Introduction: Amniotic membrane's unique combination of properties including the facilitation of migration of epithelial cells, the reinforcement of basal cellular adhesion and the encouragement of epithelial differentiation [6] together with its ability to modulate stromal scarring and its anti-inflammatory and anti-bacterial activity has led to its use in the treatment of ocular surface pathology as well as an adjunct to stem cell grafts of the corneal limbus [6-4]. We report a prospective study of 30 patients so treated.

Material and methods: We studied 31 eyes of 30 patients subjected to amniotic membrane grafts between September 1999 and May 2000. There were 25 men and 5 women with an average age of 60.1 (range 25-86) years who were followed for a mean of 7.7 (range 4-11) months. 5 groups (A to D) were observed: A: 6 eyes. Small chronic ulcers without limbal involvement. B: 4 eyes. Ulcers of at least 75% corneal area or occupying 75% of the limbus. C: 9 eyes. Corneal burns. D: 8 eyes. Painful bullous corneal dystrophies unresponsive to other treatment. E: 4 eyes. Symblepharons. Amniotic membrane was placed on the corneal lesion, epithelial surface externally [6, 15], trimmed and sutured with interrupted 10/0 nylon, removed at one month. In two patients (11, 12) inflamed conjunctiva was recessed and amnion sutured to the recessed margin. For the bullous dystrophies we removed all the corneal epithelium and either sutured the amnion to peri-limbal conjunctiva (4 eyes) or to the limbus (4 eyes). For the symblepharons the conjunctiva was dissected to reform the fornix which was lined with amniotic membrane, sutured with 8/0 vicryl. Patients were reviewed regularity.

Results: Group A: All healed within 15 days, in most with dissolution of the amnion over 2-3 months although some persisted, covered with corneal epithelium. An eye with a Descemetocoele and one with a microperforation both healed. Vision improved more than two lines in 4 of 6 eyes. Group B: 2 of 4 eyes healed, one despite detachment of the membrane after 15 days. One eye was salvaged by tarsorrhaphy over a fresh keratoplasty after perforation of a neuroparalytic ulcer on failure of three successive amnion grafts. The final cornea vascularised despite an amnion graft for a meta-herpetic ulcer. Group C: 2 of 9 eyes had limbal damage in one quadrant but 7 had vessels in at least three-quarters of the circumference. One (15) also had a limbal autograft. 3 of 9 eyes healed satisfactorily with more than 2/10 improvement in acuity in each case. 2 showed further neovascularisation despite surface healing. One old chemical burn healed satisfactorily but vascularisation remained 5 eyes failed to heal with lysis of the graft, the patient who had a limbal autograft developed a vascular pannus, and in 4 eyes neovascularisation progressed to cover the entire cornea. Group D: 3 eyes settled with loss of symptoms but in 5 the graft detached within 15 days. All eyes where the membrane had been sutured to the conjunctiva beyond the limbus failed whilst 3 of 4 in which it had been sutured anterior to the limbus succeeded, leaving a persistent whitish membrane under the epithelium. Group E: We were able to reconstruct the cul de sac in 3 out of 4 eyes. In one patient with recurrent pterygium good ocular movement was restored, previously limited by scarring. One with associated ocular surface damage from a thermal burn failed by scarring of the cul de sac a month after surgery.

Discussion: Our best results were in persistent trophic ulcers of the cornea (Groups A and B) with a success rate of 80%, comparable to those of others [49, 37, 38]. The ready availability of amniotic membrane in our facility makes amniotic membrane transplantation the main secondary treatment for such lesions, especially because of the visual improvement we obtained. Because we did not observe any improvement in corneal thickness after this treatment we advise its early use before significant stromal lysis. The technique was not sufficient to control the effect of corneal anaesthesia in two eyes [40] or in chemical burns suggesting that amniotic membrane alone is insufficient to promote corneal healing in the absence of limbal stem cells. Nevertheless, three eyes did benefit. It has been suggested [13] that the anti-apoptotic function of amnion may prevent stem cell loss in such eyes [42], thus it appears logical to offer an amniotic membrane graft first, before stem cell transplantation, which may entrain complications in the donor eye if autografted [43] or because of the rejection risk of an allograft. It may be that an amniotic membrane graft simply becomes a holding procedure allowing time to settle the eye so as to allow secondary procedures to address the underlying cause of further damage. Our treatment of bullous dystrophy only succeeded on confining the graft to within the limbus, 3 out of 4 eyes becoming comfortable. By contrast we found amniotic membrane helpful in reconstructing symblepharons in the absence of local inflammation.

Conclusion: Amniotic membrane grafting is a simple and straightforward surgical technique which should form part of the therapeutic arsenal for the treatment of ocular surface disease. Indications for the technique need further clarification for it is evident that it cannot correct all secondary pathology associated with limbal destruction. It is certainly preferable to conjunctival advancement and has proved useful in the reconstruction of the cul-de-sac.