A Comparison of the Corneal Thickness Following Descemet’s Stripping Automated Endothelial Keratoplasty and Descemet’s Membrane Endothelial Keratoplasty

Abstract Purpose To compare the central corneal thickness before and after Descemet’s stripping automated endothelial keratoplasty (DSAEK) and Descemet’s membrane endothelial keratoplasty (DMEK), and to evaluate the recipient corneal thickness following DSAEK. Methods The corneal thickness was compared between two groups of eyes following DMEK and DSAEK, performed by a single surgeon between 2015 and 2017. We evaluated the recipient corneal thickness and central corneal thickness pre- and postoperatively at 1, 3, and 6 months using anterior segment optical coherence tomography. Recipient corneal thickness was defined as the corneal thickness without graft thickness. Results We included DMEK and DSAEK eyes (n = 26 each), which were similar in terms of their etiologies. Preoperatively, there was no significant difference in the central corneal thickness between the groups (DSAEK, median [interquartile range]: 721 [606.5 to 847.8] µm; and DMEK: 690 [618 to 722.3] µm; p = 0.30). Despite the tendency of the central corneal thickness to be significantly greater (p < .01) at 6 months following DSAEK (619.5 [607.8 to 661.3] µm) compared with that following DMEK (497.5 [475.8 to 525.3] µm), there was no significant difference at 6 months between the recipient corneal thickness following DSAEK (488.5 [443.8 to 515] µm) and central corneal thickness following DMEK (p = 0.54). Conclusions DSAEK eyes display a similar tendency of stromal thinning as DMEK eyes.


Introduction
Over the decades, corneal transplantation has become one of the most established forms of organ transplantation. Particularly, endothelial keratoplasty has been widely applied to treat patients with diseased corneal endothelial cells since the introduction of the first posterior lamellar keratoplasty by Dr. Gerrit Melles in 1990. Endothelial keratoplasty displays superiority in terms of lower graft rejection and better visual quality, compared with penetrating keratoplasty. 1 There are two types of endothelial keratoplasties: Descemet's stripping automated endothelial keratoplasty (DSAEK) and Descemet's membrane endothelial keratoplasty (DMEK). Despite equivalent visual outcomes, several studies have indicated the substantial superiority of DMEK in terms of lower refractive change (hyperopic shift) or lower high-order aberrations. This difference can be attributed to the natural posterior curvature without the protrusion of the crescent-shaped DSAEK graft, residual host-Descemet membrane at the donor-host interface, the reflectivity of the interface, uneven graft thickness, donor stromal shrinkage, and irregularities and scarring at the donor-host interface. [2][3][4][5][6][7][8][9][10] However, our previous study demonstrated that DMEK eyes were not completely similar to healthy eyes (eyes without corneal disease) in terms of corneal thickness and highorder aberrations. DMEK eyes display more corneal thinning and irregular curvature with higher high-order aberrations than healthy eyes do. 11 We previously published a paper indicating that HOAs increased more after endothelial keratoplasty (EK) compared with controls, and more so after DSAEK than after DMEK. The results also showed that HOAs was inversely correlated with BSCVA after EK. 9 We reported that after DMEK improved, BSCVA improved and HOAs decreased; however, none of the eyes reached levels comparable with that of healthy controls. 11 We have previously published a study evaluating HOAs and visual function, and in this study, we would like to focus on the factors associated with thickness. Furthermore, we have reported on a significant decrease in the corneal nerve in DMEK eyes. 12 These reports have facilitated consideration of the stromal change in DSAEK eyes in terms of the stromal thinning without DSAEK grafts. We recently noticed the corneal thinning in eyes after DMEK. Clinically, the corneal thickness of DSAEK eyes is known to be thicker than that of DMEK. We questioned how the recipient corneal thickness of DSAEK eyes may change postoperatively.
Herein, we aimed to evaluate the recipient corneal thickness (RCT) following DSAEK as well as DMEK and compare it with central corneal thickness (CCT) in DMEK eyes. This novel study elucidates the incidence of host stromal thinning following DSAEK.

Study design
This retrospective study was approved by the Yokohama Minami Kyosai Review Board (approval number 1-19-11-11) and the Nihon University of Medicine Review Board (approval number P21-13-0). It was conducted in accordance with the ethical standards set forth in the 1964 Declaration of Helsinki and its subsequent amendments. The research methods followed all institutional guidelines, and all patients provided informed consent via an opt-out choice on the Yokohama Minami Kyosai Hospital website. We enrolled consecutive cases of DMEK and DSAEK from January 2015 to December 2017. The target population included patients with pseudophakic eyes without stromal scarring but with stromal edema. We excluded cases of triple DMEK and DSAEK because of differences in the procedures. Furthermore, we excluded patients with visual limitations, such as amblyopia or macular disease.

Surgical technique
All patients underwent DSAEK and DMEK by a single surgeon (T.H.), as previously described. [13][14][15][16] Briefly, for the DMEK procedure, after creating a descemetorhexis of approximately 9.0 mm in diameter under air using a reverse Sinsky hook (ASICO, Westmont IL), we inserted an appropriately undersized graft (7.75 mm, 8.0 mm, 8.25 mm, or 8.5 mm) using an intraocular lens inserter (WJ-60M1; Santen, Osaka, Japan). It was unfolded and fixated with 20% sulfur hexafluoride until 80% of the anterior chamber volume was filled. Similarly, for the DSAEK procedure, after creating a descemetorhexis of the diameter corresponding to the graft using a reverse Sinsky hook (ASICO), we inserted an appropriately undersized graft (7.75 mm or 8.00 mm) using a Busin glide (ASICO) into the anterior chamber. The graft was fixated with air injected into the anterior chamber.

Patient examinations
In addition to the standard examinations performed using slit-lamp microscopy, we evaluated the following factors preoperatively and at 1, 3, and 6 months postoperatively in all eyes: best spectacle-corrected visual acuity (BSCVA; a logarithm of the minimal angle of resolution [logMAR]) and topographic factors determined by anterior segment optical coherence tomography (CASIA, Tomey, Nagoya, Japan), namely CCT, and graft thickness. For the DSAEK eyes, aberration factors (HOAs at 6.0-mm diameter in the total cornea) were evaluated to detect the correlation between thickness factors. The RCT was defined by subtracting the graft thickness from the CCT (Figure 1(A,B)). RCT is defined as stromal thickness at the central area.

Statistical analyses
Statistical analyses were performed using IBM SPSS Advanced Statistics version 22 (IBM Japan, Tokyo, Japan). Quantitative variables were assessed for normality using the Kolmogorov-Smirnov test. The standard deviation is presented for the mean, and the interquartile range or minimum-maximum range is presented for the median, depending on the normality of distribution. P values <.05 were considered statistically significant. We used the Mann-Whitney, Fisher's exact, Friedman tests, Spearman's correlation, and a mixed-effects model. The required sample size was estimated based on an expected RR of 40%, 90% power, and an alpha value of 0.05 using G Ã Power (https://www.psychologie.hhu.de/arbeitsgruppen/allgemeine-psychologie-undarbeitspsychologie/gpower.html). The target sample size was determined to be at least 40 patients. Therefore, we assigned 20 patients to each of the DMEK and DSAEK groups.

Results
Twenty-six eyes of 26 patients (10 men and 16 women; mean age 74 [69 to 79] years) were treated by DSAEK; 26 eyes of 26 patients (6 men and 20 women; mean age 75 years [69 to 81]) were treated by DMEK. Demographic data are shown in Table 1. All eyes displayed corneal endothelial dysfunction, which was derived from Fuchs corneal endothelial dystrophy, laser iridotomy-induced bullous keratopathy, trabeculectomy, primary angle closure, pseudophakic bullous keratopathy, aphakic bullous keratopathy, corneal endotheliitis, and failed penetrating keratoplasty. There were no significant differences in demographical data between the two groups, with the exception of preoperative BSCVA.
As shown in Figure 2(A), DSAEK significantly improved the BSCVA (shown as a logMAR) at 3, and 6 months in comparison with that at the preoperative period (p < .01, Friedman test with Bonferroni correction). Furthermore, we observed no significant improvement in the BSCVA in the DSAEK group at 3 months compared to at 1 month (p ¼ .11, Friedman test with Bonferroni correction), at 1 month compared with the preoperative period (p ¼ 1.00, Friedman test with Bonferroni correction), and at 6 months compared to 3 months (p ¼ .081, Friedman test with Bonferroni correction). As shown in Figure 2(B), DMEK improved the BSCVA (logMAR) at 3, and 6 months in comparison with that at the preoperative period (p< .01, Friedman test with Bonferroni correction). We observed a statistically significant improvement in the BSCVA in the DMEK group at 3 months compared to 1 month (p < .05, Friedman test with Bonferroni correction), but no significant improvement at 6 months compared with 3 months (p ¼ .64, Friedman test with Bonferroni correction) or at 1 month compared with the preoperative period (p ¼ .06, Friedman test with Bonferroni correction). The preoperative logMAR was matched to 0, and the postoperative logMAR was evaluated by the amount of change. A mixed effects model was used for statistical analysis of the logMAR visual acuity improvement over time ( Figure 3). There was no significant difference in the variability between the two groups (p ¼ .28); however, the point-in-time variation displayed a significant difference (p < .01). The BSCVA improved gradually, but there was no difference in BSCVA between the surgical techniques ( Figure 3(A)). No significant correlation between BSCVA and corneal thickness (CCT, RCT, graft thickness) was detected postoperatively. However, when examining the amount of visual acuity change per 1 month BSCVA (log MAR), a significant difference in visual acuity recovery by surgical technique was detected at 3 to 6 months postoperatively (p < .005, Mann-Whitney U-test; Figure 3(B)).
We compared the DSAEK CCT and RCT, and DMEK CCT pre-and postoperatively; all were significantly thinner postoperatively compared to preoperatively ( Table 2). We compared the rates of change upon setting the preoperative corneal thickness at 1.0. The rate of decrease in DSAEK CCT was significantly higher than that in the DMEK CCT  Table 1.

Discussion
In this retrospective clinical study, the preoperative CCT was compared with the CCT/RCT following DSAEK and  CCT following DMEK using anterior segment optical coherence tomography. Both CCT and RCT following DSAEK, and CCT following DMEK displayed statistically significant progressive thinning postoperatively. Furthermore, the level of RCT thinning following DSAEK was similar to that following DMEK. Previously, we reported that DMEK eyes displayed temporal aberrant corneal thinning postoperatively from 3 to 6 months with an irregular curvature and highorder aberrations, compared with healthy control eyes. 11 At 1 year postoperatively, the CCT displayed a thickness in DMEK similar to that of the control eyes. In the current study, the evaluation of postoperative corneas using anterior segment optical coherence tomography enabled the precise measurement of RCT following DSAEK, which displayed a similar change in thickness to the CCT following DMEK. Although some papers have published the change in thickness after DSEK, our study reported few differences in terms of the type of surgery, observational period, or etiology. Yien et al. 17 reported a change in the thickness of donor, host, and total cornea up to three months after DSEK. However, their study lacked data between 3 and 6 months after surgery, whereas our study has indicated the    Comparisons of the rate of corneal thickness change following Descemet's stripping automated endothelial keratoplasty and Descemet's membrane endothelial keratoplasty. Changes in the rate of corneal thickness reduction over time were examined using a mixed-effects model. In between-group variation, there was no statistically significant difference between the Descemet's stripping automated endothelial keratoplasty (DSAEK) recipient corneal thickness (RCT) and Descemet's membrane endothelial keratoplasty (DMEK) (P ¼ .23; Steel-Dwass test). There were statistically significant differences between DSAEK central corneal thickness (CCT) and DMEK or DSAEK RCT (P ¼ .008 and .001, respectively; Steel-Dwass test).
importance of changes beyond 3 months after surgery. The type of surgery used also differs between the two studies in a strict definition. A further study by Ahmed et al .18 reported an improvement in the stromal thickness after DSEK, whereby the study included the restricted data of Fuchs patients, resulting in different findings from our study. In terms of the etiology, our study also differs from their study. Our biggest strength is the novel finding of stromal thinning after DSAEK, which was equivalent to the same phenomenon in eyes after DMEK. We conducted a literature search of the PubMed database on November 22, 2022, and the search yielded no similar findings.
In general, it has been thought that a thinner cornea tends to show better visual acuity.
In fact, some studies, including the DETECT study, 6 have proved the superiority of performing DMEK compared to DSAEK in terms of visual function such as HOAs. However, there has been some reports of corneal ectasia after DMEK, which means the existence of endothelial dysfunction and keratoconus at the same time. Although our results indicated the superiority of performing DMEK in terms of visual improvement, one of the advantages of selecting DSAEK is to help to restore the fragile structure of the patients' cornea after significant corneal thinning over the time after surgery, which can result in keratoectasia. 19 The strength of our study is the first evaluation of the real stromal thickness after DSAEK (Figure 1(C)). Without the precise evaluation of RCT and graft thickness at the same time, no one would discover the stromal thinning after DSAEK. In the past, Ivarsen et al. 20 suggested the negative correlation between reduced RCT and visual improvement due to the increased densitometry. Although our previous study indicated the importance of HOAs on visual outcomes, 9 this study was designed to discuss the change in visual recovery and corneal thickness without comparing HOAs or densitometry, which might be a limitation of this study.
The precise mechanisms underlying aberrant CCT thinning following DSAEK and DMEK are unknown. Recently, we evaluated the cornea before and after DMEK using in vivo confocal microscopy; the corneal nerves regenerated and the number of dendritic cells in the corneal basal epithelium decreased postoperatively, thereby indicating that DMEK might repair and normalize the corneal environment. 21 However, acute corneal thinning and dehydration following DMEK without healthy corneal nerves in bullous keratopathy may cause aberrant CCT thinning. Similar stromal changes may occur in the corneal stroma following DSAEK. Another hypothetical explanation attributes aberrant CCT following DSAEK and DMEK to the vulnerability of keratocytes, collagen fibers, and other skeletal tissues caused by bullous changes in the corneal stroma; elongated collagen fibers and partially apoptotic/necrotic keratocytes together with an altered extracellular matrix may cause aberrant CCT following immediate corneal dehydration to a normal level after DSAEK and DMEK. Apoptosis may play a role in the pathology of keratocyte death in bullous keratopathy. 22 In this study, the BSCVA improved postoperatively in both DSAEK and DMEK eyes; DMEK eyes displayed a significantly higher BSCVA than did DSAEK eyes. One possible explanation of this result is the difference in preoperative visual function between the two groups; DMEK may be selected as a surgical technique for patients with less corneal edema. Although our study did not show any statistically significant differences between these surgical techniques in terms of the amount of change in BSCVA, DSAEK eyes showed a persistent improvement after 3 to 6 months postoperatively, indicating that the visual recovery after DSAEK is generally slower than that after DMEK. This finding implies that surgeons should wait for recovery even after 3 or 6 months in DSAEK eyes.
Our study had some limitations, including, retrospective design, short observation period, and heterogeneity of causative disease of bullous keratopathy. Corneal epithelial thickness is an important factor in RCTs for DSAEK 23,24 but it is difficult to measure in eyes with severe corneal endothelial damage. The inability to evaluate corneal epithelial thickness is another limitation of this study. As this study found, visual recovery of DSAEK is gradual. Therefore, in future research, factors related to visual acuity need to be re-examined in cases that have been followed for 1 or 2 years.
In conclusion, the RCT following DSAEK displayed aberrant thinning during the study; we observed a similar thickness course of the CCT following DMEK. Researchers should elucidate the clinical significance of these thickness changes following endothelial keratoplasty in the future.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data availability statement
The data that support the findings of this study are available on request from the corresponding author [TH]. The data are not publicly available due to privacy protection in our country.