The Corneoscleral Limbus in Human Corneal Epithelial Wound Healing Harminder S. Dua, M.D., and John V. Forrester, M.D.
We studied re-epithelialization of the ocular surface in 17 human eyes (14 patients) with large corneal and conjunctival abrasions. We focused on the healing of the limbal region. During re-epithelialization, cell movement was found to occur circumferentially along the corneoscleral limbus and centripetally from the corneoscleral limbus. In no patient did the central corneal defect close before the corneoscleral limbus had first re-epithelialized completely. Normal Iimbal healing was observed to occur by circumferentially migrating tongue-shaped corneallimbal epithelium. These tongue-shaped projections developed from either side of the remaining intact epithelium and advanced along the corneoscleral limbus until they met. A centripetal movement of cells from the corneoscleral llmbus then completed the healing process. In three patients, however, the advancing conjunctival epithelium extended across the corneoscleral limbus before the tongue-shaped projections of corneal limbal epithelium had met. The surface of the cornea covered by conjunctival epithelium was thin and irregular, and later showed peripheral scarring, vascularization, and recurrent erosions. MANY PUBLISHED REPORTS on the healing of corneal epithelial wounds have come from experimental studies on animals':" and in vitro studies on animal" and human" corneal epithelial cell cultures. Animal studies have confirmed that corneal epithelial healing occurs by the processes of cell migration and cell multiplication. 6.8 Evidence derived mostly from animal
Accepted for publication Aug. 28, 1990. From the Department of Ophthalmology, University of Aberdeen, Foresterhill, Aberdeen, Scotland. Reprint requests to Harminder S. Dua, M.D., Research Division, Wills Eye Hospital, 9th and Walnut Sts., Philadelphia, PA 19107.
646
experiments has been used to indicate that the process of cell multiplication occurs predominantly at the corneoscleral limbus ,"!' The concept of limbal stem cells as the source of multiplying and migrating epithelial cells is now widely accepted.P" A continuous centripetal movement of cells from the corneoscleral limbus to the center of the cornea is believed to occur during replicative epithelial turnover. This centripetal movement from the peripheral stem cells of the cornea is exaggerated after traumatic epithelial cell loss.r" Studies on corneal epithelial wound healing in humans have also illustrated similar centripetal cell movement.":" Limbal stem cells in the rabbit cornea have been demonstrated by specific staining.P:" but no identification of these cells in humans has been made. We studied the healing of large corneal epithelial wounds involving the corneosclerallimbus in 14 patients.
Patients and Methods Fourteen men (17 eyes) with extensive ocular surface abrasions who were examined at the Eye Casualty Department of the Aberdeen Royal Infirmary over a period of 3th years were included in this study. In all but one of these patients the corneoscleral limbus was affected to a lesser or greater extent and the abrasion extended on either side to involve the cornea and conjunctiva. The corneal epithelial loss ranged from 31.8% to 85.7% of the total surface area of the cornea at initial examination as measured by planimetry of fluorescein-stained color photographs (Table). The nature of injury causing the abrasion, the percentage area of the corneal abrasion, and the percentage of limbal involvement were noted (Table). All patients were initially examined within 12 hours of injury. Eyes with chemical injuries were irrigated
©AMERICAN JOURNAL OF OPHTHALMOLOGY
110:646-656,
DECEMBER,
1990
647
Comeoscleral Limbus in Epithelial Healing
Vol. 110, No.6
TABLE CLINICAL CHARACTERISTICS OF PATIENTS WITH CORNEAL ABRASIONS CASE NO., AGE (YRS)
1,34 2,30 3,26 4,28 5,36 6,35 7,40 8,43 9, 21 10,31 11,29 12,52 13,48 14, 37
CAUSE EYE
L.E. L.E. L.E. R.E. RE. L.E. L.E. RE. L.E. RE. L.E.
OF INJURY
Alcohol" Plaster t Plaster Alcohol Acid Alcohol Ammonia Ammonia
R.E. RE.
Boiling water Plaster Alcohol
L.E. L.E. L.E.
Plaster Alcohol
RE.
Plaster
WOUND AREA AT MANIFESTATION (%)
L1MBAL INVOLVEMENT (%)
55,0 48.3 85.7 77.6
21.5 15.6 67.0 66.3
60.2 52.2 37.4 54.1 59.6 35.6 75.3 73.0 81.8 51.5 48.0 31.8 71.3
TREATMENT
Prednisolone sodium phosphate Betamethasone sodium phosphate, acetylcysteine 5%
Nil Nil
10.6 17.3 54.4 12.2 65.2 56.0 57.0 18.0 12.4 14.2 58.1
Betamethasone sodium phosphate, acetylcysteine 5% Prednisolone sodium phosphate Prednisolone sodium phosphate
Betamethasone sodium phosphate
"Alcohol-based cleaning fluids. tCement and lime plaster used in the building industry.
with saline for 20 to 30 minutes. In all eyes, one drop of benoxinate hydrochloride 0.4% was instilled into the conjunctival sac before slitlamp examination. Particulate matter, if noted, was removed from the eyelid margins, conjunctival sac, and corneal surface. One drop of 1 % fluorescein was instilled and the cornea was photographed with a fundus-anterior segment camera using a blue filter. The extent of conjunctival involvement was also photographed as far as possible by asking the patient to look in different directions and by manually retracting the eyelids. All affected eyes were treated with topical antibiotics (chloramphenicol eyedrops four times a day), mydriatic and cycloplegic eyedrops (cyclopentolate hydrochloride 1 % three times a day), and covered with a patch. Six patients also received topical corticosteroids in the form of betamethasone sodium phosphate every two hours for the first two days and then four times daily (three patients) and prednisolone sodium phosphate four times a day (three patients). Two of the patients who were treated with topical betamethasone sodium phosphate also received 5% acetylcysteine eyedrops four times a day (Table). The three
patients with bilateral abrasions and four with extensive epithelial denudation were treated as inpatients. All eyes were examined daily until the abrasion healed. The cornea and conjunctiva were photographed at each examination. One drop of benoxinate hydrochloride 0.4% was instilled for the purpose of examination if the eye was photophobic. Planimetry was used to measure the area of corneal abrasion. Photographs of the cornea with the eyes in the primary position were photocopied onto graph paper, the areas of the abrasion and the cornea were calculated, and the area of the corneal abrasion was expressed as a percentage of the total corneal area. Limbal involvement was measured by determining the number of clock hours affected. One clock hour was taken to represent 8.3% of the corneoscleral limbus. The area of conjunctival involvement was not calculated. The shape and the direction of movement of the advancing edges of the conjunctival, Iirnbal, and corneal epithelial sheets were noted every day until complete epithelial cover was reestablished. Patients were followed up for two to four months thereafter.
648
AMERICAN JOURNAL OF OPHTHALMOLOGY
December, 1990
Fig. 1 (Dua and Forrester). Case 5. Left, Right eye. Right, Left eye. The patient had an acidic solution splashed into the eyes. Note the intact rim of limbal epithelium despite 60.2% of the right cornea and 52.2% of the left cornea being abraded.
Results The pattern of corneal epithelial wound healing was determined by the extent of limbal epithelial involvement. One patient (Case 5, Table) had bilateral corneal abrasions after an accident in which an acidic solution used for washing purposes on off-shore oil installations had splashed into his eyes. Large portions of the right cornea (60.2%) and the left cornea (52.2%) were denuded of epithelium. A narrow rim of intact epithelium remained at the periphery along the corneosclerallimbus in both eyes (Fig. 1). These abrasions healed by the centripetal movement of limbal epithelium. The healing pattern was characteristic and consistent with that described previously for the healing of
large corneal abrasions with an intact corneoscleral limbus in humans." Four (left eye) and five (right eye) convex sheets of epithelium appeared at the periphery from the intact epithelium. These migrated centrally and neighboring sheets established contact with each other along their sides, which left a central quadrilateral-shaped (left eye) and a pentagonal-shaped (right eye) denuded area. Further migration of these epithelial sheets resulted in the formation of "Y" -shaped contact lines. The inferior corneoscleral limbus was affected in the remaining 13 patients (15 eyes). In seven eyes, limbal involvement extended above the horizontal meridian on one or both sides. Six eyes had a total corneal epithelial cell loss of more than 70% and limbal involvement of more than 55%. In nine eyes, less than 60% of
-----------------------1~ Fig. 2 (Dua and Forrester). Case 3. Top left, Day 4 of a healing corneal abrasion caused by plaster. The circumferentially oriented tongue-shaped epithelium at the corneoscleral limbus developing from the corneal epithelium are seen clearly (curved arrows). A small area or recess of fluorescein-stained conjunctiva extending beyond the tip of the tongue-shaped projection is also seen (arrowheads). Top right, Day 5. Note the circumferentially advancing tongue-shaped projections along the corneoscleral limbus (curved arrows), the staining conjunctival recess (arrowhead), and a broader tongue-shaped projection of conjunctival epithelium (white arrow). Middle left, Day 6. The circumferentially advancing tongue-shaped projections are clearly visible. Note the convex front formation of the centripetally advancing epithelial sheet (black arrows). The centripetally advancing conjunctival epithelium can be seen close to the nasal corneoscleral limbus (white arrow). Middle right, Day 7. The advancing conjunctival epithelial sheet has extended across the corneoscleral limbus and established contact with the upper Iimbal tongue-shaped projection (arrow). Bottom left, Week 4. The area of cornea covered by conjunctival epithelium is clearly demarcated by the contact line and shows stippled fluorescein staining. Bottom right, Week 12. The area of cornea covered by conjunctival epithelium is thinner (pooling of fluorescein dye) and buds of corneal epithelium (arrows) can be seen growing into the conjunctival epithelium.
Vol. 110, No.6
Comeoscleral Limbus in Epithelial Healing
649
650
AMERICAN JOURNAL OF OPHTHALMOLOGY
December, 1990
Fig. 3 (Dua and Forrester). Case 10. Top left, Day 3 of a right corneal abrasion caused by plaster. The limbal tongue-shaped projections are beginning to form (arrows). A small area or recess of fluoresceinstained conjunctiva extending beyond the tip of the tongue-shaped projection is also seen (arrowhead). Top right, Day 6. The tongue-shaped projections have met on the corneoscleral limbus to enclose a large central abrasion. Bottom left, Day 8. The central abrasion has assumed a quadrilateral shape and shows convex fronts.
the corneal surface and less than 55% of the corneosclerallimbus were affected. In contrast, the epithelial healing pattern was not characteristic in these patients. During re-epithelialization, migration of epithelial sheets occurred in two directions: centripetally across the corneal surface, and circumferentially at the corneoscleral limbus from the remaining intact epithelium. The centripetally migrating epithelial sheets had convex advancing edges. Limbal healing occurred by the formation of two tongue-shaped processes on the corneoscleral limbus at either end of the remaining intact epithelium (Fig. 2, top left, top right, middle left; Fig. 3, top left; and Fig. 4, top left, top right). An area or recess that stained with fluorescein was always seen on the conjunctival side of these processes (Fig. 2, top left, top right, middle left; and Fig. 3, top left). The tongue-shaped processes migrated circumfer-
entially along the corneoscleral limbus until they met (Fig. 3, top right, and Fig. 5, bottom left). A contact line could be seen at this point (Fig. 5, bottom left). The cells that formed the tongue-shaped processes did not extend further than 2 to 3 mm beyond the corneoscleral limbus on the conjunctival side. Occasionally, similar circumferentially oriented projections of conjunctival limbal epithelium were seen separated from the corneal projections by a small notch (Fig. 2, top right, and Fig. 5, top left). Complete epithelial cover of the corneal surface was not re-established untillimbal healing was first completed. Once the limbal defect was covered by the advancing tongue-shaped processes, subsequent healing proceeded with the formation of convex sheets or fronts of epithelium that met each other to give the abrasion a geometric shape and eventually form single or double "Y" -shaped contact lines, as described
Vol. 110, No.6
Comeoscleral Limbus in Epithelial Healing
651
Fig. 4 (Dua and Forrester). Case 9. Top left, Day 3 of a large abrasion caused by steam and boiling water. A sheet of conjunctival epithelium has extended across the superior comeoscleral limbus (white arrow) and established contact with a circumferentially oriented limbal tongue-shaped projection (curved arrow). Top right, Day 9. A contact line is seen between the upper conjunctival epithelial and lower corneal epithelial sheets (white arrows). Nasally, limbal tongueshaped projections are seen to develop from the upper conjunctival and lower corneal epithelium (curved arrows). Bottom left, Week 4. A recurrent erosion is seen at the point of contact of the conjunctival and corneal epithelial sheets (arrow).
for the patient in Case 5 (Fig. 3, bottom left, and Fig. 5, bottom left, bottom right). Of the 15 eyes with limbal involvement, 12 healed in this manner. In the remaining three, although the tongue-shaped processes of epithelial cells appeared at the ends of the remaining intact epithelium and migrated circumferentially, they did not make contact with each other. Instead, the advancing edge of the healing conjunctival epithelial sheet reached and extended across the corneoscleral limbus, establishing contact on either side with the circumferentially migrating tongue-shaped corneal epithelial processes (Fig. 2, middle right, and Fig. 4, top left, bottom left). The sheet of conjunctival epithelium rapidly covered the remaining defect on the corneal surface. A contact line was clearly visible between conjunctival and corneal epithelial sheets. This was well defined but did not assume a "Y"
shape (Fig. 2, middle right, bottom left, and Fig. 4, top right). The layer of conjunctival epithelium covering the cornea was thinner than the adjacent newly formed corneal epithelial layer. On slit-lamp examination, the conjunctival epithelium covering the cornea showed an irregular surface and had an exaggerated stippled appearance (Fig. 2, bottom left). On no occasion did the advancing corneal epithelial sheet migrate across the corneoscleral limbus to afford cover to bare conjunctiva. On follow-up examination two months after complete healing, one of these patients showed a wedge-shaped area of thin conjunctiva between two tongue-shaped processes (Fig. 6) and had a recurrent corneal erosion at this site. At four months, this site developed a wedgeshaped epithelial scar with superficial vessels. Four weeks after complete healing, the second of these patients still showed a thin irregular
652
AMERICAN JOURNAL OF OPHTHALMOLOGY
December, 1990
Fig. 5 (Dua and Forrester). Case 1. Top left, Day 2 of a large abrasion affecting the cornea, corneosclerallimbus, and conjunctiva caused by an acidic solution. A circumferentially oriented limbal tongue-shaped projection is seen (white arrow) and is separated from a similar projection of the adjoining conjunctiva (black arrow) by a small notch. Top right, Day 3. A large temporal and a small nasal circumferentially oriented limbal tongueshaped projection are seen (arrows). Bottom left, Day 4. A contact line (arrow) is clearly visible along the points of contact between the limbal tongue-shaped projections. The contact line can be seen to extend across the corneosclerallimbus between the points of contact of similar conjunctival projections. A well-defined triangular central abrasion remains. Bottom right, Day 5. A "Y"-shaped contact line. Small arrows point to each of the small limbs of the "Y" and a large arrow to the long limb. area corresponding to the region of conjunctival epithelial cover (Fig. 2, bottom left) and had begun to attract new vessels from the conjunctival side of the corneoscleral limbus. At 12 weeks, tiny buds of corneal epithelium had appeared all along the line of contact between conjunctival and corneal epithelial sheets (Fig. 2, bottom right). Four weeks after complete healing had occurred, the third patient had pain in the eye. A small recurrent erosion had developed at one point along the line of contact
between conjunctival and corneal epithelial sheets (Fig. 4, bottom left). The rate of healing of corneal abrasions with limbal involvement was determined by calculating the percentage area of the corneal epithelial defects on successive days. Resurfacing of the corneal epithelium, as measured by the percentage reduction in the area of the defect each day, occurred in a linear manner, which indicated a constant rate of epithelial migration throughout the healing process. The average
Comeoscleral Limbus in Epithelial Healing
Vol. 110, No.6
healing time of epithelial defects involving less than 60% of the cornea and less than 55% of the corneosclerallimbus was six days, and the average healing time of defects involving more than 70% of cornea and more than 55% of the corneoscleral limbus was 11 days (Fig. 7).
Discussion
The corneoscleral limbus has been identified as the probable source of cells for the renewal of corneal epithelium. Several in vivo and in vitro studies have reported evidence to indicate a centripetal movement of cells from the corneoscleral limbus. Ebato, Friend, and Theft" cultured human corneal epithelium from different regions of the cornea and found that the highest mitotic rate and shortest doubling time were shown by limbal cells. Buck" marked the limbal epithelium and adjacent stroma with India ink in mice and noted that the epithelium moved centrally, leaving behind the ink that remained in the stroma. We previously studied the healing of large corneal abrasions in humans and described convex sheets of epithelium that appeared at the corneoscleral limbus and migrated centrally, meeting each other to form contact lines." Lemp and Mathers" studied corneal epithelial cell movements in humans by using specular microscopy and found evidence to support the centripetal movement of epithelial cells in the normal corneal epithelium. They hypothesized that the driving force behind the central movement of epithelial cells was the preferential loss of surface cells by exfoliation from the central apex secondary to shearing forces of the upper eyelid. This concept was summarized by Thoft and Friend," who suggested that the maintenance of corneal epithelial cell mass is dependent upon a continuous proliferation and centripetal migration of epithelial cells and their subsequent exfoliation from the corneal surface. Repeated mechanical denudation of the central corneal epithelium in rabbits has shown that the healing rate of the second wound is more rapid than that of the first." This has been explained by suggesting that the younger peripheral epithelial cells migrate centripetally at the time of the first denudation and therefore are capable of responding more rapidly to the second trauma. A centripetal movement of cells was observed in all patients in our study, but, when the
653
corneoscleral limbus was affected, a circumferential migration of small tongue-shaped sheets of cells was also seen to occur. This circumferential movement of cells was unique to the corneoscleral limbus. A significant centripetal movement occurred from these cells only when complete limbal cover was re-established. There appears to be a tendency for preferential re-epithelialization of the corneoscleral limbus during the healing of human corneal epithelial wounds. In all patients with limbal involvement, complete central corneal epithelial cover was not restored until the limbal epithelium had first healed. Once limbal cover was reestablished by the circumferentially migrating tongue-shaped sheets, further healing followed the sequential pattern of convex fronts, geometric shapes, and contact line formation." Evidence indicates that a proportion of the limbal basal epithelial cells are multipotent stem cells for the corneal epithelium.P'" Schermer, Galvin, and Sun 18 first provided indirect evidence for the anatomic location of stem cells. They showed that a 64K keratin was absent in the basal cells of the rabbit corneoscleral limbus but appeared in the superficial cells in that region. They suggested that the basal cells that did not stain for the 64K keratin were stem cells. These divided and acquired the
Fig. 6 (Dua and Forrester). Case 14. Eight weeks after healing of an abrasion caused by plaster, a triangular area of conjunctival epithelium is seen interposed between limbal tongue-shaped projections (curved arrows). The patient had a recurrent erosion at this site.
654
December, 1990
AMERICAN JOURNAL OF OPHTHALMOLOGY
o
20
e
~
CI
...
ii
40
~ :z:
w
w
CI
Cl
~
w
o
40
~ :z:
~
w 60
60
u
15 IL
~
80
80
5 DAYS
2
4
6 DAYS
8
10
12
Fig. 7 (Dua and Forrester). Left, The rate of healing of small corneal abrasions affecting less than 60% of the corneal surface and less than 55% of the corneosclerallimbus (N = 9). Most abrasions healed in six days. Note the exponential decrease in wound area with time. Right, The rate of healing of large corneal abrasions affecting more than 70% of the corneal surface and more than 55% of the corneosclerallimbus (N = 6). Most abrasions healed within 11 days. Note the exponential decrease in wound area with time. 64K keratin as they migrated to the center of the cornea to become terminally differentiated. Bukusoglu and Zieske" developed a monoclonal antibody that specifically binds to basal cells in the rabbit limbal epithelium. They speculated that these cells may represent limbal stem cells. Matsuda, Ubels, and Edelhauser" observed that large corneal epithelial defects of 8-mm diameter in rabbits healed at a faster rate than small defects of 4-mm diameter, which suggests that the peripheral corneal epithelium has a higher proliferative rate, probably because of the stem cells. In our study, we noted that large wounds, which affected more than 70% of the corneal surface and more than 55% of the corneoscleral limbus, healed at a slower rate than smaller wounds with less than 55% limbal involvement. This would indicate that a proportion of the limbal stem cells were depleted as a result of limbal denudation, and the rate of healing was consequently retarded. In the rabbit experiments by Matsuda, Ubels, and Edelhauser." limbal epithelium was left intact. This could account for the faster healing rate they observed with the larger wounds. As illustrated in the patient described in Case 5 (Table) of this
study and several patients from our previous study, it would appear that a zone of cells at the corneal periphery, a few millimeters wide, is comparatively more resistant to denudation by mechanical forces and chemical agents. This zone of cells tends to adhere more firmly to the basement membrane and, when intact with a large central defect, rapidly develops into centrally migrating convex sheets. When this limbal zone of cells is lost, however, healing of the corneal surface corresponding to this segment of the corneoscleral limbus is delayed. It is likely that the basal cells of this zone are the stem cells for corneal epithelium in humans. During limbal healing, we noticed that a small recess or area staining with fluorescein was always present on the conjunctival side of the circumferentially migrating tongue-shaped processes. This would suggest that the cells of these tongue-shaped processes are derived from corneal epithelium and could well represent the migration and proliferation of stem cells along the corneosclerallimbus. These cells later develop into the centripetally migrating convex sheets, which further indicates their potential to proliferate and suggests their stem cell nature.
Vol. 110, No.6
Corneoscleral Limbus in Epithelial Healing
Occasionally, similar circumferentially oriented projections of conjunctivallimbal epithelium were seen separated from the corneal projections by a small notch. On no occasion did the corneal projections migrate centrifugally to cover adjacent denuded conjunctiva, which suggests that these cells are specific for the cornea. Although a proportion of cells located on the conjunctival aspect of the corneoscleral limbus may act as stem cells for conjunctival epithelial renewal, other locations for conjunctival stem cells have to be considered. Since conjunctival epithelial sheets migrate centripetally toward the corneoscleral limbus rather than centrifugally from the corneoscleral limbus, the conjunctival fornix would appear to be a likely site for such cells. In three patients, conjunctival epithelium migrated across the corneoscleral limbus and afforded cover to the corneal surface. In all three patients, the area of cornea covered by the conjunctival epithelial sheet was thinner and the surface irregular. This area also attracted new blood vessels and was the focus of recurrent erosions in two patients. In one patient, this area formed a small peripheral scar and in another patient it was replaced by the corneal epithelium. Conjunctival epithelium is well known to re-epithelialize corneal epithelial defects and undergo several stages of morphologic transformation to resemble corneal epithelium, a process known as conjunctival transdifferentiation.P'P:" Although rapid epithelialization of the cornea by conjunctival epithelium appears desirable, it may not be ideal, as in the three patients in whom such conjunctival epithelial cover resulted in a thin and irregular corneal surface, attracted new vessels, and caused superficial scarring and recurrent erosions. Srinivasan and associates" described recurrent erosions in rabbits when healing of the corneal epithelium occurred in the absence of limbal epithelium. Huang and Tseng" described peripheral corneal neovascularization in rabbit eyes when wound healing was allowed to proceed after complete removal of limbal epithelium. On the basis of these observations and the evidence in our study, it seems reasonable to treat such abrasions by preventing the advancing edges of conjunctival sheets from reaching the corneoscleral limbus, for example by mechanical debridement or carbolic cautery, until the circumferentially migrating tongue-shaped projections of limbal epithelium have met. This
655
would ensure corneal epithelial cover for the cornea and conjunctival epithelial cover for the conjunctiva and would probably reduce the occurrence of recurrent erosions, vascularization, and superficial scarring.
References 1. Hanna, C.! Proliferation and migration of epithelial cells during corneal wound repair in the rabbit and the rat. Am. J. Ophthalmol. 61:55, 1966. 2. Buck, R. c.: Measurement of centripetal migration of normal corneal epithelial cells in the mouse. Invest. Ophthalmol. Vis. Sci. 26:1296, 1985. 3. Crosson, C. E., Klyce, S. D., and Beuerman, R. W.: Epithelial wound closure in the rabbit cornea. A biphasic process. Invest. Ophthalmol. Vis. Sci. 27:464, 1986. 4. [urnblatt, M. M., and Neufeld, A. H.: A tissue culture assay of corneal epithelial wound closure. Invest. Ophthalmol. Vis. Sci. 27:8, 1986. 5. Ebato, B., Friend, J., and Thoft, R. A.: Comparison of central and peripheral human corneal epithelium in tissue culture. Invest. Ophthalmol. Vis. Sci. 28:1450, 1987. 6. Buck, R. c.: Cell migration in repair of mouse corneal epithelium. Invest. Ophthalmol. Vis. Sci. 18:767, 1979. 7. Kuwahara, T., Perkins, D. G., and Cogan, D. G.: Sliding of the epithelium in experimental corneal wounds. Invest. Ophthalmol. Vis. Sci. 15:4, 1976. 8. Cintron, C. E., Kublin, C. L., and Covington, H.: Quantitative studies of corneal wound healing in rabbits. Curro Eye Res. 1:507, 1982. 9. Davanger, M., and Evensen, A.: Role of the pericorneal papillary structure in renewal of corneal epithelium. Nature 229:560, 1971. 10. Kinoshita, S., Kiorpes, T. c.. Friend, J., and Theft, R. A.: Limbal epithelium in ocular surface wound healing. Invest. Ophthalmol. Vis. Sci. 23:73, 1982. 11. Singh, G., and Foster, C. S.: Influence of damage to limbal epithelial cells on the morphology of central corneal epithelium and its wound healing. ARVO abstracts. Supplement to Invest. Ophthalmol. Vis. Sci. Philadelphia, J. B. Lippincott, 1988, p. 190. 12. Thoft, R. A., and Friend, J.: The X, Y, Z hypothesis of corneal epithelial maintenance. Invest. Ophthalmol. Vis. Sci. 24:1422, 1983. 13. Lavker, R. M., Dong, G., Costsarelis, G., and Sun, T. T.: Limbal basal epithelial cells display characteristics consistent with stem cells from various stratifying epithelia. ARVO abstracts. Supplement to Invest. Ophthalmol. Vis. Sci. Philadelphia, J. B. Lippincott, 1988, p. 191. 14. Thoft, R. A., Wiley, L. A., and Sundarraj, A.: The multipotent cells of the limbus. Eye 3:109,1989. 15. Tseng, S. C. G.: Concept and application of limbal stem cells. Eye 3:141,1989.
656
AMERICAN JOURNAL OF OPHTHALMOLOGY
16. Lernp. M. A., and Mathers, W. D.: Corneal epithelial cell movement in humans. Eye 3:438, 1989. 17. Dua. H. S., and Forrester, J. V.: Clinical patterns of corneal epithelial wound healing. Am. J. Ophthalmol. 104:481, 1987. 18. Schermer, A., Galvin, S., and Sun, T. T.: Differentiation-related expression of a major 64k corneal keratin in vivo and in culture suggests limbal location of corneal epithelial stem cells. J. Cell BioI. 103:49,1986. 19. Bukusoglu, c.. and Zieske, J. D.: Characterization of a monoclonal antibody that specifically binds basal epithelial cells in the limbal epithelium. ARVO abstracts. Supplement to Invest. Ophthalmol. Vis. Sci. Philadelphia, J. B. Lippincott, 1988, p. 192. 20. Srinivasan, B. D., Worgul, B. V., Iwamoto, T., and Eakins, R. E.: The re-epithelialization of rabbit cornea following partial and complete epithelial denudation. Exp. Eye Res. 25:343, 1977. 21. Matsuda, M., Ubels, J. L., and Edelhauser, H. F.: A larger corneal epithelial wound closes at a
December, 1990
faster rate. Invest. Ophthalmol. Vis. Sci. 26:897, 1985. 22. Friedenwald, J. S.: Growth pressure and metaplasia of conjunctival and corneal epithelium. Doc. Ophthalmol. 6:184, 1951. 23. Shapiro, M. S., Friend, J., and Thoft, R. A.: Corneal re-epithelialization from the conjunctiva. Invest. Ophthalmol. Vis. Sci. 21:135, 1981. 24. Tseng, S. C. G., Hirst, L. W., Farazdaghi, M., and Green, W. R.: Goblet cell density and vascularization during conjunctival transdifferentiation. Invest. Ophthalmol. Vis. Sci. 25: 1168, 1984. 25. Kinoshita, S., Friend, J., and Thoft, R. A.: Biphasic cell proliferation in transdifferentiation of conjunctival to corneal epithelium in rabbits. Invest. Ophthalmol. Vis. Sci. 24:1008, 1983. 26. Huang, A. J. W., and Tseng, S. C. G.: Corneal wound healing in the absence of limbal epithelium. ARVO abstracts. Supplement to Invest. Ophthalmol. Vis. Sci. Philadelphia, J. B. Lippincott, 1988, p. 190.
OPHTHALMIC MINIATURE
With the gold I had received I purchased a small house on the outskirts of the fashionable quarter, furnished it according to my means, and bought a slave-a scraggy fellow with One eye, but good enough for me. His name was Kaptah. He assured me that his one eye was my good fortune, for now he could tell my would-be patients in the waiting room that he had been stone blind when I had bought him and that I had given him back partial sight. Mika Waltari, The Egyptian Helsinki, Finland, WSOY, 1983, p. 65
We studied re-epithelialization of the ocular surface in 17 human eyes (14 patients) with large corneal and conjunctival abrasions. We focused on the healing of the limbal region. During re-epithelialization, cell movement was found to occur circumferentially along the corneoscleral limbus and centripetally from the corneoscleral limbus. In no patient did the central corneal defect close before the corneoscleral limbus had first re-epithelialized completely. Normal Iimbal healing was observed to occur by circumferentially migrating tongue-shaped corneallimbal epithelium. These tongue-shaped projections developed from either side of the remaining intact epithelium and advanced along the corneoscleral limbus until they met. A centripetal movement of cells from the corneoscleral llmbus then completed the healing process. In three patients, however, the advancing conjunctival epithelium extended across the corneoscleral limbus before the tongue-shaped projections of corneal limbal epithelium had met. The surface of the cornea covered by conjunctival epithelium was thin and irregular, and later showed peripheral scarring, vascularization, and recurrent erosions. MANY PUBLISHED REPORTS on the healing of corneal epithelial wounds have come from experimental studies on animals':" and in vitro studies on animal" and human" corneal epithelial cell cultures. Animal studies have confirmed that corneal epithelial healing occurs by the processes of cell migration and cell multiplication. 6.8 Evidence derived mostly from animal
Accepted for publication Aug. 28, 1990. From the Department of Ophthalmology, University of Aberdeen, Foresterhill, Aberdeen, Scotland. Reprint requests to Harminder S. Dua, M.D., Research Division, Wills Eye Hospital, 9th and Walnut Sts., Philadelphia, PA 19107.
646
experiments has been used to indicate that the process of cell multiplication occurs predominantly at the corneoscleral limbus ,"!' The concept of limbal stem cells as the source of multiplying and migrating epithelial cells is now widely accepted.P" A continuous centripetal movement of cells from the corneoscleral limbus to the center of the cornea is believed to occur during replicative epithelial turnover. This centripetal movement from the peripheral stem cells of the cornea is exaggerated after traumatic epithelial cell loss.r" Studies on corneal epithelial wound healing in humans have also illustrated similar centripetal cell movement.":" Limbal stem cells in the rabbit cornea have been demonstrated by specific staining.P:" but no identification of these cells in humans has been made. We studied the healing of large corneal epithelial wounds involving the corneosclerallimbus in 14 patients.
Patients and Methods Fourteen men (17 eyes) with extensive ocular surface abrasions who were examined at the Eye Casualty Department of the Aberdeen Royal Infirmary over a period of 3th years were included in this study. In all but one of these patients the corneoscleral limbus was affected to a lesser or greater extent and the abrasion extended on either side to involve the cornea and conjunctiva. The corneal epithelial loss ranged from 31.8% to 85.7% of the total surface area of the cornea at initial examination as measured by planimetry of fluorescein-stained color photographs (Table). The nature of injury causing the abrasion, the percentage area of the corneal abrasion, and the percentage of limbal involvement were noted (Table). All patients were initially examined within 12 hours of injury. Eyes with chemical injuries were irrigated
©AMERICAN JOURNAL OF OPHTHALMOLOGY
110:646-656,
DECEMBER,
1990
647
Comeoscleral Limbus in Epithelial Healing
Vol. 110, No.6
TABLE CLINICAL CHARACTERISTICS OF PATIENTS WITH CORNEAL ABRASIONS CASE NO., AGE (YRS)
1,34 2,30 3,26 4,28 5,36 6,35 7,40 8,43 9, 21 10,31 11,29 12,52 13,48 14, 37
CAUSE EYE
L.E. L.E. L.E. R.E. RE. L.E. L.E. RE. L.E. RE. L.E.
OF INJURY
Alcohol" Plaster t Plaster Alcohol Acid Alcohol Ammonia Ammonia
R.E. RE.
Boiling water Plaster Alcohol
L.E. L.E. L.E.
Plaster Alcohol
RE.
Plaster
WOUND AREA AT MANIFESTATION (%)
L1MBAL INVOLVEMENT (%)
55,0 48.3 85.7 77.6
21.5 15.6 67.0 66.3
60.2 52.2 37.4 54.1 59.6 35.6 75.3 73.0 81.8 51.5 48.0 31.8 71.3
TREATMENT
Prednisolone sodium phosphate Betamethasone sodium phosphate, acetylcysteine 5%
Nil Nil
10.6 17.3 54.4 12.2 65.2 56.0 57.0 18.0 12.4 14.2 58.1
Betamethasone sodium phosphate, acetylcysteine 5% Prednisolone sodium phosphate Prednisolone sodium phosphate
Betamethasone sodium phosphate
"Alcohol-based cleaning fluids. tCement and lime plaster used in the building industry.
with saline for 20 to 30 minutes. In all eyes, one drop of benoxinate hydrochloride 0.4% was instilled into the conjunctival sac before slitlamp examination. Particulate matter, if noted, was removed from the eyelid margins, conjunctival sac, and corneal surface. One drop of 1 % fluorescein was instilled and the cornea was photographed with a fundus-anterior segment camera using a blue filter. The extent of conjunctival involvement was also photographed as far as possible by asking the patient to look in different directions and by manually retracting the eyelids. All affected eyes were treated with topical antibiotics (chloramphenicol eyedrops four times a day), mydriatic and cycloplegic eyedrops (cyclopentolate hydrochloride 1 % three times a day), and covered with a patch. Six patients also received topical corticosteroids in the form of betamethasone sodium phosphate every two hours for the first two days and then four times daily (three patients) and prednisolone sodium phosphate four times a day (three patients). Two of the patients who were treated with topical betamethasone sodium phosphate also received 5% acetylcysteine eyedrops four times a day (Table). The three
patients with bilateral abrasions and four with extensive epithelial denudation were treated as inpatients. All eyes were examined daily until the abrasion healed. The cornea and conjunctiva were photographed at each examination. One drop of benoxinate hydrochloride 0.4% was instilled for the purpose of examination if the eye was photophobic. Planimetry was used to measure the area of corneal abrasion. Photographs of the cornea with the eyes in the primary position were photocopied onto graph paper, the areas of the abrasion and the cornea were calculated, and the area of the corneal abrasion was expressed as a percentage of the total corneal area. Limbal involvement was measured by determining the number of clock hours affected. One clock hour was taken to represent 8.3% of the corneoscleral limbus. The area of conjunctival involvement was not calculated. The shape and the direction of movement of the advancing edges of the conjunctival, Iirnbal, and corneal epithelial sheets were noted every day until complete epithelial cover was reestablished. Patients were followed up for two to four months thereafter.
648
AMERICAN JOURNAL OF OPHTHALMOLOGY
December, 1990
Fig. 1 (Dua and Forrester). Case 5. Left, Right eye. Right, Left eye. The patient had an acidic solution splashed into the eyes. Note the intact rim of limbal epithelium despite 60.2% of the right cornea and 52.2% of the left cornea being abraded.
Results The pattern of corneal epithelial wound healing was determined by the extent of limbal epithelial involvement. One patient (Case 5, Table) had bilateral corneal abrasions after an accident in which an acidic solution used for washing purposes on off-shore oil installations had splashed into his eyes. Large portions of the right cornea (60.2%) and the left cornea (52.2%) were denuded of epithelium. A narrow rim of intact epithelium remained at the periphery along the corneosclerallimbus in both eyes (Fig. 1). These abrasions healed by the centripetal movement of limbal epithelium. The healing pattern was characteristic and consistent with that described previously for the healing of
large corneal abrasions with an intact corneoscleral limbus in humans." Four (left eye) and five (right eye) convex sheets of epithelium appeared at the periphery from the intact epithelium. These migrated centrally and neighboring sheets established contact with each other along their sides, which left a central quadrilateral-shaped (left eye) and a pentagonal-shaped (right eye) denuded area. Further migration of these epithelial sheets resulted in the formation of "Y" -shaped contact lines. The inferior corneoscleral limbus was affected in the remaining 13 patients (15 eyes). In seven eyes, limbal involvement extended above the horizontal meridian on one or both sides. Six eyes had a total corneal epithelial cell loss of more than 70% and limbal involvement of more than 55%. In nine eyes, less than 60% of
-----------------------1~ Fig. 2 (Dua and Forrester). Case 3. Top left, Day 4 of a healing corneal abrasion caused by plaster. The circumferentially oriented tongue-shaped epithelium at the corneoscleral limbus developing from the corneal epithelium are seen clearly (curved arrows). A small area or recess of fluorescein-stained conjunctiva extending beyond the tip of the tongue-shaped projection is also seen (arrowheads). Top right, Day 5. Note the circumferentially advancing tongue-shaped projections along the corneoscleral limbus (curved arrows), the staining conjunctival recess (arrowhead), and a broader tongue-shaped projection of conjunctival epithelium (white arrow). Middle left, Day 6. The circumferentially advancing tongue-shaped projections are clearly visible. Note the convex front formation of the centripetally advancing epithelial sheet (black arrows). The centripetally advancing conjunctival epithelium can be seen close to the nasal corneoscleral limbus (white arrow). Middle right, Day 7. The advancing conjunctival epithelial sheet has extended across the corneoscleral limbus and established contact with the upper Iimbal tongue-shaped projection (arrow). Bottom left, Week 4. The area of cornea covered by conjunctival epithelium is clearly demarcated by the contact line and shows stippled fluorescein staining. Bottom right, Week 12. The area of cornea covered by conjunctival epithelium is thinner (pooling of fluorescein dye) and buds of corneal epithelium (arrows) can be seen growing into the conjunctival epithelium.
Vol. 110, No.6
Comeoscleral Limbus in Epithelial Healing
649
650
AMERICAN JOURNAL OF OPHTHALMOLOGY
December, 1990
Fig. 3 (Dua and Forrester). Case 10. Top left, Day 3 of a right corneal abrasion caused by plaster. The limbal tongue-shaped projections are beginning to form (arrows). A small area or recess of fluoresceinstained conjunctiva extending beyond the tip of the tongue-shaped projection is also seen (arrowhead). Top right, Day 6. The tongue-shaped projections have met on the corneoscleral limbus to enclose a large central abrasion. Bottom left, Day 8. The central abrasion has assumed a quadrilateral shape and shows convex fronts.
the corneal surface and less than 55% of the corneosclerallimbus were affected. In contrast, the epithelial healing pattern was not characteristic in these patients. During re-epithelialization, migration of epithelial sheets occurred in two directions: centripetally across the corneal surface, and circumferentially at the corneoscleral limbus from the remaining intact epithelium. The centripetally migrating epithelial sheets had convex advancing edges. Limbal healing occurred by the formation of two tongue-shaped processes on the corneoscleral limbus at either end of the remaining intact epithelium (Fig. 2, top left, top right, middle left; Fig. 3, top left; and Fig. 4, top left, top right). An area or recess that stained with fluorescein was always seen on the conjunctival side of these processes (Fig. 2, top left, top right, middle left; and Fig. 3, top left). The tongue-shaped processes migrated circumfer-
entially along the corneoscleral limbus until they met (Fig. 3, top right, and Fig. 5, bottom left). A contact line could be seen at this point (Fig. 5, bottom left). The cells that formed the tongue-shaped processes did not extend further than 2 to 3 mm beyond the corneoscleral limbus on the conjunctival side. Occasionally, similar circumferentially oriented projections of conjunctival limbal epithelium were seen separated from the corneal projections by a small notch (Fig. 2, top right, and Fig. 5, top left). Complete epithelial cover of the corneal surface was not re-established untillimbal healing was first completed. Once the limbal defect was covered by the advancing tongue-shaped processes, subsequent healing proceeded with the formation of convex sheets or fronts of epithelium that met each other to give the abrasion a geometric shape and eventually form single or double "Y" -shaped contact lines, as described
Vol. 110, No.6
Comeoscleral Limbus in Epithelial Healing
651
Fig. 4 (Dua and Forrester). Case 9. Top left, Day 3 of a large abrasion caused by steam and boiling water. A sheet of conjunctival epithelium has extended across the superior comeoscleral limbus (white arrow) and established contact with a circumferentially oriented limbal tongue-shaped projection (curved arrow). Top right, Day 9. A contact line is seen between the upper conjunctival epithelial and lower corneal epithelial sheets (white arrows). Nasally, limbal tongueshaped projections are seen to develop from the upper conjunctival and lower corneal epithelium (curved arrows). Bottom left, Week 4. A recurrent erosion is seen at the point of contact of the conjunctival and corneal epithelial sheets (arrow).
for the patient in Case 5 (Fig. 3, bottom left, and Fig. 5, bottom left, bottom right). Of the 15 eyes with limbal involvement, 12 healed in this manner. In the remaining three, although the tongue-shaped processes of epithelial cells appeared at the ends of the remaining intact epithelium and migrated circumferentially, they did not make contact with each other. Instead, the advancing edge of the healing conjunctival epithelial sheet reached and extended across the corneoscleral limbus, establishing contact on either side with the circumferentially migrating tongue-shaped corneal epithelial processes (Fig. 2, middle right, and Fig. 4, top left, bottom left). The sheet of conjunctival epithelium rapidly covered the remaining defect on the corneal surface. A contact line was clearly visible between conjunctival and corneal epithelial sheets. This was well defined but did not assume a "Y"
shape (Fig. 2, middle right, bottom left, and Fig. 4, top right). The layer of conjunctival epithelium covering the cornea was thinner than the adjacent newly formed corneal epithelial layer. On slit-lamp examination, the conjunctival epithelium covering the cornea showed an irregular surface and had an exaggerated stippled appearance (Fig. 2, bottom left). On no occasion did the advancing corneal epithelial sheet migrate across the corneoscleral limbus to afford cover to bare conjunctiva. On follow-up examination two months after complete healing, one of these patients showed a wedge-shaped area of thin conjunctiva between two tongue-shaped processes (Fig. 6) and had a recurrent corneal erosion at this site. At four months, this site developed a wedgeshaped epithelial scar with superficial vessels. Four weeks after complete healing, the second of these patients still showed a thin irregular
652
AMERICAN JOURNAL OF OPHTHALMOLOGY
December, 1990
Fig. 5 (Dua and Forrester). Case 1. Top left, Day 2 of a large abrasion affecting the cornea, corneosclerallimbus, and conjunctiva caused by an acidic solution. A circumferentially oriented limbal tongue-shaped projection is seen (white arrow) and is separated from a similar projection of the adjoining conjunctiva (black arrow) by a small notch. Top right, Day 3. A large temporal and a small nasal circumferentially oriented limbal tongueshaped projection are seen (arrows). Bottom left, Day 4. A contact line (arrow) is clearly visible along the points of contact between the limbal tongue-shaped projections. The contact line can be seen to extend across the corneosclerallimbus between the points of contact of similar conjunctival projections. A well-defined triangular central abrasion remains. Bottom right, Day 5. A "Y"-shaped contact line. Small arrows point to each of the small limbs of the "Y" and a large arrow to the long limb. area corresponding to the region of conjunctival epithelial cover (Fig. 2, bottom left) and had begun to attract new vessels from the conjunctival side of the corneoscleral limbus. At 12 weeks, tiny buds of corneal epithelium had appeared all along the line of contact between conjunctival and corneal epithelial sheets (Fig. 2, bottom right). Four weeks after complete healing had occurred, the third patient had pain in the eye. A small recurrent erosion had developed at one point along the line of contact
between conjunctival and corneal epithelial sheets (Fig. 4, bottom left). The rate of healing of corneal abrasions with limbal involvement was determined by calculating the percentage area of the corneal epithelial defects on successive days. Resurfacing of the corneal epithelium, as measured by the percentage reduction in the area of the defect each day, occurred in a linear manner, which indicated a constant rate of epithelial migration throughout the healing process. The average
Comeoscleral Limbus in Epithelial Healing
Vol. 110, No.6
healing time of epithelial defects involving less than 60% of the cornea and less than 55% of the corneosclerallimbus was six days, and the average healing time of defects involving more than 70% of cornea and more than 55% of the corneoscleral limbus was 11 days (Fig. 7).
Discussion
The corneoscleral limbus has been identified as the probable source of cells for the renewal of corneal epithelium. Several in vivo and in vitro studies have reported evidence to indicate a centripetal movement of cells from the corneoscleral limbus. Ebato, Friend, and Theft" cultured human corneal epithelium from different regions of the cornea and found that the highest mitotic rate and shortest doubling time were shown by limbal cells. Buck" marked the limbal epithelium and adjacent stroma with India ink in mice and noted that the epithelium moved centrally, leaving behind the ink that remained in the stroma. We previously studied the healing of large corneal abrasions in humans and described convex sheets of epithelium that appeared at the corneoscleral limbus and migrated centrally, meeting each other to form contact lines." Lemp and Mathers" studied corneal epithelial cell movements in humans by using specular microscopy and found evidence to support the centripetal movement of epithelial cells in the normal corneal epithelium. They hypothesized that the driving force behind the central movement of epithelial cells was the preferential loss of surface cells by exfoliation from the central apex secondary to shearing forces of the upper eyelid. This concept was summarized by Thoft and Friend," who suggested that the maintenance of corneal epithelial cell mass is dependent upon a continuous proliferation and centripetal migration of epithelial cells and their subsequent exfoliation from the corneal surface. Repeated mechanical denudation of the central corneal epithelium in rabbits has shown that the healing rate of the second wound is more rapid than that of the first." This has been explained by suggesting that the younger peripheral epithelial cells migrate centripetally at the time of the first denudation and therefore are capable of responding more rapidly to the second trauma. A centripetal movement of cells was observed in all patients in our study, but, when the
653
corneoscleral limbus was affected, a circumferential migration of small tongue-shaped sheets of cells was also seen to occur. This circumferential movement of cells was unique to the corneoscleral limbus. A significant centripetal movement occurred from these cells only when complete limbal cover was re-established. There appears to be a tendency for preferential re-epithelialization of the corneoscleral limbus during the healing of human corneal epithelial wounds. In all patients with limbal involvement, complete central corneal epithelial cover was not restored until the limbal epithelium had first healed. Once limbal cover was reestablished by the circumferentially migrating tongue-shaped sheets, further healing followed the sequential pattern of convex fronts, geometric shapes, and contact line formation." Evidence indicates that a proportion of the limbal basal epithelial cells are multipotent stem cells for the corneal epithelium.P'" Schermer, Galvin, and Sun 18 first provided indirect evidence for the anatomic location of stem cells. They showed that a 64K keratin was absent in the basal cells of the rabbit corneoscleral limbus but appeared in the superficial cells in that region. They suggested that the basal cells that did not stain for the 64K keratin were stem cells. These divided and acquired the
Fig. 6 (Dua and Forrester). Case 14. Eight weeks after healing of an abrasion caused by plaster, a triangular area of conjunctival epithelium is seen interposed between limbal tongue-shaped projections (curved arrows). The patient had a recurrent erosion at this site.
654
December, 1990
AMERICAN JOURNAL OF OPHTHALMOLOGY
o
20
e
~
CI
...
ii
40
~ :z:
w
w
CI
Cl
~
w
o
40
~ :z:
~
w 60
60
u
15 IL
~
80
80
5 DAYS
2
4
6 DAYS
8
10
12
Fig. 7 (Dua and Forrester). Left, The rate of healing of small corneal abrasions affecting less than 60% of the corneal surface and less than 55% of the corneosclerallimbus (N = 9). Most abrasions healed in six days. Note the exponential decrease in wound area with time. Right, The rate of healing of large corneal abrasions affecting more than 70% of the corneal surface and more than 55% of the corneosclerallimbus (N = 6). Most abrasions healed within 11 days. Note the exponential decrease in wound area with time. 64K keratin as they migrated to the center of the cornea to become terminally differentiated. Bukusoglu and Zieske" developed a monoclonal antibody that specifically binds to basal cells in the rabbit limbal epithelium. They speculated that these cells may represent limbal stem cells. Matsuda, Ubels, and Edelhauser" observed that large corneal epithelial defects of 8-mm diameter in rabbits healed at a faster rate than small defects of 4-mm diameter, which suggests that the peripheral corneal epithelium has a higher proliferative rate, probably because of the stem cells. In our study, we noted that large wounds, which affected more than 70% of the corneal surface and more than 55% of the corneoscleral limbus, healed at a slower rate than smaller wounds with less than 55% limbal involvement. This would indicate that a proportion of the limbal stem cells were depleted as a result of limbal denudation, and the rate of healing was consequently retarded. In the rabbit experiments by Matsuda, Ubels, and Edelhauser." limbal epithelium was left intact. This could account for the faster healing rate they observed with the larger wounds. As illustrated in the patient described in Case 5 (Table) of this
study and several patients from our previous study, it would appear that a zone of cells at the corneal periphery, a few millimeters wide, is comparatively more resistant to denudation by mechanical forces and chemical agents. This zone of cells tends to adhere more firmly to the basement membrane and, when intact with a large central defect, rapidly develops into centrally migrating convex sheets. When this limbal zone of cells is lost, however, healing of the corneal surface corresponding to this segment of the corneoscleral limbus is delayed. It is likely that the basal cells of this zone are the stem cells for corneal epithelium in humans. During limbal healing, we noticed that a small recess or area staining with fluorescein was always present on the conjunctival side of the circumferentially migrating tongue-shaped processes. This would suggest that the cells of these tongue-shaped processes are derived from corneal epithelium and could well represent the migration and proliferation of stem cells along the corneosclerallimbus. These cells later develop into the centripetally migrating convex sheets, which further indicates their potential to proliferate and suggests their stem cell nature.
Vol. 110, No.6
Corneoscleral Limbus in Epithelial Healing
Occasionally, similar circumferentially oriented projections of conjunctivallimbal epithelium were seen separated from the corneal projections by a small notch. On no occasion did the corneal projections migrate centrifugally to cover adjacent denuded conjunctiva, which suggests that these cells are specific for the cornea. Although a proportion of cells located on the conjunctival aspect of the corneoscleral limbus may act as stem cells for conjunctival epithelial renewal, other locations for conjunctival stem cells have to be considered. Since conjunctival epithelial sheets migrate centripetally toward the corneoscleral limbus rather than centrifugally from the corneoscleral limbus, the conjunctival fornix would appear to be a likely site for such cells. In three patients, conjunctival epithelium migrated across the corneoscleral limbus and afforded cover to the corneal surface. In all three patients, the area of cornea covered by the conjunctival epithelial sheet was thinner and the surface irregular. This area also attracted new blood vessels and was the focus of recurrent erosions in two patients. In one patient, this area formed a small peripheral scar and in another patient it was replaced by the corneal epithelium. Conjunctival epithelium is well known to re-epithelialize corneal epithelial defects and undergo several stages of morphologic transformation to resemble corneal epithelium, a process known as conjunctival transdifferentiation.P'P:" Although rapid epithelialization of the cornea by conjunctival epithelium appears desirable, it may not be ideal, as in the three patients in whom such conjunctival epithelial cover resulted in a thin and irregular corneal surface, attracted new vessels, and caused superficial scarring and recurrent erosions. Srinivasan and associates" described recurrent erosions in rabbits when healing of the corneal epithelium occurred in the absence of limbal epithelium. Huang and Tseng" described peripheral corneal neovascularization in rabbit eyes when wound healing was allowed to proceed after complete removal of limbal epithelium. On the basis of these observations and the evidence in our study, it seems reasonable to treat such abrasions by preventing the advancing edges of conjunctival sheets from reaching the corneoscleral limbus, for example by mechanical debridement or carbolic cautery, until the circumferentially migrating tongue-shaped projections of limbal epithelium have met. This
655
would ensure corneal epithelial cover for the cornea and conjunctival epithelial cover for the conjunctiva and would probably reduce the occurrence of recurrent erosions, vascularization, and superficial scarring.
References 1. Hanna, C.! Proliferation and migration of epithelial cells during corneal wound repair in the rabbit and the rat. Am. J. Ophthalmol. 61:55, 1966. 2. Buck, R. c.: Measurement of centripetal migration of normal corneal epithelial cells in the mouse. Invest. Ophthalmol. Vis. Sci. 26:1296, 1985. 3. Crosson, C. E., Klyce, S. D., and Beuerman, R. W.: Epithelial wound closure in the rabbit cornea. A biphasic process. Invest. Ophthalmol. Vis. Sci. 27:464, 1986. 4. [urnblatt, M. M., and Neufeld, A. H.: A tissue culture assay of corneal epithelial wound closure. Invest. Ophthalmol. Vis. Sci. 27:8, 1986. 5. Ebato, B., Friend, J., and Thoft, R. A.: Comparison of central and peripheral human corneal epithelium in tissue culture. Invest. Ophthalmol. Vis. Sci. 28:1450, 1987. 6. Buck, R. c.: Cell migration in repair of mouse corneal epithelium. Invest. Ophthalmol. Vis. Sci. 18:767, 1979. 7. Kuwahara, T., Perkins, D. G., and Cogan, D. G.: Sliding of the epithelium in experimental corneal wounds. Invest. Ophthalmol. Vis. Sci. 15:4, 1976. 8. Cintron, C. E., Kublin, C. L., and Covington, H.: Quantitative studies of corneal wound healing in rabbits. Curro Eye Res. 1:507, 1982. 9. Davanger, M., and Evensen, A.: Role of the pericorneal papillary structure in renewal of corneal epithelium. Nature 229:560, 1971. 10. Kinoshita, S., Kiorpes, T. c.. Friend, J., and Theft, R. A.: Limbal epithelium in ocular surface wound healing. Invest. Ophthalmol. Vis. Sci. 23:73, 1982. 11. Singh, G., and Foster, C. S.: Influence of damage to limbal epithelial cells on the morphology of central corneal epithelium and its wound healing. ARVO abstracts. Supplement to Invest. Ophthalmol. Vis. Sci. Philadelphia, J. B. Lippincott, 1988, p. 190. 12. Thoft, R. A., and Friend, J.: The X, Y, Z hypothesis of corneal epithelial maintenance. Invest. Ophthalmol. Vis. Sci. 24:1422, 1983. 13. Lavker, R. M., Dong, G., Costsarelis, G., and Sun, T. T.: Limbal basal epithelial cells display characteristics consistent with stem cells from various stratifying epithelia. ARVO abstracts. Supplement to Invest. Ophthalmol. Vis. Sci. Philadelphia, J. B. Lippincott, 1988, p. 191. 14. Thoft, R. A., Wiley, L. A., and Sundarraj, A.: The multipotent cells of the limbus. Eye 3:109,1989. 15. Tseng, S. C. G.: Concept and application of limbal stem cells. Eye 3:141,1989.
656
AMERICAN JOURNAL OF OPHTHALMOLOGY
16. Lernp. M. A., and Mathers, W. D.: Corneal epithelial cell movement in humans. Eye 3:438, 1989. 17. Dua. H. S., and Forrester, J. V.: Clinical patterns of corneal epithelial wound healing. Am. J. Ophthalmol. 104:481, 1987. 18. Schermer, A., Galvin, S., and Sun, T. T.: Differentiation-related expression of a major 64k corneal keratin in vivo and in culture suggests limbal location of corneal epithelial stem cells. J. Cell BioI. 103:49,1986. 19. Bukusoglu, c.. and Zieske, J. D.: Characterization of a monoclonal antibody that specifically binds basal epithelial cells in the limbal epithelium. ARVO abstracts. Supplement to Invest. Ophthalmol. Vis. Sci. Philadelphia, J. B. Lippincott, 1988, p. 192. 20. Srinivasan, B. D., Worgul, B. V., Iwamoto, T., and Eakins, R. E.: The re-epithelialization of rabbit cornea following partial and complete epithelial denudation. Exp. Eye Res. 25:343, 1977. 21. Matsuda, M., Ubels, J. L., and Edelhauser, H. F.: A larger corneal epithelial wound closes at a
December, 1990
faster rate. Invest. Ophthalmol. Vis. Sci. 26:897, 1985. 22. Friedenwald, J. S.: Growth pressure and metaplasia of conjunctival and corneal epithelium. Doc. Ophthalmol. 6:184, 1951. 23. Shapiro, M. S., Friend, J., and Thoft, R. A.: Corneal re-epithelialization from the conjunctiva. Invest. Ophthalmol. Vis. Sci. 21:135, 1981. 24. Tseng, S. C. G., Hirst, L. W., Farazdaghi, M., and Green, W. R.: Goblet cell density and vascularization during conjunctival transdifferentiation. Invest. Ophthalmol. Vis. Sci. 25: 1168, 1984. 25. Kinoshita, S., Friend, J., and Thoft, R. A.: Biphasic cell proliferation in transdifferentiation of conjunctival to corneal epithelium in rabbits. Invest. Ophthalmol. Vis. Sci. 24:1008, 1983. 26. Huang, A. J. W., and Tseng, S. C. G.: Corneal wound healing in the absence of limbal epithelium. ARVO abstracts. Supplement to Invest. Ophthalmol. Vis. Sci. Philadelphia, J. B. Lippincott, 1988, p. 190.
OPHTHALMIC MINIATURE
With the gold I had received I purchased a small house on the outskirts of the fashionable quarter, furnished it according to my means, and bought a slave-a scraggy fellow with One eye, but good enough for me. His name was Kaptah. He assured me that his one eye was my good fortune, for now he could tell my would-be patients in the waiting room that he had been stone blind when I had bought him and that I had given him back partial sight. Mika Waltari, The Egyptian Helsinki, Finland, WSOY, 1983, p. 65
