Association between corneal edema and corneal astigmatism after Descemet membrane endothelial keratoplasty

Abstract

Purpose

To evaluate changes in corneal astigmatism after DMEK and determine the predictive value of preoperative central corneal thickness (CCT) for postoperative astigmatic change.

Methods

Fifty-five eyes with Fuchs’ endothelial dystrophy (FED) or pseudophakic bullous keratopathy (PBK) undergoing DMEK were analyzed. Auto kerato-refractometer (ARKRT)-derived anterior corneal astigmatic change was defined as the primary outcome in 55 eyes. Anterior-segment optical coherence tomography (AS–OCT)-derived anterior, posterior, and total corneal parameters were analyzed as secondary outcomes in 25 eyes. Vector analysis assessed magnitude and axis changes. Associations between preoperative CCT and postoperative astigmatic change were evaluated; an exploratory ROC analysis was performed for astigmatic change.

Results

Mean preoperative CCT was 738.7 ± 99.7 μm, decreasing to 517.5 ± 37.3 μm postoperatively (P < 0.001). Anterior astigmatism measured by ARKRT decreased significantly after DMEK, whereas anterior and total corneal astigmatism measured by AS–OCT showed a similar decreasing trend without statistical significance. Greater preoperative CCT showed a larger postoperative astigmatic change on ARKRT. In ROC analysis based on ARKRT measurements, a preoperative CCT of 790 μm showed modest discrimination for postoperative astigmatic change > 3.0 D (AUC = 0.749; sensitivity 67%, specificity 81%). PBK eyes showed greater anterior flattening and larger reductions in absolute astigmatism than FED eyes.

Conclusions

Preoperative corneal edema is associated with postoperative astigmatic change after DMEK. The 790 μm CCT cutoff showed modest discrimination. In eyes with marked edema, staged astigmatic correction may improve refractive outcomes.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12886-026-04920-8.

Introduction

Descemet membrane endothelial keratoplasty (DMEK) has become the preferred technique for selective replacement of diseased corneal endothelium in conditions such as Fuchs’ endothelial dystrophy (FED) and pseudophakic bullous keratopathy (PBK) [1, 2]. By transplanting only the Descemet membrane and endothelium, DMEK offers faster visual recovery, lower immunologic rejection risk, and superior optical quality compared with Descemet stripping automated endothelial keratoplasty (DSAEK) and penetrating keratoplasty [3–6].

Management of preexisting corneal astigmatism is critical for achieving optimal refractive outcomes in cataract surgery. Even with precise intraocular lens (IOL) centration, residual astigmatism can significantly reduce uncorrected visual acuity [7]. Toric IOLs are widely used to correct regular corneal astigmatism at the time of cataract extraction, with established predictability and stability [8, 9]. Accurate assessment of corneal astigmatism magnitude and axis is essential for appropriate toric IOL selection [10, 11]. However, in combined or sequential cataract surgery and endothelial keratoplasty, predicting the final corneal astigmatism remains challenging [12].

After DMEK, astigmatism may change unpredictably due to resolution of stromal and epithelial edema, remodeling of posterior curvature, and individual variations in wound healing [13–16]. In edematous corneas, preoperative keratometry often fails to reflect the true corneal shape, as swelling can induce non-uniform flattening or steepening. Consequently, toric IOL calculations based on pre-DMEK measurements risk over- or under-correction [12, 17]. Understanding the relationship between preoperative corneal edema and postoperative astigmatic change is therefore essential for surgical planning, particularly in triple DMEK procedures.

This study aimed to characterize changes in anterior, posterior, and total corneal astigmatism after DMEK and to investigate their association with corneal edema and surgical indication.

Methods

Study design and participants

This retrospective study was approved by the Institutional Review Board of Seoul National University Hospital (IRB No. H-2508-180-1672) and adhered to the tenets of the Declaration of Helsinki. The requirement for informed consent was waived due to the retrospective nature of the study.

This single-center interventional case series included 55 eyes from 51 patients who underwent Descemet membrane endothelial keratoplasty (DMEK) for either Fuchs’ endothelial dystrophy (FED) or pseudophakic bullous keratopathy (PBK). The mean follow-up duration after DMEK was 37.6 ± 24.3 months. Eyes were excluded if they had inadequate follow-up, graft failure, previous endothelial keratoplasty (DMEK or DSAEK), or a history of ocular trauma or other conditions that could affect corneal astigmatism. All patients were pseudophakic, having undergone either DMEK alone or combined with cataract surgery. The surgical indication for DMEK was persistent corneal stromal edema with associated visual impairment.

Surgical procedure

All DMEK surgeries were performed by a single surgeon (MKK) using a standardized technique described previously [18]. Donor grafts were obtained from the Eversight Eye Bank (Chicago, IL, USA) in accordance with standard protocols. Only donor tissue with a death-to-operation interval of fewer than seven days was used.

Peripheral laser iridotomy or surgical iridectomy was performed during cataract surgery prior to DMEK. The recipient Descemet membrane stripping diameter was typically 0.5–0.75 mm larger than the donor graft diameter (7.5–8.5 mm). Grafts were inserted using either the endothelium-out technique with a Jones tube (Gunther Weiss Scientific, Hillsboro, OR, USA) or, after August 2018, the endothelium-in technique with a Coronet EndoGlide insertion device (Network Medical Products, Ripon, UK) following its commercial availability in Korea. The graft was unfolded by gentle tapping and secured with an air tamponade (80–95%), with correct orientation confirmed by the S-mark.

Postoperatively, patients received systemic steroids and were prescribed topical 1% prednisolone acetate (Pred Forte; Allergan, Dublin, Ireland) and 0.5% moxifloxacin (Vigamox; Alcon, Geneva, Switzerland) four times daily. Follow-up visits were scheduled at postoperative day 1, day 2, week 1, month 1, month 3, month 6, and every 6 months thereafter.

Data collection

Baseline demographic and clinical data included age, sex, surgical indication, central corneal thickness (CCT), axial length (AXL), spherical equivalent, and follow-up duration. AXL was measured using the IOL Master 700 (Carl Zeiss, Jena, Germany), and CCT was measured with anterior-segment optical coherence tomography (AS–OCT; CASIA2, Tomey, Nagoya, Japan). Postoperative measurements were defined as those obtained at the visit when CCT was thinnest between 1 and 4 months after DMEK, as this was considered to best reflect the corneal state after maximal resolution of edema.

Anterior corneal keratometric values were obtained using both an auto kerato-refractometer (ARKRT; KR-8900, Topcon, Tokyo, Japan) and AS–OCT, while posterior and total corneal values were measured with AS–OCT. Steep, flat, and mean keratometry (Ks, Kf, Km) values and their respective axes were recorded. Astigmatism was calculated as the difference between steep and flat keratometry for anterior, posterior, and total cornea. During the study period, the AS-OCT device used in our institution was changed. To avoid potential measurement bias arising from inter-device differences, only eyes examined with the same AS-OCT platform (CASIA2) were included in the AS-OCT-based analyses. Because ARKRT data were available for all cases, ARKRT-derived anterior astigmatic change was defined as the primary outcome. AS–OCT-derived parameters were analyzed as secondary outcomes limited to the eyes examined with AS–OCT, and were used to complement the primary outcome. Vector analysis of astigmatism followed the methodology and graphical representation described by Abulafia et al. [19]

Statistical analysis

Data are expressed as mean ± SD and range. Normality was assessed using the Kolmogorov–Smirnov test. For normally distributed data, paired t-tests and Pearson correlation coefficients were used; otherwise, Wilcoxon signed-rank tests and Spearman correlation coefficients were applied. ROC curve analysis was performed, with an AUC > 0.7 considered acceptable discrimination. The optimal cut-off value was defined as the point with the maximum Youden index (J = sensitivity + specificity − 1) [20]. Multivariable linear regression was performed using postoperative absolute astigmatic change as the dependent variable and preoperative CCT, astigmatism, and surgical indication as covariates. A p-value < 0.05 was considered statistically significant. All analyses were conducted using SPSS Statistics software (version 27.0; IBM, Armonk, NY, USA).

Results

Demographic characteristics

The baseline characteristics are summarized in Table 1. A total of 55 eyes from 51 patients (43% male, 57% female) were included. The mean age was 66.8 ± 11.7 years. Surgical indications were FED in 20 eyes and PBK in 35 eyes. Mean preoperative CCT was 738.7 ± 99.7 μm. Mean AXL was 23.98 ± 2.65 mm, and mean spherical equivalent was 0.59 ± 3.06 D. The mean postoperative follow-up duration was 37.6 ± 24.3 months. The postoperative measurements included in the analysis corresponded to the time point when CCT was at its thinnest between 1 and 4 months postoperatively (mean ± SD: 1.9 ± 0.7 months).

Table 1.

Demographics of included eyes

	Mean ± SD (range)
Eyes, number	55
Age, years	66.8 ± 11.7 (33.0–86.6)
Sex, male/female	22/29
Surgical indication
FED	20
PBK	35
Central corneal thickness, µm	738.7 ± 99.7 (532–989)
Axial length, mm	23.98 ± 2.65 (20.88–32.24)
Spherical equivalent, D	0.59 ± 3.06 (-5.38–8.25)
Postoperative date*, month	1.9 ± 0.7 (1.0–3.9)

PBK = pseudophakic bullous keratopathy, FED = Fuchs’ endothelial dystrophy, N = number, SD = standard deviation

*Date when central corneal thickness was measured as thinnest between postoperative 1mo and 4mo

Keratometric and astigmatic change in anterior, posterior and total cornea

On ARKRT, the anterior cornea demonstrated a significant increase in Kf (42.21 D → 43.15 D, P = 0.003) and a significant reduction in astigmatism (2.94 D → 1.77 D, P < 0.001) after DMEK (Table 2). Vector analysis showed a decrease in centroid from 0.75 D at 171° to 0.28 D at 136° (Fig. 1). Spherical equivalent shifted toward myopia by − 1.06 D, with borderline significance (P = 0.072).

Table 2.

Preoperative and postoperative data measured by auto kerato-refractometer (n = 55)

	Preop	Postop	Δ(Post-pre)	P-value
Anterior cornea
Ks, D	45.15 ± 3.15	44.93 ± 2.28	0.02 ± 2.40	0.955
Kf, D	42.21 ± 2.65	43.15 ± 2.11	1.06 ± 2.49	0.003
Km, D	43.68 ± 2.73	44.04 ± 2.10	0.54 ± 2.26	0.085
Magnitude of astigmatism, D	2.94 ± 2.02	1.77 ± 1.28	-1.05 ± 1.85	< 0.001
Astigmatism in vector form, D	0.78 at 171°	0.28 at 136°
Central corneal thickness, µm	738.7 ± 99.7	517.5 ± 37.3	-221.1 ± 93.8	< 0.001
Spherical equivalent, D	0.59 ± 3.06	-1.22 ± 1.80	-1.06 ± 2.63	0.072

D = diopter, Ks = steep keratometry, Kf = flat keratometry, Km = mean keratometry

Values with statistical significance are shown in boldface

Fig. 1.

Double-angle plots for preoperative (A) and postoperative (B) anterior corneal astigmatism measured by auto kerato-refractometer

As a secondary investigation, AS-OCT-based analyses were limited to 25 eyes that had been evaluated using the same AS-OCT platform (CASIA2) during the study period. To assess potential bias in the AS–OCT subgroup, we compared eyes with and without eligible AS–OCT data. Surgical indication, ARKRT-derived preoperative astigmatism, and postoperative astigmatic change did not differ significantly between groups (Supplementary Table 1), suggesting no marked imbalance in variables directly related to the primary outcome. AS–OCT measurements (Table 3) revealed significant decreases in anterior corneal Ks (− 0.83 ± 1.29 D, P = 0.004) and Km (− 0.48 ± 0.99 D, P = 0.025). In the total cornea, Ks and Km decreased by − 1.14 ± 1.45 D (P < 0.001) and − 0.79 ± 1.24 D (P = 0.004), respectively. No significant changes were observed in posterior corneal curvature. Although anterior and total corneal astigmatism also decreased after DMEK, these changes did not reach statistical significance in the AS–OCT subgroup. Astigmatism magnitude decreased from 2.24 ± 2.57 D to 1.53 ± 1.07 D in the anterior cornea, from − 0.47 ± 0.26 D to − 0.41 ± 0.21 D in the posterior cornea, and from 2.42 ± 2.86 D to 1.72 ± 1.06 D in the total cornea. Vector analysis showed increases in centroid from 0.26 D at 71° to 0.64 D at 139° (anterior), 0.14 D at 8° to 0.35 D at 1° (posterior), and 0.21 D at 62° to 0.82 D at 151° (total) (Fig. 2).

Table 3.

Preoperative and postoperative data measured by anterior segment OCT (n = 25)

	Preop	Postop	Δ(Post-pre)	P-value
Anterior cornea
Ks, D	45.69 ± 2.38	44.86 ± 1.96	-0.83 ± 1.29	0.004
Kf, D	43.45 ± 2.03	43.33 ± 1.59	-0.12 ± 1.70	0.731
Km, D	44.57 ± 1.80	44.09 ± 1.70	-0.48 ± 0.99	0.025
Magnitude of astigmatism, D	2.24 ± 2.57	1.53 ± 1.07	-0.72 ± 2.28	0.130
Astigmatism in vector form, D	0.26 at 71°	0.64 at 139°
Posterior cornea
Ks, D	-6.65 ± 0.53	-6.74 ± 0.45	-0.09 ± 0.54	0.396
Kf, D	-6.18 ± 0.51	-6.33 ± 0.37	-0.15 ± 0.46	0.121
Km, D	-6.41 ± 0.51	-6.53 ± 0.39	-0.12 ± 0.47	0.211
Magnitude of astigmatism, D	0.47 ± 0.26	0.41 ± 0.21	-0.06 ± 0.35	0.443
Astigmatism in vector form, D	0.14 at 8°	0.35 at 1°
Total cornea
Ks, D	44.73 ± 2.28	43.59 ± 1.93	-1.14 ± 1.45	< 0.001
Kf, D	42.31 ± 2.19	41.87 ± 1.59	-0.44 ± 2.05	0.291
Km, D	43.52 ± 1.72	42.73 ± 1.68	-0.79 ± 1.24	0.004
Magnitude of astigmatism, D	2.42 ± 2.86	1.72 ± 1.06	-0.69 ± 2.56	0.185
Astigmatism in vector form, D	0.21 at 62°	0.82 at 151°
Central corneal thickness, µm	738.7 ± 99.7	517.5 ± 37.3	-221.1 ± 93.8	< 0.001

D = diopter, Ks = steep keratometry, Kf = flat keratometry, Km = mean keratometry

Values with statistical significance are shown in boldface

Fig. 2.

Double-angle plots for preoperative and postoperative anterior corneal astigmatism (A, B), posterior corneal astigmatism (C, D) and total corneal astigmatism (E, F) measured by anterior segment optical coherence tomography

Association between corneal edema and corneal astigmatism

Mean CCT decreased from 738.7 ± 99.7 μm to 517.5 ± 37.3 μm after DMEK (P < 0.001) (Tables 2 and 3). Greater preoperative edema (higher CCT) showed a trend toward larger changes in astigmatism (r = − 0.25, P = 0.064) and absolute astigmatism (r = 0.26, P = 0.052) on ARKRT, although these associations did not reach statistical significance (Figs. 3A, B). On AS–OCT, only posterior astigmatism was significantly correlated with preoperative CCT (r = − 0.44, P = 0.028), while anterior and total astigmatism showed no significant association (Figs. 3C, D). Similarly, greater reductions in CCT were associated with larger changes in astigmatism (r = 0.26, P = 0.053) and absolute astigmatism (r = − 0.28, P = 0.040) measured by auto kerato-refractometry (Figs. 4A, B), but not with changes in anterior, posterior, or total astigmatism on AS–OCT (Figs. 4C, D).

Fig. 3.

Correlation analysis between preoperative central corneal thickness and postoperative astigmatic change. Anterior astigmatism change measured by auto kerato-refractometer (A) and anterior, posterior, total astigmatism change measured by anterior segment OCT (C). Absolute anterior astigmatism change measured by auto kerato-refractometer (B) and absolute anterior, posterior, total astigmatism change measured by anterior segment OCT (D)

Fig. 4.

Correlation analysis between central corneal thickness change and postoperative astigmatic change. Anterior astigmatism change measured by auto kerato-refractometer (A) and anterior, posterior, total astigmatism change measured by anterior segment OCT (C). Absolute anterior astigmatism change measured by auto kerato-refractometer (B) and absolute anterior, posterior, total astigmatism change measured by anterior segment OCT (D)

The ROC analysis was based on ARKRT-measured anterior corneal astigmatism in the full cohort of 55 eyes. ROC curve analysis suggested that a preoperative CCT of 790 μm had modest discriminatory ability for identifying eyes with postoperative astigmatic change > 3.0 D (AUC = 0.749, P = 0.019; sensitivity = 67%, specificity = 81%) (Fig. 5). However, because the cutoff was derived within the present cohort, its clinical applicability remains uncertain and requires validation in an independent dataset. In multivariable linear regression analysis, only preoperative astigmatism remained significantly associated with postoperative absolute astigmatic change (β = 0.524, P < 0.001), whereas preoperative CCT (β = 0.0018, P = 0.205) and surgical indication (β = 0.152, P = 0.625) were not significant after adjustment.

Fig. 5.

Receiver operating characteristic (ROC) curve for logistic regression of preoperative central corneal thickness (CCT) and postoperative astigmatism change ≥ 3 diopters

Keratometric and astigmatic change by surgical indication

Subgroup analysis comparing FED and PBK eyes showed a greater decrease in anterior corneal Ks in PBK (− 1.21 ± 1.45 D) than in FED (− 0.27 ± 0.77 D) (P = 0.048) (Table 4). The reduction in absolute astigmatism was also greater in PBK (− 1.64 ± 1.91 D) than in FED (− 0.34 ± 1.55 D). Double-angle vector plots demonstrated a decrease in centroid from 0.87 D at 8° to 0.39 D at 160° in FED, and from 0.86 D at 161° to 0.35 D at 121° in PBK (Fig. 6).

Table 4.

Comparison of preoperative and postoperative measurement by surgical indication

	FED			PBK			P-value*
	Preop	Postop	Δ(Post-pre)	Preop	Postop	Δ(Post-pre)
Auto kerato-refractometer	N = 20			N = 35
Anterior cornea
Ks, D	44.60 ± 2.75	44.83 ± 2.49	0.23 ± 2.02	45.09 ± 2.55	44.99 ± 2.18	-0.10 ± 2.62	0.610
Kf, D	42.75 ± 2.47	43.14 ± 2.29	0.56 ± 1.99	41.81 ± 2.53	43.17 ± 2.05	1.36 ± 2.72	0.223
Km, D	43.59 ± 2.51	43.98 ± 2.32	0.39 ± 1.75	43.45 ± 2.36	44.08 ± 2.00	0.63 ± 2.49	0.695
Magnitude of astigmatism, D	2.03 ± 1.47	1.69 ± 1.12	-0.34 ± 1.55	3.46 ± 1.90	1.82 ± 1.39	-1.64 ± 1.91	0.023
Astigmatism in vector form, D	0.87 at 8°	0.39 at 160°		0.86 at 161°	0.35 at 121°
Anterior segment OCT	N = 10			N = 15
Anterior cornea
Ks, D	43.89 ± 1.61	43.62 ± 1.67	-0.27 ± 0.77	46.90 ± 2.06	45.68 ± 1.72	-1.21 ± 1.45	0.048
Kf, D	42.78 ± 1.98	42.60 ± 1.73	-0.18 ± 1.20	43.89 ± 2.00	43.82 ± 1.33	-0.08 ± 2.01	0.884
Km, D	43.33 ± 1.74	43.11 ± 1.69	-0.23 ± 0.93	45.39 ± 1.35	44.75 ± 1.41	-0.64 ± 1.03	0.315
Magnitude of astigmatism, D	1.11 ± 0.93	1.02 ± 0.46	-0.09 ± 0.75	3.00 ± 3.04	1.86 ± 1.23	-1.13 ± 2.85	0.272
Astigmatism in vector form, D	0.64 at 177°	0.31 at 173°		0.83 at 79°	1.01 at 133°
Posterior cornea
Ks, D	-6.32 ± 0.52	-6.61 ± 0.45	-0.29 ± 0.64	-6.87 ± 0.43	-6.83 ± 0.43	0.04 ± 0.43	0.134
Kf, D	-5.86 ± 0.52	-6.23 ± 0.36	-0.37 ± 0.48	-6.39 ± 0.41	-6.39 ± 0.37	-0.00 ± 0.41	0.052
Km, D	-6.09 ± 0.49	-6.42 ± 0.39	-0.33 ± 0.53	-6.63 ± 0.39	-6.61 ± 0.39	0.02 ± 0.38	0.069
Magnitude of astigmatism, D	0.46 ± 0.29	0.38 ± 0.21	-0.07 ± 0.40	0.48 ± 0.25	0.43 ± 0.22	-0.04 ± 0.33	0.838
Astigmatism in vector form, D	0.14 at 172°	0.33 at 179°		0.17 at 16°	0.36 at 2°
Total cornea
Ks, D	43.11 ± 1.49	42.39 ± 1.67	-0.71 ± 1.15	45.82 ± 2.08	44.39 ± 1.69	-1.43 ± 1.60	0.234
Kf, D	41.79 ± 2.14	41.09 ± 1.77	-0.69 ± 1.80	42.67 ± 2.22	42.39 ± 1.27	-0.28 ± 2.25	0.630
Km, D	42.45 ± 1.75	41.74 ± 1.70	-0.70 ± 1.44	44.24 ± 1.30	43.38 ± 1.36	-0.85 ± 1.13	0.770
Magnitude of astigmatism, D	1.32 ± 1.17	1.30 ± 0.60	-0.02 ± 0.90	3.15 ± 3.42	1.99 ± 1.22	-1.15 ± 3.19	0.287
Astigmatism in vector form, D	0.82 at 175°	0.63 at 176°		0.82 at 76°	1.14 at 142°
Central corneal thickness, µm	722.6 ± 87.1	511.6 ± 30.3	-210.9 ± 90.4	747.8 ± 103.8	520.9 ± 39.8	-226.9 ± 16.3	0.542

D = diopter, FED = Fuchs’ endothelial dystrophy, Ks = steep keratometry, Kf = flat keratometry, Km = mean keratometry, N = number, PBK = pseudophakic bullous keratopathy, SD = standard deviation

Values with statistical significance are shown in boldface

*P-values represent between-group comparisons of Δ(Post–pre) values for FED versus PBK

Fig. 6.

Double-angle plots for preoperative (A, C) and postoperative (B, D) anterior corneal astigmatism measured by auto kerato-refractometer in Fuchs’ endothelial dystrophy (FED) (A, B) and pseudophakic bullous keratopathy (PBK) (C, D)

Discussion

In this study of 55 eyes undergoing DMEK, corneal edema was found to substantially influence anterior corneal astigmatism based on ARKRT-derived measurements. Postoperatively, significant reductions in anterior corneal astigmatism were observed, accompanied by decreases in keratometric values and CCT. Greater preoperative CCT showed a clinically relevant association with larger postoperative astigmatic changes on univariable analysis. The 790 μm cutoff showed modest discrimination for identifying eyes at risk for changes exceeding 3 D. Subgroup analysis revealed that PBK eyes exhibited more pronounced flattening of the anterior cornea and greater reductions in absolute astigmatism compared to FED eyes. Posterior corneal parameters showed minimal change overall, and their correlation with CCT was limited to posterior astigmatism.

The differences observed between ARKRT and AS–OCT measurements are likely attributable to their distinct measurement principles. ARKRT, which is based on a Placido ring system, primarily samples the very central corneal surface and derives curvature from anterior surface reflections [21]. As a result, it may overemphasize anterior corneal changes and be more sensitive to localized central edema or irregularity, potentially exaggerating the magnitude of anterior astigmatism. In contrast, AS–OCT acquires cross-sectional tomographic images of both anterior and posterior surfaces across a broader corneal area, allowing direct calculation of total corneal power and astigmatism [22]. Thus, AS–OCT can provide complementary information on posterior and total corneal parameters. However, in the present study, AS–OCT data were available only in a limited subgroup of eyes, and therefore AS–OCT-based findings should be interpreted as secondary exploratory observations rather than definitive evidence. These methodological differences should be considered when interpreting keratometric and astigmatic changes after DMEK. However, when considering astigmatism correction in actual surgery, it is important to account for central corneal astigmatism, which most directly affects visual quality; therefore, ARKRT-derived values should also be carefully considered. The lack of statistical significance in AS–OCT-based astigmatism magnitude, despite a similar direction of change, is likely explained by the smaller sample size of the AS–OCT subgroup and the broader tomographic measurement area of AS–OCT compared with the more centrally weighted ARKRT measurements. Taken together, the present findings indicate that postoperative astigmatic reduction was most consistently supported by ARKRT measurements, whereas AS–OCT showed a similar directional tendency without consistent statistical confirmation.

The changes in centroid values observed in double-angle plot analysis are influenced not only by variations in absolute astigmatism but also by changes in axis distribution. Following DMEK, the resolution of corneal edema generally reduces absolute astigmatism and decreases variability in astigmatic axes, resulting in a greater proportion of measurements clustering within the 95% confidence ellipse. This improved alignment of vector directions can lead to an increase in the vector mean (centroid) even when the overall magnitude of astigmatism decreases. Thus, an increase in centroid after surgery does not necessarily indicate greater astigmatism but rather reflects more uniform and organized astigmatic orientation. This effect is likely more pronounced in eyes with substantial preoperative irregular astigmatism, representing an improvement in corneal surface regularity after DMEK.

The more pronounced anterior corneal flattening and greater reduction in absolute astigmatism observed in PBK eyes may be explained by differences in the pattern of edema between PBK and FED. PBK typically leads to diffuse stromal swelling that extends beyond the central cornea into the paracentral and peripheral regions [23]. This widespread edema may reduce stromal tensile strength more globally, allowing the anterior corneal curvature to undergo greater geometric remodeling once the edema resolves after DMEK. In contrast, FED is characterized by a predominantly central pattern of edema, with peripheral corneal structure relatively preserved [24]. Because peripheral stromal rigidity is maintained in FED, postoperative normalization of corneal hydration results in more limited anterior surface flattening. These spatial differences in edema distribution likely account for the greater anterior corneal shape recovery and astigmatic reduction observed in PBK compared with FED. However, the greater reduction in astigmatism observed in PBK eyes should be interpreted with caution, as PBK eyes also had larger baseline astigmatism and steeper preoperative corneal curvature. Therefore, the observed subgroup difference may partly reflect baseline corneal characteristics rather than disease category alone.

Our results highlight the importance of accounting for preoperative corneal edema when planning astigmatic correction in DMEK, particularly in triple procedures. We found that a preoperative CCT exceeding 790 μm identified eyes more likely to experience a postoperative astigmatic change of more than 3 D as edema resolves. This suggests that in eyes with pronounced edema, clinicians should be aware of the possibility of postoperative astigmatic change and may consider delaying definitive astigmatic correction until after corneal deturgescence, depending on the overall clinical context. However, this relationship should be interpreted cautiously, as the effect of CCT was attenuated after adjustment for preoperative astigmatism and surgical indication. Secondly, a postoperative myopic shift was observed; this should be considered when targeting refractive outcomes, especially in patients undergoing cataract surgery concurrently. Thirdly, the greater susceptibility of PBK eyes to preoperative steepening and astigmatic distortion compared with FED eyes indicates that the underlying pattern of edema plays a crucial role. The more diffuse and peripheral stromal edema characteristic of PBK, in contrast to the predominantly central and chronic edema in FED, likely contributes to the greater postoperative flattening and astigmatic reduction seen in PBK. To our knowledge, this study is among the first to suggest a specific CCT threshold that may help identify eyes at risk for clinically meaningful postoperative astigmatic change after DMEK.

This study has several limitations. First, its retrospective, single-center design may introduce selection bias and limit the generalizability of the findings. Second, the sample size—particularly within the FED and PBK subgroups—was relatively small, which may reduce the statistical power for subgroup comparisons. Third, astigmatism measurements were obtained using two different modalities (ARKRT and AS–OCT), each with inherent methodological differences, which may influence direct comparability of the results. In addition, AS-OCT analyses were limited to 25 eyes because only measurements obtained with the same AS-OCT platform were included to minimize inter-device variability. This reduced the statistical power of AS-OCT-based comparisons, particularly in subgroup analyses. Therefore, AS–OCT findings should be interpreted only as secondary exploratory observations. Fourth, postoperative measurements were analyzed at the time of thinnest CCT between 1 and 4 months rather than at a fixed postoperative time point. Although this approach was intended to capture the corneal state after maximal edema resolution, it may not fully reflect refractive stability and may have introduced variability across eyes. Fifth, the FED and PBK subgroup analyses were limited by baseline differences in preoperative astigmatism and keratometric values, as well as the inability to adjust for other potential confounders such as disease duration and prior interventions. The 3.0 D threshold used in ROC analysis was selected empirically within this dataset and was not predefined from prior literature. Given the modest AUC, limited sensitivity, and lack of external validation, this cutoff should not be applied as a definitive clinical risk-stratification threshold. Finally, other factors that may influence postoperative astigmatism, such as graft size, or centration were not fully controlled for in the analysis.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Abbreviations

AUC

Area Under the Curve

ARKRT

Auto Kerato-Refractometer

AS–OCT

Anterior Segment Optical Coherence Tomography

AXL

Axial Length

CCT

Central Corneal Thickness

D

Diopter

DMEK

Descemet Membrane Endothelial Keratoplasty

DSAEK

Descemet Stripping Automated Endothelial Keratoplasty

FED

Fuchs’ Endothelial Dystrophy

IOL

Intraocular Lens

IRB

Institutional Review Board

Kf

Flat Keratometry

Km

Mean Keratometry

Ks

Steep Keratometry

OCT

Optical Coherence Tomography

PBK

Pseudophakic Bullous Keratopathy

ROC

Receiver Operating Characteristic

SD

Standard Deviation

SE

Spherical Equivalent

Author contributions

Chung Young Kim: Writing – original draft, Investigation, Data curation, Formal analysis, Conceptualization. Mee Kum Kim: Writing –review & editing, Methodology, Supervision, Conceptualization.

Funding

No funding was received for this research.

Data availability

All data used is available upon reasonable request from the corresponding author Dr. Mee Kum Kim (kmk9@snu.ac.kr).

Declarations

Ethical approval and consent to participate

This retrospective study was approved by the Institutional Review Board of Seoul National University Hospital (IRB No. H-2508-180-1672) and adhered to the tenets of the Declaration of Helsinki. The requirement for informed consent was waived due to the retrospective nature of the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.