3
`Precorneal, Corneal, and
`Postcorneal Factors
`
`Vincent H. L. Lee University ofSouthern California School of
`Pharmacy, LosAngeles, California
`
`I.
`
`INTRODUCTION
`
`Designing formulations and delivery systems for topically applied ophthalmic drugs is
`Challenging. It requires a thorough understandingcof the. physiological basis of the protec-
`tive mechanisms designed by the eye to allow only 1—10% of the topically applied dose to
`be absorbed ocularly. These protective mechanisms include solution drainage, lacrimation,
`diversion of exogenous chemicals into the systemic circulation via the conjunctiva, and a
`highly selective corneal barrier to exclude exogenous compounds from the internal eye.
`Improvement of ocular drug delivery then amounts to determining the outer boundaries as
`well as the maximum duration over which these protective mechanisms can be com-
`promised without causing harm to this vital organ. According to Grass and Robinson (1), to
`significantly alter the fraction of drug absorbed into the eye, it will be necessary to either
`increase the corneal drug absorption rate constant by one to two orders of magnitude or to
`reduce the precorneal loss rate constant by a similar extent—a formidable task.
`
`This chapter discusses the role of key precorneal, corneal, and postcorneal factors in
`determining the ocular bioavailability of topically applied drugs. These factors include
`1) precorneal fluid dynamics, 2) drug binding to tear proteins, 3) conjunctival dmg absorp-
`tion, 4) systemic drug absorption, 5) resistance to corneal drug penetration, 6) drug binding
`to melanin, and 7) drug metabolism. Although drug uptake by the lens is gaining attention
`owing to a growing interest in aldose reductasc inhibitors as drugs for retarding the progress
`of sugar-induced cataracts (2,3), it will not be discussed here, since relatively little informa—
`tion exists on this topic.
`
`ll. PRECORNEAL FACTORS
`
`A. Precorneal Fluid Dynamics
`
`including aqueous solutions, oil solutions, suspensions, and
`liquid dosage forms,
`All
`liposomes, are rapidly drained from the conjunctiva] sac to the nasolacrimal duct. The
`residence time of an instilled dose ranges from 4723 min (Table 1). Such a rapid drainage
`
`59
`
`ALCON 2234
`Apotex Inc. v. Alcon Pharmaceuticals, Ltd.
`Case |PR2013-00012
`
`

`

`60
`
`Lee
`
`Table 1. Residence Time of Liquid Aqueous Ophthalmic Dosage Forms in the Conjunctival Sac
`Following Topical Instillation in Rabbit Eyesa
`
`
`
` Dosage form Drug Residence time (min) Ref.
`
`
`
`
`
`Aqueous solution
`Aqueous solution
`Aqueous solution
`
`Cromolyn Na
`Epinephrine
`Inulin
`
`6.8 x 0.46
`5.9 x 0.35 '
`7.3 t 0.71
`
`7173
`60
`60
`
`53.
`4.8 t 0.18
`Pilocarpine
`Aqueous solution
`174
`9.7 t 0.62
`Vitamin A
`Oil solution
`19
`23.0
`None
`Microspheres
`60
`6.5 t 0.57
`None
`Liposomes (neutral)b
`
`Liposomes (positive)c 6 None 4.3 1- 0.37
`
`
`
`:Residence time is defined as the time requiredforlossof95%of the instilleddose
`:Phospholipid compositionfiDipalmitoyl phosphatidylcholine
`CPhospholipid compositionvStearylamine: Lorphosphatidylcholine:cholesterol:01—tocopherol (1 .4495 :.005)
`
`rate results from the tendency of the eye to maintain the residenegol/unieatJ,—,1,Q,_uLat all
`times (4). This rate becomes even higher when the formulation is perceived to be irritating.
`The. igitaney of positively charged liposomes to the eye (5) is probably the main reason for
`its‘relatively rapid rate of clearance from the precorneal area when compared with solutions
`(6). Indeed, it is the desire to minimize the irritation potential of suspended particles to the
`eye that underlies the many attempts to render the particle size in suspensions as smallas
`
`possible. While the irritating potential versus size relationship has neverbeen dotsimented
`
`reduction in particle size does affect drug dissolution rate and bioavailability. A $1 nificant
`rank-order correlation was observed by Schoenwald and Stewart (7) between increasing
`dexamethasone concentrations in the aqueous humor and decreasing particle size over the
`range of 575—220 pm in a 1% suspension of the drug. A similar relationship was reported
`by Hui and Robinson (8) for a 0.1% suspension of fluorometholone over the size range of
`20—10.4 11m; there was no further gain in ocular drug bioavailability upon further reducing
`the particle size from 2 to 1 pm Not surprisingly, varying the particle size has no effect on
`the corneal permeability coefficient (9).
`Several factors influence the drainage rate: instilled volume, viscosity, pH, tonicity,
`and drugs.
`
`1.
`
`Insti/Ied Volume
`
`The human eye can hold about 30 pL without overflow or spillage at .thevguterrakngle,
`provided that great care is exercised and that the subject does not blink (4). This volume
`reduces to 10 11L if blinking is allowed (10). The seminal work of Robinson and his
`colleagues (11,12) based on gamma scintigraphy has established that the rate of solution
`drainage from the conjunctival sac is directly proportional to the instilled volume. In the
`rabbit, 90% of the dose is cleared within 2 min for an instilled volume of 50 11L (the volume
`delivered by most commercial ophthalmic preparations), 4 min for an instilled volume of
`25 11L, 6 min for an instilled volume of 10 11L, and 7.5 min for an instilled volume of5 11L
`(11). This volume dependency of solution drainage rate has been found to exert its expected
`effect on the percent of dose absorbed into the eye and on the pharmacological effect that
`ensues (13—16). Using a model that predicts ocular drug bioavailability from tear drug
`
`

`

`Precorneal, Corneal, and Postcomeal Factors
`
`61
`
`concentration-time data, Keister et al. (17) proposed that reducing the instilled drop would
`increase only the ocular bioavailability of drugs with low permeability four times and
`would not affect the ocular bioavailability of drugs with high corneal permeability. Since
`high corneal permeability is the exception rather than the rule, the clinical implication is
`that appropriate reduction of instilled volume and the simultaneous increase in instilled
`drug concentration should permit substantial dosage reductions without sacrifice of drug
`concentration in the eye (18).
`
`The above volume dependency of the drainage rate has also been observed for
`suspensions (3 pm in size) (19), but it has not been observed in liposomes (20). Lee et al.
`(20) reported that multilamellar, neutral liposomes prepared from phosphatidylcholine and
`cholesterol were cleared from the conjunctiva] sac of the albino rabbit, with approximately
`the same first-order rate constant, 0.45 min“l, over the instilled volume range of 10—50 11L.
`As a result,
`tlie ocular absorption of inulin from these liposomes was not affected by
`changes in instilled dose volume over the above range. The size and number of liposomes
`
`were believed to be more important factors than instilled volume influencing the extent of
`
`ocular drug absorption from liposomes.
`The volume dependency of the drainage rate for solutions has implications in multiple
`
`drop therapy in terms of 1) minimum time interval in between drops and 2) order of
`addition of drops. Using radioactive technetium (gngc) as the test substance in rabbits,
`Chrai et al. (21) showed that a 5-min interval between drops minimized drainage loss of
`drug, and that the first drug administered suffered a greater loss than the second drug.
`These findings constitute a very strong argument for combination drug products whenever
`
`multiple drug therapy is indicated.
`
`2. Viscosity
`
`Compared with reducing instilled solution volume, increasing solution viscosity is a more
`
`popular method of prolonging the residence time of an instilled dose in the conjunctival sac.
`Various polymers have been used to increase solution viscosity,
`including poly(vinyl
`alcohol), poly(pyrrolidone), hydroxypropylcellulose, and other cellulose derivatives. Based
`upon a comparison of the reduction in solution drainage rate by methylcellulose and
`poly(vinyl alcohol) and the resulting increase in aqueous humor pilocarpine concentrations
`in the albino rabbit (22,23), Patton and Robinson (23) concluded that
`it
`is the flow
`properties of the vehicle in question and its viscosity, not the concentration, that determines
`the effect of polymers on solution drainage and ocular drug absorption. It appears that the
`optimum viscosity to use is in the range of 12—15 cps, beyond which the gain in ocular
`absorption would be minimal, while the risks of inaccuracy of instillation and blurring of
`vision would increase. Even with a 100—fold increase in viscosity, the gain in ocular drug
`
`absorption is modest, being less for oil-soluble than for water-soluble drugs (1).
`The hypothesis that all polymers affect ocular drug absorption similarly so long as they
`yield the same viscosity assumes that these polymers do not interact with the corneal
`surface. Work by Saettone et al. (24,25) indicates that this may not be the case in human
`
`subjects. These investigators demonstrated that equiviscous solutions of carboxymethylcel-
`
`lulose, hydroxypropylcellulose, poly(vinyl alcohol), poly(vinylpyrrolidone), while equally
`effective in rabbits, enhanced the ocular absorption of pilocarpine as well as tropicamide to
`different extents. The most effective polymers were poly(vinyl alcohol) and poly(vinylpyr-
`rolidone). The different activity of these polymers was attributed to their influence on the
`spreading characteristics and the thickness of the medication layer over the preeorneal area
`
`

`

`62
`
`Lee
`
`in humans. Saettone et al. (26) showed that soluble, mucoadhesive polyanionic polymers,
`such as hyaluronic acid, poly(galacturonic acid), mesoglycan (a complex mixture of
`mucopolysaccharides), carboxymethylchitin, and polyacrylic acid, enhanced the ocular
`
`absorption of pilocarpine more so than poly(vinyl alcohol) of equivalent viscosity. Similar
`favorable effects with hyaluronic acid over hydroxypropylmethylcellulose and polyacrylic
`acid (Carbopol 934P) over poly(vinyl alcohol) have been reported by Camber et al. (27)
`and Davies et al. (28). Cyanoacrylate block copolymer, another mucoadhesive polymer,
`has also been found to improve the ocular absorption of pilocarpine by 53% (29).
`
`3. pH and Tonicity
`
`For stability reasons, most eye drops are formulated at pHs otheLthan pH 7.4. They are,
`therefore, potentially irritating to the eye, stimulating tear production. [acrimal gland fluid
`secretion can be stimulated by reflexes from afferent pathways arising in the cornea,
`conjunctiva, and optic nerve. In rabbits, lacrimal gland fluid stimulated by ocular surface
`reflexes was 3.0 1 0.5 til/10 min as compared with a baseline level of 1.0 x 0.2 mL/10 min.
`There was a concomitant rise in the protein concentration in tears (30), which could affect
`the amount of free drug available for corneal absorption. Conrad et al. (31) reported that
`alkalineqpflgjgdyced,greater lacrimation than acidic pHs in the albino rabbit. This is
`consistent with the lower buffer capacity of tears in the basic than in the acidic range
`(32,33). Since tears arepoorly buffered (32), a strategy to minimize the impact of induced
`lacrimation is,__to use theflmost’dilute buffer possible; i.e.,
`to keep the tonicity low. By
`progressively reducing the buffer concentration of a pH 4.5 citrate buffer from 0.11 M to
`zero, Mitra and Mikkelson (34) observed a fivefold increase in the ocular bioavailability
`of pilocarpine. In addition to buffer concentration, the buffer type used also affects the
`absorption efficiency of pilocarpine owing to its effect on the rate at which pH reequilibra-
`tion will occur (33).. Thus, a phosphate buffer (pH 4) which is resistant to pH reequilibra-
`tion near the pKa of pilocarpine owing to its high residual buffer capacity, yields a lower
`bioavailability of pilocarpine than does an acetate buffer, even though its buffer capacity is
`lower (Fig. 1) (33). Nevertheless, the phosphate buffer is still preferred because the acetate
`buffer causes excessive lacrimation and may be inherently irritating to the eye.
`
`4. Drugs
`
`Drugs that act on the lacrimal gland can affect precomeal fluid dynamics. Examples
`include epinephrine (35), pilocarpine (36), the local anesthetics tetracaine and proparacaine
`(37), certain beta-blockers (35), and the tear stimulants currently in development (38).
`Thus, epinephrine has been shown to accelerate the removal of topically applied liposomes
`from the conjunctival sac by inducing tear production (39). Induced lacrimation by
`epinephrine and pilocarpine (and possibly their prodrugs) has also been suggested as a
`reason for the reduction in ocular absorption of topically applied timolol when used in the
`same drop with either of the above two drugs (40,41). On the other hand, suppression
`of tear turnover by the topical
`instillation of five drops of 0.5% tetracaine has been
`shown to double the amount of pilocarpine absorbed in the aqueous humor of the albino
`rabbit eye (37).
`
`B. Drug Binding to Tear Proteins
`
`Although the protein content of tears is much less than that of blood, it is still appreciable
`and ranges from 0.5% total protein in rabbits to about 0.7% total protein in humans (42). Of
`
`

`

`Precorneal, Corneal, and Postcorneal Factors
`
`63
`
`(cm)
`
`
`
`
`
`ChangeinPupillaryDiameter
`
`0.20
`
`0.10 0.00\
`
`0
`
`20
`
`4o
`Minutes
`
`60
`
`Figure 1. Time course of changes in pupillary diameter following topical instillation of 25 “L of
`various formulations containing 1% pilocarpine (pH 4) to the albino rabbit eye. Key: 0, unbuffered;
`A, acetate—buffered; A, phosphate—buffered; O, citrate-buffered. (From Ref. 33.)
`
`the several proteins in tears (43,44), at least three (albumin, globttlirts,and 11392me area,
`necessary to quantitate the binding of some drugs in the tears (45), There exists, therefore,
`the possibility of drug binding to tear proteins, resulting in a reduction in the free drug
`congentrationsfor'absOrption. Such a possibility was first pointed out by Mikkelson et a1.
`(42) for pilocarpine. These investigators showed that the miotic response to topically
`applied pilocarpine in the albino rabbit was reduced about two times as the albumin
`concentration in the preeorneal fluid was increased from 0 to 3%. This problem of reduced
`drug bioavailability due to binding of drugs to tear proteins could be exacerbated when
`there is an elevation of tear proteins in certain extraocular disease states, such as corneal
`inflammation (13,46), herpes simplex infection (47), and allergic conjunctivitis (47).
`Increased loss of timolol to protein binding due to an increase in tear protein concentration
`caused by reduction in tear turnover rate by tetracaine and propracaine has been proposed
`as a reason for the reduced ocular absorption of timolol when coadministered with the two
`local anesthetics (40).
`
`C. Conjunctival Drug Absorption
`
`The conjunctiva is a vascularized, thin mucous membrane lining the inside of the eyelids
`and the anterior sclera. The conjunctiva is known to differ from the cornea in several
`
`aspects: metabolic activity (48), length and density of microvilli (49), and permeability to
`water-soluble compounds such as mannitol, inulin, and FITC-dextran (MW 20,000) (50).
`The conjunctiva possesses two important features that render it more effective in competing
`with the cornea for drug absorption: 1) a/QMIimes larger surface area in the rabbit and a 17
`times larger surface area in the human (51), and 2) a 2 to 30 times greater permeability to
`drugs (52). It is, therefore, not surprising that drug uptake by the conjunctiva is as important
`as solution drainage loss in reducing the fraction of pilocarpine available for corneal
`absorption (53). The early hint that conjunctival drug uptake is a significant preeomeal
`
`

`

`64
`
`Lee
`
`drug loss factor lies in the many observations that in spite of a 10-fold reduction in drainage
`rate by a 100-fold increase in solution viscosity,
`the maximum improvement in drug
`activity, be it miosis, inhibition of infection, or aqueous humor levels, is about twice that of
`an aqueous solution (23). To date, no attempts to reduce conjunctiva] drug absorption have
`been reported, even though this is a desirable goal from the standpoint of increasing the
`fraction of drug available for corneal absorption.
`There is now evidence that it may be possible to reduce conjunctiva] drug absorption
`in two ways: varying drug lipophilicity or changing the drug formulation. Wang et a]. (52)
`reported that although lipophilicity affected the conjunctiva] and corneal permeability to
`beta—blockers in a qualitatively same but quantitatively different way, the conjunctiva]
`permeability coefficient was less sensitive to changes in lipophilicity (log PC) compared
`with the corneal permeability coefficient. Within the log PC range of —0.62 (sotalol) and
`3:44 (betaxolol), there was only an eightfold difference in the conjunctiva] permeability
`coefficient as compared with a 48-fold difference in the corneal permeability coefficient.
`Therefore, provided that the drug candidates are sufficiently lipophilic, it should be pos-
`sible to improve corneal drug penetration without markedly affecting conjunctiva] drug
`absorption.
`In addition to its lesser sensitivity to changes in drug lipophilicity, the conjunctiva]
`permeability coefficient is also less sensitive to a given formulation change than is the
`corneal permeability coefficient (54). Formulation changes that are most effective in
`minimizing the ratio of conjunctiva] to corneal drug absorption are increasingsolutioerH,
`lowering/solution tonicity, and lowering the percentage of ethylenediaminetetraacetic acid
`OSDTA) and benzalkonium chloride in the formulation. The first hint that the conjunctiva
`and cornea are different from the standpoint of drug penetration is the different magnitude
`by which a given pH change alters the cornea] and conjunctival permeability to the four
`beta-blockers studied: atenolol, timolol, levobunolol, and betaxolol. For instance, whereas
`raising the pH from 7.4 to 8.4 increased the corneal permeability to timolol 2.4 times, it
`only increased the conjunctiva] permeability by 28%. The magnitude of increase in both
`instances was much less than the factor of 8 increase in the fraction of timolol (pKa 9.21) in
`the nonionized, preferentially absorbed form.
`_
`Besides maximizing the fraction of topically applied drug for corneal absorption,
`minim/Wing drug uptake by the conjunctiva is also desirable from the standpoint of mini-
`mizing drug absorption into systemic circulation,
`€9.1Ci15t
`in theory. Nevertheless,
`the
`resulting reduction in systemic absorption is expected to be minimal, since the conjunctiva
`" plays only a minor role, compared with the nasal mucosa, in contributing to systemic drug
`absorption (15,55).
`The vascularized nature of the conjunctiva has generated the perception that drugs
`absorbed by the conjunctiva would all be swept into systemic circulation and would not be
`available for distribution to the uveal tract underneath. This assumption has now been
`proven to be incorrect, first by Doane et al. (56) and then by Ahmed and Patton (57,58).
`Thus, an unknown fraction of the drug absorbed by the conjunctiva could lead to direct
`drug entry into the uveal tract, bypassing the cornea. This route of drug entry into the eye
`following topical dosing is called the noncorneal route, as depicted in Scheme 1. There is
`indirect evidence that
`the ratio of noncorneal
`to corneal drug absorption is increased
`whenever inefficient mixing of the instilled dose with tears occurs, as when very viscous
`solutions or when dispersed systems such as nanoparticles and liposomes are instilled
`(39,59—63). Noncorneal drug absorption is also facilitated by drug administration in an
`
`

`

`Precorneal, Corneal, and Postcomeal Factors
`
`
`65
`
`FPRECORNEAL AREA“!
`
`CONJUNCTIVA
`
`SCLERA
`
`- NTERIOR
`H AMBER
`
`
`
`
`
`INTRAOCULAR
`TISSUES
`
`
` CORNEA
`
`
`
`
`
`
`
`
`
`
`OCULAR CIRCULATION
`
`
`
`
`
`SYSTEMIC CIRCULATION
`
`CONTRALATERAL EYE
`
`
`
`
`
`Scheme 1. Ocular penetration routes for topically applied drugs. Key: 1 = transcomeal pathway;
`2 = noncorncal pathway; 3 = systemic return pathway; 4 : lateral diffusion. (From Ref. 58.)
`
`insert. This is indicated by the unequal concentrationsof timololinmflpgdgtatld inferior
`parts of the pigmented rabbit eye and very low timolol concentrations in the aqueous humor
`when a silicone cylindrical device containing timolol (1.46 x 1.94 x 0.24 mm, ID x OD x
`wall thickness) was placed in the inferior cul—de-sac (64). A smaller difference between
`timolol concentrations in the superior and inferior portions of the anterior segment tissues
`was found when the device was placed in the superior cul-de-sac.
`\
`
`0. Systemic Drug Absorption
`
`An often-neglected aspect of ocular drug therapy is systemic absorption of the topically
`applied dose that has reached the nasal mucosa following solution drainage. This is partly
`due to the fact that the drugs involved then had wide therapeutic indices or that the patient
`population who had suffered from systemic side effects "elicited by topically applied
`ophthalmic drugs was small. This situation changed dramatically when timolol, a potent
`mixed betai and betaz antagonist, was introduced into glaucoma therapy in 1978. Systemic
`risk associated with topical timolol was emphasized by Van Buskirk then (65). Six years
`later, Nelson et al. (66) reported 450 cases of serious systemic side effects attributed
`to timolol, of which 32 resulted in patient deaths. As many as 23% of the patients experi—
`enced their adverse event on the first day of timolol therapy and 33% did so within the
`
`first week (66).
`
`

`

`66
`
`Lee
`
` Drug % Applied dose Ref.
`
`
`
`Table 2. Percent of Topically Applied Dose Absorbed into the
`Systemic Circulation
`
`
`Cortisol
`30—35
`175
`Dipivalylepinephrine
`65
`176
`Epinephrine
`55
`176
`Flurbiprofen
`74
`179
`lmirestat
`50—75
`177
`Inulin
`3
`55
`Insulin, with 1% Na glycocholate
`5
`178
`Levobunolol
`46
`180
`[D-A1321Metenkephalinamide
`36
`55
`Tetrahydrocannabinol
`23
`150
`
`80 \Timolol 181
`
`
`\
`
`
`
`Restricting entry of a topically applied ophthalmic dose into the nasal cavity is
`an obvious approach to reducing the extent of systemic absorption of topically applied
`ophthalmic drugs. This objective can be achieved by nasolacrimal occlusion for 5 min, with
`or without eyelid closure (67,68), or by changes in ve\hi_clercompositionsuchasincorpora-
`tionpfflpolymers(16,69), changes in vehicle type (41,70), alteration in solution pH and
`tonicity (16), and adjustment of preservative concentration (16). Other means to reducing
`systemic drug absorption include \1)2coadministration with low doses of vasoconstrictors
`such as phenylephrine and epinephrine (40,41,69);a2) designing Ophthalmic drugs that are
`poorly absorbed into the bloodstream (71) or are rapidly inactivated in the systemic
`circulation (72,73)—so—called prodrug and soft drug approaches,
`respectively; and
`3) selecting a dosing time that minimizes systemic absorption while maximizing ocular
`drug absorption (74-77). The effectiveness of the above approaches has been reviewed
`(78). Because the majority of the above approaches aim primarily at the nasal mucosa (the
`main site of systemic drug absorption) rather than at the conjunctiva] sac, reduction in
`systemic drug absorption may not necessarily lead to enhanced corneal drug absorption.
`This has been found to be the case when epinephrine was coadministered with timolol to
`reduce the systemic absorption of timolol (41). Table 2 summarizes the percent of topically
`applied dose that has been found to be absorbed into the bloodstream of rabbits.
`
`Ill. CORNEAL FACTORS: RESISTANCE TO CORNEAL
`DRUG PENETRATION
`
`The majority of topically applied drugs enter the eye by passage across the cornea (79).
`This is an extremely inefficient process owing largely to the resistance exerted by the
`corneal epithelium to drug penetration. Generally, resistance due to metabblism is low,
`except for pilocarpine (80), fluorometholone (81), and peptides such as methionine
`enkephalin (82) and triglycine (83). The corneal epithelium contributes to over 90% of the
`corneal resistance to penetration for hydrophilie beta-blockers, decreasing to about 50% for
`the moderately lipophilic and to less than 10% for the lipophilic. This is accompanied by a
`
`

`

`Precorneal, Corneal, and Postcorneal Factors
`
`67
`
`rise in the contribution due to the corneal endothelium from less than 5% for hydrophilic
`compounds through 30% for the moderately lipophilic and to 50% for the lipophilic. A
`similar trend is seen in the contribution from the corneal stroma (84).
`A measure of-the corneal penetration efficiency of drugs is the permeability coeffi-
`cient. This is generally on the order of 0.1—4.0 x 10‘5 cm/s (52). Unless the drug is very
`lipophilic, e.g., prednisolone acetate (85), fluorometholone (85), betaxolol (86), and
`timolol ester prodrugs (87), at least a twofold increase. in the extent of corneal penetration
`occurs when the corneal epithelial barrier is absent. The degree of penetration enhancement
`can be as high as 10- to 30-fold, as is the case for S-fluorouracil, which penetrates the
`cornea poorly because of its hydrophilicity (log PC = —0.96) (52); 14-fold, as is the case for
`methionine enkephalin (82), which penetrates the cornea poorly mainly because of suscep-
`tibility to aminopeptidase-mediated hydrolysis (88); and 60—fold, as is the case for inulin,
`which penetrates the cornea poorly because of its size (MW 5000) and hydrophilicity (log
`PC = —2.90) (89).
`Changes in corneal epithelial permeability during ocular inflammation have also
`caused an increase in the permeability to such drugs as cyclosporine (90) and
`dexamethasone phosphate (91). Kupferman et al. (91) and Cox et a1. (92) demonstrated that
`dexamethasone alcohol and phosphate were absorbed across the corneal epithelium of the
`inflamed but not the noninflamed eye, even though no macroscopic changes were obvious
`in the structural integrity of the corneal epithelium in the inflamed eye. Similarly, Pavan-
`Langston and Nelson (93) reported that trifluridine, an antiviral agent, was well absorbed
`across the corneas of patients with herpetic iritis but not in the corneas of healthy subjects.
`Baum et al. (94) found that the type of injury to the cornea affected the extent of improve-
`ment in corneal drug absorption. Six times more gentamicin penetrated lye-bumed ulcers
`than corneal ulcers caused by Pseudomonas. However, for unknown reasons, a corneal
`ulcer induced by the vaccinia virus, manifested as a frank erosion of the corneal epithelium
`and Bowman’s membrane, did not affect the ocular uptake of cortisol.
`Corneal integrity can be compromised by sufficiently high concentrations of certain
`formulation excipients, such as preServatives (e.g., benzalkonium chloride and other
`cationic surfactants) and chelating agents (e.g., EDTA). Thus, benzalkonium chloride and
`other cationic surfactants have been shown to enhance the ocular absorption of drugs
`varying in melecular size and lipophilicity, including pilocarpine (95), carbachol (96),
`prednisolonf/(97), homatropine (98), inulin (99), and horseradish perOxidase (100). Grass
`et al. (101) demonstrated that EDTA at 0.5% altered the permeability of the corneal
`epithelium, probably at the intercellular junction level, to enhance the corneal permeability
`of water-soluble but not oil-soluble drugs. Moreover, EDTA reached the iris—ciliary body
`at concentrations high enough to alter the permeability of the uveal vessels,
`thereby
`indirectly accelerating drug removal from the aqueous humor.
`Interpretation of corneal penetration data usually assumes homogeneity in lipophilic
`characteristics in the corneal epithelium. In actuality, the corneal epithelium is five to six
`cell layers thick and consists of three groups of cells with unique biochemical charac-
`teristics: 1) two to three layers of flattened platelike superficial cells, 2) two to three layers
`of wing or polygonal cells comprising the intermediate zone, and 3) a single row of
`columnar basal cells. In 1979, Godbey et al. (102) made the interesting observation that
`disrupting the top two epithelial layers by treatment with 0.02% cetylpyridinium chloride,
`an ophthalmic preservative, was as effective as removing all
`layers of the corneal
`epithelium in allowing penicillin G to pass through the cornea. The implication of this
`
`

`

`68
`
`»
`
`Lee
`
`finding is that the top two layers of the corneal epithelium bear all the resistance to the
`corneal penetration of this drug.
`The above finding also raises the interesting possibility that a lipophilicity gradient
`exists across five to six cell layers within the comeal epithelium, with the most lipophilic
`layer on the tear side. This hypothesis has been confirmed by Shih and Lee (86) using a
`technique developed by Wolosin et al. (103,104) to strip off selective layers of the corneal
`epithelium. This was achieved by pretreating the cornea with 20—100 “M of digitonin for
`15 min. Such pretreatment did not affect the corneal penetration of betaxolol, a very
`lipophilic drug (log PC = 3.65). Pretreatment with 40 pM digitonin to cause exfoliation of
`the top two corneal epithelial cell layers enhanced thehcomeal penetration of timolol (log
`PC = 2.64) and levobunolol (log PC = 3.22) to the same extent as deepithelizing the corneal
`epithelium. Unlike timolol and levobunolol, atenolol (log PC = 0.15) encounters resistance
`beyond the superficial cell
`layers in its penetration across the corneal epithelium.
`Even when the intermediate zoneof wing cells was removed by treating the cornea with
`60.100 pM digitonin, the corneal permeability coefficient was only 68%of that seen in the
`deepithelized cornea. This finding is somewhat surprising, since atenolol, given its hydro-
`philic characteristics,
`is anticipated to cross the corneal epithelium via the paracellular
`pathway, the permeability of which is presumably controlled by the tight junctions in the
`superficial cells. Although direct confirmation is required, the above findings are consistent
`with the hypothesis that a lipophilicity gradient exists across the five to six cell layers
`within the corneal epithelium; i.e., the number of corneal epithelial cell layers limiting the
`corneal penetration of ocularly administered drugs is inversely related to drug lipophilicity.
`From the standpoint of maximally improving the corneal penetration of a hydrophilic drug
`such as atenolol, it will therefore be necessary 1) to modify the drug properties to match
`those exhibited by all the five to six cell layers in the corneal epithelium, or 2) to design a
`penetration enhancer capable of disrupting the integrity of all those cell layers.
`The existence of a lipophilicity gradient within the corneal epithelium has implications
`on how the parabolic relationships between corneal penetration and lipophilicity reported
`for steroids (105), n-alkyl p-aminobenzoate esters (106), substituted anilines (107), timolol
`ester prodrugs (108), and beta—blockers (84) should be interpreted. The usual interpretation
`is a shift in the rate—limiting layer from the corneal epithelium to the corneal stroma as drug
`lipophilicity is increased. In light of the new finding noted above, such a shift in the
`rate-limiting layer could have occurred within the corneal epithelium, far removed from the
`corneal stroma. Nevertheless, deciding on where in the cornea such a shift actually occurs
`must await resolution of the controversy on whether a parabolic relationship best describes
`the influence of drug lipophilicity on corneal drug penetration. As shown by Wang et al.
`(52), a sigmoidal relationship (Fig. 2) statistically described the influence of lipophilicity
`on the corneal penetration of beta-blockers better than the parabolic relationship reported
`by Sehoenwald and Huang (84), which weighted heavily on one lipophilic compound
`(penbutolol) beyond the purported maximum in their parabola. The same argument prob-
`ably holds for steroids (84). Deviation from the parabolic relationship has also been
`reported by Grass and Robinson (1) when compounds of diverse chemical structure and
`molecular size are considered. Such deviations are to be expected because not all the
`compounds selected for study utilize the usual
`transcelluiar pathway for penetration.
`Compounds that penetrate via the less common paracellular pathway include low
`molecular weight alcohols and the ionized form of such drugs as pilocarpine (109,110),
`sulfonamides (111), and cromolyn Na (101). It is important to resolve the controversy on
`
`

`

`Precomeal, Corneal, and Postcorneal Factors
`
`69
`
`cm/s)
`
`(10E5,
`
`Papp
`
`Log PC
`
`Influence of drug lipophilicity (log PC) on the permeability coefficients (Papp) of
`Figure 2.
`beta-blockers across the cornea of the pigmented rabbit. Error bars represent standard error of
`the mean fo