Endothelium-in versus endothelium-out for Descemet membrane endothelial keratoplasty graft preparation and implantation

Purpose: To evaluate the difference between endothelium-in and endothelium-out for Descemet membrane endothelial keratoplasty (DMEK) preparation and implantation using injection method. Methods: The study was a randomized, comparative, institutional, laboratory investigation. Eighteen human donor corneas were included. A total of 9.5 mm DMEK grafts were stripped and restored on the corneal base with a hinge and preserved in the deswelling medium. The grafts were detached from the hinge and either manually tri-folded with the endothelium on the inside, or allowed to spontaneously roll in phosphate-buffered saline (PBS) with endothelium outwards. The corneas were mounted on an artificial anterior chamber (AAC) and four incisions made using a 15° eye knife for the side ports and slit knife to create a 3.0 mm incision for delivery of the graft into the AAC. The grafts with endothelium-in (endo-in) were pulled into a cartridge, whereas those with endothelium-out (endo-out) were aspirated using a modified Jones tube. Both were delivered using an injection method. Central and peripheral endothelial cell density (pre- and post-delivery and after removal), time of preparation and unfolding were measured. Results: Endothelial cell loss postimplantation was 10.53% (±2.82) with endo-in (n = 9) compared to 7.56% (±14.74) with endo-out (n = 9) (p > 0.05). Preparation and unfolding time was 4.43 min (±3.43) and 0.96 min (±1.10) for endo-in compared to 1.68 min (±0.57) and 4.92 min (±4.21) for endo-out. A statistical significance between endo-in and endo-out for loading (p = 0.04) and unfolding (p = 0.023) time was observed. Conclusions: Descemet membrane endothelial keratoplasty tissues can be tri-folded (endo-in) with no significantly less cell loss as compared to endo-out. Spontaneous unfolding of endo-in may reduce overall time and surgical manipulation.