• Tips and Tricks for a Smoother Descemet’s Stripping Automated Endothelial Keratoplasty

    By Edwin S. Chen, MD, Neda Shamie, MD, and Mark A. Terry, MD
    Edited by Ingrid U. Scott, MD, MPH, Sharon Fekrat, MD, and John D. Sheppard, MD

    This article is from June 2007 and may contain outdated material.

    The development of endothelial keratoplasty (EK) has provided an appealing alternative to full-thickness penetrating keratoplasty (PK) for the treatment of corneal endothelial dysfunction. It provides a disease-targeted solution to a common indication for corneal transplantation and carries with it many benefits, including faster recovery, more stable and predictable surface topography, minimal change in corneal topography, and consistent refractive shift in comparison with PK.1,2 In addition, EK offers many theoretical advantages, such as a reduced risk of globe rupture due to the lack of full-thickness incision and a lower rejection rate due to the absence of epithelium, minimization of stromal tissue, and lack of sutures.

    In EyeNet and other publications, many variations have been described for performing this procedure, each with its own solutions to a variety of intra- and postoperative hurdles arising in this surgery that requires a novel skill set from the corneal surgeon. We will discuss here some commonly encountered complications in Descemet’s stripping automated endothelial keratoplasty (DSAEK) and our recommendations for avoiding and managing them.

    Tips for a Smooth Stripping of Descemet’s Membrane

    The first novel step encountered in DSAEK is the stripping of Descemet’s membrane. This maneuver can seem odd to first-time EK surgeons, who have spent most of their careers avoiding any manipulation of or contact with the corneal endothelium. This is usually performed using a blunt reverse Sinskey hook to first gently break through Descemet’s membrane, and then to score Descemet’s (i.e., drag the Sinskey to create a descemetorhexis) with the hook, following a previously marked template placed on the surface of the cornea. Once done, the tip of the hook is then used to peel Descemet’s membrane from the posterior stromal surface like wallpaper off a wall. Both of these steps should be done with gentle pressure to avoid engaging the stroma.

    A common mistake made by the novice EK surgeon is to apply excessive pressure during stripping due to the initially awkward feeling of this technique. This can cause sheets of stromal tissue to hang posteriorly or can render the stripping more difficult. To avoid these problems, watch the Descemet’s membrane carefully as the reverse Sinskey hook makes contact because scoring is clearly evident as it happens. If stromal tissue is engaged, this will be evident as dimpling of the corneal surface. It can be helpful to dilate the pupil preoperatively to provide a bright red reflex against which the edges of the score are unmistakable.

    If stromal tissue is pulled down despite careful effort to strip only Descemet’s membrane, one can simply disengage the tissue and reposition the reverse Sinskey hook. In our clinical experience, small peripheral stromal strands left behind do not interfere with quality of vision or with adherence of the donor lenticule. In fact, these areas are often clinically imperceptible as early as one week after surgery, and they may promote postoperative donor adhesion because they roughen the posterior corneal surface, much as peripheral scraping does.

    Tricks for an Uneventful Insertion of the Donor Graft

    Placement of the endothelial graft into the anterior chamber can be quite a challenge. Excessive handling of the donor tissue has been shown to reduce postoperative endothelial cell density. In addition, orientation of the tissue with the endothelial side facing posteriorly is necessary for correct functioning of the graft. Tissue insertion may be performed with a variety of instruments, each causing varying degrees of tissue compression. Studies with vital dye staining suggest that folds and points of compression kill endothelial cells; therefore, these features should be minimized. For this reason, when inserting the tissue, we prefer a single fold of the tissue using forceps that coapt only distally to minimize tissue damage. Insertion forceps such as the Charlie or Goosey type were specifically designed with these concerns in mind.

    Insertion can be performed with or without an anterior chamber maintainer. Although infusion into the anterior chamber may maintain its depth, efflux of fluid through the wound during or after insertion creates flow that can externalize the graft after insertion. Having extra metal instruments within the anterior chamber poses a risk for endothelial touch, and therefore damage. In addition, maintenance of a deeply formed chamber may promote unfolding of the graft in a rapid, uncontrolled fashion.

    To avoid these complications, we prefer placement of the graft without an anterior chamber maintainer. The graft is overfolded in a 60/40 “taco” configuration with the endothelium on the inside and placed into the eye in a rapid, fluid movement with the larger fold on top. Release of the tissue can be challenging, and it is imperative that the tissue is not unintentionally removed as the forceps are withdrawn. To avoid this problem, the forceps may be gently wiggled within the anterior chamber as they are removed.

    Chamber shallowing and sometimes collapse is the rule and can actually aid in keeping the graft overfolded. This 60/ 40 configuration strongly encourages the tissue to open posteriorly in the correct orientation with stroma facing stroma as the chamber is slowly deepened with balanced salt solution. As the graft reaches a nearly 90 degree angle, we halt the irrigation with balanced salt solution and use an air bubble to support and open the graft the rest of the way from inside the taco.

    This provides three benefits: 1) The tissue is allowed to open into perfect position in some cases, thereby avoiding the loss of endothelial cells and the creation of striae that can result from repositioning maneuvers. 2) The graft is prevented from sinking to the iris surface—a common occurrence when the graft opens completely without underlying air support—which might require more extensive manipulation for centration. 3) Tagging for stromal-endothelial orientation is not necessary. Some surgeons use techniques, including surgical marks and sutures, to “tag” the stromal surface for identification in the event that tissue orientation is lost. However, marks may last for months, and suture placement inevitably causes further endothelial cell loss. Because this process is performed slowly with complete surgeon control of the speed of balanced salt solution and air irrigation, the tissue can be unfolded under direct visualization in a controlled fashion.

    Occasionally, the graft can begin to unfold upside down. In these cases, chamber deepening should be stopped immediately. The upper half of the taco can be anchored against the recipient stroma with a Sinskey hook or a barbed 30-gauge needle (bent away from the bevel like a cystotome) while another Sinskey hook is used to tug the edge of the taco into a more open position with the correct orientation. Alternatively, an air bubble can be injected between the folds of the taco while the anterior half is anchored to the overlying recipient stroma. This maneuver has the added benefit of supporting the graft from underneath. Although these techniques surely kill endothelial cells, they are preferable to the alternatives: extensive intraocular manipulations to flip a graft that is upside down or externalizing the graft for reinsertion.

    Tips for Minimizing Postoperative Dislocations

    The most common postoperative complication of EK is graft dislocation, which occurs in 4 percent to 35 percent of cases in published series.3-5 Unlike deep lamellar endothelial keratoplasty (DLEK), where both donor and recipient stroma are cut, DSAEK provides one cut surface and one stripped surface. Electron microscopy studies have demonstrated an exquisitely smooth contour on the stripped stroma compared with the rough, shag carpet appearance of cut stroma.3 Hydrostatic forces between roughened, cut stromal surfaces may play a role in donor tissue adherence. Thus, the smooth contour of stripped stroma in DSAEK may be the source of the high rate of dislocation (up to 50 percent) encountered by experienced EK surgeons when simple apposition of the donor and host is performed. Several techniques have been developed to combat this drawback of DSAEK.

    Most DSAEK surgeons use an intra- and postoperative air bubble of varying sizes to support the graft. This varies from a chamber full of air to a small postoperative bubble that is used anywhere from less than 10 minutes intraoperatively to a full 24 hours postoperatively. One should bear in mind the risk of pupillary block glaucoma and optic nerve damage when deciding on the size and duration of the air bubble. We use 10 minutes of complete air fill intraoperatively and a postoperative air bubble matching the size of the graft. In combination with surface stroking and peripheral scraping, this is our standard technique and has yielded a published dislocation rate of 4 percent (the lowest rate of any complete series to date) and no cases of pupillary block glaucoma.3

    Surface stroking has also been advocated to remove fluid from the interface using instruments such as the Lindstrom LASIK roller or the Cindy Sweeper. This is performed by depressing the cornea centrally and stroking to the periphery. In addition, it must be done with the anterior chamber full of air so that fluid from the periphery is not refluxed into the interface. A large retrospective study has shown that this decreases the rate of dislocation to around 13 percent.6

    Scraping of the peripheral recipient stromal bed may be performed to roughen the surface and promote intra- and postoperative adhesion of the donor button. A large prospective study showed that this decreased dislocations to a rate equal to that of DLEK of around 4 percent and does not appear to carry any adverse side effects.3 This scraping, while clearly visible as white fibrils against the red reflex at the time of surgery, is clinically imperceptible in most cases by one week postoperatively and has no impact on visual acuity.

    Fenestration incisions have also been advocated for minimizing dislocations. Three to four paracentral stabs are made with a diamond blade or other instrument of surgeon’s choice through the corneal surface into the interface to remove residual fluid from the graft-host junction. A large retrospective study has shown that this technique, when added to surface stroking, decreases the rate of dislocations to around 6 percent.6Although interface infections utilizing this technique have not been described in the literature, they remain a theoretical risk that would likely be extremely difficult to treat.


    The development of EK has revolutionized the treatment of endothelial dysfunction. Thanks to the recent innovation of DSAEK, which simplifies this technique of corneal transplantation, EK is gaining in popularity due to its excellent surgical results. However, it requires a new skill set that can seem foreign to the traditional PK surgeon. Careful study and consideration before and during surgery can minimize intra- and postoperative complications, leading to a smooth case and swift visual recovery.


    1 Terry, M. A. and P. J. Ousley. Cornea 2005;24:59–65.

    2 Terry, M. A. and P. J. Ousley. Ophthalmology 2005;112:1541–1548.

    3 Terry, M. A. et al. Cornea 2006;25:926–932.

    4 Price, M. O. and F. W. Price. Ophthalmology 2006;113:1936–1942.

    5 Koenig, S. B. and D. J. Covert. Ophthalmology 2007;114(2):221–226.

    6 Price, F. W. and M. O. Price. J Cataract Refract Surg 2006;32:411–418.


    Dr. Chen is a cornea fellow, Dr. Shamie is an attending physician on the cornea service, and Dr. Terry is director of the cornea service at Devers Eye Institute, Portland, Ore.