Ophthalmic Solutions, the Ocular Surface, and a Unique

Therapeutic Artificial Tear Formulation

Jeffrey P. Gilbard, M.D., Scott R. Rossi, M.S., and Kathleen Gray Heyda, B.A.

In rabbit studies, we found that extendedexposure of the ocular surface to existing oph­thalmic solutions resulted in gross surface ab­normalities and decreases in conjunctival gob­let cell density. We developed an electrolytesolution (solution 15) that preserves normalgross appearance, goblet cell density, cornealepithelial glycogen levels, and ocular surfacemorphologic characteristics after extended ex­posure to the rabbit ocular surface. We createdan artificial tear formulation by adding a de­mulcent and a buffering system to solution 15and reducing its osmolarity to 162 mOsm/1. Wethen compared our artificial tear formulationto Hypotears in a double-masked, crossoverstudy involving 11 patients with dry-eye disor­ders. Our artificial tear formulation was moreeffective than Hypotears in decreasing tearfilm osmolarity and rose bengal staining, andwas preferred subjectively by an eight-to-onemargin.

IN THE FORMULATION of ophthalmic solutions,and especially in the formulation of artificialtear solutions, a major objective has been tominimize toxicity. Toward this goal, these solu­tions have been adjusted to a neutral pH and asafe osmolarity range, and strict controls have

Accepted for publication Jan. 25, 1989.From the Cornea Research Unit, Eye Research Insti­

tute (Dr. Gilbard, Mr. Rossi, and Ms. Gray Heyda), andthe Department of Ophthalmology, Harvard MedicalSchool (Dr. Gilbard), Boston, Massachusetts. This studywas supported in part by National Eye Institute grantEY03373.

The ophthalmic solution described herein has beenissued United States patent 4,775,531. Dr. Gilbard is theinventor and the Eye Research Institute of Retina Foun­dation is the assignee. Foreign patents are pending. TheEye Research Institute has a proprietary interest in thistechnology and the authors will participate in its com­mercialization.

Reprint requests to Jeffrey ·P. Gilbard, M.D., EyeResearch Institute, 20 Staniford St., Boston, MA 02114.

been established regarding the ingredients thatmay be included. More recently, there has beena trend to modify or eliminate the preservativesthat are customarily included in these solu­tions.

Pfister and Burstein! could not demonstratetoxicity to the corneal epithelium after the in­stillation of a single drop of 0.9% sodium chlo­ride. Sussman and Friedman" found, however,that instillation of a 0.9% sodium chloride solu­tion in the rabbit eye every 15 minutes resultedin severe hyperemia and photophobia, as wellas eyelid edema or corneal ulceration. Merrill,Fleming, and Girard" studied the effect of 0.9%sodium chloride on conjunctival epithelium intissue culture, and demonstrated toxicity be­ginning three to five minutes after initial expo­sure. Toxicity was not observed after instilla­tion of Hanks' balanced salt solution or acommercial balanced salt solution. Wilson,O'Leary, and Bachman' observed differences inspecular reflectance of rabbit corneas based onthe bathing solution tested. They concludedthat 0.9% sodium chloride alone resulted inincreased cell desquamation, and that the addi­tion of certain electrolytes to the sodium chlo­ride solution diminished the degree of ob­served cell desquamation.

In our rabbit models for keratoconjunctivitissicca, morphologic abnormalities, most notablydepletion of goblet cell density, are observed inthe conjunctiva almost one year before thedevelopment of morphologic abnormalities inthe cornea.v? This is consistent with earlierwork that indicated conjunctival goblet celldensity is a sensitive indicator of ocular surfacehealth.P

Discovering that all ophthalmic solutionstested depleted conjunctival goblet cell densi­ty, we sought to develop a solution that wouldpermit normal maintenance of conjunctivalgoblet cells and set a new standard for theabsence of toxicity, so that fluid could be ap­plied to the ocular surface without the aggrava­tion of ocular surface disease and with thepotential to ameliorate the surface disease.

348 ©AMERICAN JOURNAL OF OPHTHALMOLOGY 107:348-355, APRIL, 1989

Vol. 107, No.4 Artificial TearFormulation 349

Material and Methods

New Zealand white rabbits of both sexesweighing 2.5 to 3 kg were anesthetized withintramuscular xylazine, 10 mg/kg of bodyweight, and ketamine, 100 mg/kg of bodyweight; anesthesia was sustained with an addi­tional 100 mg of ketamine per hour and 10 mgof xylazine every three to four hours. Eyelidswere elevated with sutures attached to elevatedposts to form a conjunctival well. In separateexperiments the well was filled with test solu­tions such that the entire ocular surface wassubmerged. One eye of each rabbit was bathedwith a test solution, while the other eye, closedby sutures through the external eyelid skin,served as a control. Baths were exchanged withfreshly aerated solution every hour for 12hours.

At the completion of the 12-hour period, fourconjunctival biopsy specimens were taken fromthe eye that had been bathed, as well as fromthe contralateral control eye that had beenclosed. The specimens were taken at four pre­cise locations adjacent to the corneoscleral Iim-

bus with a 5-mm trephine blade, and conjuncti­val goblet cell density was counted aspreviously described.' Goblet cell densitiesfrom all quadrants were then averaged andexpressed as percent of densities in the contra­lateral control eye; "n" was used to indicatenumber of rabbits tested.

Tests were run with 20 different solutions,including five commercially available solutionsand 15 solutions formulated in our laboratory.The five commercially available solutions test­ed were: (1) Hypotears (Iolab Pharmaceuticals),(2) Tears Naturale (Alcon Laboratories), (3) Re­fresh (Allergan Pharmaceuticals), (4) Unisol(CooperVision), and (5) Balanced Salt Solution(BSS) (Alcon Laboratories). The composition,pH, and osmolarity of the solutions formulatedin our laboratory are listed in Table 1.

In a separate series of experiments we testedthe long-term effect of topically infused Unisoland our optimum solution (solution 15) (Table1). On day I, infusion tubing was implanted inone eye of four rabbits. At eight weeks, tworabbits received Model 300 Infusaid pumps(Shiley Infusaid, Norwood, Mass). In one rab­bit infusion of Unisol was begun with a flow

TABLE 1

COMPOSITION OF TESTED ELECTROLYTE SOLUTIONS AND GOBLET CELLS REMAINING RELATIVE TOCONTRALATERAL CONTROL EYE"

SOLUTION OSMOLALITY GOBLET CELLSNO. NaCI KCI CaCI2 MgCI2 NaHCO:! NaH2P04 pH (MOSM/KG) REMAINING (%j

1 111.8 20.2 0.8 0.6 27.2 1.0 7.6 299 59.3'2 116.4 18.7 0.4 0.6 25.9 0.7 7.8 316 60.3

3 97.0 47.0 48.0 7.7 309 67.3

4 100.0 32.0 7.51 258 74.2

5 140.0 8.0 15.0 7.0 292 75.9

6 103.0 10.0 14.0 7.1 295 77.27 102.0 23.0 0.8 0.6 31.0 1.0 7.4 301 89.0

8 109.8 22.0 0.8 0.6 29.2 1.0 7.3 296 89.69 101.0 23.0 31.0 7.3 300 90.9

10 89.8 42.0 0.8 0.6 47.2 2.0 7.4 325 91.5'11 93.0 42.0 46.0 7.6 311 92.512 98.0 35.0 0.8 0.6 35.0 1.0 7.2 304 94.9'13 90.0 38.0 0.6 0.5 40.0 1.0 7.5 306 95.014 100.0 24.0 32.0 7.41 296 95.815 99.0 24.0 0.8 0.6 32.0 1.0 7.4 302 99.4

'salts are expressed as mmoVI. Solution.2 is Basic Tear Solution, described by Wilson, O'leary, and Bachman.''Data are based on two conjunctival quadrants. Each solution was tested in two rabbits, except for solution 8, which was tested in six

rabbits.

ISolution brought to final pH by adding 1 M Hel.

350 AMERICAN JOURNAL OF OPHTHALMOLOGY April, 1989

rate of 2.4 ml/day, and in the second rabbitinfusion of solution 15 at 2.6 ml/day. Infusionswere continued until 20 weeks, at which timeslit-lamp examinations were performed. Allfour rabbits were then killed for morphologicstudy, counting of goblet cell density, andmeasurement of corneal epithelial glycogenlevels, as previously described.' The contralat­eral untreated eyes, and the two rabbits withinfusion tubing but no pump or infusion,served as controls.

For our clinical studies, solution 15 was con­verted to an artificial tear formulation by add­ing a demulcent and a sodium borate and boricacid buffering system. Furthermore, since pa­tients with dry-eye syndrome have increasedtear film osmolarity, 9 and hypotonic solutionshave been shown to be more effective thanisotonic ones in decreasing osmolarity to thenormal range after instillation in the eye asdrops.!?:" we reduced proportionately the elec­trolyte concentrations so that the final osmolar­ity of our artificial tear formulation was 162mOsm/1. Selection of 11 women with dry-eyedisorders was based on a characteristic clinicalhistory and signs, and increased tear film os­molarity. Patients ranged in age from 33 to 65years (mean ± S.E.M., 54 ± 3 years). Informedconsent was obtained from all participants.

In this double-masked, prospective, cross­over study, eyes were randomly assigned totreatment with our artificial tear formulation orHypotears. Patients were instructed to use ourartificial tear formulation in one eye andHypotears in the contralateral eye at least sixtimes a day. After eight weeks, solutions wereswitched so that the eye that had received ourartificial tear formulation received Hypotearsand vice versa. During this 16-week study, thepatients were examined five times: at the begin­ning of the study and then at weeks 4, 8, 12,and 16. Tear osmolarity measurernents.v":" vis­ual acuity testing, and slit-lamp examinationwere performed at all visits. The initial exami­nation and the examinations at weeks 8 and 16also included the instillation of rose bengaldye. Rose bengal staining was diagrammed,photographed, and scored using the method ofvan Bijsterveld.!' Subjective evaluation of thesolutions and preferences were elicited on allfour return visits. Patients returned all solu­tions at each return visit, and the volume ofsolution remaining was measured to determinecompliance.

Results

Rabbit eyes bathed for 12 hours withHypotears, Tears Naturale, and Refresh had amarkedly abnormal gross appearance (Fig. 1),and only 8.4% (n = 1),39.1 % (n = 1), and 47.8%(n = 1) of goblet cells remained, respectively.Eyes bathed with Unisol and BSS retained81.1 % (n =1) and 88.9% (n = 6) of goblet cells,respectively.

By modifying the electrolyte composition insolutions formulated in our laboratory, wewere able to demonstrate that the maintenanceof normal goblet cell density was dependentupon the electrolyte composition of the bathingsolution (Table 1). Eyes bathed with solution 15retained a normal gross appearance after 12hours of bathing (Fig. 1) and a goblet celldensity that was 99.4% (n = 2) of that in thecontralateral, unbathed control eyes (Table 1).

In separate experiments, infusion tubing wasimplanted in one eye of four rabbits, and two ofthese rabbits received a continuous infusion ofeither Unisol or solution 15 from weeks 8through 20 after implantation of the tube. After20 weeks, the eye receiving Unisol showedinjection of the bulbar and inferior tarsal con­junctiva, abnormal fluorescein staining of thesuperior bulbar conjunctiva, and abnormal rosebengal staining of the superior bulbar conjunc­tiva and cornea. The eye receiving solution 15remained uninflamed and there was no abnor­mal staining of the surface. Transmission elec­tron microscopy of the conjunctiva in theserabbits demonstrated cell desquamation andswelling in the eye receiving Unisol; the con­junctival morphologic features of the eye re­ceiving solution 15 were normal (Fig. 2). Con­junctival goblet cell density and cornealepithelial glycogen were decreased after Unisolinfusion, but were unchanged from controlsafter infusion of solution 15 (Figs. 3 and 4).

Before beginning the clinical protocol, solu­tion 15 was modified as previously described tocreate the artificial tear formulation. Rabbiteyes bathed in this formulation for 12 hourshad a goblet cell density measuring 102.4% (n= 3) of the contralateral control eyes.

The 16-week, double-masked, crossoverstudy had two baselines, one at the initial visitand the other at the crossover point eightweeks into the study. Mean tear osmolarity and

Vol. 107, No.4 Artificial Tear Formulation 351

Fig. 1 (Gilbard, Rossi, and Gray Heyda). Gross appearance of rabbit eyes after 12 hours of bathing. Top left,Hypotears. Top right, Tears Naturale. Bottom left, Refresh. Bottom right, Solution 15.

rose bengal staining scores were not signifi­cantly different at these two time points so dataat these points were merged. Tear osmolarityincreased after four weeks of treatment withHypotears, whereas tear osmolarity decreasedafter four weeks of treatment with our artificialtear formulation and was significantly lowerthan osmolarity in the Hypotears-treated eyesat this time point (P < .01). Comparing tearosmolarity after four and eight weeks of treat­ment to baseline data, we found that our for­mulation produced a significant decrease intear film osmolarity (P < .025), whereasHypotears did not (Fig. 5). Tear osmolarity ineyes receiving Hypotears for eight weeks beganto decrease.

Ocular surface disease, as indicated by meanrose bengal staining scores, increased aftertreatment with Hypotears and decreased aftertreatment with our artificial tear formulation.The difference in the change in rose bengal

staining after treatment with these two solu­tions is significant (P < .05) (Fig. 6).

In this study with 11 patients and four returnvisits, there were 44 opportunities for patientsto express a subjective preference. On 25 occa­sions patients expressed a preference for ourartificial tear formulation and on two occasionsa preference for Hypotears; on 17 occasionsthey declined to commit themselves. At thecompletion of the 16-week study, eight patientspreferred our artificial tear formulation, onepatient preferred Hypotears, and two patientsremained uncommitted (Table 2).

Discussion

We demonstrated that the electrolyte compo­sition of ophthalmic solutions is critical for themaintenance of ocular surface health as indicat-

352 AMERICAN JOURNAL OF OPHTHALMOLOGY

.••...~.•....."'I

April, 1989

Fig. 2 (Gilbard, Rossi, andGray Heyda). Inferior bulbarconjunctiva (X 7,500). Top,From eye receiving Unisol in­fusion for 12 weeks. Bottom,From eye receiving solution15 infusion for 12 weeks. Cellswelling and desquamationare evident in the eye thatreceived Unisol.

ed by conjunctival goblet cell density and cor­neal epithelial glycogen. It is not sufficient for asolution to be preservative-free to be nontoxicto the ocular surface.

It has been shown that decreased goblet celldensity is an important pathologic change inkeratoconjunctivitis sicca, both in humans":"and in our rabbit model. 5,6 In our rabbit model,

the loss of goblet cells was progressive andproportional to the severity of disease. Rabbitswith greater decreases in their tear production,caused by permanent interruption of the func­tion of greater amounts of glandular tear secre­tory tissue, had higher increases in tear filmosmolarity and greater decreases in goblet celldensity. The present study indicates that solu-

Vol. 107, No.4 Artificial Tear Formulation 353

0""""'---

100 ---------

ALGD+

SOLUTION 15

ALGD+

UNISOLALGDalone

c:

'"coooa~ 75_0.. ::=c:'" 0 50~~0.0.w><-'":~ 25e(;U

1P-ALGD+

SOLUTION 15

ALGD+

UNISOLALGDalone

100~c0

~ 750.><~

!! 50a;o4i 25:is0Cl

1P- 0

Fig. 3 (Gilbard, Rossi, and Gray Heyda). Effect ofartificial lacrimal gland duct (ALGD) and infusionsolutions (Unisol and solution 15) on goblet celldensity-.All artificial lacrimal gland ducts were im­planted for 20 weeks, and all solutions were infusedcontinuously between week 8 and week 20.

Fig. 4 (Gilbard, Rossi, and Gray Heyda). Effect ofALGD and infusion solutions on corneal epithelialglycogen. All artificial lacrimal gland ducts wereimplanted for 20 weeks, and all solutions were in­fused continuously between week 8 and week 20.

tions currently considered therapeutic for dry­eye disorders actually exacerbate the loss ofconjunctival goblet cells. Unisol had a moredevastating effect on the ocular surface after 12weeks of infusion than after 12 hours of bath­irig. This suggests that the more compliantdry-eye patients are with the use of currentophthalmic solutions, and the longer they usethese solutions, the greater the potential forthe solution to exacerbate their disease. In our12-week continuous infusion experiment, abuffered saline solution (Unisol) not only re­produced the loss of goblet cells seen in ourrabbit models for dry-eye disease, but also

reproduced the depletion of corneal epithelialglycogen. Moreover, the extent of ocular sur­face abnormalities that developed after 12weeks of Unisol infusion exceeded the ocularsurface abnormalities (decrease in goblet celldensities, decrease in corneal glycogen levels,and conjunctival morphologic abnormalities)seen in our most severely affected dry-eye rab­bit model" after 12 weeks of disease. The impli­cations of this are particularly remarkablegiven that Unisol was considerably less toxic inour 12-hour experiments than were the othersolutions usually used as attempted treatmentfor dry-eye disorders.

In the double-masked, crossover study of our

Fig. 5 (Gilbard, Rossi, and GrayHeyda). Tearosmolarity after treat­ment with Hypotears and our arti­ficial tear formulation.

ATF

1" ," ,, ".", -,, ., '.

,'1' ''''...

"" "'''''Ip<O.Oll • 1

""" HYPOTEAAS

325

~<5E~wen 320+1

>~

~...Jo~ 315o~w~

~w::; 310 .........------_-----_>-- ----4

BASELINE 4 Wks 8Wks

WEEKS OF TREATMENT

354 AMERICAN JOURNAL OF OPHTHALMOLOGY April, 1989

0.5

_-----_.1""""AA'.i-> 1

• ----- lp<0.05l

ATF

Fig. 6 (Gilbard, Rossi, and GrayHeyda). Change in rose bengalstaining after treatment withHypotears and our artificial tearformulation.

BASELINE

WEEKS OF TREATMENT

artificial tear formulation and Hypotears in thetreatment of dry-eye disorders, our formula­tion was significantly better than Hypotears inreducing tear film osmolarity. We were sur­prised that tear osmolarity increased after fourweeks of treatment with Hypotears. We postu­late that the detergent (benzalkonium chloride)in Hypotears may disrupt the lipid layer, there­by increasing tear film evaporation. It is possi­ble that tear film osmolarity decreased aftereight weeks of Hypotears therapy because ofthe continued delivery of this solution or be­cause of an increased rate of tear secretion from

TABLE 2

PREFERENCES OF DRY-EYE PATIENTS INDOUBLE-MASKED CROSSOVER COMPARISON OF

HYPOTEARS AND ATP

FINAL

PATIENT NO. 4WKS BWKS 12WKS 16WKS

1

2

3 H H

4 ATF

5 ATF ATF

6 ATF ATF ATF

7 ATF ATF ATF

8 ATF ATF ATF ATF

9 ATF ATF ATF ATF

10 ATF ATF ATF ATF

11 ATF ATF ATF ATF

*ATF, our artificial tear formulation; H, Hypotears; -, no

preference.

B Wks

the irritative effect of this artificial tear as indi­cated by the increased rose bengal staining inthese eyes. As might be predicted from our12-hour bathing experiments, rose bengalstaining actually increased after treatment withHypotears, whereas it decreased after treat­ment with our artificial tear formulation.

Not only did our formulation objectively out­perform Hypotears with regard to tear filmosmolarity and rose bengal staining, but pa­tients preferred our artificial tear formulationover Hypotears by an eight-to-one margin. Ourunique therapeutic artificial tear formulationappears to represent a new benchmark in fluidsupplementation therapy for dry-eye disor­ders.

References

1. Pfister, R. R., and Burstein, N.: The effects ofophthalmic drugs, vehicles, and preservatives oncorneal epithelium. A scanning electron microscopestudy. Invest. Ophthalmol. 15:246, 1976.

2. Sussman, J. D., and Friedman, M.: Irritation ofrabbit eye caused by contact-lens wetting solutions.Am. J. Ophthalmol. 68:703, 1969.

3. Merrill, D. L., Fleming, T. c., and Girard, L. J.:The effects of physiologic balanced salt solutions andnormal saline on intraocular and extraocular tissues.Am. J. Ophthalmol. 49:895, 1960.

4. Wilson, G., O'Leary, D., and Bachman, W.: Cellexfoliation and light scatter by the corneal epitheli­um. ARVO Abstracts. Supplement to Invest. Oph­thalmo!. Vis. Sci. Philadelphia, J. B. Lippincott,1983, p. 118.

5. Gilbard, J. P., Rossi, S. R., and Gray, K. L.: A

Vol. 107, No.4 Artificial Tear Formulation 355

new rabbit model for keratoconjunctivitis sicca. In­vest. Ophthalmol. Vis. Sci. 28:225, 1987.

6. --: Tear film osmolarity and ocular surfacedisease in two rabbit models for keratoconjunctivitissicca. Invest. Ophthalmol. Vis. Sci. 28:374, 1988.

7. Kinoshita, S., Kiorpes, T. c., Friend, J., andTheft, R. A.: Goblet cell density in ocular surfacedisease. A better indicator than tear mucin. Arch.Ophthalmol. 101:1284, 1983.

8. Friend, J., Kiorpes, T., and Thoft, R. A.: Con­junctival goblet cell frequency after alkali injury isnot accurately reflected by aqueous tear mucin con­tent. Invest. Ophthalmol Vis. Sci. 24:612, 1983.

9. Gilbard, J. P., Farris, R. L., and Santamaria, J.:Osmolarity of tear microvolumes in keratoconjunc­tivitis sicca. Arch. Ophthalmol. 96:677, 1978.

10. Gilbard, J. P., and Farris, R. L.: Tear osmolari­ty and ocular surface disease in keratoconjunctivitissicca. Arch. Ophthalmol. 97:1642, 1979.

11. Gilbard, J. P., and Kenyon, K. R.: Tear dilu­ents in the treatment of keratoconjunctivitis sicca.Ophthalmology 92:646, 1985.

12. Gilbard, J. P., and Farris, R. L.: Ocular surfacedrying and tear film osmolarity in thyroid eye dis­ease. Acta Ophthalmol. 61:108, 1983.

13. Gilbard, J. P., Gray, K. L., and Rossi, S. R.: Anew technique for storage of tear microvolumes.Invest. Ophthalmol. Vis. Sci. 28:401, 1987.

14. Van Bijsterveld, O. P.: Diagnostic tests in thesicca syndrome. Arch. Ophthalmol. 82:10, 1969.

15. Ralph, R. A.: Conjunctival goblet cell densityin normal subjects and in dry eye syndromes. Invest.Ophthalmol. 14:299, 1975.

16. Nelson, J. D., and Wright, J. c. Conjunctivalgoblet cell densities in ocular surface disease. Arch.Ophthalmol. 102:1049, 1984.

17. Nelson, J. D., Havener, V. R., and Cameron,J. D.: Cellulose acetate impressions of the ocularsurface. Arch. Ophthalmol. 101:1869, 1983.

OPHTHALMIC MINIATURE

Persons whose business lies with morbid eyes are apt to consider theworld an aviary of owls, and put a prohibition upon a practice whichmillions pursue without injury to their sight and to the great advantage ofeverything else. Fontenelle was told that coffee was a slow poison. "Veryslow, indeed," he replied, "for it has been 80 years in killing me." Theprudent plan is to pay attention to sensations and not neglect their warningin the vain hope that it may be neglected with impunity.

SpectaclesQuart. Rev. 87:1850