Clinical appearance of failed Descemet stripping automated endothelial keratoplasties. A, Case 22. The graft is detached (arrows), and an air bubble is present in the anterior chamber. B, Case 28. The cornea is diffusely edematous in primary graft failure.
Endothelial cell loss. A, Case 8. The corneal stroma is thickened and edematous. A linear Descemet stripping automated endothelial keratoplasty scar is present (arrows), and there are 3 endothelial cells per high-power field (hematoxylin-eosin, original magnification ×100). B, Case 5. There are 0 endothelial cells per high-power field in the edematous graft (hematoxylin-eosin, original magnification ×100).
Retained Descemet membrane. A, Case 15. A piece of periodic acid–Schiff [PAS]–positive, partially folded Descemet membrane is sandwiched by the recipient and donor corneal stroma. The graft remains adherent to the central portion of the host stroma (PAS, original magnification ×25). B, Case 15. Electron microscopic examination shows retained Descemet membrane (arrows) present in the host-donor interface (original magnification ×1900). Inset shows that the interface contains electron-dense fibrillar material with scattered pigment granules (original magnification ×19 000). C, Case 9. Residual Descemet membrane is present peripherally, and there are 12 endothelial cells per high-power field in the cornea (PAS, original magnification ×25). D, Case 9. Descemet membrane has prominent posterior nodular excrescences (arrows), consistent with Fuchs endothelial dystrophy, indicating recipient Descemet membrane (PAS, original magnification ×100).
Fibrocellular tissue. Case 34. A, Proliferation of keratocytes with myofibroblastic differentiation is present on the peripheral stromal bed (arrows). The central portion of the stromal bed is smooth and hypocellular (thick arrows) (hematoxylin-eosin, original magnification ×25). Inset shows the proliferation of myofibroblasts at the periphery (hematoxylin-eosin, original magnification ×100). B, The fibrocellular tissue stains positive for the myogenic marker smooth-muscle actin (peroxidase antiperoxidase, original magnification ×100).
Epithelial implantation. A, Case 27. The lenticule is covered by multilayered epithelium on the anterior surface, along with the margin, and on the posterior surface (periodic acid–Schiff [PAS], original magnification ×25). Inset, These epithelial cells stain positive for cytokeratins AE1/AE3. The clefts are artifact (peroxidase antiperoxidase, original magnification ×100). B, Case 27. Transmission electron microscopic examination shows multilayered squamous epithelial cells with surface microvillae (arrows) present on the posterior surface of the Descemet membrane (original magnification ×2900). C, Case 11. A proliferation of nonkeratinized stratified squamous epithelium is present, forming a cyst between the recipient and the donor stroma (hematoxylin-eosin, original magnification ×25). D, Case 11. This cyst contains fluid with PAS-positive material in the lumen (original magnification ×100).
Fungal keratitis. A, Case 6. The graft is mostly detached. Residual Descemet membrane with posterior nodular excrescences is present at the host stromal surface. Periodic acid–Schiff (PAS)–positive fungal elements were observed at the host-donor interface (original magnification ×25). B, Case 6. Higher magnification shows yeast on the surface of the stromal bed (PAS, original magnification ×100). C, Case 12. The graft is detached. The corneal stroma is inflamed and ulcerated, with fungal hyphal elements and polymorphonuclear leukocytes (PMNs) in both recipient and donor sides (arrows). There are numerous PMNs present in the anterior chamber (thick arrows) (PAS, ×4). D, Case 12. Higher magnification shows fungal hyphae in the stroma (PAS, original magnification ×40).
Concomitant laser in situ keratomileusis (LASIK) scar. Case 4. A Descemet stripping automated endothelial keratoplasty lenticule (arrows) is partially detached from the posterior corneal surface, and a scar from previous LASIK surgery (thick arrows) is present anteriorly (hematoxylin-eosin, original magnification ×25).
Customize your JAMA Network experience by selecting one or more topics from the list below.
Zhang Q, Randleman JB, Stulting RD, et al. Clinicopathologic Findings in Failed Descemet Stripping Automated Endothelial Keratoplasty. Arch Ophthalmol. 2010;128(8):973–980. doi:10.1001/archophthalmol.2010.140
To evaluate the clinical features of and histologic findings from failed Descemet stripping automated endothelial keratoplasty (DSAEK).
This retrospective observational case series evaluated 47 consecutive corneal specimens from 42 patients who underwent either penetrating keratoplasty or repeated DSAEK for failed DSAEK. Clinical information was obtained for the cases. Sections of the specimens were examined using light microscopy. Immunohistochemical staining was performed for cytokeratins AE1/AE3 and for the myogenic marker smooth-muscle actin when indicated. Transmission electron microscopic examination was performed in some cases.
Graft survival ranged from 0.5 to 34 months. Histologic examination showed that 94% of the specimens (44 of 47) had endothelial cell loss. Residual host Descemet membrane (19%; 9 of 47), fibrocellular tissue (19%; 9 of 47), epithelial implantation (15%; 7 of 47), and fungal infection (4%; 2 of 47) were also identified. Immunohistochemical stains were positive for AE1/AE3 in the epithelial implantations and for smooth-muscle actin in cells in the fibrocellular proliferations.
The principal cause of failed DSAEK is endothelial cell loss. Residual host Descemet membrane, fibrocellular tissue at the edge of the lenticule, and epithelial implantation are common histologic findings. Fungal infection may occur in the setting of DSAEK.
In contrast to penetrating keratoplasty (PK), endothelial keratoplasty (EK) allows for selective replacement of diseased corneal endothelium. With the explosion of popularity in the past 10 years, EK has evolved in nomenclature and technique, with published studies including posterior lamellar keratoplasty, deep lamellar EK, Descemet stripping EK, and, most recently, Descemet membrane EK. The simplification of donor tissue preparation using an automated keratome has been referred to as Descemet stripping automated EK (DSAEK).1 Donor tissue preparation using a microkeratome rather than meticulous manual hand dissection has led to the wide acceptance of DSAEK as the preferred EK procedure. In addition to increased efficiency and ease of donor tissue preparation, DSAEK compared with PK allows for earlier visual recovery, earlier refractive stability, more predictable postoperative refractive outcomes, avoidance of wound- and suture-related complications (including postkeratoplasty astigmatism), and a reduced risk of intraoperative and late suprachoroidal hemorrhage.2
However, there are issues with DSAEK regarding complications of keratoplasty. Multiple published series1,3 found that the top 3 most common complications included graft dislocation, endothelial rejection, and primary graft failure. In a recent review2 of published DSAEK articles, the average graft failure rate was 5% (range, 0%-29%), a higher rate than what had been published for conventional PK. Although several alterations in the DSAEK surgical technique have occurred, from increased incision size to new insertion techniques and new tissue insertion devices, the failure rate with EK remains concerning. To better understand the causes of DSAEK failure, we evaluated the clinicopathologic characteristics in 47 cases of failed DSAEK in 42 patients who subsequently underwent PK or repeated DSAEK. The objective of this study was to evaluate the clinical features of and histologic findings from failed DSAEK, which may lead to an increased success rate for this procedure.
Corneal grafts from 47 failed DSAEK cases in 42 patients who subsequently underwent PK or repeated DSAEK were obtained between January 1, 2007, and December 31, 2009, in the L. F. Montgomery Ophthalmic Pathology Laboratory, Emory University, Atlanta, Georgia. The specimens were routinely processed for light microscopic examination by fixation with formalin, 10%, dehydration in increasing grades of alcohol, xylene clearing, paraffin embedding, sectioning at 7-μm thickness, and staining with hematoxylin-eosin or periodic acid–Schiff. Specimens with suspected epithelial implantation were immunostained using the avidin biotin complex method4 for the anticytokeratin antibodies AE1/AE3 (Dako Corp, Carpentaria, California). Grafts with fibrocellular tissue at the edge of the lenticule were immunostained with smooth-muscle actin (Dako Corp). Seven cases were also examined using transmission electron microscopy. Briefly, a portion of each specimen was fixed in glutaraldehyde, 2.5%, and postfixed in 0.1M cacodylate buffer and osmium tetroxide, 1%. The specimens were dehydrated and embedded in epoxy resin, and 0.1-μm-thick sections were cut and stained with toluidine blue. Ultrathin sections were obtained and were stained with uranyl acetate and lead citrate. Examination was performed using a transmission election microscope (JEOL 100CX II; JOEL Ltd, Tokyo, Japan). After Emory University institutional review board approval was granted, the medical records of the corresponding patients were retrospectively reviewed.
Forty-seven failed grafts from 42 patients were examined. The clinical features of these cases are given in Table 1. Of the 47 cases, 15 underwent PK and 32 underwent repeated DSAEK. The average age of the patients was 71 years. The most common preoperative diagnosis was Fuchs dystrophy, followed by pseudophakic bullous keratopathy. Twelve eyes underwent concurrent DSAEK and cataract extraction with intraocular lens implantation. The survival duration of the grafts ranged from 0.5 to 34 months (mean, 5 months). The complications that led to DSAEK failure included detachment of the graft, graft rejection, and primary graft failure (Figure 1).
The histologic findings from the PK and repeated DSAEK specimens from the 47 cases are given in Table 2 and Table 3, respectively. The thickness of the host corneal stroma measured 400 to 800 μm (mean, 511 μm) in the 15 PK specimens, and the thickness of the failed DSAEK lenticules in all 47 cases ranged from 40 to 300 μm (mean, 171 μm). The features of failed grafts were categorized as (1) endothelial cell loss, (2) retained host Descemet membrane, (3) fibrocellular proliferation at the edge of the lenticule, (4) epithelial implantation, and (5) fungal infection. In addition, 3 grafts had melanin pigment granules in the interface and 1 had a concomitant laser in situ keratomileusis (LASIK) scar. These characteristics occurred randomly across the study. The pathologic findings provided clues to the causes of the DSAEK failures.
Comparison with a nomogram derived from age-matched corneas from the L. F. Montgomery Laboratory of Ophthalmic Pathology5 showed that 44 of the 47 cases had far fewer endothelial cells than did age-matched normal corneas. Eighteen grafts had 0 endothelial cells per high-power field, and 26 had varying degrees of endothelial attenuation (1-5 endothelial cells per high-power field) (Figure 2). Thus, corneal endothelial attenuation or absence was the most common pathologic finding in failed DSAEK. Residual Descemet membrane on the anterior stromal interface of the failed graft was identified in 9 of the 47 cases (Figure 3). In these cases of retained host Descemet membrane, most of the lenticules remained adherent to the host stroma. Another common feature of failed DSAEK that occurred in 9 of the 47 cases was the presence of fibrocellular tissue at the edge of the lenticule, sometimes extending into the interface. These cells in the fibrocellular tissue stained positive for smooth-muscle actin (Figure 4). This likely represented a wound-healing response, similar to what occurs at the edge of a LASIK flap. Epithelial implantation was present in 7 cases. The epithelial implantation manifested as a sheet of stratified squamous epithelium along the edge and extending onto the back of the lenticule or as an implanted island of epithelium in the interface between the donor and host stroma. The proliferated epithelial cells immunostained positive for AE1/AE3 (Figure 5). Two failed grafts contained fungal elements in the corneal stroma, anterior chamber, or host-donor interface (Figure 6), and 1 case contained both a DSAEK lenticule and a LASIK flap (Figure 7). In addition, there were melanin pigment granules at the host-donor interface in 3 cases.
The pathologic characteristics of failed DSAEK have been described in previous articles.6-11 These studies were based on 2 to 19 cases and showed that corneal endothelial loss is the most common finding in failed DSAEK, consistent with the present findings. Incomplete recipient Descemet membrane stripping and fibrocellular tissue associated with the wound were also reported.7-11 Compared with the previous series, the current one is the largest to date (47 cases) and includes for the first time, to our knowledge, immunohistochemical and ultrastructural findings along with a novel finding of fungal infection as a cause of failed DSAEK.
Results of previous studies7,8 have noted that attenuation or absence of endothelial cells after DSAEK was the principle factor contributing to persistent edema and graft detachment, which is in accord with the present findings. Endothelial cell loss occurs during lenticule preparation, surgical placement of the lenticule, or postoperatively. In contrast to PK or deep lamellar EK, the corneal lenticule for DSAEK is harvested by using a microkeratome and an artificial anterior chamber from the corneoscleral rim. The donor tissue preparation step may lead to donor endothelial cell damage from mechanical trauma during placement in the artificial anterior chamber, air exposure, elevated pressure in the artificial chamber, or trauma during the donor trephination step.12,13 Also, endothelial cell loss may occur during storage as a function of preservation to surgery time.14 Endothelial trauma also occurs during the various surgical steps of EK from donor tissue folding, forceps injury, wound compression during tissue insertion, and iris and posterior chamber lens trauma if an anterior chamber collapse occurs. There is a learning curve with this new surgical procedure: various modifications to the surgical technique and instrumentation may simplify the surgery and reduce endothelial damage in the future.15-19
Graft dislocation or detachment is the most frequently reported complication of DSAEK. Dislocated grafts have been rebubbled or repositioned; however, those approaches may cause subsequent endothelial cell loss.20,21 In addition, endothelial damage may occur when the lenticule is detached and touches the iris. Nonsurgical treatments, such as postoperative face-downward positioning, may be helpful to reposition or reattach the lenticule.22 Eye rubbing is also a risk factor for graft dislocation, especially in the early postoperative period.23 Terry et al24 found that selectively roughening the peripheral recipient stromal bed can promote donor edge adhesion and help prevent donor dislocation. The present findings of fibrocellular tissue at the edge of the lenticule that sometimes extends onto the anterior surface of the lenticule indicate that wound healing occurs at the edge.
Incomplete removal of the recipient Descemet membrane may also occur in DSAEK. There is controversy regarding whether it is necessary to completely remove the recipient Descemet membrane before insertion of the donor graft. Mondloch et al25 reported a failed DSAEK that exhibited delaminated, retained Descemet membrane on the recipient stromal bed, and 2 previous case series of failed DSAEK7,8 showed a high incidence of retained Descemet membrane, which may play a role in lenticule nonadherence. However, another review of 10 PK specimens from patients with previous DSAEK indicated that retained Descemet membrane was not a critical factor for lenticule dislocation.10 In the present study, we found that the grafts can remain attached to the host even if retained Descemet membrane is present at the interface, suggesting that retained Descemet membrane may not seem to hinder graft adhesion.
Fibrocellular tissue at the margin of the lenticule extending into the interface was found in 9 of 47 cases in this series. Cells in this fibrocellular tissue immunostained for smooth-muscle actin, and ultrastructural examination showed intracytoplasmic filaments with fusiform densities, indicating myofibroblastic differentiation. This represents fibrocellular scar formation as occurs in the hypercellular peripheral scar of a LASIK flap26 and may contribute to adherence of the lenticule to the recipient stroma. Dislodged epithelial tissue or fragments of iris may be introduced onto the edge of the lenticule during lamellar surgery, thus leading to proliferation of stromal keratocytes and myofibroblastic differentiation with peripheral scar formation.27 In contrast, the central portion of the DSAEK scar is a lamellar, horizontal, hypocellular wound that is histologically and ultrastructurally similar to the central portion of a LASIK wound and allows for optical clarity.26 There are few keratocytes and no myofibroblasts in the center of the donor-host interface. There is a lack of healing in this area, and we suspect that DSAEK lenticules have the potential to detach in this area. This is similar to a lack of healing in the central portion of the LASIK wound, which allows the LASIK flap to be lifted many years after the procedure. This also allows for a potential space for fungal infection, as occurred in one of the 2 fungal infection cases in this series.
Eccentric trephination of the donor lenticule may introduce donor epithelium during insertion, suggesting an invasive epithelial implantation process. This has been confirmed by Saelens et al28 using an in situ hybridization test. In addition, 2 of us (W.B.L. and H.E.G.) have prepared DSAEK lenticules from eye bank corneas and have found histologic evidence of the introduction of donor epithelium onto the lenticule (unpublished data, 2008). Alternatively, host epithelial cells may adhere to the stromal surface of the folded lenticule as it is inserted into the anterior chamber or may be introduced by intraocular instruments during insertion or rebubbling procedures. When suspected clinically, careful pathologic examination of the graft and immunohistochemical staining for keratins with AE1/AE3 may be helpful to confirm the epithelial cells, as we found in the present cases.
Infection is a rare complication in DSAEK surgery. Although herpes simplex virus epithelial keratitis and cytomegalovirus endotheliitis have been reported in failed DSAEK cases,29,30 fungal keratitis has not been reported. We had 2 cases of fungal infection in this series of failed DSAEK. Fungal infection associated with DSAEK may be due to (1) transmission of the fungus from donor corneal tissue into the recipient, (2) the recipient ocular surface harboring latent fungal elements without clinical disease, or (3) contamination of surgical instruments. In case 12, the patient developed symptoms of redness, photophobia, decreased vision, and corneal infiltrate at the third postoperative week. Cultures of the donor rim were negative, although cultures of the patient's cornea demonstrated the presence of Candida albicans. A concurrent recipient's eyelid culture also confirmed the same pathogen. The patient was treated with hourly topical amphotericin B and gatifloxacin along with oral fluconazole. The infiltrate became larger, with increased corneal ulceration, early corneal neovascularization, anterior chamber inflammation, and development of a hypopyon. A PK was urgently performed to remove the infectious region. Histopathologic examination showed the presence of fungal hyphae consistent with the cornea culture results. In case 6, the patient remained asymptomatic except for the lack of graft clearance with a persistent epithelial and stromal edema in follow-up. Cultures of the donor rim were positive for Candida glabrata, the same pathogen as found in the DSAEK graft. In both cases, the use of topical corticosteroids after DSAEK likely potentiated the fungal infection.
This study demonstrates the histopathologic findings after failed DSAEK. The most common findings associated with DSAEK failure were endothelial cell loss or attenuation, retained recipient Descemet membrane fibrocellular tissue at the edge of the lenticule, epithelial implantation, and infection. We hope that these findings will help with the understanding of the causes of DSAEK failure and will lead to procedures that will increase its success rate.
Correspondence: Hans E. Grossniklaus, MD, MBA, L.F. Montgomery Ophthalmic Pathology Laboratory, BT428 Emory Eye Center, 1365 Clifton Rd, Atlanta, GA 30322 (firstname.lastname@example.org).
Submitted for Publication: December 28, 2009; final revision received January 6, 2010; accepted January 7, 2010.
Financial Disclosure: None reported.
Funding/Support: Supported in part by grant NEI P30EY06360 from the National Institutes of Health and by an unrestricted departmental grant from Research to Prevent Blindness Inc.