Endothelial cell loss and visual outcome of deep anterior lamellar keratoplasty versus penetrating keratoplasty: a randomized multicenter clinical trial.

Objective: To compare endothelial cell (EC) loss, visual and refractive outcomes, and complications after deep anterior lamellar keratoplasty (DALK) and penetrating keratoplasty (PK).

Design: Randomized, multicenter clinical trial.

Participants: Fifty-six eyes of 56 patients with a corneal stromal pathology not affecting the endothelium were randomized to DALK or PK.

Methods: The DALK procedure was performed according to Anwar's big-bubble technique. Patients underwent an ophthalmic examination preoperatively and 3, 6, and 12 months postoperatively.

Main Outcome Measures: Endothelial cell loss, refractive and topographic astigmatism, spherical equivalent, uncorrected visual acuity, and best spectacle-corrected visual acuity (BSCVA) were measured, and complications were recorded.

Results: Endothelial cell loss was significantly higher after PK compared with DALK procedures performed without perforation of Descemet's membrane (12 months: 27.7% ± 11.1% vs. 12.9% ± 17.6%). The BSCVA was significantly better in the PK group at 3 and 6 months after surgery but was not significantly different 12 months after surgery (0.39 ± 0.3 logarithm of the minimum angle of resolution [logMAR] in DALK and 0.31 ± 0.3 logMAR in PK). At 12 months postoperatively, refractive and topographic astigmatism in the DALK and PK groups were -3.37 ± 2.3 diopters (D) and -3.76 ± 2.1 D (P = 0.53), and 3.57 ± 2.3 D and 4.16 ± 2.0 D (P = 0.34), respectively. (Micro)perforation of the Descemet's membrane occurred in 32% (9/28) of the DALK eyes, and 18% (5/28) of the patients required conversion to PK. Endothelial cell loss was not significantly different between DALK and PK when cases with perforation of Descemet's membrane were included in the (intention-to-treat) analysis (12 months: 19.1 ± 21.6 vs. 27.7 ± 11.1 P = 0.112). Rejection episodes were reported in 1 patient in the DALK group (epithelial rejection) and 3 patients in the PK group (all endothelial rejections). No graft failure occurred.

Conclusions: One year after DALK performed without perforation of Descemet's membrane, EC loss is significantly lower, whereas the BSCVA is comparable to that in the PK group. In addition, no endothelial rejection occurred in the DALK group. However, Descemet's membrane perforation remains a major complication in DALK and warrants improvements to standardize the big-bubble technique.