Descemet membrane detachment: risks in donor tissue processing

Descemet Membrane Detachment Risk Factors in Donor Tissue: When the Patient You Never Met Decides Your Surgical Outcome

The actual data, when you sit down and read it without the comforting preamble, says something rather different. The variables that genuinely threaten a DMEK graft are not necessarily the clocks ticking between death and the eye bank fridge. They are the chronic diseases the donor carried for decades before they died. This is where the paradigm deficit in corneal research shows up most sharply: we have spent years optimizing cold-chain logistics while quietly underestimating the biology embedded in the tissue itself.

Descemet membrane detachment after keratoplasty is no longer framed as a single surgical mishap. Eye banks and anterior segment surgeons increasingly recognize it as a multifactorial event, governed in part by the donor’s systemic and ocular history. Stripping the endothelium–DM complex from a donor button looks, on paper, like a mechanical exercise. In practice, it is a confrontation with a basement membrane whose adhesion strength and handling behavior were shaped years earlier by glycemic control, renal function, prior intraocular manipulation, and age.

The relevant distinction is between what makes tissue difficult to prepare and what makes a prepared graft difficult to unfold, position, or keep attached. Those problems overlap, but they are not interchangeable. A membrane that resists peeling is not automatically a membrane that will detach after implantation. A graft that scrolls tightly is not necessarily a graft with poor endothelial quality. Treating all of these behaviors as one generic form of tissue failure is how useful donor information gets flattened into a single label.

Systemic Comorbidities and Tissue Fragility: The Role of Diabetes and Chronic Disease

The uncomfortable finding from donor tissue analyses is that the comorbidities we treat as background epidemiology become, in the eye bank, foreground surgical risk. Donor type 2 diabetes, chronic kidney disease, heart failure, and a history of cataract surgery are not politely statistical curiosities. They can influence how the DM behaves under stripping forceps, how much resistance the technician encounters, and how readily the tissue tolerates manipulation.

Diabetes in particular reshapes the donor cornea in ways that eye bank technicians feel before surgeons see them. Studies examining DMEK graft preparation report that diabetic donor tissue exhibits a significantly higher degree of central adhesions, meaning the DM does not separate cleanly from the posterior stroma in the conventional plane. Pseudophakic donor eyes, by contrast, show greater graft fragility: the tissue strips, but tears more readily during manipulation. These are not the same problem, and conflating them has been a quiet source of preparation failure.

An adherent diabetic membrane and a brittle pseudophakic membrane require different handling strategies. The first may resist the initial separation and demand more deliberate progression through the cleavage plane. The second may appear to cooperate before failing at the point where traction, a change in angle, or an attempt to enlarge the prepared area places stress on the tissue. A uniform peeling protocol cannot fully account for both behaviors, even when the same instruments are used.

Chronic kidney disease and heart failure round out the systemic risk profile. Both conditions may be associated with changes in tissue hydration, microvascular integrity, and the chronic inflammatory environment, all of which can affect the material properties of the DM–stromal interface. The biochemical mechanism by which diabetes specifically weakens or strengthens that interface is not fully elucidated. That uncertainty matters: a consistent clinical association does not automatically reveal a single molecular pathway.

What can be said with greater confidence is that diabetic donor tissue often demands more time, more careful force control, and a lower tolerance for treating resistance as a technical nuisance. A difficult peel is not always evidence that the technician has chosen the wrong movement. Sometimes it is the tissue reporting its history.

The donor’s metabolic memory — decades of systemic disease compressed into a thin posterior membrane — may matter more to preparation than the hours recorded on a storage log.

The practical consequence is not that every diabetic or chronically ill donor should be excluded. That would turn a risk signal into an unusable rule. The consequence is that systemic history needs to become operational information. If it remains buried in a recovery record that never reaches the person preparing the graft, the eye bank is discarding one of the few clues available before the tissue is placed under traction.

The Contralateral Effect: Predicting Graft Tearing Based on Paired Donor Tissue Performance

One of the more clinically actionable findings in the DMEK literature concerns paired donor eyes. If the first eye of a pair tears during manual preparation, the odds of a tear occurring during preparation of the contralateral eye increase by an odds ratio of 7.2. That is a substantial increase in odds, not a near-certain prediction of bilateral failure. The distinction is more than statistical housekeeping. It determines how the result should be used at the bench.

A first-eye tear should change the status of the second eye from routine to high-alert. It should prompt a review of the preparation plan, the instrument choice, the amount of traction being applied, and the intended allocation of the tissue. It should not be interpreted as proof that the second eye will fail, nor as a reason to abandon it automatically. An odds ratio of 7.2 identifies a strong association between paired-eye performance; it does not replace preparation judgment with a deterministic rule.

The timing of this information is important. Paired-eye performance is not fully knowable at the moment of tissue recovery. It becomes available only after one eye has undergone preparation. The first eye’s behavior is therefore a post-recovery finding that can inform the handling of the second eye, not a donor characteristic that can be entered into the allocation system in advance.

That makes the contralateral signal unusually useful and unusually easy to misuse. The second eye should not simply be stripped with the same technique that has just failed on the first. But neither should it be treated as biologically unusable by default. The more defensible response is to regard it as a different processing problem: one that may justify senior handling, an alternative preparation strategy, or reassignment to a graft format with different mechanical demands.

The published mean peel time of 10.1 ± 3.3 minutes suggests that preparation often falls within a recognizable working range, but a mean is not a deadline and it cannot neutralize donor-specific behavior. A prolonged peel may reflect difficult adhesion, operator hesitation, or a tissue defect that is already developing. Once the first eye has torn, the second eye’s preparation should be treated as a new decision rather than a routine repetition.

The same point applies to tissue allocation. When the first eye of a pair fails, the second eye is often considered compromised by default, even though the observed event may reflect a combination of tissue biology and mechanical handling. With an odds ratio of this magnitude, the second eye deserves explicit flagging. In some programs, such tissue may be redirected toward DSAEK or ultrathin-DSAEK formats, where the manual stripping demands are different and the consequences of a peripheral defect may be less severe. That is a risk-management decision, not proof that the second eye is unsuitable for every use.

Age-Related Biomechanics: Balancing Endothelial Adherence and Spontaneous Scrolling

Donor age is the variable the field talks about most, and the one it understands least. Conventional wisdom splits neatly along a generational fault line. Older donor tissue, particularly tissue from donors over 65, may peel faster. There is genuine biomechanical logic here: age-related thickening of the posterior nonbanded layer can alter the cleavage plane. Yet the same tissue may scroll less aggressively once it is inside the anterior chamber.

Younger donor tissue, particularly tissue from donors under 50, can behave in the opposite direction. It may peel with relatively normal adhesion but scroll tightly once unscrolled on the table. The result is a preparation that looks favorable at one stage and becomes difficult at the next. What appears to be a healthier graft from an endothelial standpoint may demand more work during deployment.

Spontaneous tight scrolling in younger donor tissue is not a cosmetic inconvenience. It creates genuine intraoperative unscrolling difficulties and can elevate the risk of postoperative graft detachment. The surgeon is left fighting tissue that wants to roll itself back into a scroll as soon as air or gas is introduced, regardless of how carefully it was positioned. The problem is mechanical persistence: the graft does not simply need to be placed; it needs to remain in the intended configuration long enough for adherence to develop.

This is also where the consensus around donor age becomes a paradigm deficit. The threshold for acceptable endothelial cell density — above 2500 cells/mm² as a starting reference — is a cell-count metric, not a biomechanical one. A 45-year-old donor with excellent endothelial counts may produce a graft that is more difficult to handle than a 72-year-old donor with comparable counts. The cell count answers one question. It does not tell the surgeon how the tissue will scroll, unfold, or respond to air.

Age therefore works better as a context variable than as a standalone quality score. It may help predict broad handling tendencies, but it should not be asked to perform the work of a direct assessment of scroll behavior. An allocation system that treats age as a proxy for all forms of tissue quality risks favoring a convenient number over a clinically relevant description.

The more useful profile would combine age with preparation behavior:

• whether the DM separates readily or resists the initial peel;
• whether the tissue tears under traction or tolerates manipulation;
• whether the graft forms a tight scroll after preparation;
• whether the scroll can be opened and kept stable during implantation;
• whether the endothelial count supports the intended use without obscuring handling risks.

The honest answer is that age may become a secondary predictor of detachment risk once scrolling tendency and preparation behavior are taken into account. Eye banks that report both donor age and a qualitative assessment of expected scroll behavior — something many currently do not — give surgeons a more actionable profile than age alone.

Processing Protocols and the Limits of Manual Peeling Techniques

Manual SCUBA preparation — the submerged corneal underwater approach — has become an eye bank standard for DMEK tissue processing. Reported tear rates hover around 6%. That number is low enough to feel reassuring and high enough to be operationally serious. Six percent of prepared tissue represents a nontrivial volume of donor corneas that cannot be used as intended, or that arrive in the operating room with a defect the surgeon discovers during the procedure.

The honest framing is that manual peeling remains constrained by the underlying tissue biology. Hydrodissection, pneumatic dissection, and forceps-based peeling each modify the mechanics of separation. Whether any of these approaches completely eliminates DM tearing in diabetic or pseudophakic donor categories remains unresolved in the available research. The evidence does not justify presenting one method as a definitive solution, and it does not establish that all alternatives perform equally across every high-risk tissue type.

That uncertainty should not be confused with therapeutic pessimism. Different tools may still change the probability, location, or severity of a tear. They may also alter how much direct traction is transferred to the membrane, how clearly the cleavage plane can be followed, or how much manipulation is required after the initial separation. But the research question is not closed. A technique that appears advantageous in one preparation setting may not eliminate failure when the underlying tissue has unusually strong adhesion or unusual fragility.

What is better established is the role of operator experience. Peel time and tear rate are influenced by the technician’s familiarity with DMEK preparation, and operator variability is itself a hidden variable in outcome reporting. An eye bank processing only a small number of DMEK preparations may have difficulty reproducing published averages because the learning curve remains active. The issue is not simply whether a technician knows the sequence of movements. It is whether the technician can recognize when the tissue is deviating from the expected plane and adjust before resistance becomes a tear.

This makes centralization of high-risk preparation a more defensible strategy than universal in-house capability. Centralization does not remove biological risk, and it does not make the efficacy of a particular tool settled. It does, however, concentrate experience where donor-specific behavior is most difficult to interpret. A protocol can specify the instrument. It cannot substitute for the judgment required when the tissue stops behaving like the protocol’s average example.

Debunking Post-Mortem Variables: Storage and Enucleation Intervals in Graft Detachment

It is time to retire the assumption that death-to-enucleation time, enucleation-to-processing time, and total storage duration automatically explain postoperative graft detachment. They do not appear to do so in any statistically detectable way in the analyses described here. Studies of DSAEK tissue processing and implantation have examined these intervals and found no significant difference between grafts that later detached and grafts that did not. The temptation to blame old tissue for a postoperative detachment is, the data suggests, often a retrospective confabulation.

That does not make storage conditions irrelevant to tissue quality. It means that elapsed time, considered by itself, is a poor explanation for the specific event of graft detachment. A well-documented interval can be operationally reassuring without being a useful predictor of whether the graft will separate after keratoplasty. Conversely, a longer interval should not be treated as the decisive cause when the more informative variables may lie in donor disease, tissue handling, or graft biomechanics.

This distinction matters for both allocation and quality control. If post-mortem intervals do not predict detachment in the relevant analyses, then the logic of minimizing every interval as though it were an independent detachment lever deserves scrutiny. Timing remains important for logistics, traceability, and the broader preservation process. It simply should not displace the biological and mechanical signals that are closer to the failure mechanism.

The variables that matter most in the preparation and implantation pathway include donor systemic disease, prior ocular surgery, age-related scrolling behavior, and the observed performance of the paired eye. They are not all available at the same moment. Systemic and ocular history may be available during recovery and record review. Age is available at allocation. Scroll behavior and paired-eye performance emerge during preparation. That sequence is crucial: a risk model should distinguish what can be known before processing from what must be learned during processing.

The donor cornea is a biological record of decades of systemic disease. The eye bank’s job is to read that record before the surgeon has to fight the tissue.

The challenge this leaves on the table is direct. Translational vision science has spent considerable energy optimizing the logistics of tissue procurement while treating the donor’s biological history as background noise. The evidence suggests that the background noise is part of the signal. Until donor systemic profiling becomes a structured input into tissue allocation, and until preparation outcomes are fed back into the handling plan for paired tissue, failure rates will continue to reflect inherited workflow more than the biology actually present in the graft.

The practical shift is not to abandon timing, age, or standardization. It is to stop pretending that any one of them can stand in for tissue behavior. Enucleation and storage intervals describe the journey. Diabetes, prior surgery, paired-eye performance, adhesion, fragility, and scrolling describe what arrives. In DMEK, that difference is where preparation succeeds or tears apart.