Long-Term Endothelial Outcomes and Dislocation Risks in Small Versus Large Grafts for Descemet’s Stripping Endothelial Keratoplasty

Introduction

Bullous keratopathy (BK) is an end-stage condition of corneal endothelial failure, characterized by corneal edema and opacity resulting in profound visual impairment [1]. Its pathogenesis involves dysfunction of the endothelial cell layer, which fails to maintain corneal dehydration. This allows aqueous humor to infiltrate the stroma and subepithelial space, leading to edema and the formation of epithelial bullae [2]. These bullae rupture spontaneously during normal eyelid blinking, exposing nerve endings and causing severe pain, epiphora, and photophobia [3]. Repeated cycles of rupture and healing can further induce corneal neovascularization and scarring, contributing to irreversible vision loss and a significant reduction in quality of life [4].

Consequently, effective surgical intervention is required. Although penetrating keratoplasty (PKP) was once the standard procedure, it has been largely superseded by less invasive alternatives such as Descemet’s stripping endothelial keratoplasty (DSEK) [5]. By replacing the diseased endothelium through a small incision while preserving the recipient’s corneal stroma and epithelium, DSEK facilitates faster visual recovery, reduces surgically induced astigmatism, and can lower the risk of immune rejection compared to PKP [6]. Furthermore, the use of manually dissected grafts enhances the accessibility and cost-effectiveness of DSEK, particularly in resource-limited settings [7].

As this technique becomes more widely adopted, optimizing its surgical parameters becomes increasingly crucial. Among these, donor graft diameter is an important consideration, potentially influencing anterior chamber anatomy, graft-host apposition, and long-term endothelial cell survival [8]. However, comparative evidence regarding the impact of graft diameter on clinical outcomes in BK patients undergoing DSEK remains insufficient. Therefore, this study aimed to systematically compare the postoperative outcomes between small grafts (7.0 mm <diameter ≤8.0 mm) and large grafts (8.0 mm <diameter ≤9.0 mm) in patients with BK undergoing DSEK, with the goal of providing evidence-based guidance for optimal graft selection.

Material and Methods

STUDY DESIGN AND PATIENTS:

Patients with BK who underwent DSEK at our hospital were enrolled. Based on the graft diameter prepared during surgery, patients were divided into 2 groups: a small-graft group (7.0 mm <diameter ≤8.0 mm) and a large-graft group (8.0 mm <diameter ≤9.0 mm). The study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of the General Hospital of the Northern Theater Command of the People’s Liberation Army (Ethics Approval No. Y[2024]284). The requirement for written informed consent was waived for this retrospective study.

The inclusion criteria were: (1) diagnosis of bullous keratopathy confirmed by medical history, clinical ocular manifestations, and slit-lamp microscopy; (2) no improvement with conservative treatment and meeting indications for corneal endothelial transplantation; (3) availability for postoperative follow-up for at least 12 months. Exclusion criteria were: (1) presence of other ocular conditions that could affect postoperative visual recovery, such as lagophthalmos, severe dry eye disease, macular pathology, or optic atrophy; (2) psychiatric or psychological disorders impairing compliance with ophthalmic examinations; (3) age <18 or >90 years; (4) severe allergic constitution, systemic connective tissue disease, significant cardiovascular or cerebrovascular disease, pregnancy, lactation, or planned pregnancy.

CORNEAL DONOR SOURCES:

With informed consent from the donor families in accordance with ethics guidelines, all donor corneas were obtained from fresh, healthy deceased donors within 2 h of death and were used for surgery within 48 h. Preoperatively, each donor cornea underwent thorough examination using slit-lamp and corneal endothelial microscopy to confirm a smooth and transparent surface and a well-organized endothelium with a cell density of ≥2000 cells/mm².

SURGICAL PROCEDURE:

All surgical procedures were performed by the same experienced corneal specialist. Diameter selection was determined by the surgeon according to recipient corneal diameter, anterior chamber depth, and donor tissue availability, aiming to optimize graft-host apposition and anterior chamber clearance. Patients in both groups received the current standardized preoperative protocol of our center, which included ocular surface cleaning, lacrimal irrigation, eyelid trimming, and conjunctival sac irrigation. Intraocular pressure (IOP) was maintained within the normal range, and broad-spectrum antibiotic eye drops were administered preoperatively. For all patients, the pupil was constricted with topical pilocarpine nitrate before graft insertion to facilitate attachment and prevent pupillary block. Topical anesthesia was achieved using 2% proparacaine hydrochloride eye drops, supplemented by a retrobulbar block with 2% lidocaine to soften the globe. Under sterile conditions, the host bed was prepared by trephination. A 5-mm tunnel incision was created, accompanied by 2 corneal side ports. Descemet’s membrane was scored and stripped using a Sinskey hook, after which sterile air was injected to restore the anterior chamber. The donor cornea was mounted and trephined to match the intended graft diameter using a punching technique. The graft was carefully inserted into the anterior chamber via a plastic glide and unfolded with balanced salt solution. The corneal incision was closed with interrupted 10-0 nylon sutures. After ensuring full positioning of the graft, additional balanced salt solution and sterile air were injected to facilitate adhesion, during which the “double-chamber sign” was observed. Gentle corneal compression was applied to eliminate excess air. Tobramycin-dexamethasone ointment was applied at the end of the procedure.

POSTOPERATIVE MANAGEMENT:

From the first day, postoperative management included intravenous 20% mannitol (250 mL twice daily) as needed. Anti-infective and anti-rejection therapy was also initiated postoperatively. Levofloxacin and prednisolone acetate eye drops were administered 4 times daily, with tobramycin-dexamethasone ointment applied at night. Anti-rejection therapy involved either tacrolimus (twice daily) or cyclosporine eye drops (4 times daily). Prior to complete re-epithelialization, bovine basic fibroblast growth factor eye drops were administered 4 times daily to promote healing, with therapeutic soft contact lenses considered for persistent epithelial defects. Graft adherence and IOP were closely monitored. Sutures were typically removed at 3 to 6 months postoperatively.

OBSERVATION INDICATORS:

During hospitalization, patients were assessed daily for best-corrected visual acuity (BCVA), IOP, and slit-lamp examination of corneal epithelial healing, graft-host apposition, graft clarity, suture status, and anterior chamber reaction. Scheduled outpatient follow-ups at 1, 3, 6, and 12 months included BCVA, IOP, refractive error, ECD, anterior segment photography, and graft apposition. Corneal topography and other examinations were performed as needed. ECD was measured by specular microscopy, with the average of 3 readings recorded.

Since preoperative ECD was unmeasurable due to corneal opacity, the 1-month postoperative value was used as the baseline for calculating subsequent endothelial cell loss (ECL) rate. BCVA was converted to the logarithm of the minimum angle of resolution (logMAR) for analysis. Graft transparency was graded from 0 (completely transparent) to 5 (completely opaque) on slit-lamp examination, with grade 1 defined as mild superficial haze and grade 2 as mild stromal haze with visible pupil and iris. Postoperative secondary glaucoma was defined as an IOP >25 mmHg or a 30% increase from the preoperative baseline within 1 year postoperatively [9], requiring intervention. Graft rejection was diagnosed in a previously clear graft after ≥2 weeks based on ≥2 of the following signs: Khodadoust line, keratic precipitates, anterior chamber reaction, stromal edema, or ciliary injection. Graft dislocation was defined as partial separation from the host bed.

STATISTICAL ANALYSIS:

Data were analyzed using IBM SPSS Statistics (version 29.0). A complete case analysis was performed without imputation of missing data. Normality of continuous variables was assessed with the Shapiro-Wilk test. Homoscedasticity was analyzed with the Levene test. Normally distributed data are presented as mean±standard deviation (SD) and compared with the independent-samples t test. Non-normally distributed data were analyzed using the Mann-Whitney U test. Categorical data were compared with the Pearson’s chi-square or Fisher’s exact test. The risk ratio (RR) and its 95% confidence interval (CI) were calculated using the logarithmic method. Additionally, ordinal data were analyzed using nonparametric rank-sum tests. Statistical comparisons between groups were conducted independently at each follow-up time point. Graft survival was compared using Kaplan-Meier analysis, with an event defined as the need for secondary surgical intervention due to severe postoperative complications, and the log-rank test was used to assess differences. All tests were two-sided, with a P value <0.05 considered statistically significant.

Results

PATIENT BASELINE CHARACTERISTICS:

A total of 62 patients were enrolled in the study and divided into a small-graft group (n=28) or a large-graft group (n=34). The 2 groups were comparable in preoperative demographic and clinical characteristics, with no significant differences observed (Table 1). Specifically, the mean age was similar (small-graft group: 67.14±8.52 vs large-graft group: 65.59±11.73 years, P=0.560), and the groups were balanced in terms of sex (14 [50.00%] vs 18 [52.94%] female, P=0.818). Preoperative BCVA did not differ significantly between groups (1.93±0.88 vs 1.75±0.87 logMAR, P=0.414). Furthermore, the prevalence of a history of glaucoma (6 [21.43%] vs 10 [29.41%], P=0.475) and the mean graft thickness (101.70±3.44 vs 101.06±4.84 μm, P=0.507) were also comparable. No significant differences were found in the distribution of crystalline lens status (P=0.276) or the types of surgical procedures performed (P=0.923).

POSTOPERATIVE OUTCOMES:

The postoperative outcomes of the 2 groups are summarized in Table 2. ECD decreased over time in both groups. At the 12-month follow-up, the ECL rate was significantly lower in the large-graft group (20.07±8.17%) compared to the small-graft group (24.34±7.20%, mean difference: −4.27%, 95% CI: −8.23% to −0.07%, P=0.035), despite comparable outcomes at the 1-, 3-, and 6-month follow-ups. BCVA improved progressively in both groups throughout the follow-up period, with no significant differences between the small-graft and large-graft groups at any time point (at 12 months: 0.540±0.260 vs 0.489±0.268 logMAR, P=0.452). Regarding clinical outcomes, graft transparency at 12 months was favorable and comparable between the groups (P=0.595), with most patients (96.43% in the small-graft group and 91.18% in the large-graft group) achieving a grade 0 transparency. Representative cases from each group are presented in Figure 1 (small-graft DSEK) and Figure 2 (large-graft DSEK), each illustrating the preoperative condition and postoperative outcomes at 1 day, and 1, 3, 6, and 12 months.

COMPLICATIONS:

A significant difference was observed in the incidence of graft dislocation, which occurred predominantly within the first 24 h and was significantly higher in the large-graft group (11 eyes [32.35%]) than in the small-graft group (3 eyes [10.71%], RR: 3.02, 95% CI: 0.93 to 9.77, P=0.043, Table 2). All dislocated grafts were reattached following reinjection of air or gas.

The overall incidence of secondary glaucoma was comparable between the groups (small-graft group: 21.43% vs large-graft group: 29.41%, P=0.475). Based on the time of onset, early-onset glaucoma (within 2 weeks postoperatively) occurred in 2 (7.14%) and 4 (11.76%) patients in the small-graft and large-graft groups, respectively, while late-onset glaucoma (beyond 2 weeks) was observed in 4 (14.29%) and 6 (17.65%). Elevated IOP was controlled by tapering topical corticosteroids in 3 patients in the small-graft group and 4 in the large-graft group. An additional 2 and 3 patients, respectively, required supplementary IOP-lowering medication. Ultimately, surgical intervention was necessary for 1 patient (glaucoma drainage device) in the small-graft group and 3 patients (1 glaucoma drainage device and 2 trabeculectomies) in the large-graft group.

Graft rejection occurred in 2 eyes (7.1%) in the small-graft group and 4 eyes (11.8%) in the large-graft group (P = 0.540). All patients received hourly topical corticosteroids, followed by a once-daily maintenance dose continued for 6 months once the condition was controlled. The graft rejection was controlled with this treatment in 1 eye from the small-graft group and 2 eyes from the large-graft group. The remaining patients (1 eye in the small-graft group and 2 eyes in the large-graft group) experienced graft failure and subsequently underwent PKP.

SURVIVAL ANALYSIS:

Graft survival rates were 96.43% (27/28) in the small-graft group and 94.12% (32/34) in the large-graft group, corresponding to an overall rate of 95.16% among all patients (Figure 3). The mean survival time was 11.71±0.281 months for the small-graft group and 11.82±0.246 months for the large-graft group. Comparison of Kaplan-Meier survival curves using the log-rank test indicated no statistically significant difference between the groups (P=0.788).

Discussion

DSEK has become one of the preferred surgical interventions for BK, effectively restoring corneal transparency and visual function by replacing the dysfunctional endothelial layer [10]. However, the influence of graft diameter on postoperative outcomes in BK patients has not been fully elucidated. This study was conducted to compare the clinical outcomes between small- and large-diameter grafts in patients undergoing DSEK for BK, with the aim of providing evidence-based guidance for optimal graft selection.

In this study, due to severe preoperative corneal opacity, ECD could not be assessed by specular microscopy. Therefore, the 1-month postoperative ECD was used as the baseline for calculating subsequent ECL rates. Although this method may not fully capture the impact of surgical manipulation on early endothelial injury, the lack of a significant intergroup difference in ECD at 1 month (P=0.129) may indicate that the early postoperative endothelial cell status was comparable. At the 12-month postoperative follow-up, the large-graft group maintained a lower ECL rate, suggesting a potential advantage for long-term ECD preservation. Long-term graft survival is critically dependent on the preservation of ECD [11]. Throughout the postoperative period, a dynamic pattern of ECL was observed: initially, the small-graft group had a lower rate of ECL, possibly attributable to reduced mechanical compression during insertion through a smaller corneal incision [12]. However, the large-graft group subsequently had a more gradual rate of ECL, ultimately resulting in superior ECD preservation at the later stage. One possible explanation for this finding is that the larger graft, by virtue of its greater initial endothelial cell count, provides a richer pool of functional cells [13,14]. This larger cell reserve may confer greater resilience against postoperative stressors such as inflammation and IOP fluctuations [14,15]. The mean ECD at 12 months postoperatively in this study was consistent with published data for DSEK at a comparable follow-up duration [16], and the overall graft survival rate of 95.16% closely aligns with rates reported in a previous study [17]. These previous studies collectively affirm the reliability of the DSEK technique and the validity of our results. Regarding visual rehabilitation, BCVA improved markedly and comparably in both groups throughout the follow-up period. This indicates that the restoration of endothelial function and corneal transparency by DSEK is the primary driver of visual improvement in BK [18], with graft diameter playing a comparatively minor role. Fluctuations in vision observed during the follow-up period were likely influenced by factors such as changes in corneal curvature following suture removal and the gradual resolution of graft edema [19].

The incidence of graft dislocation was significantly higher in the large-graft group (P=0.043) in this study, with RR of 3.02 suggesting that larger grafts may be associated with an elevated dislocation risk. Graft dislocation is a common complication of DSEK, primarily resulting from its reliance on air tamponade in the anterior chamber for initial attachment [20,21]. A larger graft diameter may increase the challenge of achieving perfect apposition with the host bed and reduce interface friction, thereby elevating the risk of dislocation. Although the dislocation rates observed in our study fall within the range reported for DSEK [22,23], our findings suggest that graft diameter may be a specific risk factor. Consequently, when planning for a larger graft diameter in resource-limited settings where manual DSEK is commonly performed, a choice supported by the potential for enhanced long-term ECD survival in our study and previous research [24], surgeons should pay particular attention to optimizing surgical steps such as incision construction, viscoelastic use, and air injection techniques to mitigate dislocation risk [25]. However, given the marginal statistical significance, limited sample size, and wide CI, this association should be interpreted with caution and warrants validation in larger prospective studies. Furthermore, no statistically significant differences were observed between the 2 groups in the incidence of secondary glaucoma or graft rejection, with the rates consistent with the ranges reported in previous studies [26,27]. This suggests that the occurrence of these complications is more closely related to patient-specific factors such as pre-existing glaucoma and postoperative medication regimens, rather than being primarily determined by graft diameter.

This study has several limitations. First, as a single-center retrospective study, we could not provide a patient enrollment flow diagram or detailed screening timelines. Additionally, graft diameter was not randomly assigned but was determined by surgical judgment, and some potentially important covariates were not systematically recorded. However, all inclusion and exclusion criteria as well as available baseline data were clearly reported. The balanced baseline characteristics between groups were observed, which may help reduce potential selection bias. Second, the statistical power was limited by the relatively small sample size, reducing the ability to detect subtle differences or rare events. Given the available data, conducting either a multivariable or a repeated-measures analysis would carry a high risk of overfitting and would be underpowered for detecting meaningful effects. Furthermore, as an exploratory study, to preserve the ability to detect meaningful signals, adjustment for multiple testing was not performed, although this may have increased the Type I error rate. Moreover, the single-center, single-surgeon design may affect generalizability. However, it could maximize technical consistency, thereby strengthening the internal validity of the graft diameter comparison. Finally, certain unmeasured or unadjusted variables, such as anterior chamber depth and intraoperative complications, that could influence outcomes were not controlled for in this study. Despite these limitations, this study provides clinically relevant preliminary evidence on graft diameter selection in DSEK. Future large-scale prospective studies should standardize the documentation of all study processes and relevant covariates, employing stricter statistical controls to robustly verify these preliminary findings.

Conclusions

This study provided clinically relevant comparative evidence that graft diameter may be an important determinant of postoperative endothelial dynamics and complication profiles in BK patients undergoing DSEK. While larger grafts were associated with superior long-term endothelial cell preservation, this benefit may be offset by a higher observed incidence of early graft dislocation, without corresponding differences in visual recovery, graft transparency, or overall graft survival at 12 months. These findings highlight a clinically meaningful trade-off between endothelial longevity and early postoperative stability, underscoring the need for individualized graft diameter selection based on patient anatomy, surgical expertise, and risk tolerance. Collectively, these results refine current decision-making in DSEK and provide valuable guidance to optimize surgical outcomes in clinical practice.

Data Availability Statement

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.