Corneal Epithelial Regeneration and Adhesion Following Acid Burns in the Rhesus Monkey

Corneal epithelial regeneration and adhesion following acid burns in the rhesus monkey

Lawrence W. Hirst, Jerry A. Fogle,* Kenneth R. Kenyon, and Walter J. Stark

For study of corneal adhesion after chemical injury, sulfuric acid was used to burn the central and peripheral corneal epithelium in rhesus monkeys. The adhesion properties of the regenerating epithelium were sequentially followed by clinical assessment and by serial corneal biopsy specimens examined by transmission electron microscopy. The peripheral burns healed un- eventfully, but the central corneal burns residted in persistently loose adhesion for more than 8 weeks. The pathogenesis of the defective epithelial adhesion appeared to be initially the persistence of the original basement membrane that could not be utilized by regenerating epithelium. Later, poor adhesion was related to the slow rate of basement membrane regeneration and also to fragmentation of the Bowman's layer substrate and to mild stromal edema. (INVEST OPHTHALMOL VIS SCI 23:764-773, 1982.)

Key words: corneal epithelium, basement membrane, adhesion, acid, ultrastructure, erosion model

From the Cornea Service, The Wilmer Institute, The The clinical significance of poor corneal epi- Johns Hopkins Hospital, Baltimore, Md. (L. W. H. thelial adhesion and spontaneous epithelial and W. J. S.), and Department of Cornea Research breakdown lies not only in the considerable and Morphology Unit, Eye Research Institute of the pain suffered by such patients but also in Retina Foundation, and Department of Ophthalmol- ogy, Harvard Medical School, Boston, Mass. (K. R. K. the frequency with which these problems in- and J. A. F.). terfere with proper healing after physical Supported in part by National Eye Institute grant EY trauma, epithelial herpes infection, chemical 02476-01 (L. W. H.); Academic Investigator Award burns, and corneal edema. Defects of the EY 00156 from the National Eye Institute, NIH; the ocular surface epithelium also provide access William F. Milton Fund of Harvard University; Grant-in-aid from Fight for Sight, Inc., New York; for pathogenic organisms to enter the corneal Research to Prevent Blindness, James Adams Award stroma. (K. R. K.); Biomedical Research Support Grant PHS Study of the regenerating epithelial layer 5S07RR05527 from the National Eye Institute; the Mas- after corneal surface damage and the impor- sachusetts Lions Eye Research Fund, Inc. (K. R. K. tant role of its basement membrane was ad- and J. A. F.); a Bausch & Lomb Fellowship and a Heed Fellowship (J. A. F.); and the National Eye In- vanced by the observations of Khodadoust 1 stitute Core Facility Grant EY 10765 (The Wilmer et al. in 1968. Numerous studies have since Institute). contributed to our knowledge of the function Submitted for publication July 21, 1980. of basement membrane complexes in such Reprint requests: Dr. L. W. Hirst, The Wilmer Insti- conditions as recurrent erosive disorders, an- tute, The Johns Hopkins Hospital, 600 N. Wolfe St., terior corneal dystrophies,2 and chemical and Baltimore, Md. 21205. 3 6 *Present address: The University of Virginia, Char- thermal injuries. " lottesville, Va. Our purpose was to produce an experi-

Downloaded from iovs.arvojournals.org on 09/23/2021 Volume 23 Number 6 Corneal epithelial regeneration and adhesion 765

mental model of poor epithelial adhesion and spontaneous breakdown in the corneal epithelium, using the monkey, the animal whose corneal anatomy is most similar to that of the human. We elected to use acid as the traumatizing agent in the production of epithelial defects, so as to produce lesions predominantly con- fined to the anterior cornea. The clinical course of epithelial recovery and adhesion was thereafter correlated with light microscopic and transmission electron microscopic examinations.

Materials and methods Nine adult rhesus monkeys were sedated with phencyclidine (Sernylan) and the lids were sepa- rated with a speculum. One drop of topical pro- paracaine hydrochloride (0.5%) was applied to each treated eye. A cornea trephine without its piston was held in an Allis clamp and centered on the corneal epithelial surface, with care to obtain firm ©1978 surface contact without torque. THE JOHNS HOPKINS UNIVERSITY In group 1, 14 corneas were burned centrally, using a 6.5 mm diameter trephine for containment Fig. 1. Diagram of trephine well and area of cen- without involving peripheral cornea (Fig. 1). In tral cornea burned. group 2, three corneas were burned peripherally in a ring fashion, involving a 1.75 mm-wide zone over at least 8 weeks, and these samples were of the corneal epithelium immediately adjacent to fixed in glutaraldehyde-formalin and examined by the limbus. This was achieved with a double- light microscopy and transmission electron mi- barreled trephine (6.5 mm inner diameter, with croscopy. an outer diameter of 10 mm) (Fig. 2). The outer The adhesion of the epithelium at the surround- trephine protected the conjunctiva, while the ing edge of the biopsy site was tested by applica- inner trephine formed a barrier protecting the tion of gentle traction with jewelers' forceps under central 6.5 mm of corneal epithelium from the acid the microscope. Epithelial adhesion was judged as in the peripheral annular well. being (1) loose, if the epithelium lifted off the edge A 0.5 ml volume of 47% (18N) sulfuric acid was of the biopsy site with virtually no traction, or if transferred by pipette to this trephine well. After a biomicroscopic examination revealed bullous ele- 10 sec exposure, the acid was rapidly diluted by vation of the epithelium, or (2) tight, if any de- vigorous irrigation with balanced salt solution. The tectable traction was necessary before epithelial opaque area of epithelium was gently debrided separation, often with fragmentation, could be with a Bard-Parker blade or a moist cotton-tipped achieved. applicator. The monkeys were again sedated for To study the effect of the acid burn on the un- examination at 2 to 3 days after the burns and derlying corneal stroma and endothelium, an ad- thereafter at weekly intervals for a minimum of ditional three cynomolgus monkeys underwent eight observations. Serial slit-lamp examinations central corneal burns to both eyes with the were performed, and slit-lamp photographs were method described above. Central clinical specular taken when possible. At biweekly intervals, su- microscopy was performed prior to the burns and perficial corneal biopsies were performed with a 1 established the presence of a normal endothelial mm Elliott biopsy trephine, and a razor-blade mosaic. Central pachometry with Haag-Streit slit- knife was used to excise each specimen at the level lamp and Hedbys-Mishima centration attachment of the superficial stroma. Four biopsy specimens was performed 1 day after burn and at 1, 2, 5, and of each cornea were taken at intervals extending 8 weeks after burn. Eyes were obtained for his-

Downloaded from iovs.arvojournals.org on 09/23/2021 Invest. Ophthalmol, Vis. Set- 766 Hirst et al. December 1982

Fig. 3. Slit-lamp photograph of central cornea 1 week after acid burn, showing initial bullous epithelial elevation and stromal edema.

Fig. 2. Diagram of double-barreled trephine posi- tion on cornea and resultant peripheral corneal acid burn.

tologic examination of the corneal endotheliuin immediately after burn and at 1 day and 1, 2, 5, Fig. 4. Slit-lamp photograph of monkey cornea 3 and 8 weeks after burn. The eyes were placed in weeks after central acid burn, showing total resto- glutaraldehyde-formalin, the corneas were ex- ration of epithelial cover with clear stroma, and cised, and comeal endothelium was examined by one biopsy site visible (arrows). scanning and transmission electron microscopy.

Results weeks after burn, and three of them remained Central burns. The closure of the epithe- nonadherent for 17 weeks of follow-up. lial defects occurred without incident in the Although the stroma appeared biomicro- 14 eyes with central burns (group 1) over a scopically normal and there was no anterior- range of 2 to 10 days after burn, with no chamber reaction in 13 of the 14 eyes after an episodes of subsequent spontaneous epithe- initial 1 to 2 weeks of stromal edema (Figs. 3 lial breakdown. In seven of the 14 eyes there and 4), sterile stromal ulceration occurred in was prolonged poor epithelial adhesion per- one eye of the 14 with intact epithelium. sisting through 7 weeks. In these seven eyes, Histologic and ultras tructural studies of the apparent initial reestablishment of epi- biopsy samples taken immediately after the thelial adhesion was followed about 2 weeks acid application showed an intact basement later by recurrence of poor adhesion for up to membrane, with only minimal epithelial de- 14 weeks of follow-up. The other seven of the bris remaining adherent (Fig. 5). Bowman's 14 central-burn eyes failed to establish firm layer and the stroma appeared unaffected. adhesion at all for a follow-up period of 8 In the seven of 14 eyes that failed to rees-

Downloaded from iovs.arvojournals.org on 09/23/2021 Volume 23 Number 6 Corneal epithelial regeneration and adhesion 767

Fig. 5. Monkey cornea, immediately after acid burn, showing residual basement membrane (asterisks) and unaffected Bowman's layer (B). (x 10,000.)

Fig. 6. Central cornea 2 weeks after acid burn, with total epithelial cover. Original basement membrane appears somewhat diffuse and devoid of attachment complexes (brackets) and early disruption of Bowman's layer (circles), (x50,000.)

tablish adhesion, the basal cells did not form produced was in many cases thickened and hemidesmosoma] attachment complexes with otherwise aberrant in appearance (Figs. 7 the underlying original basement membrane and 8). In some instances, a plane of nonad- (Fig. 6). Also, the basal cells showed no at- herence within the zone of aberrant base- tempt to manufacture new basement mem- ment membrane was apparent, with resul- brane until the slow disintegration of the old tant separation of the entire layer of overlying basement membrane began at about 6 weeks epithelium (Fig. 9). after injury. The new basement membrane In contradistinction, the seven eyes that

Downloaded from iovs.arvojournals.org on 09/23/2021 Invest. Ophtkalmot. Vis. Sci. 768 Hirst et al. December 1982

Fig. 7. Central cornea 16 weeks after acid burn, with nonadhering epithelium showing redun- dant accumulation of multilaminated new basement-membrane material (asterisks), (x 12,000.)

Fig. 8. Central cornea 8 weeks after acid burn; discontinuous segments of new basement membrane and hemidesmosomal complexes (asterisks) over areas of nonadhering fibrillar- granular debris (circle). (x30,000.)

Downloaded from iovs.arvojournals.org on 09/23/2021 Volume 23 Number 6 Corneal epithelial regeneration and adhesion 769

Fig. 9. Central cornea 4 weeks after acid burn. Inset, Phase-contrast microscopy showing loose epithelium with subepithelial vacuolation and disorganization of Bowman's membrane, (paraphenylenediamine [PPDA]; X200.) Main figure, Original basement membrane has been lost, with physical separation of regenerating epithelium from substrate (arrows). Basement membrane interface consists only of loose granular debris (circles). (X40.000.)

had established tight adhesion at 8 weeks ap- stantial increase in corneal thickness from peared to do so by promptly engaging the preburn values (mean 0.42 ± 0.01 mm) to original basement membrane in attachments 0.53 ± 0.05 mm 1 day after burn and persis- achieved with new hemidesmosomal attach- tently increased thickness of 0.57 ± 0.02 mm ment complexes. In these cases, no evidence at 5 weeks after burn. This corresponds to the of new basement membrane production was scanning and transmission electron micro- observed in any specimen (Fig. 10). scopic findings of central endothelial vacuol- Regardless of whether epithelial reattach- ization immediately after acid burn, attenu- ment was loose or tight, abnormalities of ated endothelium at 1 week after acid burn, Bowman's layer became evident at about 3 to and areas of absent endothelium at 5 weeks 4 weeks after the burn. The normally dense after burn, with recovery to an attenuated en- network of collagen became disrupted by dothelial layer without vacuolation at 8 weeks. vacuoles, fibroblastic proliferation, and even- Anterior stroma was accellular for the first 2 tual scarring (Figs. 10 and 11). This de- weeks, but posterior stroma remained normal rangement of Bowman's membrane usually throughout the 8 weeks of observation. occurred sooner in the corneas with initially Peripheral burns. The healing of the pe- tight adhesion than it did in the corneas with ripheral annular burns in the three eyes pro- initially loose epithelial adhesion after the ceeded promptly and with none of the prob- burn. lems of epithelial nonadhesion that occurred Pachometric examination indicated a sub- with the central burns. The defect was cov-

Downloaded from iovs.arvojournals.org on 09/23/2021 Invest. Ophthalmol. Vis. Set. 770 Hirst et at. December 1982

Fig. 10. Central cornea 10 weeks after acid burn. Inset, Normal-thickness regenerated epithelium (phase-contrast microscopy). Fibrocellular scarring of Bowman's layer and superficial stroma is evident. (PPDA; X450). Main figure, Normal-appearing, closely apposed epithelial basement membrane and Bowman's layer surface, (x40,000.)

ered by epithelial regeneration derived from abrasions. Although the original basement both the adjacent corneal and conjunctival membrane remained in situ after burn, epi- epithelial surfaces (Fig. 12). Tight adhesion thelial adhesion was delayed for up to 8 was established by 3 weeks, and there were weeks in all central burns. This is in sharp no associated stromal defects. contrast to the rapid utilization of original The light microscopic and ultrastructural basement membrane for corneal healing in the findings in serial biopsy specimens were con- rabbit after scraping of the epithelium, and sonant with these clinical observations. The the establishment of firm adhesion within 1 epithelial cover included conjunctival epi- week, as described by Khodadoust et al.' We thelial cells, containing melanosomes. The suppose the reason that adhesion may be de- original basal lamina was engaged in the for- layed despite retention of the original base- mation of new hemidesmosomal attachment ment membrane is that the acid may well complexes (Fig. 13). There was no evidence alter the character of this original basement of new basement membrane formation, and membrane, making it less suitable as an Bowman's layer and stroma remained intact attachment site for the regenerating epithe- (Fig. 14). lium. The prolonged persistence of this func- tionally useless basement membrane may ac- Discussion count for the delay in secretion of new base- The epithelial regeneration in all the cen- ment membrane with concomitant delay of tral and the peripheral ring burns appeared epithelial readhesion. A possible contribut- to follow an uneventful course that was ing cause to the poor epithelial adhesion in parallel to the course seen in human corneal this model is the mild but persistent stromal

Downloaded from iovs.arvojournals.org on 09/23/2021 Volume 23 Number 6 Corneal epithelial regeneration and adhesion 771

Fig. 11. Central cornea 6 weeks after acid burn. Inset, Phase-contrast microscopy shows epithelial cover with loose fibrillary scarring of Bowman's layer. (PPDA; x450.) Main figure, New basement membrane and attachment complexes (brackets) have reformed beneath undis- turbed regenerated epithelium. Disruption of Bowman's layer fibrils suggests nonadherence to stromal substrate tissue (circle). (X25,000.)

edema as a result of corneal endothelial damage. The resultant migration of fluid through the cornea could further jeopardize the formation of stable adhesion complexes for the regenerated epithelium. The persistent failure of epithelial adherence after the burn was also apparently based on significant changes occurring in Bowman's layer and stroma. Continuing failure to use the new basement membrane for epithelial adherence in the seven eyes (50%) that could not reattach to the old basement membrane^ and the subsequent loose adhesion of the Fig. 12. Slit-lamp photograph of monkey cornea 1 seven other eyes that initially did form at- week after peripheral ring burn; epithelial re- tachment complexes to the original basement generation is complete, with some melanocytosis membrane, can possibly be attributed to (arrows). an inability of the anchoring fibrils of the hemidesmosomal attachment complexes to epithelial attachment to the dermis and skin obtain a firm substratum in the disrupted and has been well demonstrated by Heaphy and changing Bowman's layer. The importance of Winkelmann.7 This plane of possible separa- this infraepithelial bond for achievement of tion at the level of superficial Bowman's layer

Downloaded from iovs.arvojournals.org on 09/23/2021 Invest. Ophthalmal. Vis. Sci. 772 Hirst et al. December 1982

Fig. 13. Peripheral cornea 3 weeks after peripheral ring burn. Inset, Slightly thinned epithelial layer with normal Bowman's layer, (PPDA; X450.) Main figure, Normal-appearing interface of epithelium and Bowman's layer, and attachment complexes (circles), (x25,000.)

Fig. 14. Peripheral cornea 6 weeks after peripheral ring burn, with normal-appearing interface of epithelium and Bowman's layer, and attachment complexes with compact Bowman's layer (x 25,000.)

has not been appreciated prior to the present our findings include demonstration of poor study, possibly because experiments under- epithelial adhesion in clinical thermoker- taken in rabbits demonstrate rapid reattach- atoplasty4 because of abnormal subepithelial ment to stroma taking place, with no inter- basement-membrane material and debris,8 vening Bowman's layer. Clinical analogies to the regenerating but poorly adhering corneal

Downloaded from iovs.arvojournals.org on 09/23/2021 Volume 23 Number 6 Corneal epithelial regeneration and adhesion 773

of peripheral cornea could account for the absence of clinically detectable corneal edema within this area at any time and lack of subsequent epithelial adhesion problems con- •< Bosal Cell ,_ Basement Membrone tributed to by this edema. Complex We have developed a model of poor cor- •< Bowman's Loyer neal epithelial adhesion in monkeys. Our model demonstrates the separation of epithelium from stroma at two levels. Faulty adhesion at either of the two levels of Fig. 15. Representation of adhesion zone of cor- separation—between epithelial cells and the neal epithelium consisting of basal cells, basement basement membrane, or within Bowman's membrane complex, and Bowman's layer. The po- layer or an abnormal basement-membrane tential plane of separation between epithelium and layer—could result in clinical syndromes of basement membrane and between basement mem- poor epithelial attachment (Fig. 15). brane and Bowman's layer are evident. REFERENCES epithelium in the diabetic patient after vi- 1. Khodadoust AA, Sitverstein AM, Kenyon KR, Dowl- trectomy,9 and the recurrent corneal erosion ing JE: Adhesion of regenerating corneal epithelium: syndrome.2 the role of basement membrane. Am J Ophthalmol 65:339, 1968. The peripheral annular burns were in- 2. Fogle JA, Kenyon KR, Stark WJ, Green WR: Defec- duced in such a way that the total area of tive epithelial adhesion in anterior corneal dystrophy. epithelial surface burned (approximately 39 Am J Ophthalmol 79:925, 1975. mm2) was the same as that of the central 3. Matsuda H, Smelser GK; Epithelium and stroma in burns. In the annular-burn series there was alkali-burned corneas. Arch Ophthalmol 89:396,1973. 4. Fogle JA, Kenyon KR, Stark WJ: Damage to epithe- rapid epithelial cover within 2 days or less lial basement membrane by thermokeratoplasty. Am and immediate firm adhesion of the epithe- J Ophthalmol 83:392, 1977. lium, with reutilization of existing basement 5. Pfister RR, Burnstein N: The alkali-binned cornea. I. membrane. The speed of epithelial cover was Epithelial and stromal repair. Exp Eye Res 23:519, possibly enhanced in the annular series by 1976. the greater length of regenerating epithelial 6. Hirst LW, Kenyon KR, Fogle JA, Hanninen L, Stark WJ: Comparative studies of corneal surface injury in margins. Therefore, although the total area to the monkey and rabbit. Arch Ophthalmol 99:1066, be covered by epithelium was the same as 1981. that in the central burns, the available length 7. Heaphy MR, Wilkelmann RK: The human cutaneous of adjacent migrating epithelial edges in the basement membrane-anchoring fibril complex: prep- ring burns was 55 mm, as compared with the aration and infrastructure. J Invest Dermatol 68:177, 1977. 22 mm circumference in the 6.5 mm central 8. Kenyon KR, Fogle JA, Stone DL, Stark WJ: Regen- acid burns. The total lack of vascularization, eration of corned epithelial basement membrane fol- absence of inflammatory infiltrate, and main- lowing thermal cauterization. INVEST OPHTHALMOL tenance of a normal Bowman's layer could VIS SCI 16:292, 1977. also be factors in the benign course of this 9. Kenyon KR, Wafai Z, Michels RC, Conway B, To- lentino F: Corneal basement membrane abnormality type of burn. Finally, possible protection of in diabetes mellitus. INVEST OPHTHALMOL VIS SCI corneal endothelium by increased thickness 17(ARVO Suppl.):245, 1978.

Downloaded from iovs.arvojournals.org on 09/23/2021