Pinnita Prabhasawat, MD, Department of Ophthalmology, Siriraj Hospital, Faculty of Medicine, Mahidol University, 2 Prannok Road, Bangkoknoi, Bangkok 10700, Thailand.

Tel: (662) 411-2006 Fax: (662) 411-1906 E-mail: sippb@mahidol.ac.th

Supported in part by the grant of Faculty of Medicine, Siriraj Hospital

Presented in part at the 7th International Conference of the Asia Pacific Association of Surgical Tissue Banking, Kuala Lumpur, Malaysia, November 1998, in part at the Thai-Japan Joint Meeting, Bangkok, Thailand, January 1999 and in part at the fourth Annual Meeting of the Ocular Surface and Tear Conference, Miami, Florida, May 1999.

Methods: Preserved human amniotic membrane transplantations were performed in 84 eyes of 78 patients for corneal surface reconstruction. The indications were limbal stem cell deficiency from Steven-Johnson syndrome, chemical burn and herpes keratitis (27 eyes), bullous keratopathy (26 eyes), persistent epithelial defect and dellen (17 eyes), band keratopathy (11 eyes), preparing for prosthesis (1 eye), corneal ulcer (1 eye) and acute chemical burn (1 eye).

Results: Success was noted in 83.3% (70/84) eyes, partial success in 13.1% (11/84) eyes, and failure in 3.6% (3/84) eyes for an average follow-up of 10.5 months (3 to 29 months). No patient developed major immediate post-operative complications.

Conclusion: Amniotic membrane transplantation can reduce inflammation, promote corneal epithelial healing, and decrease irritation in corneal surface problems.

AMT = Amniotic membrane transplantation

EBSS = Eale’s balanced salt solution

ED = Epithelial defect

EDTA = Ethylene diamine tetrachloroacetic acid
LT = Limbal transplantation

PED = Persistent epithelial defect

PKP = Penetrating keratoplasty

PO = Punctal occlusion

Human amniotic membrane was first used as a graft for treating conjunctival defects (DeRoth 1940, p.522-5) and pterygium (Panzardi 1947, p.332-6) since 1940. In 1993, preserved human amniotic membrane was successfully used in corneal surface reconstruction in rabbit eyes by Kim and Tseng (1995). Since then, this technique has been applied as an adjunct in the treatment of many ocular surface disorders. In corneal surface reconstruction, the amniotic membrane is reported to have been used in persistent epithelial defects (Lee et al. 1997; Kruse et al. 1999; Azuara-Blanco et al 1999), combined with limbal transplantation in Steven-Johnson syndrome, and chemical burns (Tsubota et al 1996; Shimazaki et al. 1997; Tseng et al. 1998; Tsubota et al 1999), bullous keratopathy (Pires et al. 1999) and for reducing haze in following phototherapeutic keratectomy (PRK) (Wang et al. 1997; Choi et al. 1998).

An ideal graft for use in ocular surface reconstruction would not only, to a maximum, promote healing while minimizing scarring, but would also yield cosmetically acceptable results and be relatively easy to perform. Human amniotic membrane clearly has shown promise along these lines. To further investigate its applicability, we have used this technique for corneal surface reconstruction in the treatment of several severe eye conditions encountered in a Bangkok public hospital.

This study was approved by the committee for protection of human subjects in research at the Faculty of Medicine, Siriraj Hospital, Mahidol University. From September, 1997 to March, 2000, amniotic membrane transplantation was performed in 78 patients (84 eyes) with various corneal surface disorders after obtaining their informed consents. Their clinical data, indications for surgeries and goal of treatments in each group are summarized in Tables 1 and 2.

The patients were divided into 2 groups depending on the indications. The first group of 21 cases (27 eyes) showed limbal stem cell deficiency, caused by Steven-Johnson disease (13 cases, 18 eyes), chemical burn (7 cases, 8 eyes), and HSV keratitis (1 case, 1 eye). In this group, 12 patients (12 eyes) received amniotic membrane transplantation (AMT) combined with limbal transplantation in the same or subsequent operation, of which 7 patients (7 eyes) received autograft and 5 patients (5 eyes) received allograft limbal transplantation. Nine patients (13 eyes) received AMT alone because of their financial difficulties of obtaining systemic cyclosporin which is indispensable for allograft limbal transplantation. The second group consisted of other corneal diseases such as bullous keratopathy (26 cases, 26 eyes), persistent epithelial defect, dellen and descemetocele (17 cases, 17 eyes), band keratopathy after removal of calcium deposit (11 cases, 11 eyes), preparing for prosthesis (1 case, 1 eye), corneal ulcer (1 case, 1 eye), and acute chemical burn (1 case, 1 eye).

After obtaining the informed consent, the patient was subjected to the same routine admitting preoperative tests. One surgeon (PP) performed all surgeries using different techniques depending on the indications.

After retrobulbar injection or topical anesthetic drop (2% tetracaine eye drop), the base of the epithelial defect was debrided and covered with a piece of amniotic membrane sutured to the edge of the defect with interrupted 10-0 nylon. A bandage lens was applied until the defect was healed. The procedure was similar to that previously described (Lee et al. 1997).

Clinical outcome was categorized into success, partial success, and failure based on the criteria as in Table 1. When all goals of surgery were achieved in terms of relief of pain and irritation, reduction of inflammation and scaring and promotion of epithelial healing, the outcome was defined as "success”. When part of these goals were achieved, e.g., relief of pain but still some inflammation or recurrence of the pain symptom exist, the outcome was defined as "partial success". When none of the surgical goals was met, the outcome was defined as "failure".

In limbal stem cell deficiency patients, 18 out of 27 eyes (66.7%) were rated as success, and these patients felt more comfortable without irritation or inflammation, and showed improved vision. The visions improved from hand motion to at least 20/200 in 4 eyes of AMT combined with limbal allograft transplantation and in 5 eyes of AMT combined with limbal autograft transplantaion. The vision improved from hand motion to at least 10/200 or more than 2 snellen lines in 6 eyes of AMT alone (Table 3). Seven out of 21 eyes (25.9%) were rated as partial success, with decreased inflammation, but mild fibrovascular tissue invasion into the grafts or slight recurrence of symblepharon. Importantly, all these patients with partial success felt more comfortable and satisfied with the outcome to the extent they would like to have the other eye operated. Two eyes (7.4%) were rated as failure. One of them showed recurrent conjunctival keratinization due to severe keratoconjunctivitis sicca. The other case still showed the same inflammation as before surgery. One eye showed corneal graft rejection after 9 months post triple surgery, combined amniotic membrane, limbal and corneal transplantation, however amniotic membrane graft still being quiet and indicating no sign of rejection. Intriguingly, all cases rated as partial success or failure suffered from Steven-Johnson syndrome.

In painful bullous keratopathy, 24 out of 26 eyes were rated as success and became pain-free with a stable corneal surface after AMT. The partial success cases were pseudophacic and traumatic bullous keratopathy. In both cases, pain and irritation were relieved in the first 1.5 and 2 months after surgery, but recurred thereafter, and required the use of contact lens to relieve pain and irritation. Nine out of eleven eyes with band keratopathy, which were rated as success, felt much more comfortable without irritation or pain and no recurrence after removal of calcific band keratopathy with 3% EDTA and the whole cornea covered with the amniotic membrane. Two patients were rated as partial success due to recurrence of band keratopathy at postoperative 5 months and 7 months, respectively, but they did not have any irritation and remained pain-free.

All 17 cases with persistent epithelial defect, dellen or descemetocele were rated success and their corneal surfaces were healed after AMT. The causes for their surface breakdowns were diverse and included herpes simplex keratitis, corneal ulcer, lid abnormalities from tumor removal, corneal surface irregularity, post radiation, following corneal graft rejection and post phacoemulsification. The average duration for healing epithelial defects was 2.0 weeks (ranging from 1 to 4 weeks). Among them, 2 cases had failed with bandage contact lens treatment prior to the amniotic membrane graft. In the five cases with descemetocele and dellen, the corneal stromal thickness increased after the healing promoted by amniotic membrane graft.

In one case, which suffered from irritation and pain, that prevented this blind patient from wearing eye prosthesis, AMT was successful to eliminate such irritation and pain that he did not undergo conjunctival flap. In the case with superficial corneal ulcer following candidiasis superficial keratectomy and amniotic membrane graft successfully restored the corneal surface integrity. The epithelial defect and ulcer was completely healed in 1 week despite prior failure following medical treatments for more than 3 weeks. However, in one case with acute (2 weeks after burns) stage 4 burn with a 9 mm epithelial defect with severe ischemia, amniotic transplantation failed to decrease inflammation and to promote epithelial healing.

Late complications were those which developed after 1 month of surgery and the epithelial surface had healed completely. Three eyes developed corneal ulcer. The first case suffering from Steven-Johnson syndrome developed corneal ulcer in 1 month after surgery due to contamination of the eye drop. The second case with pseudophacic bullous keratopathy developed corneal ulcer in 2 months after surgery as a result of the dust in the eye. The third case, suffering from limbal deficiency and persistent corneal epithelial defect due to chemical burn, received autolimbal transplantation and amniotic membrane transplantation and penetrating keratoplasty 2 months later. These procedures resulted in a clear and stable corneal graft 3 months after penetrating keratoplasty, but developed bacterial corneal ulcer which was successfully treated with topical antibiotics. Three eyes developed minute corneal perforation in Steven-Johnson syndrome in 2 months, 2 months and 9 months, respectively. In these three eyes, corneal perforation was less than 1 mm in diameter without signs of infection and was successfully treated with glue and contact lens.

An HIV-positive 36 year-old male, diagnosed with Steven-Johnson syndrome and severe keratoconjunctivitis sicca, complained of painful irritation and inability of opening his eyes. The vision was 20/200 in both eyes, and the conjunctiva was marked injected in both eyes and symblepharon was noted in the temporal conjunctiva of the left eye (Fig.a). Despite occlusion of both upper and lower puncta, he remained symptomatic with marked inflammation. AMT was performed in both eyes in two settings to cover the whole cornea and bulbar conjunctiva 5 millimeters from the limbus as a graft. After surgery, his vision improved to 20/80, he felt much more comfortable, and there was no conjunctival injection in both eyes (Fig. b) at a follow-up of three months except at the previous symblepharon area in the left eye.

A 50 year-old female suffered from irritation and blurred vision for many years due to limbal deficiency from herpes simplex keratitis and severe keratoconjunctivitis sicca in her right eye, for which she had received a penetrating keratoplasty 4 years ago. Her visual acuity in her right eye was counting finger at 2 ft. Moderate conjunctival injection was noted in both eyes, and all four puncta were occluded for treating severe keratoconjunctivitis sicca. Conjunctivalization was noted 360⁰ of her cornea (Fig. c). AMT was performed to cover the whole cornea and bulbar sclera 5-7 mm from the limbus in her right eye. After surgery, her eye had no irritation and the bulbar conjunctiva was smoothly wettable without inflammation. Three months later, allograft limbal transplantation was performed resulting in improved vision of 20/50, clear cornea without vascularization, and quiet conjunctiva (Fig. d) for a follow-up of 29 months.

A 66 year-old female suffered from irritation and tearing in her right eye due to corneal decompensation following extracapsular cataract extraction and anterior chamber lens implantation 5 years ago. Her vision was hand motion, and the cornea was edematous with epithelial bullae (Fig. e). Following epithelial scraping and AMT, the epithelial defect healed in 1 week, and the transplanted amniotic membrane became transparent and difficult to be identified in one month (Fig. f). Eight months after surgery, her cornea remained edematous and bullae recurred, but she still enjoyed the comfort without irritation for the entire follow up of 18 months.

A one-eyed 87 year-old female developed persistent corneal epithelial defect from herpes simplex keratitis for more than 2 weeks, which was refractory to treatment including bandage contact lens and frequent tear substitutes. Her vision dropped to 3/200, and the defect showed progressive thinning (Fig.g). AMT was performed to cover the epithelial defect, which showed healing at postoperative day 6 and completed in 3 weeks. At that time, amniotic membrane partially dissolved, but still attached at the lesion (Fig.h) and totally dissolved after 2 months. Her vision improved to 20/50, the corneal stroma became thickening and the corneal epithelium did not show any recurrent breakdown for 15 months of follow-up.

Loss of corneal stem cells due to destruction by external or internal injuries such as chemical burn, severe infection, and Steven-Johnson syndrome cause the limbal stem deficiency disease. Corneal surface gets conjunctivalization and vascularization. This can cause discomfort, photophobia, irritation, inflammation, decreased vision and recurrent corneal epithelial defect. It is a difficult disorder in which conventional penetrating keratoplasty has a limited role. Tseng (Tseng et al 1998) recently showed that amniotic membrane transplantation alone can be used to reconstruct corneal surfaces with partial limbal deficiency, i.e., there is part of the limbus retaining healthy limbal epithelial stem cells. In this study, when limbal deficiency is not severe, AMT alone can restore corneal surface smoothness and eliminate photophobia – e.g. case 1 - and improve the visual outcome. As shown by the result 6 eyes in the group of AMT alone had visual improvement - at least 2 snellen lines - even they still had corneal vascularization (Table 3). Furthermore, in this group, except 2 patients (7.4%), all others felt more comfortable, moist, less irritation, less keratinization and pleased with the result and asked to do the other eye as well. The reason why the patients feel more moist in their eyes and less keratinization might be because the amniotic membrane can increase goblet cell density of conjunctiva (Prabhasawat et al. 1997). Besides, cyclosporine is not required in AMT, which is a benefit for developing countries like Thailand for which the high expenses of cyclosporine represent a serious problem. However, when the limbal stem cell deficiency becomes total and diffuse, AMT alone is not sufficient (Tseng et al. 1998) and requires additional limbal stem cell transplantation for corneal surface reconstruction (Tsubota et al. 1996; Shimazaki et al. 1997; Tsubota et al. 1999). Here we also support this concept as all 5 eyes receiving AMT combined with limbal allografts and 5 eyes receiving AMT combined with limbal autograft had visual improvement and symptomatic relief while 9 eyes receiving AMT alone could not improve the vision.

In bullous keratopathy, the cornea is edema due to endothelial dysfunction. The corneal epithelium also becomes unhealthy and gets an irregular surface causing ocular pain and surface breakdown. This generally requires penetrating keratoplasty to relief the ocular discomfort and to improve vision. Lee (Lee et al. 1997) first reported that amniotic membrane can be used as an alternative to conjunctival graft to relief pain and recurrent erosion. We noted here amniotic membrane is an excellent alternative for countries like Thailand where there is a shortage of corneal donors, to relief pain and reduce surface inflammation while the patient is on the waiting list for corneal transplantation. Although the mechanism remains unclear, we are intrigued by the finding that symptomatic relief persists even there remains marked edema and recurrent bullae several months following AMT. This result is the same as the report of Pires (Pires et al. 1999).

When there is a persistent epithelial defect, corneal ulcer and perforation can develop. The conventional methods to treat the defect include the use of contact lens, conjunctival flap (Gundersen 1958; Insler et al. 1987) or tarsorrhaphy (Welch et al. 1988). Consistent with what has been reported by Lee (Lee et al. 1997) and Kruse (Kruse et al. 1999), our result also noted an overwhelming success of treating this indication. Surprisingly, AMT not only can promote epithelial healing, but also increase the corneal thickness in the case of dellen and descemetocele. This prevents corneal perforation, thus solving the surface healing problem refractory to medical treatments and contact lens wear. Furthermore, it also gave a good cosmetic result and improved vision (Case 4).

We did not note any sign of graft rejection, and no patient became clinically worse than before the surgery in term of visual acuity or inflammation. Only one case developed mild entropion after symblepharon lysis in Steven-Johnson syndrome, and this complication slowly resolved 2 months later without the need of surgical correction. Immediate complications were not serious and treatable. For example, bacterial conjunctivitis was successfully treated with topical antibiotics. The cause of infection was not directly linked to the preparation of amniotic membrane as microbial cultures of the amniotic membrane during preparation or following surgeries were all negative. Late complications may or may not be related to the procedure itself as they can be derived from the underlying disorders. For example late occurrence of corneal ulcer and perforation developed after the surface had healed. Furthermore, it is well known that diseases such as bullous keratopathy, penetrating keratoplasty and Steven-Johnson syndrome have weak ocular surface defense and are prone to develop corneal ulcer. Among all ocular surface disorders in this study, it is worth noting that Steven-Johnson syndrome is the disease which developed most complications and showed the lowest success.

In summary, this study has shown that the amniotic membrane can be used effectively to treat the corneal surface problems, ease to obtain and no need of immunosuppresive treatment like other organ transplantation. Future studies to understand more about the action of amniotic membrane transplantation may lead to other clinical uses.

Table 1 Criteria for success, partial success and failure in each category