Three patients with intractable PED were chosen for transplantation. The intractable PED was diagnosed after five months of standard medical and surgical treatment; preservative-free artificial tear, serum therapy, epidermal growth factor (EGF) solution, therapeutic contact lens, and amniotic membrane transplantation - temporary and permanent. These patients had nearly total limbal deficiency in both eyes due to several diseases. A blood type-matched living related donor underwent epithelial peeling, followed by denuded amniotic membrane transplantation (AMT) in the nondominant eye; then the corneal epithelial cells were expanded in vivo. Informed consent for the surgical procedure was obtained from each patient and their participating relatives. We performed in vivo cultivated corneal epithelial transplatation in three eyes in three patients with severe ocular surface disease: the primary diagnosis was a chemical burn in two cases and SJS in one (Table 1). The eyes received cultivated corneal epithelial transplantation for the purpose of covering the corneal surface and avoiding the complications that accompany PED.

AM was prepared at Bioland Korea (Cheonan, Korea) with following method. After informed consent, human AM was obtained at the time of cesarean section delivery. Under sterile conditions, the AM was washed 3 times in 1,000 mL of normal saline containing antibiotics (80 mL of Gentamicin sulfate), and all blood clots were removed. The AMs were frozen in preservative solution (Dulbecco's modified eagle's medium [DMEM] and glycerol mixed solution) and stored at -70℃; they were thawed just before use.

The amnion cells were removed by gentle scrubbing with a 20% alcohol soaked scraper. The AM was trephinized into 8 mm-diameter round excisions. First, the epithelial cell layer of the healthy donor cornea was removed with a No.15 Bard Parker blade, then the denuded round AM was transplanted onto the corneal surface with the epithelial side up using a #10-0 Nylon (Fig. 1A). A therapeutic bandage contact lens was then applied to protect the epithelium. Postoperatively, 0.3% ofloxacin (four times per day) and artificial tears (frequently) were applied topically until epithelialization was completed. A few weeks later (2-7 weeks), the expanded epithelium with the underlying AM of the donor cornea was harvested with a spatula and crescent knife. This harvest procedure was similar to superficial lamellar keratectomy; however, the stromal dissection was performed at a more superficial layer, under the AM graft. The postoperative management was the same as the in vivo cultivation.

To remove fibrovascular tissue from the patient's ocular surface, a superficial lamellar keratectomy was performed with a crescent knife and diamond burr. The cultivated corneal epithelial cell sheet on the AM was transferred onto the corneal surface; it was then sutured with 10-0 nylon. Before the temporary AM graft was covered, epithelial samples were obtained from the donor's cornea at the time of the transplantation procedure; they were then processed for microscopic examination and immunohistochemical staining. Then another AM was applied temporally with the epithelial side down over the cultivated epithelium; 20% autoserum was applied every 2 hr and 0.3% ofloxacin was used four times a day for 2 weeks after surgery.

The peripheral rim of these transplants were partly excised before surgery; they were then prepared for examination by light and JOEL JEM-1200CX transmission electron microscopy. The density of the adhesion complex such as the hemidesmosome, was evaluated as previously described (10). The hemidesmosome density was calculated based on the average of more than 5 points at 40,000 magnification. In addition, immunohistochemical staining was performed in case 1. The monoclonal antibodies against p63 and Integrin β4 were purchased from Santa Cruz (Santa cruz Biotechnology, Santa Cruz, CA, U.S.A.), and the polyclonal antibody against Connexin 43 was purchased from Chemicon (Millipore Corporation, Billerica, MA, U.S.A.). Immunohistochemical staining was performed to evaluate the expression of different molecular markers that have been proposed to identify epithelial basal cells and differentiated cells. In brief, the expanded corneal epithelium were fixed in 2% paraformaldehyde in PBS at 4℃ for 10 min and then permealized with 0.2% TritonX-100 in PBS at room temperature for 10 min. the endogenous peroxidase were quenched with 0.3% H₂O₂ in 0.5% horse serum in PBS and incubated with 5% horse serum to block the non-specific sites. Monoclonal antibodies against p63 (1:1,000), integrin β4 (1:200), or polyclonal antisera against Connexin 43 (1:200) was applied and incubated for 1 hr at room temperature, followed by incubation with biotinyated second antibodies, anti-mouse or anti-rabbit IgG, using a Vectastain Elite ABC Kit, according to the manfacturer's protocol. The samples were finally incubated with DAB peroxidase substrate to give a brown stain and counterstained with hematoxylin. After washing with PBS, the samples were mounted and analysed.

Epithelial integrity, recurrence of erosion, and decreased size of the PED on the patient's cornea were evaluated postoperatively (1st day, 3rd day, and every week) using slit lamp examination. In addition, the presence of corneal erosion, opacity, and neovascularization on the donor's cornea were evaluated. A clinical success was defined as intact epithelium that was maintained for longer than 4 weeks.

The cultivated epithelial sheet on the donor cornea was well differentiated and multilayered in appearance; they looked normal under the light microscope. They showed markedly stratified epithelia with cuboidal basal epithelial cells and a scanty cytoplasm. The epithelialization on the AM of the donor cornea was completed 2 weeks after expansion. After that time, they showed a well-formed hemidesmosome under the electron microscope (Fig. 1B, C). The calculated hemidesmosome density was 4.3±1.2/2.25 µm, 4.0±1.6/2.25 µm, and 4.2±0.6/2.25 µm at 2, 3, and 7 weeks of culture, respectively.

On immunohistochemical staining of case 1, p63, connexin 43, and integrin β4, were expressed in the epithelial sheet representing well-developed adhesion complexes, intercellular junctions, and the existence of epithelial cells (Fig. 2).

The epithelium that was expanded on the donor cornea was transparent and clear just like normal corneal epithelium. Clinically, the PED of the recipient's cornea was healed with these epithelial sheets (Fig. 3, 4). However, a bandage contact lens was needed to protect the cornea because of its fragility. In detail, the surface of cornea was covered with in vivo expanded epithelium, but a slight small sized erosion was present. The in vivo grown corneal epithelium functioned successfully in all cases; the clinical success rate was 100%. The PED in all cases was covered by 2 weeks after surgery (average 2 weeks). The transplanted epithelium was maintained for 5, 4, and 27 weeks on each of the patient's cornea, respectively (average 12 weeks). The PED recurred in cases 1 and 2 (SJS and severe alkali burn patients). A clear surface was maintained for 27 weeks in case 3, and was followed by a penetrating keratoplasty procedure. The donor cornea recovered completely without any complications at 1 week, 10 days after harvest. Complications such as recurrent erosion, corneal opacity, and neovasculization were not observed.