The transcorneal permeability of sulfonamide carbonic anhydrase inhibitors and their effect on aqueous humor secretion

The transcorneal permeability of sulfonamide carbonic anhydrase inhibitors and their effect on aqueous humor secretion

The Transcorneal Permeability of Sulfonamide Carbonic Anhydrase Inhibitors and Effect on Aqueous Humor Secretion THOMAS H. MAREN*, LYDIA AND HENRY ...

2MB Sizes 0 Downloads 23 Views

The

Transcorneal Permeability of Sulfonamide Carbonic Anhydrase Inhibitors and Effect on Aqueous Humor Secretion THOMAS

H. MAREN*, LYDIA AND HENRY

JANKOWSKA; GACTAM F. EDELHALTSER

Their

SAXYAL

Depccrtment of Pharmacology and Therapeutics, University oj Flor%da College of Medicine, Gainesville, FE 32610, U.S.A. and Department of Physiology, Medical’College of Wisconsin, Milwaukee, WI: 1J.S.A. (Received 15 October 1981 and accepted 1 December

1982, New York)

Eleven sulfonamide carbonic anhydrase inhibitors of varied chemical and physical

types were studied with respect, to transcorneal permeability and reduction of intraocular flow and pressure. Using the isolated rabbit cornea, a constant drug concentration on the epithelial side and 6 ml solution in the endothelial chamber, first order rate constants (k,,) ranged from 0.1-40 x 10e3/hr, roughly proportional to their lipid solubility. Drugs on the high side of this range were generally water insoluble and had phg’s too high to yield sodium salts at useful pH: therefore, the actual amount of drug delivered was small. We sought compounds which combined low pK,, good lipid solubility, and high activit,y against the enzyme. Trifluormethazolamide (TFM) has a pKa of 6.6, ether partition coefficient of 6, and a K, of 2 x lo-@ M. ki, is 3 x 10m3/hr. TFM and five other compounds were also studied in viva for their ability to penetrate the eye into the anterior and posterior chambers. These rate constants were roughly proportional to those measured in vitro; however: significant differences in accession to the two chambers were observed. as a function of varying physico-chemical properties of the drugs. A 3 o/0 solution of TFM (100 mx) applied to the rabbit eye for 25 min generated 67 mM in the anterior chamber and 0.07 mw in the posterior. Tissue distribution of TFM (and its metabolite) showed a relativeiy high concentrat,ion in the ciliary body 6 hr after dose. Intraocular pressure was reduced by 4 mmHg. Wit,h 10 min exposure this concentration of TFM reduced pressure by about 1.7 mmHg. Although the use of this drug is limited by its chemical instability and the length of exposure needed. the principle of treating glaucoma by the topical use of carbonic anhydrase inhibitors appears feasible. Key worcls: su!fonamides; glaucoma.; intraocular presswe: cornea1 permeability.

1. Introduction Since the introduction of parenteral carbonic anhydrase inhibitors for the treatment of glaucoma, the question has been asked repeatedly whether topical administration of sulfonamides also could reduce aqueous humor formation. In his first paper on acetazolamide Becker (1955) addresses this point, citing negative results and suggesting r,hat the drug may not penetrate to the secretory site. Two other studies gave conflicting results (Foss, 1955; de Feo, Piccinelli, Putzolu and Silvestrini; 1975), but adequate controls were not included or the time course of intraocular pressure (IOP) measured. Because of its much great’er lipid solubility, we tried (with Dr Norman Ballin, cited in Maren, 1967) many years ago to use topical ethoxzolamide dissolved in peanut oil. However, it also failed to lower IOP in rabbit or man. Recently, we and Podos (personal communication, and see below) obtained negative results with methazolamide on IOP in rabbits when given at the limit of its solubility in a neutral solution. We report initial studies on the transcorneal penetration of 11 carbonic anhydrase mhibitors, and show how this function and the basic physicochemical properties of * To whom UOl4-4835/83/040457+23

16

s03.00/0

correspondence

should

be addressed

0 1983 Academic

Press Inc. (London)

Limited

RER36

T. H. HAREN

458

ET AL.

the drugs relate to the lowering of aqueous secretion and IOP by the topical route. The problem may be introduced by an analysis of the properties of acetazolamide and related drugs in this context. In the ideal condition, a constant and maximum concentration of drug on the cornea is maintained. That value for acetazolamide (Gout) is given by the solubility, 8 mM at pH 7% The loss of drug from the aqueous humor will proceed by bulk flow, the rate constant for which (k,) is about l/hrl. Entry of drug to the aqueous occurs by transcorneal diffusion, which we have newly measured in vitro and in vivo (see below) and for which Ici, = 2 x 10e3/hr in the living rabbit. The equilibrium concentration in anterior aqueous (C,) will then be c

a

=

‘in

(Gout) 4%

= 16 ,UM,

(1)

AS will become evident, this will generate considerably lower (by 5-h) concentrations in the posterior chamber (Ch) ; in any case equation 1 gives an ideal maximal figure since equilibrium is not achieved. We have shown previously that to lower aqueous humor secretion substantially at least 8 ,/AM free drug (methazolamide or acetazolamide or any drug with K, % lo-’ M against carbonic anhydrase) is required at the ciliary process (Ma,ren, Hayword, Chapman a,nd Zimmerman, 1977). Under these conditions it would be impossible for acetazolamide to lower IOP, even if exposed idefinitely to the cornea. Similar calculations for ethoxzolamide2 and methazolamide, accounting for their different solubilities and permeabilities (see below), yield similar results. We shall return to some of these issues in the Discussion. It is evident that the solubility of these drugs (all weak acids) can be increased by converting them to their sodium salts (at pH 7-10); this is particularly true for acetazolamide which has two pK,‘s, 7.4 and 9.1. The solubilit,y at pH 8.6 is 76 rnM, but application of such a solution to the cornea for 25 min generated only 30 ,UM in the anterior chamber and did not lower pressure. Even at pH 9 (10 y. solution, 0.45 M) no pressure drop was observed in normal rabbits when drops were given every 5 min x 5, although there was some blunting of the pressure rise in water loaded animals (Stein et al., 1983). Methazolamide at pH 8.4 also failed, as did ethoxzolamide. In the case of the latter drug, delivery was severely limited by low water solubility, despite very high activity against the enzyme and very favorable lipid solubility. As noted above, application in a lipid phase did not work; a not surprising result since the drug must diffuse through a predominantly aqueous environment. Accordingly, we sought new sulfonamide carbonic anhydrase inhibitors with different properties from those now in parenteral use. Our pilot studies suggested lipid solubility as a guide to cornea1 penetrability, and pKa as a guide to aqueous solubility at physiological pH. It became apparent that the cornea1 permeability rate constants varied by some 400.fold among the compounds tested, and that, aqueous solubility could be increased greatly by lowering pK,. The result is to increase both cornea1 permeability and drug solubility (terms of the numerator in equation 1) so that a calculated steady-state C, at the same pH (7%) increases to 1200 ,U~Mfor trifluor1 If flow is reduced by the drug, k, will be half this value. Thus, from equation 1, Gi, will be twice the value here in Table I. See also footnotes to Table III for effect of reduction in k, on effective residence times. 2 Less ethoxzolamide than 8 ,UM should be necessary, since its K, = - 1OF’ IM. However. the present data suggest that only about 1 PM-ethoxzolamide reaches the posterior chamber in the experiments shown.

‘TOPICAL methazolamide

as compared

of certain

of the

AND

EYE

469

PRESSURE

of 16 PM for acetazolamide. This of IOP and aqueous flow following

to that

leads to reduction

penetrability tration

SULFONAMIDES

greatly

topical

increased

adminis-

drugs.

2. Materials

and Methods,

Dmgs The sources of the various compounds are given in Table 1. For in vitro studies of transcornea! penetration they were dissolved directly (as the parent acid) in the buffered glutathione bicarbonate Ringers solution (see below). For in vivo work. the compounds were usually prepared by adding NaOH to the acid form to reach the desired pH. Properties of’many of these compounds are given by Maren (1967). Compounds 8 and 9 are new and were made by addition of the appropriate trihaloacetic anhydride to compound 6, which along with compound 7 (methazolamide) is described by Young, Wood, Eichler, Vaughan and Anderson (1956). Compound 8 was made in our laboratory by Mrs Ruth Palenik, and by Mr Dennis Vogel at the Frick Laboratory at Princeton. Compound 9 was also made by Vogel. In his procedure the trihaloacetic anhydride in the corresponding trihaloacetic acid was added to compound 6, and the resulting reaction mixture left overnight or briefly heated. chemical

and

enzymic

procedures

The pK, of the drugs were determined by titration with NaOH. Solvent/buffer partition ratios were determined by shaking the drug dissolved in phosphate buffer at, pH 7.2 with et,her or chloroform. Equilibrium was achieved within 10 min. Drug in both phases was analyzed by the method of Maren, Ash and Bailey (1954) ; recovery was essentially complete in all cases. Drug concentrations in blood, aqueous humor and tissues were analyzed in the same way. The inhibitory activity (K,) of each drug against carbonic anhydrase was assayed using the same method, except in barbital (pH = 7.9,50 mM) buffer; for this test the enzyme used was pure human red cell carbonic anhydrase C. T’runscorneal

penetration

of drugs

This was studied in vitro by the method of O’Brien a,nd Edelhauser (1977), using excised rabbit corneas with epithelium either intact or scraped off. Solutions used and details of techniques are given in that paper. The pH was 7.6 and exposed area was 1.2 cm2 in the rabbit and 3.2 cm2 in cat. Concentrations of drug of 40-2000 ,ul~ w-ere placed in the epithelial chamber. and samples of fluid were collect’ed from the endothelial chamber at 30.min intervals up to 4 hr. Both chambers contained 6 ml. Drug was analyzed by the method of Maren et al. (1954). The rate of appearance of drug (in ,&r/hr) divided by the concentration in the epithelial chamber yielded a first order rate constant (k,,) for each drug. An absolute permeability constant (P) may be obtained by taking into account the volume (6 ml) of the endothelial chamber and the cornea1 area. The flux into the endothelial chamber is J = mmol/hr/cmz and Since

P = J/mmol/cm3 ki,

=

epithelial

mmol/cm3/hr

P = ki,

in endot*helial

mmol/cma

of epithelial

endothelial

The concentration of drug in the aqueous rabbit was also studied. The same procedures and ‘Intra-ocular pressure’. The first order

solution.

chamber cornea1

solution solution

volume

area

humor following topical application in the intact were used as described below under ‘Sampling’ rate constant, ki,, was calculated by dividing the

460

T. N. NARES

ET

AL.

rate of a,ppearance in anterior aqueous by the concentration applied to t,he cornea and measured at the end of the exposure time (Gout). These times were in the linear portion of the accession curve usually 1625 min: for 90% equilibrium between C,,, and Cg, time is 7f.M80 min when k, is 22@5/hr, the limits observed depending if drug exited by diffusion plus flow, or by flow alone with rate cut in half by carbonic anhydrase inhibition (cf. equation 4 below). The k,, obtained from the in vivo experiments theoretically should be 24 times that in vitro, the ratio of the in vitro chamber (6 ml) to the anterior aqueous volume (0.25 ml). Ideally P is the same in vitro as in vivo since this defines the permeability characteristics of the cornea to a given drug. It will be shown however that P measured in vivo falls below that in vitro, and the probable reasons put forward. Aqueous

humor

flow

This was studied in anesthetized cats by the method of Oppelt (1957). Solutions of drug ( - 95 ml) were applied topically to corneas in various concentrations, and at times as noted under Results. Cats were immobilized in the stereotaxic apparatus, with heads upright. The lower lid was held up with hemostats and forceps making a little pocket into which solution of the drug, or phosphate buffer as a control, were placed. Great care was taken to cover the whole cornea with solutions. The contralateral eye served as control and received saline or phosphate buffer. Particular care was taken to check for leaks around the infusion needles. This was recognized by application of fluorescein to the cornea; or marked discrepancies in the data which would yield an abnormally high and spurious flow rate. Such experiments were discarded. Sampling

of jiuids

Rabbits were anesthetized with intraperitoneal pentobarbital (30-60 “g/kg). They then were placed on their sides and one eye was exposed to the drug in the following manner: The lids were opened widely with hemostats, and a small toothed forceps was used to immobilize the eyeball. The entire cornea was covered with 05 ml of the drug solution. At, given times after steady state exposure to such solutions, they were washed off with running water or saline. (In many experiments, 95 ml of fresh saline then was applied to the cornea, and after several minutes, was withdrawn and analyzed for drug. In virtually all cases, results were negative.) The 26-gauge tip of a 5 cm needle attached to a 50 or 100 $1 Hamilton syringe was placed in the posterior chamber 1 mm below the eorneo-scleral junction. The tip was visible through the iris. The sample was withdrawn slowly with extreme care to avoid trauma. The Hamilton unit was replaced with a 1 ml tuberculin syringe and 27 gauge needle, which was placed in the anterior chamber t’hrough the cornea. About 200 ,~l of aqueous was then withdra,wn. Innfraocular

pressure

White New Zealand rabbits of either sex (2.536 kg) were used. They were given urethane solution (666 g/kg) in a marginal ear vein. When necessa,ry, one drop of @.!I’/~ proparacaine HCl was used. Pressure was measured using a Digilab Model 30 Pneuma-Tonometer. A well was made in the cul-de-sac by extending the eyelids, and the drug solution (0.5 ml) was placed on the cornea as described above. Various exposure times (10-60 min) were used. The solution was then washed from the cul-de-sac with large volumes of warm water or saline. The control eye was washed in the same manner. The anesthetic wore off after l-2 hr, so that subsequent measurements were made only with the local application of proparacaine. no general anesthesia was used. One drop of In a few experiments lasting 10 min, proparacaine was placed on the cornea. In those cases, animals were calmed and put into restraining boxes. They were lying down on one side with one eye exposed to the drug. The lids were opened widely and the fur surrounding them was held gently with hemostats to keep eyes open wide throughout the exposure. The whole cornea was covered with solution. Responses to the drugs (notably trifluormethazolamide) were the same as in the procedure given above and are not reported separately.

Solubibit~,

pR, lipid

partititm, structure and in vitro cornea1 permea,bility carbonic anhydrase inhibitors Partition coefficient at pH 7%

Water solubilitv of free acid Type

Acetazolamide

(2)

(3)

I&* (4)

0.7 0.3 @l 0.3+ 0.47

3.2 1.1 0.4 1.0 1.2

811

(1)

InM

Ether __ Buffer

CHCl, __ Buffer

of 11

isi, x 103/hr Cornea Intact pH 7.6

KI x IOS vs.

No epithelium

c.a. (M)

(8)

(91

(5)

16)

7.4 7.4 7.1 6.0 3.4

0.14 46 5.5 0.03 10-3

10-S 0.05 0.03 3x 10-4 10-h

0.37 0.40 043 oc29: 0.1

7 7 6 7f a

5 8 0.24

7.7 7.4 6.6 7.0

0.12 0.6 6 56

0,014 0.06 @3 4

056 1.9 2.8 8

12 13 13 28

0.01

0.04

8.0

140

25

40

28

0.1

@l

04

7.3

25

10

13

26

101

(7)

type

N--N M II II R--N-~, ,X--SO~NH~ S K CH,CO acetalzolamide iso C,H,CO phenyi-CO BrCH,CO 5. phenyl SO, I. 2. 3. 4.

Wethazoiamide

R--N=k‘ , 6. 5. 8. 9. 0.

! 1 @2 2 0.2

type

,&OINH, S

H H CH,CO methazokmide CF,CO cc1,co Ethoxzolamide

27 1.21 2.3 0,081’

II7

10 1 2 2

:*M,Q C-SO,NH,

1. Chlorzolamide

N-N '1

,S,C-SOzNH2

Sources of compounds: iaboratory and Chemistry * For dibasic compounds :s 8-9. For compounds + 50 g/l as Sa salt at 1 In cat cornea 1.5 x 1 40 g/l as Sa. salt at 11 12 g/i as Na salt at

1, 5, 6, 7: 11, Lederle Laboratories. Department, Princeton University. 1-5 t’his is the first macroscopic e-1 1: pK, is for SO&H, + $O,NH-. pH 7+%+0. lO+/hr for intact and 17 x 10-3/hr pH &O. pH 8.4.

2. 3. Alcon Laboratories. 10 from Upjohn Co. pK, obtained by titration,

without

epithelium.

4, 8, 9, this The

second

pK,

462

T. H. MARES

ET

AL.

3. Results A. Chemistry

and permeability

Eleven sulfonamide carbonic anhydrase inhibitors of different structural types were studied with respect to certain chemical properties, potency against the enzyme and cornea1 permeability in vitro (Table I). Six of the compounds were studied in viva for transcorneal permeability into the anterior chamber (Table II). Additionally, data on these six compounds at the maximum physiological pH (8.3) were analyzed with respect to their equilibrium ratio CJC,,, and the cornea1 residence times required to generate two levels of C, (Table III). For two of the compounds, tissue distribution was studied (Tables IV and V). The compounds will be considered in turn, since the findings yield the underlying pharmacokinetics for selection of drugs to be tried topically for reduction of IOP. (1) Acetazolamide. This and the related structures 2-5 are characterized by a 1,3,4-thiadiazole ring substituted with a free -SO,NH, group at the 5 position (Table I). The -SO,NH, moiety confers carbonic anhydrase inhibitory activity (Maren, 1967). Additionally and importantly in the present context, the 2 position is linked to an-NH-group conferring dibasicity. The relationship between the ionization of this proton and that of a -SO,NH, proton has been discussed (Lindskog, 1969). Acetazolamide is poorly soluble in water but, its 2 pK,‘s are such that it can form a 10 y0 solution at pH 9 and a 17 y0 solution at pH 8.6. Ether and CHCl, partition coefficients are low. Column 7 shows cornea1 permeability is also low, ki, = 037 x 10e3 hr in vitro. When the cornea1 epithelium is removed, permeability through the denuded cornea increases 20-fold. As seen in Table I, this is typical of drugs with low lipid solubility. Table II shows that the transcorneal rate constant at pH 7.8 in vivo is 2-O x 10e3/hr,

TABLE

Transcorneal

permeability

in

II

vivo of six sulfonamides Accession rate

Compound 1 4 7 8 9 10

my applied pH 7.5-7.8 8 130 25 120 99 ti72* 0.06 0045*

to anterior chamber (w/hr) 16 15G 200 1680

kin x 103/hr

2 1.2 8 14

56

78

15

330

From experiments in which drugs were applied to the cornea for 60 min) and the anterior chamber sampled a,t the end of this concentrations applied to the cornea remained st,eady. The accession for each compound with S.E.M about 25 %. * The concentrations noted reflect a decrease from those applied (5 min) of the experimental period. For both these compounds the surface is about 10 min.

let-25 min (except for compound 1. time. Except as noted below. the rates are means of 5-10 experiments and are measured at the mid-point half-time for deca.y from the cornea1

TOPICAL

SULFONAMIDES

AK’D

i ’I

EYE

PRESSURE

163

464

T. M. MAREN

ET

AL

2,5-times that in vitro3. Accession rate to the anterior chamber is relatively low at pH 7.8; c, is 16 ,UM after 1 hr of exposure and C, was < 1 FM. Accession increases as pH rises, due to increased solubility (Table III). S’mce acetazolamide is already largely ionized (72 %) at pH 7.8, increasing pH does not greatly change the charge relations of the molecule and thus ki, drops only 30% ( compare Tables II and III). As noted in the Introduction, acetazolamide does not lower pressure when applied in high concentrations. Table III shows why, since even the theoretical equilibrium concentration in the anterior chamber is 53 PM, and this does not generate enough drug posteriorly to inhibit enzyme at the physiological level. (2) 2-isopentenyl amino 1,3,4-thiadiazole-5-sulfonamide. This is a higher homologue of acetazolamide which predictably has the same pK,, and inhibitory activity, lower water solubility, and considerably higher lipid solubility. Despite this last feature, its cornea1 permeability was no greater than that of acetazolamide. This indicates that another factor is involved in cornea1 permeability, and consideration of combined dat’a of Table I suggests that the dissociable PITH proton on the 2 position retards permeability. As will be evident, all five compounds (l-5) of the acetazolamide type which have this component have iow permeability constants; no matter what other alterations there are on the molecule. (3) 2-benzoylamino-1,3,4-thiadiazole-5-sulfonnmide. Replacement of the methyl group in acetazolamide with pnenyl leads to major changes. In columns 2, 4, and 5 of Table I we see the decrease in aqueous solubility, but marked increase in lipid solubility compared to compound 1. In compound 3, pK,, decreases since phenyl is an electron-withdrawing substituent and promotes ionization of the KH proton. A 1 oh solution of compound 3 at pH 8.2 may be made, so that this compound would probably be active topically. Activity against carbonic anhydrase is × greater than acetazolamide. However, cornea1 penetration, like that of compound 2. was not superior to acetazolamide despite much better ether and chloroform partition. This is further evidence that ionization of the NH group retards permeability. (4) Bromacetazolamide. Changing the CH, group of acetazolamide to CH,Br did not increase the lipid solubility. However, there is a marked drop inpK,, so that a soluble sodium salt of bromacetazolamide is achieved at pH 7.4. The ki, is slightly reduced in vitro due to greater ionization of KH and the in vivo ki, is 1.2 x 10-3/hr. The net result yielding C, in equation I is 240 ,UM, much higher than other drugs mentioned thus far. This is the result of setting C,,, = 200 rnw, a 6 o/0 solution (Table III). We show later that local administration of bromacetazolamide under the artificial circumstance of 1 h cornea1 application does lower aqueous pressure and flow ~ the first topical sulfonamide reported to do so. (5) Benzolamide. This drug has been extensively studied as a specific renal carbonic anhydrase inhibitor since, like p-aminohippurate, it is secreted and accumulated by renal tubules (Maren, 1967). Unlike drugs 1-4, the NH moiety is linked to SO, (rather than CO) ; this greatly enhances ionization so that the pKal drops to 3.4. Table I shows that lipid solubility and cornea1 permeability are extemely low, as might be predicted. Although C,,, is made high by using the soluble sodium salt at 5 %, even at, equilibrium 3 As noted under Methods, the theoretical ratio of in viva/in vitro ki, is 24: the volume ratio of the endothelial chambers. This would be the case if I’, the cornea1 permeability constant of a given drug: is the same in vitro as in viva; see Materials and Methods section. Experimentally the ratio of the two rate constants is less than 24, ranging from 4 to 10 (compare Tables I and II). The Sam relation is observed when p is calculated. We attribute this difference to two factors; (a) lack of mixing in the ill viva experiment and (b) sequestration of the relatively hydrophilic drugs (1, 4. 7, 8) in the stroma at the early time periods used in viva. Xote that these compounds have the lower in viva/in vitro ki, ratios, and the lipid soluble pair 9 and 10 have the higher ratios.

TOPICAL

SULFOSAMIDES

AND

EYE

PRESSURE

465

C, is low, about 15 ,UM, from equation 1. The data are given as an example of the effect of pKitl on Ic,,. Note also that when the epithelium is removed, the k,, of this ionized lipophobic drug rises 70-fold. Clearly, the barrier is at this site. The data reveal that the observed penetration rate is due to the ionized form of the drug. At pH 7.6 only 1 part in 16000 is non-ionized, and for this to account for the observed rate constant (kin) of 1 x 10-4/hr (Table I), the ki, (non-ionized) would be i.B/hr. This is most unlikely since the corresponding value for the most permeant and largely undissociated member of this series, compound 10, is 40-times less. This was investigated (6) 5-imino-4 methyl-a-1,3.4-th’ za d iazoline-Z-sulfonamide. because it is the chemical precursor of the acyl compounds 7-9 (Young et al., 1956) and their hydrolysis product. Its K, of 1O-7 M is higher than its acyl derivatives; thus it resembles precisely its analog des-acyl acetazolamide (CL 5343, Maren, 1967). Lipid solubility and cornea1 penetration are poor. However, the tissue concentration of this compound that’ may be partly responsible for the ocular effect following cornea1 application of trifluormethazolamide (see Section 8). (7) Methazolamide. The general pharmacology of this drug has been described (Maren et al., 1977). In the present context it has favorable properties of lipid

TABLE

Distribution

of methazolamide

rabbit eye following (pH 7.9) for 25 min*

in

Time Tissue Cornea II+

Ciliary body Lens Sclera Retina Vitreous Anterior aqueous Post aqueous RBC Plasma

0 26 1 22.4 3% 12.7 41 2.4 0.5 11.9 3.1 8.6 0.5 1.0 0.6 96 17 23 5 32 4 0.5 0.1

IV

2 19 @7 5.0 1.5 9.0 1.9 2.1 1.5 3.3 0.7 7.0 15 0.2 @l 29 5.0 5.0 1.0 31 4.0 0.3 0.1

after

application,

cornea1

application

of 25 ,rnM

(hr)

6

18

54

3.8 o-8 9.2 3.0 7.0 1.7 2.0 1.0 2.6 0.8 5.4 1.0 0.4 0.1 1,2 0.3 2.27 0.3 30 3.0 0.6 0.2

$5 2.2 3.2 1.5 9.0 4.0 0.6 0.6 5.4 1.4 1.8 @6

1.1 0.3 1.0 @3 1.3 0.4 < 0.3

1.1

1.3 1.1 1.4 o-5 < 0.4

0.6 < @4

<04

< 04

i 0.4

28 5.0 0.6 0.1

26 2.0 < 0.4

* Data are means calculated from four to eight measurements fs.~. (/cmol/kg or 1). ’ Means calculated from two or three measurements: the third was below the limit of the method, 04 ,lLM. In each case the numbers given are for the treated eye minus the opposite eye. The latter number includes a small blank for t.he method as applied to tissues, and the drug content of the red cells in some of the tissues. The opposite eye values never exceed 20 o/o of the treated eye. The red cell data show that 2 pmol of drug were absorbed into the body. This is 16 y0 of the dose applied (0.5 ml of 25 miwsolution).

466

T.H.MAREB

ET AL.

solubility and cornea1 penetration, which results from eliminating the -xH-pr&on from the C-2 position of the acetazolamide series. The first proton titrated on this molecule is one of the pair on - SO,NH,, so the pK, is that of the &sulfonamide group itself. This is an important distinction between the methazolamide and acetazolamide series; in the latter both the sulfonamide and the - NH protons are titrated. The ether partition and cornea1 permeability in vitro are 4-5 times greater than for acetazolamide. Water solubility is twice that of acetazolamide, 11-60 mM at pH 7.5-8.4. From a saturated solution of pH 8.3, the equilibrium concentration in anterior chamber is 180 pM (Table III). Table IV shows the observed values for 25 min of cornea1 exposure to be 96 ,UM and 23 ,uM for anterior and posterior aqueous, respectively. Table IVshows the distribution of methazolamide in ocular tissues for 54 hr after topical application. Perhaps the most interesting finding is that drug persists in the ciliary body for 6 hr. This is not bound to carbonic anhydrase, since enzyme concentration in this tissue is only 0.3 ,uM (Maren, 1967). The concentration found (about 9 ,UU~M) if all were free to react with carbonic anhydrase is borderline for pressure lowering; 8-16 ,UM free in plasma or posterior aqueous is required (Maren et al., 1977). We found no effect on pressure in the regimen of Table IV; S. Podos of Mount Sinai Medical Center gave one drop of 15 rnM methazolamide every 5 min for 1 hr in the rabbit eye and obtained no effect (pers. comm.). In the cat a modest decrease in aqueous flow occurred after 1 hr of cornea1 exposure of 25 mM-drug (not shown). Attention was then turned to compounds in this series which would retain the desirable lipophilic properties but also yield a high C,,, in the form of soluble sodium salt. (8) TrijCuormethazoZamide. The conversion of CH,CO- to CF,CO- in methazolamide provided sufficient electron-withdrawing power to decrease the pK, of the 5SO,NH, group by almost one unit. The overall decrease in pK, from that of sulfanilamide and other homocyclic sulfonamides (pK, = 10-l 1; Northey, 1948) is enormous and results from combined electron-withdrawing effects of the heterocyclic ring and fluorine atoms. Compound 8 appears to have the lowest ph& (6.6) of any known sulfonamide for the group R-SO,NH,. Table I shows that compound 8 has good properties of lipid and water solubility, cornea1 penetration, and enzyme inhibition. It forms a soluble sodium salt (4%) at pH 7.8. Table II shows the high rat.e of passage of this drug into the anterior chamber in vivo. Table III shows the high equilibrium concentration in the anterior chamber and that 4 min residence time on the cornea should generate 200 ,uM inside. We show below that application of a 3 y0 (100 mM) solution to the cornea reduces IOP and aqueous flow. However, there is a drawback in that the compound hydrolyzes spontaneously to compound 6 and trifluoracetate with a half-time of about 1 hr at pH 7.8 and 25°C. This has little effect on the permeability studies since only early times are used, i.e. less than 25 min. The solutions are analyzed in terms of compound 6, since the hydrolysis is complete by the time the analysis is carried out. There is evidence that it is largely compound 8 itself that crosses the cornea in vivo. Samples of anterior aqueous were withdrawn after 10 and 25 min of cornea1 exposure and their carbonic anhydrase inhibitory activity measured. They were then heated or allowed to stand overnight, whereupon this activity decreased five-fold showing that the conversion to compound 6 occurred principally after the sample was taken. Additionally, solutions withdrawn from the cornea1 surface at these times have (within 50 %) the inhibitory activity of the original compound. The transcorneal rates in vivo

TOPICBL

SCLFOSAMIDES

AKD

TABLE

Distribution

of

PRESSPRE

tri$uormethazolamide i,n rabbit eye following 100 rn*M (pH 7.8) for 25 min

Cornea Iris Ciliary body Lens Sclera Retina Vitreous Anterior aqueous Posterior aqueous Red cells Plasma

467

V

~rnOl/l

Data are * Analyzed is compound

EYE

cornea1 application

of

or kg*

0 time

2 hr

6 hr

736 70 26 7 56 <6 19 682 70 16 <5

80 19 55 5 19 24 w 10: 27 39 15

9.4 7.9 10.4 1.2 7 7.4 1.2 2 12 26 <5

means from 3rlO animals for each tissue or time. for total drug: TFM (Compound 8) and its metabolite compound 6. At 2 hr about 80% 6. See Text. At 18 hr, fluids were < 2 ,uM and tissues < 5-10 q, except for red cells, 24 ,ux.

are in line with those of compound 8 in vitro, but too high for compound 6 (see Table I). However, the 2 hr tissue samples (Table V) contain a combination of compounds 8 and 6, with the latter predominating. Since compound 6 is 15as active against the enzyme, calculations of fractional inhibition at 2 hr and past must take ‘this into account. Note from Table V that, like methazolamide, compounds 8 and/or 6 appear to be retained in the ciliary body at least for 6 hr. We return to this point below in connection with estimates of fractional inhibition. (9) Trichlormethazolamide. This has similar properties to its trifluorine analogue, but differs by being much less water soluble and more lipid soluble. The rate constanbs for cornea1 penetration are high (Tables I and II). However, the low water solubility of the free acid dictates that 99 y0 must be converted to the sodium salt to yield a 1 y0 solution at a pH of 855. Such alkaline solutions are irritating and cause an initial rise in IOP. This may be avoided by decreasing the pH to 82 and concentration to 95 “//o (15 mM). Tables II and III show how the low water solubility of this compound compromises effectiveness. Trichlormethazolamide is similar to compound 8 in hyd.rolysing to compound 6, but in this case with a half-time of about 6 hr. We feel that this limits its potential, in addition to the unfavorable balance betweenph’, and water solubilitp ofthe free acid. (10) Ethoxzolamide. This drug has been used for many years in the parenteral treatment of glaucoma. It is different from acetazolamide or methazolamide in structure and properties (Maren, 1967). It is exceedingly lipid soluble and permeant through the cornea - IOO-times greater than acetazolamide (Tables I and II). However, its water solubility is very low and pK, high (Table I), therefore the maximum usable concentration (Gout) is 0.12 mM at pH 8.3. The theoretical equilibrium concentration (Ca) is 14 ,LLM (Table III). However, this cannot be approached because Gout declines rapidly, a property of highly lipid-soluble drugs; see footnote to Table II. After cornea1 application to four rabbits (one eye each) of 0.1 mm-solution (pH 8.2)

468

T. H. MARES

ET

AL.

for 13 min, a.nd the mean Gout for the period measured is 70 ,UM, C’& was 2.4 ,UM (I S.E. 1.5) and C, 1.6 ,UX ( f S.E. 64). The outside solution was removed and 50 min later c, was 0.9 ,UM and C, was 0.6 ,UUM.There was no pressure lowering ; these concentrations are small for a pharmacological effect even for a compound of such high activity against the enzyme. Note, however, the very high ratio of C&/C, compared with less lipid soluble drugs; we return to this question in Section B. The ocular pharmacology of this compound (and chlorzolamide, see below) is quite different from those of compounds 1-9, because of far greater lipid solubility. As noted Analyses of the drug concentration in red above, Guta C, and C, all decline rapidly. cells suggest that Gout declines because of rapid uptake into the conjunctival vessels ; thus less drug is available for the eye. Additionally, C, declines rapidly (k, = 1.7/hr), about twice the rate of bulk flow, indicating a mode of exit by diffusion. Thus there are four components mitigating against this compound: low water solubility of the acid, high pK,, rapid loss from conjunctiva into the blood, and rapid decay from the chambers. (11) Chlorzolamide (CL 13580). Although the structure is quite different, compound 11 resembles ethoxzolamide in chemical and physical properties (Table I). Its high lipophilicity promotes cornea1 penetration in vitro (Table I) and in vivo (ki, = 0*3/hr; not shown), but low water solubility makes it impossible to achieve high Co,, with this drug. [It may be noted that Maren’s review (1967) contains the serious error of giving the pK, of chlorzolamide (CL 13580) as 6.6 rather than 7.3.1 The same factors analyzed above for ethoxzolamide apply here, except that with chlorzolamide its lower pK, and higher water solubility make possible the use of a 2 mM solution at pH 8.0. The half-life of drug applied to the cornea (Gout) was 8 min, and k, is 2.5/hr, much greater than bulk flow. Negative results were obtained on intra-ocular pressure measured over 4 hr when 2 rnM wa’s applied for 10 min to the rabbit cornea and then washed off (n = 6); in this protocol C, was 98 ,uM ( f S.E. = 14 ,UM) at the end of application and 13 ,uM (k S.E. = 7 pM) i hr later. It is interesting to observe that for the highly lipophilic compounds 10 and 11, there is little difference in penetration with or without the epithelial layer, whereas with lipophobic compounds (e.g. No. 5) the difference is as great as 70.fold.

FIG. 1. Appearance and decay of sulfa&amide rabbit cornea for 15 min. From Robson and T&rich

in the eye (1942).

following

application

of 1400

rnM to the

TOPICAL

B. Permeability

SULFOKAMIDES

and distribution

AP;D

EYE

469

PRESSURE

in viva

Figures l-4 show the ocular appearance and decay of four sulfonamides following topical application to the cornea. Figure 1: included as a reference standard, replots data for sulfacetamide (not a carbonic anhydrase inhibitor) obtained over 40 years ago (Robson and Tebrich, 1942) and illustrates some of the principles ancl calculations used. The transcorneal rate constant, Ki,, may be approximated by the concentration in the anterior chamber at zero time, divided by the applied concentration and the time of application. c kin = a (2) c0ut.t. Thus,

from data of Fig. 1

k, = 7 rnMl0.25 In

1400

hr

= O.O20/hr.

rnM

The decay from the anterior chamber yields a half-time of 36 min: or k, = l.l4/hr, rather similar to that obtained for p-aminohippurate and Na+, and taken to be a measure of flow (Kinsey and Barany, 1949). Sulfacetamide has a p&C, of 5.6 (PJorthey, 1948) and is rather lipid insoluble (Rieder, 1963). However, it penetrates the cornea reasonably well in comparison to the other compounds listed in Table I. Because of its solubility as the sodium salt, high concentrations may be applied and relatively large amounts are found in the eye. At l+ hr cornea and iris have about twice the concentration of drug in the aqueous. There is no known receptor for t,his compound in the eye. From the 11 compounds studied in vitro we chose four for exploration of both their ocular pharmacology (present section) and effect upon aqueous humor pressure and flow (Section D). Figure 2 gives our data for bromacetazolamide. As calculated from hour t; for decay from the equation 2, ki, = 00012/hr (Table II). F or the initial anterior chamber is 42 min, yielding a k, of 0.96 hr. Thereupon the rate of e&x slows? and anterior chamber levels do not decline from 1.5 hr. This is a possible consequence

FIG. following

2. Appearance and topical application

decay of bromacetazolamide of 5 y0 (165 mM) to the

in cornea

anterior for 1 hr.

(0) and posterior pH = 7.4.

aqueous

(0)

470

T.H.MdRESETAL

of the drug reducing aqueous flow, combined with continued entrance from cornea1 stroma, which is a property of hydrophilic compounds, such as fluoresein (Maurice, 1980). From Fig. 2 we may also calculate an apparent rate constant (Pi,) for accession to the posterior chamber, while recognizing that the value has no anatomical meaning, since we do not know the contributions of the various pathways, i.e. from anterior chamber or from conjunctiva through the iris root. To compare the several drugs, we use the applied concentration to cornea as a base, dividing this into the rate of appearance in the posterior aqueous, analagous to the equation 2 for anterior aqueous. Thus, from the initial appearance of bromacetazolamide in posterior aqueous at 0 time of Fig. 2, and the concentration used, k, = ~6 Per = 3.6 x 10P5/hr In 165 mM

(3)

This is 0.03 of the corresponding value, I$,, for the anterior chamber. A net decay constant from the posterior chamber, le,, may be calculated from the lower curve of Fig. 2 as @B/hr. Again, since drug is entering the posterior chamber at the same time, this value is only given as a practical means of comparing the several compounds.

FIG.

following

3. Appearance and decay of trichlormethazolamide in anterior (0) and posterior topical application of (05%) 15 rn~ to the cornea for 10 min. pH = 8.3.

aqueous

(0)

Figure 3 yields the accession rate constant ki, for trichlormethazolamide of O.O8/hr. The half-life for decay from the anterior chamber is 48 min, yielding k, of 0.84/hr. This is less than the rate of sulfacetamide exit, probably reflecting the fact that trichlormethazolamide has slowed the rate of formation (Section D). Entry to the posterior chamber, calculated as in equation 3 above, yields klin = 16 x 10e3/hr. This chamber. Net is 0.2 that of the equivalent constant kin for entrance to the anterior decay from the posterior aqueous appears slower than for bromacetazolamide, k, = 0.25/hr. Figure 4 shows data for trifluormethazolamide. Entrance rate constant (kin) is O.O14/hr. The k, is 0+34/hr, the same for trichlormethazolamide. Accession to posterior chamber, k’i, is 1.7 x 10e3/hr, 612 that of k,,. Net loss from the posterior chamber as k, is O%/hr. From the in vivo experiment with ethoxzolamide described in Section A, 10, we ma,y calculate roughly k,, as 0.2/hr and klin as O.l4/hr. Note that the direct comparison

TOPICAL

Fro. following

4. Appearance 3 Y0 (100

and mx)

RCLFOKAMIDES

decay topically

of trifluormethazo;~mide applied to the cornea

AND

for

EYE

471

PRESSURE

in the anterior 25 min. pH

(0) and = 7%

posterior

aqueous

(0)

between these numbers4 demands that the latter be divided by 5. We arrive then at the interesting factor of only 7 for the difference in accession rates to the two chambers. For bromacetazolamide, the least lipid soluble ofthese compounds, the corresponding factor, from the data given above, is 170. The accession rate constants to both chambers increase with lipid solubility among whilst for kfi, it is these four compounds; however for ki, the increase is 165fold, 4000.fold. This suggests that the route to the posterior chamber is in large part through tissue, both through iris from anterior chamber and iris root from conjuctiva. Table IV gives the tissue distribution of methazolamide, which was studied as a stable and prototypic drug of the series. From the low concentrations in the posterior chamber and ciliary body, it is not surprising that there is no effect on aqueous humor dynamics. The persistence of drug in ciliary body for 18 hr is unexplained. It is of interestf that of the 12.5 ymol of methazolamide placed on the eye, 2 pmol or 16 o/0 were absorbed into the body (see footnote to Table IV). Table V shows that following the high initial concentrations oftrifluormethazolamide in the anterior chamber (see also Fig. 4), drug also reaches the iris, ciliary body and posterior chamber in reasonably large concentrations. It may be recalled that this drug is hydrolyzed in vitro to compound 6 with a half-life of about 1 hr; the analysis of Table V does not distinguish between the parent drug and the breakdown product in the tissues. It is possible that trifluormethazolamide remains bound as such to the enzyme. Data are needed on this point, as well as analyses of the ciliary processes themselves. For the present we may roughly calculate as follows: About 8 ,LLMfree methazolamide in contact with ciliary process enzyme is sufficient for the physiological effect (Maren et al., 1977). The present data for trifluormethazolamide shows that this is probably exceeded by a considerable margin; thus even with the lower inhibitory activit’y of compound 6 (Table I), it is likely that in the regimen of Table V, complete physiological inhibition can be achieved. The fractional inhibition (i) is equal to

4 The rate constants direct comparison the chambers.

4, value

and /cl, show the overall permeabilities of kl, must be divided by 5, the ratio

from the outside to each fluid. For of volumes in the anterior to posterior

472

T. H. MAREN

ET

AI,.

where I,,,, is the concentration of unbound drug available to the enzyme, and K, is the in vitro inhibition (dissociation) constant against the enzyme. For compound 6; K, is 0.1 ,u~, so if the concentration in the posterior chamber (27 ,u~) or ciliary body (55 PM) at 2 hr is truly free, i = O-996-0.998. This is borne out by the data on pressure and flow, to follow (Section D). At 18 hr, where there was still an apparent pharmacological effect, the ciliary body concentration was below the limit of detection, which unfortunately is 5-10 ,uM for compound 6 in tissues. Thus i = 0.99 could escape detection; more remains to be done on this important point. Table V illustrates other points of interest. Although the concentration in the vitreous is relatively low, the amount of drug present is quite high (since its volume is six-times that of the anterior chamber), showing that vitreous is to some degree a sink for drug administered topically. Lens concentration is low, in agreement with our earlier finding that only very lipid soluble drugs penetrate the lens (Friedland and Maren, 1981). Concentration of drug in cornea is high, as expected, and initial decay is parallel to that in anterior chamber, as in Fig. 1. Finally, we may ask how much drug was absorbed into the body. In the experiment of Table V the dose administered was 50 pmol, or 15 mg = 5 mg/kg. If this were an intravenous dose, we could expect a plasma concentration of 10-20 ,ug/ml or about 70 ,u~. Less than 5 ,u~ was found. The total amount in the red cells, assuming a 3 y0 red cell volume in a 3 kg rabbit, was 3.5 bhrnol. This represents binding to carbonic anhydrase since 90 o/0 of the enzyme is in red cells (Maren, 1967), and there are no other known receptors. Free drug is negligible as shown by plasma data; thus the sequestration in red cells represents nearly all the drug that has been absorbed, 3.5/50 x 100 or 7 %. This is about half that observed with methazolamide (Table IV). No systemic pharmacological effects are apparent from 7 o/0 of 5 mg/kg = 0.35 mg/kg of acetazolamide; thus none would be expected from topical trifluormethazolamide or compound 6 which has the same or less activity (Table I). C. Theoretical relations in the eye

among

framcorneal

permeability,

residence

times, and

concen-

tration

The purpose of this section is to show for the several drugs the relations among conc,entrations applied to the eye, the time applied and the concentrations generated in the anterior chamber, Although we do not, know precisely the relations between the latter numbers and lowering of flow or pressure, this represents a start, to the In the case of systemic carbonic anhydrase problem of topical administration. inhibition, the relations are between unbound drug in plasma in diffusional equilibrium with posterior aqueous, which in turn is in equilibrium with enzyme in the ciliary process (Maren, 1967; Maren et al., 1977). Following topical administration, the posterior chamber concentration may be lower than free drug in tissue water of the ciliary process (and thus may underestimate inhibition) since it may arrive at the tissue first. At any rate, our data suggest that about 200,uM in the anterior aqueous is necessary for the full pharmacological effect of compounds 4, 8, and 9 (Section D). This agrees with calculations showing > 99 y0 enzyme inhibition (Section B) based on drug concentrations in the posterior segment, when 200 ,L&Mare present, anteriorly. The equations for opposing first order reactions are used t,o treat the data. A useful and clear derivation is given by Davson (1967). The expression

TOPICALSULFOSAMIDESANDEYE is integrat’ed

PRESSURE

453

to yield lncl-&j=-k,.l

ca is the concentration of drug found in the anterior aqueous. OUt is t’he concentration applied to the cornea. e ki, is the first order rate constant for entry into anterior aqueous as shown Table II above. ,$& is the measured rate constant for drug leaving the anterior chamber. c r = 2 at equilibrium, c out

(4)

in

k. also = 2, ka

t = time of drug application to yield C,. Note that these are times preceeding zero time, when drug is washed off. Table 111, column 3 shows the equilibrium Cr, that would be generated when saturated solutions of drugs at pH 8.3 are placed in the cornea. These values are calculated from Ca k, = Gout kin, t’he values of k,? k,, and Gout being known experimentally for each drug. For acetazolamide and methazolamide, whose K, and physico-chemical properties a,re very broadly in the range of compounds 4, 8 and 9. the equilibrium value is < 200 ,KVI, and they are inactive in the topical tests. Ethoxzolamide, because of its lower K, and much higher lipid solubility, and permeability, should be active at C, less than 200,~~~; but the equilibrium value is only 14,u+~, and it is inactive topically. The active drugs 4, 8 and 9 have high equilibrium C,; note also that for trifluormethazolamide 200 ,LLM is only 6 ‘A of the equilibrium concentration (column 4). Table III, column 6 gives the residence times (t) on the cornea required to yield 200 ,uM in the anterior chamber. This is done by using equation 4 where we substitute 200~~~ for C, and experimental values of k,, kin and CcUt. Again the most effective drug is trifluormethazolamide, with t = 3.9 min. At a lower pH (7%) and Gout of 100 ,u,M; as used in physiological studies to follow (also Stein et al., 1983), t = 13.4 min. Column 7 of Table III gives residence times to generate a much lower equilibrium concentration in the anterior chamber, 25 ,UM. Such a concentration would be consistent with pharmacological activity in the case of compounds of high lipid solubility, in which the rate constant for entrance into the posterior aqueous, kli,,, approaches that’ for the anterior, kin. In section B we show that k’Jkin ranges from 0.03 to nearly unity among the four compounds studied. Rearrangement of equation 4 yields the following, so that C& may be calculated for any given cornea1 residence time t: C, = Co,, T( 1 - e&i t, (5) At pH 7% and 100 mM trifluormethazolamide (Gout), equation 5 yields for 1 min of exposure 16.5 yM, and for 15 min 221 ,UM. This portion of the curve relating t and Ci, is nearly linear, since we are far from the equilibrium value of 800 ,uM (C7, = Gout r). We have not defined precisely the effective range oft (or Cg). but data to follow (Section D) suggest that for this drug it would fall between these broad limits, t = l-15 min and C, = 16221 ,UM. It is hoped that this model will be extended to the case where C& is replaced by the concentration of free (unbound) drug in the eiliary processes, Then the direct, relations a,mong drug disposition, enzyme inhibition and pharmacological effect will be apparent.

474

T. H. MAREN

ET

AI,

FIG. 5. The effect of bromacetazolamide (0) and trifluormethazolamide (0) on aqueous flow in the cat. Each drug was topically applied to the cornea for 1 hr during pentobarbital anesthesia. Following this the head was placed in a stereotaxic inskument and needles connected to the infusion system were attached to the eye as described by Oppelt (1967). Anesthesia was continued. Blood pressure was monitored throughout. Average data over a 3-hr period are given. Each point represents the mean difference between control and treated eyes for 9-25 measurements. k S.E. of means vary 6-20 o/0 for control eyes and S-16 o/0 for treated eyes.

D. EJect ox jbw

and pressure ; correlation

with drug concentrations

Figure 5 gives dose-response data on changes in aqueous flow in the cat during topical application of trifluormethazolamide and bromacetazolamide. The control flow averaged 16 pl/min and measurements were made for 2 hr after treatment. Full effects of the two drugs are elicited by 2 y0 (70 ,u~) and 5 y0 (165 PM) respectively. From the shape of the curves, the trifluormethazolamide compound is about 4-times as active as bromacetazolamide, corresponding to a lo-fold margin in cornea1 permeability (Tables I and II). It has been amply shown that complete carbonic anhydrase inhibition reduces aqueous flow by 40-50 o/0 (Oppelt, 1967 ; reviewed by Maren, 1967). In the type of experiment shown in Fig. 5 in cat using 100 mM-trifluormethazolamide for 1 hr, the concentration of drug at 140 min was 9 ,UM in the anterior chamber and 43 PM in the iris-ciliary body sample. As shown above, these numbers are consistent with inhibition of carbonic anhydrase over 99 %. This degree of penetration is far less than seen in rabbit, as evident from comparison with Fig. 4. Our current work with other drugs support the idea that penetration in vivo is much less in cat than in rabbit. Reasons for this remain to be discovered. Maurice and Mishima (1983) have shown that cornea1 permeability for several drugs is some lo-fold higher in man than rabbit. It should be noted that our in vivo data (Table II) are those of cornea-scleral permeability. Figures 6 and 7 show the lowering of intraocular pressure in rabbits by two of the drugs following topical application. The degree of pressure lowering is about the same as that for systemic carbonic anhydrase inhibitors (Becker, 1955; Wistrand, Rawls and Maren, 1960). The bromacetazolamide experiment (Fig. 6) shows a gradual reduction in pressure, consistent with the slow accession of drug to the posterior segment (Fig. 2). With this drug, however, t’here was clouding of the cornea the following day, and about 40% of the treated eyes showed a rise in pressure at zero time. These animals were excluded from the series. It may be recalled that this drug

TOPICAL

SULFONAMIDES

pressure in rabbits following Control (0). Standard errors

t, t 0"

EYE

475

PRESSURE

HO”,S

on

FIG. 6. Intraocular 1 hr (n = 67) (0).

AND

0 '5‘

FIG, 7. Intraocular pressure in rabbits following for 25 min (n = 8) (0). Control (0). The standard

topical administration were 0.5-19 mmHg.

3

2

of bromacetazolamide

4

.i “/b for

'5

HOURS

topical errors

administration of trifluormethazolamide ranged between 04-1.0 mmHg.

3 “/b

has properties of an irreversible alkylating inhibitor (Kandel, Wong, Kandel and Gornald, 1968) unlike the others we are studying. Figure 7 shows the lowering of IOP by trifluormethazolamide. The onset of effect is relatively rapid, consistent with the rate of penetration of drug into the posterior segment (Fig. 4 and Table V). The effect is sustained: note from Table V that at 6 hr there is still an inhibitory concentration of drug in the ciliary body. By 18 hr, drug concentrations were below the limit of the method, which (for compound 6) is 5-10 ,LUN for tissues, and 2 PM for fluids. Surprisingly, however, the treated eyes had consistently low pressure 18-24 hr after treatment (12.9f0.9 vs. 17.7 + 1,1 mmHg in control eyes, Fig. 7.) As noted in Section C, 5-10 ,LhM could still be an effective level; on the other hand, the late effect may be an artifact of handling or minimal damage. More remains to be done on this point. In the type of experiment shown in Fig. 7, some 30 y0 of animals were excluded from the series; these showed pupillary constriction, slight redness, 1-2 points flare under slit lamp and initial rise in pressure. Protein in the aqueous was of the order of 10 mg/ml. These animals showed a later fall in pressure; however they constitute a second population which appear to complicate the picture unnecessari1y.j TriAuormethazolamide, although a useful experimental compound, is not destined for human use and we are not primarly concerned with these reactions. Representative animals ’ This is not a prostaglandin effect. The pressure rise or lowering intraperitoneal indomethacin given 1 hr earlier, in a group of six rabbits

was not examined

blocked by 50 mg/kg by Dr Amir Bar-Ilan.

476

T.H.MARES

ET AL.

used in Fig. 7 were carefully observed and showed no pupillary constriction or redness, no flare under slit lamp, and aqueous protein < 1 mg/ml. Minimum contact time for pressure lowering with 3% trifluormethazolamide is 5-10 min. Following IO min of exposure, the anterior chamber drug concentration was 635 ,LDI and posterior 18 pM. The mean fall in pressure over 5 hr was 1.7 mmHg, which persisted until the next day (n = 4). Irritation was minimal or absent. In a companion study, Stein et al. (1983) found that drops of 2.5 o/0 suspension of trifluormethazolamide given 5 min x 5 lowered pressure 2.3 mmHg for 1 hr in normal rabbits. Anterior chamber drug concentration averaged 50 ,XIM. In the conditions of that study there was no irritation in any of the animals. Trichlormethazolamide is effective in lowering pressure following IO min of exposure to a 15 mM-solution. There is a fall of 3.5 mmHg over a 3 hr period. With 60 min exposure there is a drop of 4.5 mmHg for the same period (not shown). Figure 3 shows high drug concentrations in the chambers at the 15 mM dose. [However, due to the low water solubility ofthe free acid (Table I), the sodium salt of trichlormethazolamide was too alkaline (pH 8.3) for effective use and caused initial rises in pressure and some irritation.] Of interest, however, was the sustained effect in lowering pressure for as long as 48 br, comparable to trifluormethazolamide. The relatively low solubility of trichlormethazolamide is compensated by high permeability (Tables I and II). However, there is a further limitation to this compound, as noted above: its susceptibility to hydrolysis. Ethoxzolamide did not lower pressure when applied to the eye at its maximum concentration (0.12 mx) for 10 min. The pharmacological data in Section A, 10 give the basis for this failure. Unsuccessful trials with certain of the other compounds are described briefly in Section A. As controls for these procedures, sulfacetamide, a related sulfonamide of pK, 56 (similar to bromacetazolamide) which does not inhibit carbonic anhydrase, was used as a 2 o/0 solution of its sodium salt. No reduction in IOP was found in five animals after I hr of cornea1 exposure and measurements made over the next 24 hr. This control procedure did not elicit any irritation. This suggests that the irritation observed in some drug treated animals was not solely a result of experimental procedures. 4. Discussion The reduction of intraocular pressure by topical use of carbonic anhydrase inhibitors is clearly feasible. The experimental findings agree well with theoretical considerations derived from measurements of cornea1 permeability, activity of the drugs against the enzyme, and the concentrations in the eye necessary to produce a reduction in flow. These data wggest that earlier failures occurred because existing drugs marketed for parenteral use did not have the correct physico-chemical properties for topical use. There is nothing in the underlying physiology of the system or pathology of glaucoma that mitigates against topical use of such drugs. What emerges from this work is the importance of both lipid and water solubility, and their balance, in the design of drugs for topical ocular administration. The matter of lipid solubility is not surprising; indeed it is this basic principle in pharmacology that led to our (unsuccessful) trial of topical ethoxzolamide to reduce pressure in rabbits. Our present data (Tables I and II) show that this very lipid soluble drug is highly permeable and that there is little difference in this property between normal

TOPICAL

SULFOSAMIDES

AND

EYE

PRESSUR,E

477

lipid insoluble drugs are relatively and de-epithelialized corneas. By contrast, impermeant in intact cornea, and show nearly 100 times increase in de-epithelialized corneas. However, lipid solubility is often associated with water insolubility; thus ethoxzolamide can only be applied in aqueous concentrations at most 0.12 mM. It is clear that very high lipid solubility (cf. compounds 10 and 11) may confer diffusional properties inimical to pharmacological effect, specifically rapid absorption into the conjunctival vessels and dissipation from the aqueous faster than bulk flow. Therefore, we have searched for compounds with moderate water and lipid solubility as a means to introduce much higher drug concentrations to the cornea1 surface and yet confer some degree of cornea1 and iridial permeability. Sulfonamides are weak organic acids and, in their fully protonated form (shown in Table I), are relatively insoluble in water. When they are converted to their sodium salts, they are soluble; however, most drugs of this class have such high PK, (> 7) that the solutions of sodium salt are too alkaline for topical use. We sought drugs of lower pK,, but if ionization is too great and particularly if it is a result of proton loss at the C2 nitrogen, permeability will be low (cf. compounds l-5). Nevertheless, the greatly increased water solubility of compound 4 (bromacetazolamide) in its ionized form did lead to activity in vivo. The transcorneal permeability of the ionized compounds 4 and 5 shows that these species do penetrate the cornea, although conventional teaching in pharmacology generally suggests the opposite. (It may be noted that pilocarpine, a base ofpK = 7.1, is dispensed in the ionized form at pH about 5.5.) The relative permeabilities of the ionized and non-ionized forms of several of the compounds of Table I are now being studied. An important goal was to increase ionization of the R-SO,NH, group to achieve greater solubility of drugs at physiological pH. We show (Section A, 8) as an example bow this was achieved for trifluormethazolamide, which appears to have the lowest pK, (6.6) of any known R-SO,NH,-R-SO,NHpair. The solubility of the fully protonated or acid form of the pair is also a factor in determining the final solubility of the compound. If it is granted that the sodium salt of an ionized compound is infinitely soluble, the total solubility of a monobasic compound at any pH (8,) is given by: S, = 8, [I + 10(pH-pK,)], where S, is the solubility of the acid form. For trifluormethazolamide 8, = 7.9 mM. The predicted solubility at pH 7.8 is then 133 mM or 3.9 %. This is borne out by experiment. Compound 9 or trichlormethazolamide is far less satisfactory because of low S, (0.2 mM) and higher pK, (7.0). Trifluormethazolamide has the twin advantages of water and lipid solubility, neither of which are found in acetazolamide or methazolamide, and only the latter is found in ethoxzolamide. The quantitative advantages of trifluormethazolamide are seen in Tables I-III. As Table III suggests, residence times of 4-108 min yield 200 ,UM drug in the anterior chamber, for compounds 4,s and 9. Figures 24 show that this generates concentrations in the posterior chamber of the order of 30 ,UVI, which in contact with secretory tissue should inhibit < 99 o/0 enzyme for a drug of KI = lo-* M (Maren, 1967). Table V shows that concentrations in the ciliary body following trifluormethazolamide are also adequate for functional inhibition of the enzyme and may remain for at least 6 hr. More needs to be learned about localization of the drug at or near the enzyme site. Our data show a 400.fold range of k,, values in vitro (Table I) and nearly that in vivo (Table II). The ki, values are converted to permeability constants (P) by the relation endothelial chamber volume P = kin cornea1 area

478

T. H. MAREK

ET

AL

as shown in the Materials and Methods section. From Table I; P would range from 5 x 10-4-2000 x 10e4 cm/hr. The calculation of P from in vivo Ic,, data (Table II) yields a somewhat lower result (range 2-700 x lop4 cm/hr) for reasons already discussed.3 This range of P values is broadly in agreement with those cited by Mishima (1981) for a variety of other compounds. The data on accession to the aqueous chambers reveal that hydrophilic drugs penetrate very slowly to the posterior, compared to anterior. From Section B the ratios of k’,,/t& for the four drugs studied, in order of increasing lipophilicity, are bromacetazolamide (0*33), trifluormethazolamide (0.12), trichlormethazolamide (0.2), and ethoxzolamide (0.7).4 These data suggest that the bromacetazolamide type traverses the cornea much better than (for example) iris tissue, but that a major route to the posterior segment for a lipophilic drug is through the conjunctiva, sclera and iris root. The predictions from these chemical and pharmacological data are borne out by our pressure and flow studies. If the sulfonamides are delivered in adequate amounts to the posterior segment, it makes no difference whether they arrive from the blood or from the anterior chamber. Figures 5 and 6 show the same effects of the topical drugs on aqueous humor formation and intraocular pressure as shown for systemic acetazolamide and its congeners (Oppelt, 1967 ; Wistrand et al., 1960). Current studies with rabbits in our laboratory by Dr Amir Bar-Ilan show further that aqueous humor flow, as measured by ascorbate dilution, and HCO,- concentration in both chambers are reduced by 5-10 min cornea1 exposure to 100 mM-trifluormethazolamide. Of the three compounds tested in vivo, however, none are likely to be satisfactory for clinical use. Bromacetazolamide is too toxic and not active enough. Trichlormethazolamide requires too high a pH for safety and is unstable. Trifluormethazolamide is chemically unstable and is sometimes irritating in solution although not in suspension. A stable compound with the chemical properties of trifluromethazolamide would be a candidate for glaucoma therapy. Finally, a successful drug of this type would have to work at shorter cornea1 residence times than shown here. Nevertheless, these experiments suggest that with further exploration of compounds linked to new syntheses, the topical treatment of glaucoma with sulfonamides is a reasonable goal, if not a likely probability. ACKNOWLEDGMENT We thank Ness I. Pessah and Marianne This research 00933 and HL

was supported 22258.

by the

E. Antoine for National Institutes

expert analytical of Health grants

work. EY 02227,

EY

REFERENCES Becker,

B. (1955). The effect of the carbonic anhydrase inhibitor, acetazolamide, composition of the aqueous humor. Am. J. Ophthalmol. 40, 129-36. Davson, H. (1967). Physiology of th.e Cerebrospinal Fluid. Churchill, London. de Feo, G., Piccinelli, D.. Putzolu, S. and Silvestrini: B. (1975). Effects of topically drugs on intraocular pressure in rabbits. ilr:neimittelforschung 25, 806-9. Foss, R. H. (1955). Local application of Diamox. Am. J. Ophthdmol. 39. 3369. Friedland, B. R. and Maren, T. H. (1981). The relation between carbonic anhydrase and ion transport in elasmobranch and rabbit lens. Exp. Eye Res. 33, 545-61. 3 See footnote, 4 See footnote,

p. 464. p. 471.

on the

instilled

activity

TOPICAL

SULFOKAMIDES

AND

EYE

PRESSURE

479

Kandel. S. I.. Wong, S. C., Kandel, M., and Gornald, A. G. (1968). Irreversible inactivation of bovine carbonic anhydrase B by bromacetazolamide. J. Biol. Ch,em. 243, 2437-49. Kinsey, V. E. and Barany, E. H. (1949). The rate of flow of aqueous humor. II. Deviation of rate of flow and its physiological significance. Am. J. OphthaEmoZ. 32; 189-201. Lindskog, S. (1969). On the mechanism of sulfonamide inhibition of carbonic anhydrase. In CO, I Chemical, Biochemical, and Ph.yaiological Aspects. (Eds, Forster, R. E.. Edsall: J. T.. Otis. A. B. and Roughton, F. J. W.) Pp. 157-165. NASA. Washington: DC. Maren, T. H. (1967). Carbonic anhydrase: chemistry, physiology, and inhibition. Physiol. Rev. 47, 5955781. Maren, T. H., Ash, V. I. and Bailey, E. M. (1954). Carbonic anhydrase inhibition. II. A method for determination ofcarbonic anhydrase inhibitors, particularly of Diamox. Bull. Johns Hopkins Hosp. 95, 244-55. Maren, T. H., Haywood, J. R., Chapman, S. K. and Zimmerman; T. Z. (1977). The pharmacology of methazolamide in relation to the treatment of glaucoma. Invest. OphthaZmoZ.

16: 730-42. Maurice. D. &I. (1980). Factors influencing the penetration of topically applied drugs. In Cl’linical Pharmacology of the Anterior Segment, (Ed. Holly. F. .J.) Pp. 21-32, Little, Brown and Co., Boston. Maurice. D. M. and Mishima, S. (1983). Ocular pharmacokinetics: absorption, distribution. excretion. In Handbook of Experimental Pharmacology, Vol. 69, Pharmacology of the Eye, (Ed. Sears, M. L.) Springer-Verlag, Heidelberg (In press). Mishima, S. (1981). Clinical pharmacokinetics of the eye. Invest. Ophthalmol. Vis. Sri. 21, 504-4 1. Northey, E. H. (1948). The Sulfonamides and Allied Compounds. Reinhold Publ. Corp., N.Y. O’Brien, W. J. and Edelhauser, H. F. (1977). The cornea1 penetrat,ion of trifluorthymidine, adenine arabinoside, and idoxuridine: a comparative study. Invest. Ophthalmol. 16, 1093-103. Oppelt’, W. W. (1967). Measurement of aqueous humor formation rates by posterior-anterior chamber perfusion with inulin. Normal values and the effect of carbonic anhydrase inhibition. Invest. Ophthalmol. 6, 76-8. Rieder, .J. (1963). Physikalisch-chemische und biologische Untersuchungen an Sulfonamiden. Ar~~~eim.itte2forschung 13, 81-103. Robson, J. IM. and Tebrich, W. (1942). Penetration and distribution of sodium sulfacetamide in ocular tissues of rabbits. Br. Med. J. 1, 687-90. Stein, A., Pinke, R., Krupin, T.. Podos, S. M., Serle, J. and Maren, T. (1983). Topical carbonic anhydrase inhibitors and aqueous humor dynamics in rabbits. Am. J. Ophthalmol. (In press). Wistrand, P. J., Rawls, J. A. and Maren, T. H. (1960). Sulfonamide carbonic anhydrase inhibitors and intra-ocular pressure in rabbits. Acta Pharmacol. Toxicol. 17, 337-55. Young, R. W.; Wood, K. H., Eichler, J. A., Vaughan, J. R.: and Anderson, G. W. (1956). 1.3,4 thiadiazole - and thiadiazoline sulfonamides as carbonic anhydrase inhibitors. Synthesis and structural studies. J. Am. Chem. Sot. 78, 4649-54.