Iowa User on 06/17/2015
pharmacokinetics and biocompatibility of device implantation in a rabbit model.
MATERIALS AND METHODS Materials
Ganciclovir was obtained from Syntex Laboratories, Palo Alto, Calif, in powdered form. Polyvinyl alcohol (molecular weight 76 000 to 78 000 d, 98% hydrolyzed) was obtained from Aldrich Chemical Co, Mil¬ waukee, Wis, while ethylene vinyl acetate (Elvax 40 w) was obtained from Du Pont, Bedford Park, 111. Sodium phosphate, sodi¬ um hydrogen phosphate, and sodium chlo¬ ride
were
obtained from Fisher Scientific
Co, Cincinnati, Ohio. Diffusion chambers were purchased from Crown Glass Co, Somerville, NJ. Diffusion Cell Studies
Polyvinyl alcohol membranes were pre¬ pared using 2% (weight to volume) PVA
solutions in deionized water. The solution poured so that 200 mL completely covered a silicone-coated glass slab measur¬ ing 35 46 cm. After drying for 16 hours, the PVA film was removed and heated at 190°C for 4 hours. Ethylene vinyl acetate membranes were prepared by compressing 3.5-g pellets under 4 metric tons at 190°C to a thickness of 0.6 mm. Diffusion cell studies were performed in jacketed glass diffusion cells. The internal diameter of these cells was 1 cm, and the volume of each compartment was 3 mL. Temperature was maintained at 37°C using a circulating water bath. Membranes were soaked in isotonic buffered saline (pH 7.4) for 30 minutes and later inserted between the chambers. A 0.025% solution of ganci¬ clovir in isotonic buffered saline was added to the donor compartment and the buffer solution added to the receptor compart¬ ment. The contents of the receptor com¬ partment were periodically removed for analysis with high-performance liquid chromatography (HPLC). The apparent perme¬ ability coefficient (flux divided by the prod¬ uct of donor concentration multiplied by was
area) was calculated using the linear portion of the total receptor content (adjusting for sampling dilution) vs time profile. Device Construction and Release Studies
ganciclovir devices prepared by coating a 6-mg pellet (diameter, 2.5 mm) of ganciclovir in 300 µL Sustained-release
were
of a 10% PVA solution. Two types of de¬ vices were prepared that released ganciclo¬ vir at either 5 µg/h (series 1) or 2 µg/h (series 2). To prepare series 1 devices, 3-mm EVA discs coated in 10% PVA were fixed to the top and bottom of dried pellets using 2% PVA solution (Fig 1). To prepare series 2 devices, pellets were coated on three sides with a film of prepressed EVA and later capped with a 3-mm EVA disc coated in 10% PVA. After the addition of EVA, both series 1 and series 2 devices were completely coated in 10% PVA, al¬ lowed to dry overnight, and heated to 190°C for 4.75 hours. The devices were placed in 10 mL of isotonic buffered saline at 37°C. One-milliliter samples were periodically removed for analysis, and the entire receptor solution changed every 20 days to maintain sink conditions.
HPLC
Analysis
Ganciclovir concentration was deter¬ mined with HPLC using a 25-cm 2-µ C18 reverse-phase column supplied by Cole Scientific (Calabasas, Calif). The mobile phase was 97% aqueous (0.02% ammonium acetate; pH, 4.0) and 3% acetonitrile, with a flow rate of 1.0 mL/min. The detection wavelength was 254 nm. Under these condi¬ tions, the retention time of ganciclovir was 7.8 minutes, and the detection limit, 0.2 mg/L.
Spectroscopy The near-infrared spectra of completed Near-Infrared
devices
were
obtained
over
the range of
1000 to 2500 nm, and these spectra were with those of unheated PVA, EVA, and ganciclovir using the same range. Second-order derivatives were cal¬
compared
compared. Device Implantation All animal work was performed
culated and
in accor¬ dance with Association for Research in Vi¬ sion and Ophthalmology guidelines. Twen-
1000 100
ft*
10H
Maximum Concentration After Intravitreal Injection
*
¿
» .
»
1-
Assay Detection Limit 0.1-
25
50
75
Time, d
Fig 4. Mean concentration of ganciclovir in the rabbit vitreous after device implantation. Standard deviations are indicated. Solid tri¬ angles indicate series 1 eyes. For points up to 6 to 9. On day 56, the ganciclovir day 42, concentration in five eyes was less than the detectable limit, and the mean concentration of the remaining four eyes is indicated. On day 70, the ganciclovir concentration in all eyes was less than the detectable limit. Open triangles indicate series 2 eyes. For each data point, three animals were used (n 6). —
=
Time, d
Fig 3. Release of ganciclovir from two types of devices. Solid triangles indicate device 1 (mean release rate, 5.2 µg/h), and open trian¬ gles, device 2 (mean release rate, 1.9 µg/h). Standard deviations are also indicated. —
1.— Diagram of a series 1 implantable sustained-release device that delivers ganci¬ clovir at 5 µg/h.
Fig
Position of the device in the eye.
Fig 2.
=
Fig 5.—Histologie specimen of the scierai implant site. There is evidence of a chronic inflamma¬
tory reaction with multinucleated giant cells (red pointer) to the silk suture used to secure the implant to the sclera. Around the implant site there is less evidence of inflammation (yellow
pointer).
—
Downloaded From: http://archopht.jamanetwork.com/ by a University of Iowa User on 06/17/2015
ty-five New Zealand white rabbits weighing 2 to 3 kg were used in this study.
Anesthesia consisted of intramuscular keta¬ mine hydrochloride (40 mg/kg of body
weight), xylazine hydrochloride (10 mg/kg of body weight), and topical proparacaine hydrochloride (0.5%). Pupillary dilation was achieved with topical 2.5% phenylephrine hydrochloride and 1% tropicamide drops. Baseline flash (cone and rod response) and flicker (cone response) electroretino¬ grams (ERGs) were obtained under lightadapted conditions. Graphic measurements were recorded, and the waveform ampli¬ tudes and implicit times were calculated. Because of the poorly developed pars plana in the rabbit, cryotherapy was applied to the superior 180° of the peripheral retina to prepare a site for device implantation and vitreous sampling.
Four weeks after cryopexy, a second ERG was obtained and, under sterile condi¬ tions, a device was inserted. After place¬ ment of a lid speculum, a 360-degree con¬ junctival peritomy was performed. The rectus muscles were secured, and the area of cryopexy was then localized and marked with scierai diathermy. Devices were im¬ mersed in sterile saline before placement of a 7-0 silk suture through the external PVA supporting tag. A microsurgical vitreal reti¬ nal blade was used to create a 5-mm sclerotomy site through the area of cryopexy. Penetration into the vitreous was verified with direct visualization of the retraction blade. Devices were placed into the vitre¬ ous cavity and secured to the sclera with the 7-0 silk suture. Remaining defects in the sclera were closed with 7-0 silk suture in an interrupted fashion. Polymyxin sulfate, bacitracin zinc, and neomycin sul¬ fate were instilled into eyes twice daily for 3 days after each procedure. The implanted device is shown in Fig 2.
Experimental Groups Animals were divided into three experi¬ mental groups. In each of the first nine rabbits (series 1), a placebo device (contain¬ ing no ganciclovir) was implanted into the right eye and a series 1 device (releasing ganciclovir at a rate of 5 µg/h) was implant¬ ed into the left eye. Vitreous samples (200 µ ,) were obtained from the drug eye with a 20-gauge needle on a tuberculin syringe and stored at -60°C until analyzed with HPLC. Samples were taken 1, 3, 7, 14, 21, 28, 42, 56, and 72 days after surgery. In a second experimental group, consist¬ ing of 15 animals, a series 2 device (releas¬ ing ganciclovir at a rate of 2 µg/h) was implanted into both eyes. Three animals (six eyes) were killed on each of the follow¬ ing days after implantation: 10, 30, 40, 70, and 80. Vitreous samples and devices were removed for analysis. Retinal examination with indirect oph¬ thalmoscopy was performed on each eye before vitreous sampling (series 1) and im¬ mediately before death (series 1 and 2). Samples were analyzed with reverse-phase HPLC. Electroretinography was repeated 1 and 2 months after device insertion in series 1 eyes and immediately before killing in series 2 eyes.
The third experimental group consisted one rabbit in which an uncoated 6-mg pellet of ganciclovir was inserted into each eye to assess the potential retinal toxicity in the event of device failure. Retinal function was assessed with ERG before surgery and 2, 11, and 48 days after surgery. The animal was then killed. Animals from all groups were killed with an overdose of intravenous sodium pento¬ barbital. Histopathologic analysis was per¬ formed on eyes from each group. of
RESULTS Diffusion Cell
Ganciclovir
permeated
a
2% PVA
membrane, heated for 3 hours at 190°C, at a mean rate (±SD) of
4.86 ±0.44 µg/h per square centime¬ ter, which corresponds to an apparent diffusion coefficient of 0.019 cm/s. Per¬ meation of ganciclovir through EVA membranes was too slow to measure. Device Release Rate
The two devices released ganciclovir at rates of 1.9 ±0.3 µg/h and 5.2 ±0.5 µg/h (Fig 3). Release rates remained constant until more than 90% of the drug had been released. Results of liquid chromatography analysis indi¬ cated that ganciclovir was not chemi¬
cally altered during device
con¬
struction.
Near-Infrared Spectroscopy
All
peaks found in the near-infrared
spectra of the devices
were
attribut¬
able to ganciclovir, PVA, EVA, indicating no chemical change during the preparation of the devices. Similar¬ ly, the second-order derivative failed to show any thermal degradation of ganciclovir or the polymers. or
Device
Implantation
In series 1, the mean concentration of ganciclovir in the vitreous was 16.0 ±7.1 µg/h for the first 42 days. When sampled on day 56, the ganciclo¬ vir concentration in five eyes was undetectable, but the mean concentration in the remaining four eyes was 12.8 ±2.4 mg/L. Seventy-two days af¬ ter device implantation in series 1 eyes, no drug was detected. In series 2 eyes, the mean vitreous concentration was 9.4 ±4.6 mg/L for 80 days (Fig 4). The devices appeared to be well tolerated in both groups, with no evi¬ dence of intraocular inflammation ob¬ servable with indirect ophthalmosco¬ py. However, five series 1 treatment eyes developed lens opacification, two of which developed a diffuse white cataract. Three developed focal nonprogressive lens opacification. Retinal detachments occurred in two series 1
Downloaded From: http://archopht.jamanetwork.com/ by a University of Iowa User on 06/17/2015
treatment eyes. No series 2 eyes or series 1 control eyes developed observ¬
able abnormalities. After surgery, both flash and flicker wave amplitudes decreased slightly for all animals, but no significant differ¬ ence in amplitude between control and treatment groups was evident. The ERGs of two series 1 eyes were flat secondary to retinal detachment. The only other significant change in retinal function of treatment eyes was an in¬ crease in flicker b-wave latency (30 ms to 34 ms, P