Purpose
Longstanding corneal edema can lead to anterior stromal scarring that may limit visual acuity following Descemet stripping endothelial keratoplasty (DSEK). The ideal time to perform DSEK to prevent permanent changes is unclear. Our aim is to determine whether earlier DSEK is associated with improved visual outcomes.
Design
Cohort study.
Methods
Setting : Institutional. Study Population : Total of 120 eyes of patients who underwent DSEK for corneal edema following cataract surgery (CE); 87% of eyes had a diagnosis of Fuchs dystrophy. Intervention : Post-DSEK visual acuity was compared in patients who underwent DSEK ≤6 months vs >6 months after CE. Linear and logistic regression were performed to assess the relationships between DSEK timing and best spectacle-corrected visual acuity (BSCVA) while accounting for baseline preoperative patient characteristics. Main Outcome Measures : Postoperative best-corrected visual acuity 6 months after DSEK (POM6 BSCVA).
Results
Median CE-to-DSEK time was 8.62 (interquartile range [IQR] 12.28) months (n = 120). Overall median preoperative VA and POM6 BSCVA were 0.54 (IQR 0.68) and 0.24 (IQR 0.16), respectively. Median POM6 BSCVA was better in patients with CE-to-DSEK time ≤6 months (median 0.18, IQR 0.19) vs >6 months (median 0.30, IQR 0.21) ( P = .014). A significant relationship was found between CE-to-DSEK time and POM6 BSCVA (coefficient = 0.002, P = .033), accounting for preoperative vision and pachymetry. Patients who underwent DSEK ≤6 months after CE were more likely to achieve POM6 BSCVA better than 20/40 (odds ratio = 3.73 P = .035).
Conclusion
Performing earlier DSEK for pseudophakic corneal edema appears to be associated with improved vision. Further prospective study is warranted to determine the optimal time to perform DSEK in patients with pseudophakic corneal edema.
Pseudophakic bullous keratopathy (PBK) is a complication of intraocular surgery in which injury to corneal endothelial cells results in failure of the pumping mechanism and subsequent corneal edema. Although PBK may only occur in 0.16%–1% of cataract surgery cases, patients with preexisting endothelial disease, such as Fuchs endothelial dystrophy, are at increased risk of corneal decompensation, with an incidence up to 22%. Treatment of corneal edema after cataract surgery ranges from conservative measures such as topical hypertonic saline and bandage soft contact lenses to surgical therapy with corneal transplantation. Descemet stripping endothelial keratoplasty (DSEK) is currently the most commonly performed surgical therapy for patients with pseudophakic corneal edema that persists despite a trial of nonoperative management.
Although DSEK has repeatedly been shown to improve visual acuity (VA) and vision-related quality of life, many eyes do not achieve best spectacle-corrected visual acuity (BSCVA) of 20/20 postoperatively. This may be partially attributed to structural alterations in the anterior cornea that can develop over time secondary to chronic stromal edema and that may be irreversible despite endothelial replacement. Although the specific amount and duration of corneal edema that results in such changes is unknown, patients who undergo earlier DSEK may potentially have improved visual outcomes. However, the relationship between the timing of DSEK in patients with pseudophakic corneal edema and visual outcomes is unclear.
In this study, we aimed to investigate the relationship between the timing of DSEK and post-DSEK VA in patients with pseudophakic corneal edema. We hypothesized that earlier DSEK would be associated with better VA.
Methods
We conducted a retrospective cohort study of patients undergoing DSEK for pseudophakic corneal edema by 6 surgeons at Wills Eye Hospital between January 1, 2009 and January 8, 2015. Eligible patients were identified by CPT code (65757) and chart review. Inclusion criteria were as follows: (1) age >18 years, (2) undergoing DSEK for persistent corneal edema following phacoemulsification, and (3) at least 6 months of follow-up time after DSEK. Exclusion criteria were as follows: (1) history of advanced glaucoma or filtering or tube shunt surgery; (2) prior retinal and corneal (including corneal scarring) pathology, presence of an anterior chamber intraocular lens, or other pathology that might limit visual acuity; (3) DSEK performed for failed prior penetrating or endothelial keratoplasty; (4) DSEK performed concurrently with another intraocular procedure (eg, exchange of intraocular lens); and (5) cases in which graft failure or rejection occurred within 6 months of DSEK. Approval from the Wills Eye Hospital Institutional Review Board was obtained prior to commencement of the study. The study was Health Insurance Portability and Accountability Act compliant and adhered to the tenets of the Declaration of Helsinki.
Preoperative data collected included patient age and sex, diagnosis of Fuchs dystrophy, cataract surgery date, preoperative BSCVA, preoperative ultrasound-based central corneal thickness (CCT), and complications. The “CE-to-DSEK time” was calculated by determining the duration (in months) between phacoemulsification and DSEK. Postoperative BSCVA at 6 (POM6 BSCVA) and 12 months (POM12 BSCVA), as well as “final postoperative BSCVA,” defined as BSCVA at the last follow-up visit, were recorded.
Patients were divided into 2 groups: (1) patients who underwent DSEK surgery that was performed within 6 months of cataract surgery (CE-to-DSEK time ≤6 months) and (2) patients who underwent DSEK surgery that was performed >6 months after cataract surgery (CE-to-DSEK time >6 months). Medians and interquartile ranges (IQR) were reported for all continuous variables. Mann-Whitney U tests (for continuous variables) and χ 2 tests (for categorical variables) were used to compare baseline characteristics and post-DSEK outcomes according to DSEK group. Univariate and multiple linear regression were used to assess the relationships between preoperative factors and POM6 BSCVA. Simple and multiple logistic regression with backward and forward selection of covariates were performed to determine if the timing of DSEK and/or other preoperative factors was related to achieving a POM6 BSCVA better than 20/40. All visual acuity data were converted to logarithm of the minimum angle of resolution (logMAR) scale for the analysis. P values less than .05 were considered statistically significant. STATA 12.1 (StataCorp LP, College Station, Texas, USA) was used for the statistical analysis.
Results
Charts of 1445 DSEK surgeries were reviewed. Of these, 120 eyes of 114 patients met the inclusion and exclusion criteria and were included in the study. Fifty-four (45%) of the patients were male and median age was 71.4 (IQR 10.9) years. A total of 87% patients had a diagnosis of Fuchs endothelial dystrophy. All patients had visually significant central corneal edema prior to DSEK. Overall median preoperative logMAR BSCVA and corneal thickness were 0.54 (IQR 0.68) and 672.5 μm (IQR 80.0 μm), respectively. In 63% of DSEK cases, the surgeon prepared the donor tissue at the time of surgery whereas 37% of cases used precut donor tissue from an eye bank. No patients were excluded because of graft failure or rejection within the first 6 postoperative months.
Median CE-to-DSEK time was 8.62 (IQR 12.28) months (range, 1.1–63.7 months). Forty-four (37%) and 76 (63%) eyes underwent DSEK within 6 months of and >6 months after cataract surgery, respectively ( Table 1 ). In the earlier DSEK group, 25% of eyes underwent DSEK 3 months or less after CE. In the later DSEK group, 45%, 43%, and 12% of eyes had a CE-to-DSEK time between 6 and 12 months, between 12 and 36 months, and greater than 36 months, respectively. The proportions of eyes with Fuchs dystrophy in the earlier and later DSEK groups were 88% and 84%, respectively ( P = .528).
| Overall (N = 120) | CE-to-DSEK Time ≤6 mo (N = 44) | CE-to-DSEK time >6 mo (N = 76) | P b | POM6 BSCVA Better Than 20/40 (N = 63) | POM6 BSCVA 20/40 or Worse (N = 57) | P b | |
|---|---|---|---|---|---|---|---|
| Age (y) | 71.4 [10.9] | 70.9 [10.3] | 73.5 [10.4] | .2516 | 68.2 [11.1] | 73.7 [10.3] | .006 |
| Sex (n, %) | |||||||
| Male | 54 (45%) | 19 (43%) | 35 (46%) | .761 | 30 (48%) | 24 (42%) | .544 |
| Female | 66 (55%) | 25 (57%) | 41 (54%) | 33 (52%) | 33 (58%) | ||
| Surgeon (n, %) c | 1/18/54/25/17/5 (0.8/15/45/20.8/14.2/4.2) | 0/4/24/11/3/2 (0/9.1/54.6/25/6.8/4.6) | 1/14/30/14/14/3 (1.3/18.4/39.5/18.4/18.4/4) | .232 | 1/8/33/11/7/3 (1.6/12.7/52.4/17.5/11.1/4.8) | 0/10/21/14/10/2 (0/17.5/36.8/24.6/17.5/3.5) | .442 |
| Pre-DSEK BSCVA, logMAR (Snellen) | 0.54 [0.68] (20/115) | 0.70 [0.80] (20/100) | 0.47 (0.40) (20/59) | .000 | 0.50 [0.70] (20/63) | 0.54 [0.60] (20/70) | .411 |
| Pre-DSEK CCT (μm) | 672.5 [80.0] | 700.5 [99.0] | 664 [72.5] | .203 | 670 [95.0] | 677.5 [91.0] | .292 |
| CE-to-DSEK time (mo) | 8.62 [12.28] | 3.78 [1.92] | 14.3 [17.03] | .000 | 6.1 [12.03] | 10.27 [12.97] | .010 |
| POM6 BSCVA, logMAR (Snellen) | 0.24 [0.16] (20/35) | 0.18 [0.19] (20/30) | 0.30 [0.21] (20/40) | .014 | 0.18 [0.08] (20/30) | 0.36 [0.14] (20/46) | .002 |
| POM12 BSCVA, logMAR (Snellen) (n = 82) | 0.20 [0.16] (20/32) | 0.18 [0.20] (20/30) | 0.26 [0.12] (20/36) | .065 | 0.18 [0.18] (20/30) | 0.3 [0.22] (20/40) | .000 |
| Last follow-up (mo) | 27.77 [28.07] | 35.00 [33.13] | 24.95 [23.93] | .087 | 22.95 [30.8] | 28.25 [28.47] | .229 |
| Final VA (logMAR, Snellen) | 0.22 [0.22] (20/33) | 0.18 [0.19] (20/30) | 0.30 [0.23] (20/40) | .002 | 0.18 [0.08] (20/30) | 0.38 [0.28] (20/48) | .000 |
a Continuous variables reported as median [interquartile range].
b Mann-Whitney U test for continuous variables, χ 2 test for categorical variables.
c Distribution of numbers and percentages of patients operated by each of 6 surgeons.
At postoperative months 6 (n = 120) and 12 (n = 82), median logMAR BSCVA was 0.24 (IQR 0.16) and 0.20 (IQR 0.16), respectively. Overall median follow-up time at last visit was 27.8 (IQR 28.07) months. Median logMAR BSCVA at 6 months was 0.18 (IQR 0.19) in patients who underwent DSEK within 6 months of cataract surgery (n = 44). Patients who underwent DSEK >6 months after cataract surgery (n = 76) had a median logMAR BSCVA at 6 months of 0.30 (IQR 0.21) ( P = .014). On univariate regression analysis of preoperative factors, older age, lower CE-to-DSEK time, and CE-to-DSEK time >6 months were associated with worse POM6 BSCVA ( P < .05) ( Table 2 ).
| Univariate Models | Multivariate Models | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Model 1 a | Model 2 b | |||||||||||
| Coefficient (m) | P | 95% Confidence Interval | Coefficient (m) | P | 95% Confidence Interval | Coefficient (m) | P | 95% Confidence Interval | ||||
| Age | 0.004 | .006 | 0.001 | 0.006 | 0.005 | .003 | 0.002 | 0.009 | 0.005 | .001 | 0.002 | 0.009 |
| Male sex | 0.048 | .062 | −0.002 | 0.098 | 0.019 | .473 | −0.034 | 0.073 | ||||
| Surgeon | 0.067 | .643 | −0.218 | 0.351 | 0.037 | .774 | −0.216 | 0.287 | ||||
| Pre-DSEK BSCVA (logMAR) | 0.021 | .330 | −0.021 | 0.061 | −0.004 | .30 | −0.064 | 0.056 | ||||
| Pre-DSEK CCT (μm) | −0.000 | .921 | −0.000 | 0.000 | 0.000 | .800 | −0.000 | 0.000 | ||||
| Increased CE-to-DSEK time (mo) | 0.002 | .045 | 0.000 | 0.004 | 0.002 | .033 | 0.000 | 0.004 | 0.002 | .023 | 0.000 | 0.004 |
| CE-to-DSEK time ≤6 mo | −0.064 | .013 | −0.116 | −0.014 | 0.055 | .074 | −0.0114 | 0.005 | −0.062 | .024 | −0.116 | −0.008 |
a Initial multivariate model including all variables regardless of level of significance.
b Final multivariate model following backward elimination of variables if level of significance was >.05. Forward elimination resulted in an identical model.
Patients who underwent earlier DSEK had a median logMAR BSCVA at 12 months (n = 82) and at last follow-up of 0.18 (IQR 0.20) and 0.18 (IQR 0.19), respectively, compared to 0.26 (IQR 0.12) ( P = .065) and 0.30 (IQR 0.23) ( P = .002) in patients in the later DSEK group. Last follow-up time was similar between the earlier vs later DSEK groups. The Figure displays the change in median VA over time between patients undergoing DSEK 6 months or less vs greater than 6 months after cataract surgery.
When patients with POM6 BSCVA better than 20/40 (n = 63, 53%) were compared with patients with POM6 BSCVA of 20/40 or worse (n = 57, 48%), there was no statistically significant difference between preoperative VA or corneal thickness ( P > .05). Patients with POM6 BSCVA better than 20/40 had a significantly shorter median CE-to-DSEK time compared with patients with POM6 BSCVA of 20/40 or worse (6.1 months vs 10.27 months, P = .010). CE-to-DSEK time less than 6 months and younger age were found to be a significant predictor of POM6 BSCVA better than 20/40 ( P < .05) ( Table 3 ).
