Graft detachment is a common complication after DMEK surgery [24, 25]. The detection of an early graft detachment after DMEK surgery is essential for a successful outcome as it might demand a rebubbling [1, 19]. In this study on 45 eyes undergoing uncomplicated DMEK surgery, 25 (55.6%) showed graft detachment, with 11 (24.4%) eyes needing rebubbling due to detachment ≥ 1/3 of the graft’s area (group 3). Those eyes had a greater CCT prior to and 1 week after DMEK compared to eyes without any detachment (group 1). One week after uncomplicated DMEK, 90.0% of patients in group 1 had a CCT smaller than 600.0 µm and an improved BCVA. A CCT greater than 700.0 µm and an increase in CCT at 1 week only occurred in graft detachment (groups 2 and 3). Stromal ripples were present in all groups prior to DMEK surgery; severe posterior stromal ripples 1 day and 1 week after surgery were only present in a patient in group 3.

While the reported occurrence of graft detachments varies in the literature (2.0–82.0%) [1, 9,10,11,12,13,14], our findings match those reported by Muijzer et al. (50.8% graft detachment, 26.2% rebubbling) and Guindolet et al. (36.0% rebubbling) [26, 27]. Yeh et al. also reported overall detachment of approx. 50.0% over the course of 6 months, with 33.0% showing graft detachment of more than 1/3 of the graft’s area [28]. Differences in reported detachment rates could relate to possible risk factors reported in the literature, including donor characteristics, such as donor age or low endothelial cell density; recipient factors, such as recipient age; and surgical parameters, such as descemetorhexis diameter, graft decentration, anterior chamber tamponade agent, postoperative intraocular pressure, and surgeon experience [16]. Overall, agreement on these risk factors across reports was weak [16].

Reports on the CCT 1 week after DMEK surgery are scarce. Guindolet et al. reported that the preoperative CCT was not significantly different between the group that required rebubbling (624 ± 92 µm) compared to the one that did not (660.0 ± 120.0 µm; P = 0.340) [27]. Contrarily, we found greater preoperative CCT values in eyes in group 3 (746.8 ± 95.8 µm, group 3) compared with eyes in group 1 (665.0 ± 74.4 µm, group 1; P = 0.041). While CCT in eyes without rebubbling matched those in our study without any detachment, CCT in eyes with rebubbling differed greatly. This could have been caused by a wider variety of included patients, since only graft preparation failures and inverted grafts were excluded [27]. CCT measurements 1 week after DMEK were not reported [27].

Coco et al. reported absolute CCT values at approx. 3 days after DMEK only in relation to the presence of stromal ripples, not in relation to graft detachment nor subsequent rebubbling (no stromal ripples: 624.3 ± 80.2 μm; mild stromal ripples: 707.6 ± 62.9 μm; moderate/severe stromal ripples: 757.4 ± 125.1 μm; P < 0.001 one-way ANOVA) [22]. While an exact comparison to our data is impossible, the CCT range matched our findings.

Muijzer et al. performed pachymetry mapping showing an increase in local corneal thickness in zones with a graft detachment [26]. One day after surgery, the mean corneal thickness over all zones was higher in eyes that developed a graft detachment compared with subjects without a graft detachment (745.0 ± 82.4 µm vs. 805.4 ± 98.03 µm, P = 0.015) [26]. Importantly, within subjects who developed a graft detachment, the corneal zones in which the graft detached were thicker compared with zones in which the graft remained attached; conversely, the thickness of corneal zones in which the graft remained attached in the detachment group did not differ compared with the corneal thickness in subjects without a graft detachment [26]. While predictability of graft detachment using this method was valid, no correlation to the necessity for rebubbling was made [26]. Furthermore, absolute values on the CCT were not reported for any time [26]. Dirisamer et al. found that corneal thickness was increased in corneal quadrants with detached grafts compared with adjacent corneal quadrants with attached grafts [29]. A greater CCT in eyes with graft detachment as detected in our data supports this finding.

In our study, a relative and absolute increase in CCT after 1 week was only present in eyes in groups 2 and 3. Interestingly, Guindolet et al. reported that a decrease in the CCT after 1 day was associated with a lower risk for rebubbling, and that an increase of 20.0% of the CCT was associated with an increased risk for rebubbling [27]. This is supported by our data on the 1-week examination, where a decrease in relative/absolute CCT and a CCT of less than 700 µm was associated with graft attachment without the need for rebubbling. In a subgroup analysis regarding preoperative CCT, Guindolet et al. showed no difference in the rebubbling rate between patients with CCT greater than 700.0 µm and those with a CCT less than 700.0 µm [27]. While in our study the CCT was greater in eyes in group 3 compared to eyes in group 1, all groups contained eyes with CCT greater 700.0 µm prior to surgery. Guindolet et al. suggested introducing indices of CCT increase to predict a detachment risk in clinical practice [27].

Our study showed that in groups 2 and 3, the area and maximum detachment distance increased from 1 day to 1 week after surgery, followed by a consecutive decrease to 6 months after surgery. In a prospective study, Yeh et al. described a “biphasic” adherence pattern: initial attachment within the first hour after surgery in the majority of eyes, followed by a partial detachment in approx. 1/3 of the eyes at 1 week and a recovery of graft adherence at 1 to 6 months [28]. In eyes with less than 1/3 of the graft’s area detached, we found similar results: both max. area and distance of detachment were largest 1 week after surgery and decrease afterwards. Also, our data agreed with a 100.0% negative predictive value of AS-OCT scans 1 week and 1 month after DMEK surgery: If graft attachment (group 1) or graft detachment in less than 1/3 of the graft’s area (group 2) was recorded, a detachment of over 1/3 of the area did not occur after 6 months [28]. Hence, the 1-week and 1-month postoperative AS-OCT scan showed good sensitivity. Spontaneous graft detachment after 1 month after DMEK was not reported by groups previously mentioned, yet rare cases of late graft detachment exist [30]. We would argue that in cases of good visual recovery and inconspicuous slit-lamp examination, a AS-OCT is not mandatory on examinations past 1 month after surgery.

In a study on 111 eyes receiving DMEK surgery, Kramer et al. found a difference of BCVA improvement of 0.22 logMAR in favor of eyes with spontaneous graft adherence without intervention compared to eyes needing rebubbling 6 months after surgery (P = 0.048) [31]. While in our study no difference was found after 6 months, BCVA was higher 1 week after DMEK surgery in eyes without detachment (group 1) compared to eyes receiving rebubbling (group 3, P < 0.001). One month after surgery, no difference between groups was found, hinting at a quick recovery in eyes receiving rebubbling. Yeh et al. reported 44.0% of eyes receiving rebubbling reached a BCVA better or equal to approx. 0.2 logMAR 6 months after DMEK, while in our study, 75.0% of patients receiving rebubbling reached a BCVA better or equal to approx. 0.2 logMAR 6 months after surgery. Overall, a swift and excellent visual recovery has been described for DMEK, with and without rebubbling [15, 32, 33]. Furthermore, Dunker et al. found no relationship between the timing of rebubbling (i.e., within 1 week or longer) and incidence of graft failure, likely leading to BCVA decline [16].

An assumed predictive biomarker for graft detachment is the presence of stromal ripples, irregularities in the posterior corneal profile that assumed the shape of a ripple briefly after DMEK [22]. Coco et al. found that the presence of stromal ripples was significantly associated with the risk of graft detachments requiring rebubbling at any time, the risk of detachment of previously attached grafts, the risk of detachment worsening over time, and a higher CCT in mild or severe ripples compared with the absence of ripples [22]. In their study, OCT imaging was performed 2.9 ± 2.4 days after surgery. Consequently, when posterior stromal ripples are present, patients should be monitored more closely and managed on an individual basis as is the case with larger graft detachments [22]. Furthermore, stromal ripples were positively correlated with the CCT of patients after DMEK surgery [22]. However, whether they appeared first and lead to a problem with graft attachment or if they were a consequence of subclinical detachment is still not known [22]. While Coco et al. did not show if posterior stromal ripples were present before surgery, we looked for ripples in preoperative OCT images. In all three groups of our study, stromal ripples occurred prior to surgery. In some cases, complete resolution of severe stromal ripples occurred within 1 day after surgery; in others, an increase of severity over the course of 1 week was seen. One eye showed severe posterior stromal ripples both 1 day and 1 week after surgery. This very eye needed subsequent rebubbling. As our data reinforces, it remains uncertain if stromal ripples were the cause or consequence of a graft detachment. Further studies are needed to evaluate the cause of stromal ripples, and if preoperative or intraoperative factors influence their development [22].

As a limitation to our study, we chose to only include uncomplicated DMEK cases with limited concomitant diseases. That allowed us to reduce confounding factors yet limited the analysis of additional risk factors. Even though AS-OCT is the most common tool used for CCT measurement post DMEK, as it also allows the reliable visualization of graft detachments, other tools of CCT measurement such as ultrasound pachymetry or Scheimpflug camera (Pentacam) imaging have to be evaluated for their agreement with AS-OCT CCT values in attached and detached DMEK situations and for their suitability to predict different graft conditions, as we did in this study for AS-OCT.

If 1 week after uncomplicated DMEK CCT is < 600 µm and has decreased from before surgery, BCVA has improved, and there are no posterior stromal ripples, a graft detachment ≥ 1/3 and the need for rebubbling are very unlikely. In all other cases, meticulous slit-lamp and OCT inspection of the peripheral graft for detachments should be advised.