JOURNAL OF OCULAR PHARMACOLOGY Volume 8, Number 4, 1992 Mary Ann Liebert, Inc., Publishers
Intravitreal Ganciclovir Pharmacokinetics in Rabbits and Man ASHTON,1 ID. BROWN,1 P.A. PEARSON,1 BLANDFORD,1 T.J. SMITH,2 R. ANAND,3 S.D. NIGHTINGALE,3 and G.E. SANBORN4
P. D.L.
department of Ophthalmology, Kentucky Clinic,
University of Kentucky, Lexington, Kentucky
2Present Address: New England Glaucoma Research Foundation, Boston,
Massachusetts
Southwestern Medical Center, Dallas, Texas 4Richmond Retina Associates, Richmond, Virginia
ABSTRACT
Cytomegalovirus (CMV) retinitis occurs in immune-compromised patients and can be treated by repeated intravenous or intravitreal injections of ganciclovir (GCV) or foscarnet. Due to toxicity and complications these modalities are not ideal. The development of alternative
administration
routes
is
pharmacokinetic data. Devices giving pseudo zero order intravitreally first in rabbits and
hindered
by
a
lack
of
release of GCV were implanted then in patients with AIDS associated CMV retinitis as part of a Phase I clinical trial. Steady state intravitreal GCV levels were obtained immediately after death and the elimination rate constants were calculated Normalizing for retinal assuming first order pharmacokinetics. surface area, distribution volume and anatomic volume, the retinal elimination rate constants were calculated. These were found to be This cm-2hr-1 in rabbits and 0.015 cm-2hr-1 in man. 0.017 indicates that the rabbit eye is a good model for studying intravitreal pharmacokinetics of ganciclovir and suggests a common elimination mechanism which may be trans-retinal. INTRODUCTION
Cytomegalovirus retinitis (CMV) occurs in approximately 20% of patients with AIDS (1) and is presently treated with high doses of either ganciclovir or foscarnet. Although both treatments are initially effective the retinitis frequently progresses while the patient is undergoing treatment (mean time to progression of 70-80
ganciclovir and foscarnet produce requiring treatment withdrawal (2). In an attempt to reduce systemic exposure intravitreal injections can also be given (3). The frequency of injections is determined Study of human by the half-life of ganciclovir in the vitreous. intravitreal pharmacokinetics is extremely difficult as the repeated intravitreal sampling required carries risks to the days). sever
As importantly, systemic toxicities
both often
343
patient (including retinal detachment, cataract formation and endophthalmitis). Despite these concerns attempts have been made to investigate the elimination of GCV from the vitreous in man. Using data from one patient, Henry and co-workers estimated the intravitreal half-life of ganciclovir to be 13.3 hours with a distribution volume of 11.7 ml (3). Using data reported by Jabs the intravitreal half-life can be calculated to This is also based on data from one patient. These reports indicate that frequent intravitreal injections (1 or 2 each week) of GCV are needed to maintain therapeutic drug levels. In an attempt to maintain therapeutic levels in the vitreous, sustained release devices of ganciclovir were implanted first in rabbits (4) and then in patients (5). These devices are composed of a 6 mg pellet of ganciclovir encased in two inert, nonerodible polymers, This paper presents polyvinyl alcohol and polyvinyl acetate. pharmacokinetic data obtained using this drug delivery system. and coworkers be 8.1 hours.
(4)
MATERIALS AND METHODS
Sustained devices were release prepared as previously described (5). Briefly, 6 mg pellets of GCV were completely coated in polyvinyl alcohol (PVA, permeable) then partially coated in
ethylene vinyl
acetate (EVA, impermeable) and finally completely The coated pellets were then heat treated at PVA. 190°C for 4.75 hours. The release of ganciclovir from these devices was found to be constant (zero order) until less that 10% of GCV remained. The release rate could be controlled by varying the thickness of the PVA, the duration of the heat treatment and the integrity of the EVA coat. Both PVA and EVA are biologically inert and nonerodible and the devices do not undergo any gross changes while implanted in the eye. Devices releasing GCV at 2 ug/hr (4 month duration) were implanted into the vitreous of 15 rabbits. Electroretinogram (ERG) examinations were performed immediately prior to cryopexy. After 4 weeks ERGs were repeated and a 5 mm sclerotomy site created through Devices were then inserted into the vitreal the area of cryopexy. cavity and sutured to the sciera before the wound was closed. Three animals (six eyes) were sacrificed 10, 30, 40, 70 and 80 days Devices were then removed and a sample of after implantation. vitreous (100 ul ) taken. The GCV concentration of the vitreous and the amount of GCV remaining in the device was then determined by HPLC using a reverse phase C-18 column with uv detection at 254 nm. The mobile phase was 0.025% ammonium acetate at pH 4.0 with 2.5% acetonitrile and a flow rate of 1.0 mL/min. Under FDA IND number 34,108 a phase I clinical trial was performed in eight informed volunteers. These patients had sight threatening CMV retinitis and were intolerant of iv ganciclovir. IOn no patients had vitrectomy, lensectomy or cap[sulotomy been Devices were implanted into 11 eyes of previously performed. these patients and follow up examinations (over periods of 5 to over than 50 weeks) included bimonthly fundus photography. Samples of vitreous (100 ul) were obtained from 6 eyes following the death of the patient and one sample was obtained when a rhegmatogenous detachment was repaired. These samples were obtained 11, 12, 35, Six devices were 35, 37, 69 and 70 days after implantation. removed following patient death, these were taken 11, 12, 37, 69, HPLC analysis was used to determine 70, 143 after implantation. the concentration of GCV in the vitreous and the amount of GCV remaining in the device. Histopathology was performed on two eyes and showed well healed scierai incisions with no evidence of an coated
again in
344
inflammatory
response.
(6).
This
work
has
been
previously
described
Mathematical Treatment the elimination of GCV from the order process with respect to vitreous rate of elimination of GCV is given by
first
Assuming
Re
VdkC
=
vitreous
to
concentration,
be a the
[1]
hr-1),
where Re is the elimination rate (ug V¿ is the distribution volume (ml) of GCV in the vitreous, C is the vitreous concentration and k is the elimination rate constant of GCV (ug
(hr-1).
ml-1)
Rewriting [1] gives
Re where k^ is the product volume.
a
kjC
=
[2]
rate constant with units ml hr-1. of the elimination constant and
This constant is distribution
the
Assuming that the ratio of volume of distribution to anatomic volume is similar in the rabbit and man a second rate constant (k2) can be defined as follows
*2
kl/va
=
[3]
where Va is the anatomic volume of the vitreous. constant by the surface area of the retina gives K, rate constant K
where a
SR
k2/SR
=
Dividing this the
retinal
[4]
K therefore represents is the surface area of the retina. of the elimination rate constant per unit area of the normalized to account for differences in distribution
measure
retina
volume. The radius of the human eye is 1.2 cm while the radius of the is approximately 0.6 cm (7) and the vitreous volume cornea The radius of the rabbit eye is approximately 3.9 ml (8). approximately 0.9 cm (9) and the vitreous volume 1.5 ml (10). The radius of the rabbit cornea is 0.7 cm (11). The surface area of the retina can be approximated to be that of a sphere minus the section occupied by the cornea and this can be readily calculated.
RESULTS AND DISCUSSION
After implantation both flash and flicker wave amplitudes decreased slightly for all animals but this was not significant. There was however an increase in flicker b-wave latency (30 is to Devices appeared to be well tolerated with no 34 ms, P < 0.05). evidence of poor wound healing, inflammation or damage to the anterior segment observable by slit lamp examination. Histologie The devices, examination showed no evidence of inflammation. releasing drug at a constant rate, maintained a mean intravitreal GCV level of 9.4 + 4.6 ug/mL in the rabbit over the 80 day duration The mean release rate of the devices was found of the experiment. to be 2.0 + 0.3 ug/hr.
345
In the
patients
clinical study the
intravitreal concentration in
mean
found to be 2.0 + 1.3 ug/mL and the mean release rate 1.9 + 0.4 ug/hr. The terms k.-^, (rate constant, ml k2, and K (elimination rate constant normalized for volume, (elimination rate constant normalized for volume and retinal were calculated for rabbits and man surface area (table was
hr-1) hr-1)
hr-1mm-2)
1).
This work indicates that the rabbit is a good model for the of the intravitreal pharmacokinetics of GCV in the diseased human eye. The mathematical treatment described assumes that elimination is a first order process and that this process is not affected by either device implantation or chronic exposure to GCV. It is further assumed that the ratio between anatomic volume Va and distribution volume V
Intravitreal Ganciclovir Pharmacokinetics in Rabbits and Man ASHTON,1 ID. BROWN,1 P.A. PEARSON,1 BLANDFORD,1 T.J. SMITH,2 R. ANAND,3 S.D. NIGHTINGALE,3 and G.E. SANBORN4
P. D.L.
department of Ophthalmology, Kentucky Clinic,
University of Kentucky, Lexington, Kentucky
2Present Address: New England Glaucoma Research Foundation, Boston,
Massachusetts
Southwestern Medical Center, Dallas, Texas 4Richmond Retina Associates, Richmond, Virginia
ABSTRACT
Cytomegalovirus (CMV) retinitis occurs in immune-compromised patients and can be treated by repeated intravenous or intravitreal injections of ganciclovir (GCV) or foscarnet. Due to toxicity and complications these modalities are not ideal. The development of alternative
administration
routes
is
pharmacokinetic data. Devices giving pseudo zero order intravitreally first in rabbits and
hindered
by
a
lack
of
release of GCV were implanted then in patients with AIDS associated CMV retinitis as part of a Phase I clinical trial. Steady state intravitreal GCV levels were obtained immediately after death and the elimination rate constants were calculated Normalizing for retinal assuming first order pharmacokinetics. surface area, distribution volume and anatomic volume, the retinal elimination rate constants were calculated. These were found to be This cm-2hr-1 in rabbits and 0.015 cm-2hr-1 in man. 0.017 indicates that the rabbit eye is a good model for studying intravitreal pharmacokinetics of ganciclovir and suggests a common elimination mechanism which may be trans-retinal. INTRODUCTION
Cytomegalovirus retinitis (CMV) occurs in approximately 20% of patients with AIDS (1) and is presently treated with high doses of either ganciclovir or foscarnet. Although both treatments are initially effective the retinitis frequently progresses while the patient is undergoing treatment (mean time to progression of 70-80
ganciclovir and foscarnet produce requiring treatment withdrawal (2). In an attempt to reduce systemic exposure intravitreal injections can also be given (3). The frequency of injections is determined Study of human by the half-life of ganciclovir in the vitreous. intravitreal pharmacokinetics is extremely difficult as the repeated intravitreal sampling required carries risks to the days). sever
As importantly, systemic toxicities
both often
343
patient (including retinal detachment, cataract formation and endophthalmitis). Despite these concerns attempts have been made to investigate the elimination of GCV from the vitreous in man. Using data from one patient, Henry and co-workers estimated the intravitreal half-life of ganciclovir to be 13.3 hours with a distribution volume of 11.7 ml (3). Using data reported by Jabs the intravitreal half-life can be calculated to This is also based on data from one patient. These reports indicate that frequent intravitreal injections (1 or 2 each week) of GCV are needed to maintain therapeutic drug levels. In an attempt to maintain therapeutic levels in the vitreous, sustained release devices of ganciclovir were implanted first in rabbits (4) and then in patients (5). These devices are composed of a 6 mg pellet of ganciclovir encased in two inert, nonerodible polymers, This paper presents polyvinyl alcohol and polyvinyl acetate. pharmacokinetic data obtained using this drug delivery system. and coworkers be 8.1 hours.
(4)
MATERIALS AND METHODS
Sustained devices were release prepared as previously described (5). Briefly, 6 mg pellets of GCV were completely coated in polyvinyl alcohol (PVA, permeable) then partially coated in
ethylene vinyl
acetate (EVA, impermeable) and finally completely The coated pellets were then heat treated at PVA. 190°C for 4.75 hours. The release of ganciclovir from these devices was found to be constant (zero order) until less that 10% of GCV remained. The release rate could be controlled by varying the thickness of the PVA, the duration of the heat treatment and the integrity of the EVA coat. Both PVA and EVA are biologically inert and nonerodible and the devices do not undergo any gross changes while implanted in the eye. Devices releasing GCV at 2 ug/hr (4 month duration) were implanted into the vitreous of 15 rabbits. Electroretinogram (ERG) examinations were performed immediately prior to cryopexy. After 4 weeks ERGs were repeated and a 5 mm sclerotomy site created through Devices were then inserted into the vitreal the area of cryopexy. cavity and sutured to the sciera before the wound was closed. Three animals (six eyes) were sacrificed 10, 30, 40, 70 and 80 days Devices were then removed and a sample of after implantation. vitreous (100 ul ) taken. The GCV concentration of the vitreous and the amount of GCV remaining in the device was then determined by HPLC using a reverse phase C-18 column with uv detection at 254 nm. The mobile phase was 0.025% ammonium acetate at pH 4.0 with 2.5% acetonitrile and a flow rate of 1.0 mL/min. Under FDA IND number 34,108 a phase I clinical trial was performed in eight informed volunteers. These patients had sight threatening CMV retinitis and were intolerant of iv ganciclovir. IOn no patients had vitrectomy, lensectomy or cap[sulotomy been Devices were implanted into 11 eyes of previously performed. these patients and follow up examinations (over periods of 5 to over than 50 weeks) included bimonthly fundus photography. Samples of vitreous (100 ul) were obtained from 6 eyes following the death of the patient and one sample was obtained when a rhegmatogenous detachment was repaired. These samples were obtained 11, 12, 35, Six devices were 35, 37, 69 and 70 days after implantation. removed following patient death, these were taken 11, 12, 37, 69, HPLC analysis was used to determine 70, 143 after implantation. the concentration of GCV in the vitreous and the amount of GCV remaining in the device. Histopathology was performed on two eyes and showed well healed scierai incisions with no evidence of an coated
again in
344
inflammatory
response.
(6).
This
work
has
been
previously
described
Mathematical Treatment the elimination of GCV from the order process with respect to vitreous rate of elimination of GCV is given by
first
Assuming
Re
VdkC
=
vitreous
to
concentration,
be a the
[1]
hr-1),
where Re is the elimination rate (ug V¿ is the distribution volume (ml) of GCV in the vitreous, C is the vitreous concentration and k is the elimination rate constant of GCV (ug
(hr-1).
ml-1)
Rewriting [1] gives
Re where k^ is the product volume.
a
kjC
=
[2]
rate constant with units ml hr-1. of the elimination constant and
This constant is distribution
the
Assuming that the ratio of volume of distribution to anatomic volume is similar in the rabbit and man a second rate constant (k2) can be defined as follows
*2
kl/va
=
[3]
where Va is the anatomic volume of the vitreous. constant by the surface area of the retina gives K, rate constant K
where a
SR
k2/SR
=
Dividing this the
retinal
[4]
K therefore represents is the surface area of the retina. of the elimination rate constant per unit area of the normalized to account for differences in distribution
measure
retina
volume. The radius of the human eye is 1.2 cm while the radius of the is approximately 0.6 cm (7) and the vitreous volume cornea The radius of the rabbit eye is approximately 3.9 ml (8). approximately 0.9 cm (9) and the vitreous volume 1.5 ml (10). The radius of the rabbit cornea is 0.7 cm (11). The surface area of the retina can be approximated to be that of a sphere minus the section occupied by the cornea and this can be readily calculated.
RESULTS AND DISCUSSION
After implantation both flash and flicker wave amplitudes decreased slightly for all animals but this was not significant. There was however an increase in flicker b-wave latency (30 is to Devices appeared to be well tolerated with no 34 ms, P < 0.05). evidence of poor wound healing, inflammation or damage to the anterior segment observable by slit lamp examination. Histologie The devices, examination showed no evidence of inflammation. releasing drug at a constant rate, maintained a mean intravitreal GCV level of 9.4 + 4.6 ug/mL in the rabbit over the 80 day duration The mean release rate of the devices was found of the experiment. to be 2.0 + 0.3 ug/hr.
345
In the
patients
clinical study the
intravitreal concentration in
mean
found to be 2.0 + 1.3 ug/mL and the mean release rate 1.9 + 0.4 ug/hr. The terms k.-^, (rate constant, ml k2, and K (elimination rate constant normalized for volume, (elimination rate constant normalized for volume and retinal were calculated for rabbits and man surface area (table was
hr-1) hr-1)
hr-1mm-2)
1).
This work indicates that the rabbit is a good model for the of the intravitreal pharmacokinetics of GCV in the diseased human eye. The mathematical treatment described assumes that elimination is a first order process and that this process is not affected by either device implantation or chronic exposure to GCV. It is further assumed that the ratio between anatomic volume Va and distribution volume V
