Histopathology of posterior lamellar endothelial keratoplasty graft failure.

PURPOSE

To describe the histopathologic findings and relevant clinical details of 3 patients undergoing penetrating keratoplasty (PK) after failure of posterior lamellar endothelial keratoplasty (EK).

METHODS

Retrospective clinicopathologic case series. Patients 1 and 2 underwent EK for pseudophakic bullous keratopathy. Patient 3 underwent EK for persistent corneal edema secondary to a Descemet membrane (DM) detachment after cataract extraction. Patient 1 had persistent diffuse corneal edema and broad, long-standing iridocorneal adhesions that precluded repeat EK. Patient 2 had high intraocular pressure and severe anterior chamber inflammation 1 day postoperatively with subsequent noncorneal clearing and elected PK over repeat EK. Progressive corneal edema with resultant poor visual acuity after EK was the reason for PK in patient 3.

RESULTS

Histopathologic examination disclosed thickened, edematous corneas with attenuated endothelium consistent with graft failure caused by endothelial decompensation in all 3 cases. Although various degrees of posterior lamellar graft detachment were observed in each instance, significant parts of each graft remained adherent to the host stroma or to segments of residual host DM. The wounds in the adherent areas, although discernible, were relatively inconspicuous, resembling those seen at the flap-stromal interface after laser in situ keratomileusis. The donor graft endothelium was atrophic in all cases, and a delicate retrocorneal fibrous membrane was present in 2 cases. Most of the graft in cases 1 and 2 remained adherent, with small areas of peripheral detachment. In contrast, the graft in case 3 adhered peripherally but had separated from the stroma centrally, forming a thin cleft.

CONCLUSIONS

Histopathology suggests endothelial decompensation, incomplete graft adherence, and the formation of retrocorneal fibrous membranes as possible etiologies for EK failure. The adherence of endothelial grafts to residual host DM suggests that it may not be necessary to remove optically clear DM before endothelial graft placement. The inconspicuous nature of the EK interface suggests that it may not play a large role in image degradation, although more study is needed.