Article Text

Download PDFPDF
A fuller picture? National registry studies and the assessment of corneal graft outcomes
  1. Emma J Hollick
  1. Ophthalmology, King's College Hospital NHS Foundation Trust, London, UK
  1. Correspondence to Emma J Hollick, Ophthalmology, King's College Hospital NHS Foundation Trust, London SE5 9RS, London, UK; emma.hollick{at}

Statistics from

Numerous corneal graft studies have demonstrated that the Descemet’s membrane endothelial keratoplasty (DMEK) potentially delivers better patient outcomes than Descemet’s stripping endothelial keratoplasty (DSEK). Performed successfully, DMEK has been shown to deliver better and more rapid recovery of vision, although this difference has narrowed with the introduction of thinner DSEK tissue.1 2 DMEK also has lower rejection rates than DSEK.3 4

However, DMEK is considered a more challenging technique than DSEK with higher rates of early graft failure. One major randomised controlled trial (RCT) reported that surgical failures requiring retransplantation occurred in 7% of DMEK cases, compared with 0% of DSEK cases.5 Early graft failure can be caused by iatrogenic damage due to prolonged donor manipulation or upside-down orientation. Donor grafts may be difficult to unfold due to fibrin or haemorrhage sticking the DMEK scroll together and, in such cases, either the donor is reinserted after the fibrin has been removed or the donor material is lost. DMEK has also been shown to be associated with higher levels of rebubbling, usually performed if more than one third of the graft is detached.6 7 In two recent RCTs comparing EK techniques, 24% of DMEKs were rebubbled compared with only 4% of DSEK cases.5 8 A recent large cohort reported a DMEK rebubbling rate of 32%.6 Rebubbling is inconvenient for the patient and not always successful: in almost a third of rebubbled cases reported by Dunker the grafts failed.9 Patients who have had a rebubbling procedure have been shown to have a reduced endothelial cell count,9 particularly if rebubbling is performed more than once.10 The risk of rebubbling has been shown to be higher if there have been surgical complications.9

To mitigate the complications associated with DMEK, multiple techniques and approaches have been developed by surgeons, reflected in the literature. A preoperative laser iridotomy may reduce the risk of haemorrhage and surgeons may mark their grafts to minimise the risk of incorrect orientation.11 Primary graft failure and loss of donor tissue can be reduced if surgery is supervised by a trained DMEK surgeon and a standardised technique is used.12 We have reported a low rebubble rate of below 3% with a standardised technique.13–15 Early rebubbling has been shown to improve visual outcomes and reduce the risk of fibrosis and corneal scarring.16

Notwithstanding such progress, appropriate management of the learning curve associated with DMEK remains a challenge. Higher rates of complication have been reported for surgeons in the DMEK learning curve, whether defined in terms of duration of experience or volume of cases undertaken.12 17–19 A study across a cohort of 2485 cases showed graft detachment rate of 34% for novice DMEK surgeons, compared with 22% for experienced surgeons.18 A separate study showed 37% of surgeons reporting primary graft failures in their first 10 DMEK cases.12 Data from six US eye banks have shown that the risk of graft failure reduces as surgeons move up the learning curve, such that the proportion of early graft failures in DMEK progressively align with failure rates in DSEK.19 Another study suggests that the average surgeon may require 45 DMEK cases to pass through the learning phase.20

In this context, the results of the Dutch Registry study reported in this issue are of particular interest. Dunker et al report the results of the first 7 years of DMEK surgery recorded by the Dutch Registry, a national registry capturing all grafts performed in the Netherlands (excluding a single private clinic).9 A previous paper from the same Dutch Registry study confirms an overall failure rate higher than that in most other studies: they report a 1-year graft survival of only 85% after DMEK,21 compared with the 92%–100% DMEK survival rate elsewhere reported.19 For novice surgeons, the Dutch Registry study paper in this issue reports that a quarter of cases failed in the first few months following surgery. Early graft failure reduced to between 4% and 9% for more experienced surgeons.

Large National Registry studies may provide a corrective to a potential selection bias in previous DMEK reporting. For understandable reasons, cohort studies on graft outcomes tend to originate from specialised units. Surgeons with poor results may equally be less likely to publish postoperative outcomes. RCTs comparing graft techniques may also be more likely to be performed in high-volume specialised units. By contrast, as the authors explain, the Dutch Registry study data comes from a ‘heterogeneous group of medical centres, including high and low volume as well as specialised and non-specialised centres’.21 Data from the Dutch and other national registries may therefore provide a fuller picture of outcomes across an entire surgical population and should make the results more generalisable.9 21–25

While this study reports improved outcomes as surgeons gain more years of experience, the comprehensive datasets provided by national registries may allow other comparisons which could be equally instructive. For example, it may be possible to separate out the results of the high and low volume and specialised and non-specialised centres that the registry data encompasses. It would also be interesting to compare whether the duration and intensity of progression through the DMEK learning curve impact outcomes: to what degree might a lower number of cases in a shorter period of time accelerate a novice surgeon’s progress?

But the Dutch Registry study may also prompt more challenging—and immediate—questions for the surgical community and, potentially, professional bodies. Worldwide, it is estimated that only 1 in 70 patients requiring a corneal graft undergoes a transplant each year.26 Given the global shortage of donor corneas, is its loss in one in four DMEK cases by less experienced surgeons acceptable? If it is not, might policy changes be required, such as concentrating DMEK training in high-volume units under experienced supervision or requiring utilisation of donor tissue for multiple recipients?

DMEK is a technique that can deliver a step-up in outcomes compared with the established DSEK approach. As with many surgical innovations, it comes with risks, which individual surgeons and institutions are developing means of mitigating. Progress is being made in terms of accelerating DMEK skills-transfer, via advances in technology and simulation practice.27 28 A standardised approach with a trained supervisor should reduce the waste of donor cornea associated with surgeon learning. Over time it might be expected that organic spread of such surgical best practice will lead to improved outcomes and encourage more surgeons to move to DMEK for suitable cases. However, the Dutch registry study and previous reports from other national registries indicate that successful outcomes overall may be significantly lower than previously reported, and—to some extent—may be being pulled down by the performance of less-experienced surgeons on the learning curve of the technique.9 21–24 Due to the scarcity of donor corneas this raises the question of whether such a gradualist approach is sufficient. As well as further research, collective discussion across the surgical community and professional bodies may be required to address whether this merits a response in terms of guidance or regulation.



  • Competing interests None declared.

  • Provenance and peer review Commissioned; internally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • Clinical science
    Suryan Dunker Bjorn Winkens Frank van den Biggelaar Rudy Nuijts Pieter Jan Kruit Mor Dickman on behalf of the Dutch Cornea Consortium Marjolijn C Bartels Yanny Y Cheng Bart T H van Dooren Cathrien A Eggink Annette J M Geerards Tom A van Goor Ype P Henry Ruth Lapid-Gortzak Chantal M van Luijk Ivanka J van der Meulen Carla P Nieuwendaal Siamak Nobacht Monica T P Odenthal Lies Remeijer Jeroen van Rooij Nathalie T Y Santana Remco Stoutenbeek Mei L Tang Thijs Vaessen Nienke Visser Robert H J Wijdh Robert P L Wisse Michel J W Zaal