In this issue, Straiko and Associates highlight the increasing use of Descemet stripping automated endothelial keratoplasty (referred to as DSEK or DSAEK) for treatment of endothelial dysfunction, including eyes with a failed penetrating keratoplasty (PK). DSEK has undergone extremely rapid adoption, especially in the United States, where in 2009 it constituted more than 40% of all corneal transplantations, a 10-fold increase in 5 years. Key factors driving widespread use are quick visual recovery and superior safety profile relative to conventional PK, benefits that derive from the small incision and use of air rather than sutures to hold the DSEK graft in place. Mirroring the increase in DSEK use, the diversity of proposed technique variations continues to expand. However, it is often difficult to separate the wheat from the chaff, because most new variations have been evaluated only in small case series without concurrent comparison groups, let alone validation in prospective, randomized trials.

Whether performed as a primary procedure or regraft, the most frequent endothelial keratoplasty complication is early graft detachment. Therefore, identification of the root causes and best methods to prevent detachment are of particular interest.

When treating a failed PK, Straiko and associates suggest that the DSEK diameter should be the same or smaller to avoid the irregularities common on the posterior surface of the PK wound and to prevent graft detachment. Using this approach in 17 cases, they observed 1 detachment requiring air reinjection (5.8% detachment rate). Conversely, other surgeons have achieved similarly low detachment rates using oversized grafts. Since originating use of DSEK to treat failed PK in 2004, our center has performed 57 consecutive cases with 3 detachments (5.3%) using a diameter approximately 1 mm larger than the failed PK. This detachment rate with oversized grafts is remarkably similar to that of Straiko and associates with smaller grafts, and we have observed that DSEK grafts can fold and conform extraordinarily well even to very irregular PK posterior wound surfaces. Likewise, in 31 consecutive cases, Busin and Patel observed only 1 detachment (3.2%) using oversized grafts (Patel A. Descemet stripping automated endothelial keratoplasty (DSAEK) following penetrating keratoplasty (PK). Paper presented at the European Society of Cataract and Refractive Surgery (ESCRS) Meeting. September 7, 2010; Paris, France.). Furthermore, in 74 cases pooled from 9 centers, Clements and associates found that undersized grafts were associated with a somewhat higher detachment rate than oversized or same-sized DSEK grafts under failed PK. When different surgeons achieve similar detachment rates with opposite approaches to graft sizing, it suggests that graft size is not a root cause of DSEK detachment from a failed PK.

Some have suggested that residual Descemet membrane affects graft attachment. When treating Fuchs dystrophy, it is essential to strip Descemet membrane to remove guttata, which scatter light and cause visual distortion. However, to minimize the risk of wound dehiscence when treating a failed PK, we believe that it is safer to leave Descemet membrane intact, if it does not have guttata or other light-scattering abnormalities. The only 2 areas with substantial strength in nonvascularized PK wounds are Descemet membrane and Bowman layer. In eyes with a previous relaxing incision in the wound to control astigmatism, only Descemet membrane remains strongly healed. Highlighting the fragility of the healed PK incision, we have observed 1 wound dehiscence in 56 cases (1.8%), and Busin and Patel (Patel A. Descemet stripping automated endothelial keratoplasty (DSAEK) following penetrating keratoplasty (PK). Paper presented at the European Society of Cataract and Refractive Surgery (ESCRS) Meeting. September 7, 2010; Paris, France.) observed 1 in 31 cases (3.2%). In contrast to Straiko and associates’ imaginative misinterpretation of our original technique paper, our case of wound dehiscence did not occur while stripping Descemet membrane outside the wound margin, something we have never done. It occurred while placing a traction suture before even entering the eye. Therefore, we are concerned by Straiko and associates’ recommendation that surgeons should strip Descemet membrane and roughen the peripheral stroma to avoid DSEK detachment from a failed PK. Even careful scoring runs the risk of Descemet membrane tearing into the wound and weakening it, and stromal roughening could weaken the wound further. In addition, it is not clear that the extra effort confers any benefit. With stripping and roughening, Straiko and associates observed a 5.8% detachment rate, whereas in 49 cases without any stripping or roughening, we have observed 2 detachments (4.1%). Likewise, Clements and associates reported no detachments in 5 cases of failed PK treated without stripping Descemet membrane.

If neither the relative graft sizing nor residual Descemet membrane is a root cause of DSEK detachment, then what is? Assuming the graft did not sustain major surgical trauma, root causes of detachment seem to be lack of a complete initial air fill to press the graft firmly into place, hypotony in the early postoperative period, eye rubbing, or a combination thereof. Consistent with these proposed causes, Straiko and associates note that their detachment occurred in an eye with a wound leak that resulted in early postoperative hypotony. In a soft eye, every blink allows the eyelids to indent the corneal surface, pressing the donor inward, away from the recipient cornea, and this allows fluid into the interface, leading to detachment. Therefore, preventing wound leaks and not placing endothelial keratoplasty grafts in eyes with hypotony are keys to safeguarding against detachment.

Preventing graft detachment and wound dehiscence are not the only concerns in eyes with failed PK. One of the most important preoperative evaluations is to determine whether the anterior chamber has sufficient space to place a DSEK graft without danger of anterior synechia or iris touch.

Finally, regrafts have notoriously poorer survival rates than primary grafts, so optimizing graft survival is an important consideration. Small-diameter PKs have a relatively small reserve of healthy endothelial cells and are associated with reduced survival rates. Therefore, we are concerned that Straiko and associates’ recommendation to frequently undersize the DSEK graft could affect long-term survival negatively. Although eyes with Fuchs dystrophy may have a substantial reserve of peripheral recipient endothelial cells, as suggested by Balachandran and associates, this reserve is likely already to have been depleted by migration onto a failed PK. Because the posterior corneal surface is larger than the anterior surface (evident when performing trabeculectomies and similar surgeries), it easily can support a larger diameter DSEK than PK without undue proximity to the limbus, and large (8.5 to 9.0 mm) primary DSEK grafts have excellent 5-year survival. Further study is needed to assess the effect of DSEK diameter on long-term survival for various treatment indications.

We believe it is important for surgeons to maintain an open mind in evaluating the root causes of complications and best means of prevention until well-controlled studies document superiority of certain techniques over others. Certainly, meticulous attention to wound construction and surgical detail are key to achieving the low detachment rates reported by Straiko and associates and others.