Relationship between Steroid Permeability across
`Excised Rabbit Cornea and Octanol-Water
`Partition Coefficients
`
`RONALD D. SCHOENWALD" and RICHARD L. WARD
`Received July 13, 1977, from the Ophthalmic Biopharrnaceutics Group, Alcon Laboratories, F o r t Worth, T X 76101.
`publication September 22,1977.
`
`Accepted for
`
`Abstract Permeability rates were determined across excised rabbit
`corneas for 11 steroids. Permeability coefficients for each steroid were
`calculated, and their logarithms were plotted against their respective log
`octanol-water partition coefficients. A parabolic relationship resulted,
`with an optimum log permeability and coefficient observed at a log PO
`of 2.9. From these experimental results, an improvement in ophthalmic
`bioavailability of dexamethasone acetate as compared to dexamethasone
`is predicted and correlates with literature results.
`Keyphrases Permeability-of various steroids across excised rabbit
`corneas, related to octanol-water partition coefficients
`Partition
`coefficients, octanol-water-related
`to permeability of various steroids
`across excised rabbit corneas 0 Steroids, various-permeability
`across
`excised rabbit corneas related to octanol-water partition coefficients
`
`The amount of drug that ultimately penetrates the
`cornea is often largely determined by physiological factors
`such as the extent of drainage, blinking, andlor tearing
`during the first 4-6 min after topical dosing (1). The
`time-honored approach to resisting drainage has been
`through the use of viscous solutions. Water-soluble poly-
`mers such as methylcellulose, hydroxypropyl methylcel-
`lulose, and polyvinyl alcohol impart a slight lowering of
`surface tension and an increase in viscosity to ophthalmic
`solutions. The increase in viscosity prolongs contact of the
`drug in the eye, thereby resisting drainage (2,3); a lowering
`of the surface tension improves mixing with the precorneal
`tear film (4). In rabbits, a two- to threefold improvement
`in pilocarpine bioavailability has been found for viscous
`aqueous solutions (2, 3).
`Another approach to improving ophthalmic bioavail-
`ability has been to increase drug absorption into the eye.
`For example, it was reasoned (5) that the enhancement of
`the lipophilicity of epinephrine through the use of a dipi-
`valyl ester would facilitate penetration through the lipoidal
`layers of the cornea, Prodrug concentrations of 0.1%, ad-
`ministered twice daily for 1 month, significantly reduced
`intraocular pressure in hypotensive patients (6). The
`prodrug is reported to be about 100 times more effective
`than epinephrine in the management of glaucoma and
`about 100-400 times weaker than epinephrine in affecting
`the cardiovascular system of dogs and cats (7).
`In terms of improved efficacy, the final evaluation of an
`analog or prodrug must come from in uiuo studies. How-
`ever, optimization of permeability based on molecular
`modification can be assessed more reliably from a model
`devoid of extraneous physiological factors (8, 9). Such a
`model should focus primarily on enhanced permeability
`as a function of molecular modification.
`The purpose of this study was to determine the rela-
`tionship between the permeability of various structurally
`related steroids across an excised rabbit cornea and their
`octanol-water partition coefficients.
`
`706 I Journal of Pharmaceutical Sciences
`
`EXPERIMENTAL
`
`tritium-labeled steroids and one 14C-labeled steroid
`Materials-Ten
`were purchased'. Identification of each steroid was confirmed by TLC
`using two solvent systems2. This step was accomplished by dissolving
`unlabeled steroids3 in methanol along with the labeled steroid and de-
`termining Rl values for each by scintillation4 and UV5 spectroscopic
`methods. Purification consisted of removing volatile, labile tritium, using
`low temperature vacuum evaporation of alcoholic and then aqueous
`solvents (1 1). Monitoring the aqueous distillate for radioactivity indicated
`that removal of labile tritium by solvent evaporation was complete after
`three distillations6. All other chemicals were reagent grade.
`Excised Corneal Preparation Procedure-A 1-1.5-kg young albino
`rabbit was sacrificed by injecting 20-30 ml of air into the marginal ear
`vein. The intact eye along with the lids and conjunctival sac was then
`enucleated. According to a published procedure (12, 131, the exposed
`cornea of the enucleated eye was placed carefully on a specially designed
`corneal holder, which maintained the cornea curvature and held the eye
`in place. Various tissues of the eye were dissected away, leaving the cor-
`nea, a small ring of scleral tissue, and palpebral conjunctiva, which was
`tied to the corneal holder.
`The conjunctiva and scleral tissue served as a gasket and permitted
`the cornea to be suspended within the block system without inducing
`significant trauma or curvature distortion to the corneal endothelium
`or epithelium (12, 14). The block system7 consisted of two cylindrical
`compartments separated by the cornea. The compartment adjacent to
`the endothelial surface of the cornea was designated the internal side,
`whereas the compartment adjacent to the epithelial side was referred to
`as the external side.
`Study Procedure-Within 20 min of death, the cornea was mounted
`within the preheated block system and clamped into place; 6.0 ml of
`preheated (37") glutathione Ringers solution8 was added to the inter-
`nal side. Then 6.0 ml of preheated glutathione Ringers solution, con-
`taining labeled and unlabeled steroids, was added to the external com-
`partment. These solutions were prepared about 1 hr prior to the start of
`the experiment.
`
`From New England Nuclear, Boston, Mass.: 6,7-3H-prednisolone, lot 747-276;
`1,2-3H-hydrocortisone, lot 853-184; 7-3H-testosterone, lot 772-277; 1,2-3H-des-
`oxycorticosterone, lot 853.169 1,2-3H-cortexolone, lot 951-024; 6,7-3H-triam-
`cinolone acetonide, lot 635-258; 6,7-3H-dexamethasone, lot 998-003; and 6 7 -
`3H-dexamethasone acetate, lot 690-1348. From Amersham/Searle, Arlington
`Heights, Ill.: 4-'4C-progesterone, lot CFA.148. batch 42; 1,2-3H-fluorometholone,
`lot TRQ.974, batch 21474; and 6,7-3H-prednisolone "-acetate, lot TRQ.915, batch
`20806.
`* Solvent systems specific for each steroid and suggested by the manufacturer
`were used; methylene chloride-acetone (4:l) was suggested by Stahl (10) for general
`steroid use and also was used.
`Progesterone (Aldrich, lot 021957). hydrocortisone (Sigma, lot 25C-0319),
`prednisolone (Sigma, lot 23C-1900), desoxycorticosterone (Aldrich, lot 0318471,
`testosterone (Aldrich, lot 112457), cortexolone (Aldrich, lot 06181). triamcinolone
`acetonide (Sigma, lot 26C-0211), dexamethasone (Alcon RPA 5399), dexamethasone
`acetate (Sigma, lot 83C:-1700), fluorometholone (Farmila, lot 56/015), and pred-
`nisolone acetate (Organon, lot 92305).
`The silica gel was scraped from the origin to the solvent front (Quanata/Gram,
`Fairfield, N.J. LQDT prescored 20 X 20-cm TLC plates) in 1-cm increments. Every
`centimeter of silica gel was placed in a separate vial, scintillation fluid was added,
`and the vials were counted.
`Visual location of spot with the short wavelength of B UV lamp (Ultra-Violet
`Products, San Gabriel, Calif.).
`6 Less than 1.5% total radioactivity was recovered in distillate.
`7 The block system and corneal holder were purchased from Mr. Harold Eick
`through the cooperation of Dr. H. F. Edelhauser, Department of Physiology,
`Medical College of Wisconsin, Milwaukee, Wis.
`This solution differed from the glutathione bicarbonate Ringers solution used
`in Ref. 12 in that sodium bicarbonate was replaced with 0.77 g of sodium biphos-
`phatehter to maintain the pH at 7.2 f 0.2 during the experiment.
`
`IPR Page 1/3
`
`Santen/Asahi Glass Exhibit 2020
`Micro Labs v. Santen Pharm. and Asahi Glass
`IPR2017-01434
`
`

`

`The concentration of steroidg used in each experiment was significantly
`lower than the drug's reported solubility in water at 25 or 37" (15). The
`specific activities of each steroid test solutionlo were adjusted so that
`nanogram levels could be detected on the internal side. This condition
`required relatively high specific activities for practically insoluble ste-
`roids. The block system containing the excised cornea was placed on a
`hot plate" preadjusted to maintain the solution temperature within
`35-37", The temperature was periodically measured to assure that the
`solution temperature remained within the specified range. The block
`system and the hot plate were surrounded by Styrofoam for insulation
`against temperature fluctuations.
`As soon as the solutions were added to each side, circulation of fluids
`was induced by slowly bubbling air at a rate of approximately two or three
`bubbles per second. Samples of 0.1 ml were taken12 from the internal side.
`The first sample was withdrawn within 1 min of adding the solution to
`the external side. This sample represented the zero-time sample and also
`served as a control. If the conjunctiva was accidently perforated during
`the surgical procedure, then fluid would be exchanged rapidly between
`compartments within this 1st min. Under these conditions, the zero-time
`sample would yield a counting rate well above background and indicate
`that the kinetic data were invalid. Subsequent samples were taken every
`0.5 hr for 4 hr.
`All samples were immediately placed into a scintillation vial13 con-
`taining 10 ml of scintillation solution14. A like volume of solution was
`removed and discarded from the external side at each time increment.
`Thus, the cornea remained in hydrostatic equilibrium on both sides of
`the system. All vials containing tracer and scintillation solution were dark
`adapted for 18 hr or more and countedl5. The external standardization
`method (16) was used to determine counting efficiencies and, therefore,
`permitted the quantity of steroid present in each sample to be calculat-
`ed.
`Partition Coefficient-All partition coefficients are expressed as
`averaged log octanol-water and were taken from Leo et al. (17). For
`dexamethasone acetate, no value was listed; however, the acetate ester
`could be estimated readily from the parent molecule, dexamethasone (17,
`18). An octanol-water system was chosen because of the large volume of
`data that has been generated for correlation studies (17,18).
`
`RESULTS
`All corneas were judged clear at the end of each 4-hr experiment"j.
`Figure 1 represents a typical plot of data for the penetration of dexa-
`methasone acetate and dexamethasone across an excised cornea. A short
`lag time was observed, representing the time required for drug to reach
`steady state in the cornea. Following the lag time, the data followed a
`linear relationship. The individual experiments yielded highly linear
`correlations as judged by the Pearson r (>0.975). In addition, no trends
`in nonlinearity were observed with time. This result indicated that the
`steady-st,ate permeability rate was constant with time as required by Eqs.
`1-3.
`The transfer of substances across membranes by simple diffusion has
`been described (20) as a simplification of Fick's law according to:
`
`(Eq. 1)
`where dqldt is the permeability rate or the rate of drug quantity pene-
`
`The conrentration (micrograms per milliliter) of each steroid placed into the
`external compartment was: prednisolone, 16.7; hydrocortisone, 267; testosterone,
`25; progesterone, 0.90; desoxycorticosterone, 133; dexamethasone, 83.3; dexa-
`methasone aretate, 1.33; triamcinolone acetonide, 8.33; prednisolone acetate, 8.33;
`cortexolone, 261; and fluorometholone, 1.17.
`l o The specific activity (microcuries per milligram) of each radioactive steroid
`tracer was: prednisolone, 236; hydrocortisone, 15.4; testosterone, 119; progesterone,
`151.1; desoxycorticosterone, 26.5; dexamethasone, 41.2; dexamethasone acetate,
`2554; triamcinolone acetonide, 353; prednisolone acetate, 603; cortexolone, 12.1;
`and fluorometholone, 9:21.
`I ' Model 4812, 15 X 1.5 cm, Cole Parmer, Chicago, Ill.
`l 2 Eppendorf pipettor, Rrinkmann Instruments, Westbury, N.S.
`l 3 With Polyseal core liners, Kimble. Toledo, Ohio.
`I4 Handifluor Srintillar, Mallinckrodt, St. Louis, Mo.
`LS-230 liquid scintillation counter, Beckman Instruments, Fullerton, Calif.
`l 6 Each cornea was judged clear if typewritten print could be identified through
`the cornea at the end of each experiment. The clarity of the cornea is related to its
`hydration level: a normal cornea measures 78% (19). As the hydration level increases,
`the thickness of the cornea also increases; however, linear steady-state plots of drug
`penetrating the cornea generally result as long as the hydration level is below 85%.
`These latter corneas are cloudy and typewritten print cannot be read through them.
`Measurements were carried out for fluorometholone, prednisolone acetate, and
`progesterone and were 82.2 f 0.6% ( n = 4), 82.6% (n = l), and 82.6 f 0.8% (n = 4),
`respectively.
`
`a 8 r
`
`T
`
`,-
`0 ,/'
`e
`1
`
`3
`
`I
`I
`
`J
`a
`
`2
`HOURS
`of dexamethasone acetate (0) and dexa-
`Figure 1-Permeability
`methasone ( 0 ) across a n excised rabbit cornea. Each point represents
`a mean of four determinations. The vertical bars indicate 1 SD bars that
`are absent were smaller than the circle. The slope of the line represents
`the steady-state permeability rate.
`
`trating the cornea at time t , D is the diffusion coefficient of steroid
`through the excised cornea (square centimeters per second), A is the area
`of the absorbing surface of the cornea, T is the thickness of the cornea,
`(PC) is the partition coefficient (ratio of the drug concentration in the
`barrier membrane to the i n vitro testing solution), CE is the drug con-
`centration on the external side, and CI is the drug concentration on the
`internal side.
`For these experiments, samples taken from the internal side were
`considered negligible in drug concentration when compared to the ex-
`ternal side1?. Therefore, CI was eliminated from the equation, and CE
`was considered a constant through the time course of the experiments.
`Therefore, Eq. 1 can be rewritten as:
`
`When q is plotted versus t , the linear portion of the plot represents the
`steady-state permeability rate. The least-squares slope of each individual
`experiment was determined and averaged. Each averaged slope value was
`divided by an averaged valueIs of A , 1.089 cm2. T o compare the results
`from each steroid experiment, each averaged slope value was also divided
`by the CE value used for each steroid. Consequently, the final value
`represented a permeability coefficient [D(PC)/T] with units of square
`centimeters per second. When expressed in logarithmic form, this rela-
`tionship becomes:
`
`(Eq. 3)
`
`log (P,,,,)
`
`D
`= log - + log (PC)
`T
`Figure 2 illustrates the parabolic nature of the data when the logarithms
`of permeability and partition coefficients are plotted against one another.
`The data could be best represented by a second-order power seriesls:
`log (Pperm) = -0.28 (log PI2 + 1.7 log P -7.0
`(Eq. 4)
`
`where log (Pperm) is the logarithm of the permeability coefficient (cen-
`timeters per second), and log P is the logarithm of the octanol-water
`partition coefficient.
`An optimum log PO value of 2.9 was found by setting d log (Pp,,,)/d
`log P equal to zero and solving for log PO (21). Figure 2 predicts a decrease
`in permeability once a partition coefficient of 2.9 is reached. According
`to Flynn and Yalkowsky (221, limited solubility is often responsible for
`the parabolic shape of many structure-activity curves. As a consequence,
`the solubility of progesterone was determined carefully under the exact
`test conditions of the permeability experiments (same solution, tem-
`perature, etc.). A solubility of 14.1 f 1.46 Fg/ml (n = 3) was obtained for
`progesterone. Therefore, to be certain that solubility would not influence
`the results, a concentration of 0.90 pg/ml, which was greater than 1 log
`unit below the drug's solubility, was chosen for study.
`If the progesterone results are excluded from consideration, the re-
`maining 10 steroids approach a plateau, as predicted by the kinetic model
`
`17 After 4 hr, samples taken from the internal side were 3% or less of the con-
`centration on the external side, with the exception of dexamethasone acetate which
`was 7.4%.
`18 Dr. H. Edelhauser, Department of Physiology, Medical College of Wisconsin,
`Milwaukee, Wis., personal communication.
`19Nonlinear regression was performed using the BMDX85 program on a
`CDC6400 computer.
`
`Vol. 67, No. 6, June 19781 787
`
`IPR Page 2/3
`
`

`

`L L
`
`I- z
`tf!
`V
`G -4.5-
`W 8
`>
`I-
`=’ ! -5.0-
`z a:
`a Lu
`
`I T
`
`LOG PARTITION COEFFICIENT
`Figure 2-Computer-genernted curuilinear relationship between the
`k ~ g permeability coefficient of 11 laheled steroids and their respective
`log octnnol-watcr partition coefficients. Prom left to right, the steroids
`are prednisolone, hydrocortisone, dexamethasone, fluorornetholone,
`triamcinolone ncetonide. prednisolone acetate, cortewolone, desoxy-
`cortirostrwnc, dcxamethasone aretate, testosterone, and progesterone.
`The, twrticaI hars wprc’spnt I SD.
`
`of‘ Yalkowsky and Flynn (‘23). However, the value obtained for proges-
`terone is statistically different ( p < 0.05) from desoxycorticosterone,
`dexamethasone acetate, and testosterone, all of which are associated with
`maximum permeability. To determine if the carbon-14 label resided with
`progesterone during the experiment, an aliquot was taken from the ex-
`ternal side lollowing each experiment and chromatographed on silica gel
`thin-layer plates. Results showed that 95% of the label was associated
`with the progesterone molecule.
`
`DISCUSSION
`Equation 3 requires that the data follow a linear relationship with re-
`spect to the logarithms of permeability and partitioning coefficients.
`However, Fig. 2 shows a linear increase in log (P,,,,)
`but begins to level
`off‘ with log Pc‘ for cort.exolone, desoxycorticosterone, and dexamethasone
`acetate. As discussed by Sinkula and Yalkowsky (241, the increase in
`permeability cannot go on indefinitely with an increase in the hydro-
`phobicity of the homolog series; several reasons were given for the decline.
`Strictly speaking, progesterone and testosterone do not belong to the
`corticosteroid class of compounds; they lack a 17n-hydroxyl group.
`Therefore. if they are deleted from Fig. 2, the kinetic model best describes
`the remaining data.
`The parabola in Fig. 2 identifies a specific optimal partition coefficient
`for steroids. expressed as log PO, and predicts that a decreased intrinsic
`corneal penetration results if the optimal partitioning behavior is either
`decreased or increased. The kinetic model (23), which suggests that a
`plateau is a more precise relationship, identifies only a lower limit above
`which an intrinsic opt.imal penetration is theoretically predicted. Re-
`gardless of whether the parabola or the kinetic interpretation best de-
`scribes the results, the data can he useful in the design of optimally per-
`meahle ophthalmic drugs. For practical purposes, optimal penetration
`can he considered to occur when d log (Pperm)/d log P begins to approach
`zero, which in Fig. 2 is represented by a log P of about 2.5-3.0.
`Optimal penetration is desirable because a more rapid penetration rate
`leads to higher peak concentrations and lower quantities lost to naso-
`lacrimal drainage. This condition permits a lower dose to be administered
`without sacrificing drug activity and promotes a lowered potential for
`systemic side effects. Even though t,he kinetic model predicts a plateau,
`ophthalmic bioavailability would not always be expected to follow the
`same relationship with an increase in log P. According to Eq. 2, the
`penetration rate is a function of the drug concentration as well as the
`partition coefficient. However, as molecular modification produces a more
`hydrophobic analog, aqueous solubility decreases. Also, the residence
`time of nondissolved drug particles in the eye is limited; therefore, ex-
`pulsion of‘the particles by the eye may take place before solubilization
`may occur.
`
`700 I Journal of Pharmaceutical Sciences
`
`Under these conditions, the partition coefficient and solubility would
`tend to cancel one another in terms of increased penetration. Conse-
`quently, improved ophthalmic bioavailability would reach an upper limit
`or perhaps decrease if a low enough tear solubility is produced. If the
`parent drug is very soluble, however, then molecular modification may
`reduce solubility as partitioning is improved. But because the adminis-
`tered therapeutic dose never approaches its tear solubility, bioavailability
`is enhanced. Dipivalylepinephrine represents an example of this type.
`The data generated in this study showed that the addition of an acetate
`functional group to the 21-hydroxy position of prednisolone and dexa-
`methasone improved the log partitioning by 2 and 1.7 times over each
`parent drug. Based on permeability calculations, prednisolone and
`dexamethasone esters penetrated the excised cornea 7.4 and 8.2 times
`faster than the respective parent drugs. The observed improvement in
`in uitro permeability may not necessarily extrapolate to a significant
`improvement in ophthalmic bioavailability. However, when tested in uiuo,
`a large difference in ophthalmic bioavailability was observed for dexa-
`methasone acetate in comparison to dexamethasone.
`In a study by Kupferman et al. (25), 0.1% petrolatum base ointments
`of I4C-dexamethasone and 14C-dexamethasone acetate were dosed in
`volumes of 50 p1 to normal rabbits. After sacrificing animals at fixed times,
`the drug was assayed in aqueous and corneal tissue samples. The area
`under the corneal dose-time curve was 4.5 times greater for the acetate.
`Aqueous humor levels were detected for the ester but not for the parent
`drug. This latter study established, to a limited extent, a correlation
`between the in uitro corneal permeability model and an improvement
`in in V ~ U O ophthalmic bioavailability.
`
`REFERENCES
`(1) S. S. Chrai, T. F. Patton, A. Mehta, and J. R. Robinson,J. Pharm.
`Sci., 62. 1112 (1973).
`(2) S. S. Chrai and J. R. Robinson, ibid., 63,1218 (1974).
`(3) T. F. Patton and J. R. Robinson, ihid., 64, 1312 (1975).
`(4) D. A. Benedetta, D. 0. Shah, and H. E. Kaufman, Inuest.
`Ophthalrnol., 14,887 (1975).
`(5) A. Hussain and J. E. Truelove, J . Pharm. Sci., 65,1510 (19761.
`(6) M. B. Kaback, S. M. Podos, T. S. Harbin, A. Mandell, and B.
`Becker, Am. J. Ophthalmol., 81,768 (1976).
`(7) D. A. McClure, in “Prodrugs as Novel Drug Delivery Systems,”
`T. Higuchi and V. Stella, Eds., American Chemical Society, Washington,
`D.C., 1975, pp. 224-235.
`(8) M. Katz and Z. J. Shaikh, J . Pharm. Sci., 54,591 (1965).
`(9) B. J. Poulsen, E. Young, V. Coquilla, and M. Katz, ibid., 57,928
`( 1 968).
`(10) E. Stahl, “Thin-layer Chromatography,” Springer-Verlag, New
`York, N.Y., 1965, p. 262.
`(11) S. S. Chrai and J. R. Robinson, Am. J. Ophthalrnol., 77, 735
`(1974).
`(12) D. S. Hull, J. E. Hine, H. F. Edelhauser, and R. A. Hyndiuk, In-
`uest. Ophthalmol., 13,457 (1974).
`(13) H. F. Edelhauser, J. R. Hoffert, and P. 0. Fromm, ihid., 4,290
`(1965).
`(14) 8. E. McCarey, H. F. Edelhauser, and D. L. Van Horn, ihid., 12,
`410 (1973).
`(15) P. Kabasakalian, E. Britt, and M. D. Yudis, J . Pharm. Sci., 55,
`642 (1966).
`(16) C. H. Wang and D. C. Willis, “Radiotracer Methodology in Bio-
`logical Science,” Prentice Hall, Englewood Cliffs, N.J., 1965, pp. 134,
`135.
`117) A. Leo, C. Hansch, and D. Elkins, Chem. Reu., 71,525 (1971).
`(18) C. Hansch, in “Drug Design,” E. J. Ariens, Ed., Academic, New
`York, N.Y., 1975, pp. 284-290.
`(19) B. 0. Hedbys and S. Mishima, Erp. Eye Res., 5,221 (1966).
`(20) D. S. Riggs, “The Mathematical Approach to Physiological
`Problems,” Williams & Wilkins, Baltimore, Md., 1963, p. 185.
`(21) C. Hansch and J. M. Clayton, J. Pharm. Sci., 62,1(1973).
`(22) G. L. Flynn and S. H. Yalkowsky, ibid.. 61,838 (1972).
`(23) S. H. Yalkowsky and G. L. Flynn, ihid., 62,210 (1973).
`(24) A. A. Sinkula and S. H. Yalkowsky, ibid., 64,181 (1975).
`(25) A. Kupferman, M. V. Pratt, K. Suckewer, and H. Leibowitz, Arch.
`Ophthalrnol., 91,373 (1972).
`
`IPR Page 3/3
`
`