Editor:

We read the recent results of “Descemet Stripping Automated Endothelial Keratoplasty under Failed Penetrating Keratoplasty,” written by Straiko and associates, with great interest. They describe modifying their standard Descemet stripping automated endothelial keratoplasty (DSAEK) surgical technique by careful stromal scraping to avoid the graft–host interface and purposely sizing the DSAEK donor tissue to fit inside the diameter of the penetrating keratoplasty (PK). With this technique, only 1 graft dislocation (5.9%) was noted. Price and Price reported a retrospective series of 7 eyes, with visual improvement in 6 of 7 cases and a partial donor detachment in 1 case. Covert and Koenig published a retrospective series of 7 eyes with visual improvement in most eyes, but 2 of 7 primary graft failures and donor dislocation in 3 of 7 eyes. Lee and associates retrospectively reported on 8 patients, most of whom had improved vision, despite 3 of 8 requiring repositioning or rebubbling and 1 primary graft failure. Clements and associates report a large multicenter collaboration with 97 eyes and a 31% dislocation rate, but only 2% primary graft failure.

Despite generally improved graft clarity and visual outcomes, considerable variation in surgical technique and incidence of complications, including dislocation and graft failure, exist. At our institution, 22 consecutive eyes of 19 patients treated with DSAEK after failure of previous PK by 2 surgeons were reviewed. The mean preoperative best visual acuity at the time of surgery was worse than 20/180. The mean follow-up period was 8.2 ± 7.5 months (range, 1 to 26 months). All except 2 cases in our series had improved visual outcomes, with 8 cases (36.3%) achieving a best visual acuity of 20/40 or better and 15 (68.2%) achieving a best visual acuity of 20/60 or better. Only 1 dislocation (4.5%) was noted, which required rebubbling with subsequent reattachment.

All our grafts were punched using a trephine the same measured diameter of the failed PK. Descemet membrane was not removed or scraped in any case in our series, but all cases had venting incisions. We used a glide-and-pulling technique through a 4.5-mm limbal clear corneal incision. Although we attempted to produce identically sized grafts, intraoperative variability in attempted versus achieved cut diameter was noted, and some PK–host overlap of the DSAEK tissue was noted in many cases.

DSAEK after failed PK demonstrated favorable clinical results in our series. Graft–host interfaces were well apposed, as demonstrated clinically and on optical coherence tomography, although unlike Straiko and associates, we did not use preoperative optical coherence tomography to aid in determining DSAEK graft size. Of 22 cases evaluated, only 1 had a significant complication, developing an acute allograft rejection, although is unclear whether this complication was related to the surgical method of DSAEK. All other cases in our series had clear grafts, with no evidence of decompensation. Our dislocation and failure rates are similar to the those in the series reported by Straiko and associates, despite significant differences in technique.

Although the results of DSAEK after failed PK are promising, longer follow-up and prospective comparisons may be helpful in determining the relative value of different surgical methods to treat endothelial failure in selected PK patients.