What Is DMEK Surgery? Procedure, Recovery, and More

Descemet membrane endothelial keratoplasty, known as DMEK, is a partial-thickness corneal transplant that replaces only the innermost layer of cells lining the back of the cornea. Unlike older transplant techniques that swap out most or all of the corneal tissue, DMEK grafts a thin membrane roughly 10 to 15 micrometers thick, which means the eye’s own structural tissue stays intact. That selective approach translates to faster visual recovery, sharper final vision, and a dramatically lower risk of graft rejection compared with full-thickness transplants. The trade-off is a technically demanding surgery and a recovery period with its own quirks, including a signature instruction you may have heard about: lying face-up for hours after the operation.

Why DMEK Is Performed

The cornea’s endothelial layer is a single sheet of cells on the back surface that pumps fluid out of the cornea to keep it clear. When those cells die off faster than normal, the cornea swells and vision turns hazy, glary, and unreliable. The most common reason for this is Fuchs endothelial corneal dystrophy, a genetic condition in which endothelial cells gradually deteriorate. Fuchs dystrophy accounted for about 36% of all corneal transplants performed in the United States in 2021, making it the leading indication for corneal transplant for the ninth year running at that point.¹ The vast majority of those patients received endothelial keratoplasty rather than a full-thickness graft.

DMEK is also used for bullous keratopathy, a condition where the endothelium fails after previous eye surgery (commonly cataract surgery), and for a handful of other endothelial disorders. It is not appropriate for every type of corneal disease. Problems that affect the front layers of the cornea, such as scarring from infection or keratoconus, still require different transplant approaches because DMEK only replaces the back layer.

How the Surgery Works

DMEK is performed under local anesthesia and usually takes somewhere between 25 and 60 minutes depending on whether it is done alone or combined with another procedure.² The basic sequence involves three steps: removing the patient’s damaged endothelial layer, inserting the donor graft, and using an air or gas bubble to press the graft against the back of the cornea so it can adhere.

The donor tissue is a delicate sheet of Descemet membrane with endothelial cells attached. It naturally curls into a tight scroll once separated from the donor cornea, which creates one of the procedure’s biggest technical challenges: getting that scroll to unfold correctly inside the eye. Surgeons have developed a range of techniques to coax the graft open. One common method involves injecting short bursts of balanced salt solution to spin the scroll continuously, rotating it until it opens in the correct orientation, then finishing with gentle tapping maneuvers on the corneal surface.³ In more difficult situations, such as eyes that have had previous vitrectomy surgery, the usual tapping technique may not work because the eye cannot maintain enough pressure. In those cases, surgeons have used suction-based approaches with a small cannula to pull the graft open in a more controlled way.⁴

Once the graft is unfolded and properly oriented, the surgeon fills most of the anterior chamber with an air bubble. That bubble acts as a tamponade, holding the tissue in place while biological adhesion takes over during the first few days.

Pre-loaded Grafts and Eye Bank Preparation

Traditionally, the surgeon or a surgical team member prepared the donor tissue in the operating room just before insertion. That step adds time and introduces a point where the graft can be damaged. Increasingly, eye banks now offer pre-stripped, pre-stained, and pre-loaded grafts — sometimes called “P3 DMEK” — that arrive ready to inject directly into the eye. A study comparing pre-loaded grafts to surgeon-prepared tissue found no significant difference in corneal thickness outcomes, graft failure rates, or the frequency of needing a rebubble procedure afterward, though the rate of partial graft detachments was somewhat lower with pre-loaded tissue.⁵ Pre-loaded grafts have helped standardize the procedure across different surgical centers and reduce one of the steeper parts of the learning curve.

Recovery and the Supine Positioning Question

After DMEK, patients are typically told to lie flat on their backs (supine) so that the air bubble floats upward against the graft. In the first 48 hours, the bubble covers more than 85% of the graft surface when the eye is looking straight up or slightly tilted, which is the ideal scenario for adhesion.⁶ By 72 hours, the bubble has shrunk enough that coverage starts to drop at all gaze angles.

How long you actually need to lie flat has been a point of real debate. Many surgeons have prescribed up to five days of strict supine positioning, which patients understandably find miserable. A randomized controlled trial directly compared five days of supine positioning against just one day. Using sensor data to confirm that patients actually complied, the study found no meaningful difference in graft detachment between the two groups: the average area of detachment was about 29% in the one-day group and about 29% in the five-day group.⁷ Another study went further, sending patients home after surgery in an upright position with no instructions to remain supine at all, as long as the air fill was near-total at the time of surgery.⁸

The emerging picture is that a near-complete air fill at the end of surgery matters more than days of strict face-up positioning. That said, most surgeons still advise at least some supine time during the first 24 hours and will tailor instructions based on how the bubble looks at the first postoperative check. You will also have blurry vision while the air bubble is still large enough to cover the visual axis, which typically clears within a few days as the bubble dissolves.

Visual Outcomes

Vision after DMEK tends to improve quickly by transplant standards. In a ten-year follow-up study comparing DMEK, DSAEK (a slightly thicker partial transplant), and full-thickness PK, the median time to reach driving-level vision was about eight months for DMEK, roughly 12 months for DSAEK, and nearly 38 months for PK.⁹ At both five and ten years, DMEK eyes had a higher chance of achieving good acuity than either alternative. A separate two-year comparison using detailed corneal clarity measurements confirmed that DMEK consistently outperformed DSAEK in best-corrected vision at every time point measured.¹⁰

Patients notice the difference in daily life, not just on an eye chart. Glare, hazy vision, blurred vision, and daily fluctuations in clarity were the most commonly reported symptoms before surgery. All of them dropped significantly after DMEK, though mild residual glare and some fluctuation in vision persisted in a portion of patients.¹¹ Quality-of-life scores measuring vision-related function and glare both improved substantially after surgery, whether DMEK was performed alone or combined with cataract removal.¹²

Graft Rejection Risk

One of DMEK’s most striking advantages is how rarely the body rejects the transplanted tissue. In a comparative study, only about 1% of DMEK eyes experienced a rejection episode at two years. That compared with 12% for DSEK (a related partial transplant technique) and 18% for full-thickness PK. DMEK carried roughly 15 times less risk of rejection than DSEK and 20 times less risk than PK.¹³ The likely reason is simple geometry: DMEK grafts are so thin that they expose far less foreign tissue to the immune system.

Complications Worth Knowing About

While rejection is rare, DMEK has its own set of complications. The most common is graft detachment, where part of the transplanted membrane separates from the back of the cornea in the days after surgery. If the detachment is large enough to affect vision, the surgeon performs a “rebubble” procedure: injecting another air bubble to press the tissue back into place. Rebubble rates vary widely depending on surgeon experience and patient characteristics. In one study of 52 eyes using imported pre-cut donor tissue, about a third needed rebubbling.¹⁴ Risk factors included older recipient age, diabetes, and low eye pressure in the first hours after surgery. Experienced surgical centers report rebubble rates that decrease substantially over time as technique improves, with one multi-surgeon center cutting their rate roughly in half from the early phase of their practice to a later period.²

Elevated eye pressure is another concern. The air bubble itself can cause a spike in pressure if it blocks fluid drainage, a condition called pupillary block. One long-term study found that about 16% of patients experienced some form of pressure elevation in the first year, with the most frequent cause being the steroid eye drops used to prevent rejection rather than the air bubble itself. Roughly 4% went on to develop post-DMEK glaucoma within the first year, rising to about 7% by three years.¹⁵ Patients with pre-existing glaucoma were at significantly higher risk. In most cases, pressure was managed by tapering the steroid drops and starting or adjusting glaucoma medications.¹⁶

Endothelial Cell Loss Over Time

Here is the part that gets less attention in the initial conversation about DMEK but matters a great deal long-term. The transplanted cells do not regenerate. From the moment the graft is placed, the endothelial cell count starts declining, and it drops fastest in the first few months. One mathematical model estimated that cell density halved roughly every three months during the early postoperative phase, then slowed considerably, with the late-phase half-life stretching to about 11 years.¹⁷ That model predicted the graft would remain functional (above a critical cell density threshold) for roughly 18 years on average, offering some reassurance about durability.

Real-world data paints a consistent picture. In one series tracking patients out to a decade, the average cell density had dropped to around 730 cells per square millimeter by year ten, well below the starting count but still enough to keep the cornea clear in most cases.¹⁸ A broader ten-year comparison found that endothelial cells are actually lost faster after DMEK and DSAEK than after full-thickness PK in the early years, though the rates tend to converge over the long haul. By ten years, only about 3% of DMEK eyes maintained a density above 1,000 cells per square millimeter, compared with 8% for DSAEK and 18% for PK.⁹

This faster initial cell loss is the main caveat to DMEK’s otherwise impressive track record. It is one reason some younger patients and surgeons think carefully about the trade-off between excellent early vision and the possibility of needing a repeat transplant decades later.

Combining DMEK With Cataract Surgery

Because Fuchs dystrophy and cataracts frequently coexist — both are age-related — surgeons often face the question of whether to perform cataract removal and DMEK together in a single session (called “triple DMEK”) or stage the operations separately. A systematic review and meta-analysis found that combined surgery offered slightly better vision at the one-month mark compared with DMEK alone, though the advantage disappeared by three to six months. Complication rates, cell loss, and graft failure rates were not significantly different between the two approaches.¹⁹

Clinical experience supports that conclusion. In a large series, triple DMEK was not associated with higher risks than DMEK alone, and the air reinjection rate was essentially identical between the two groups. Notably, the reinjection rate in both groups dropped dramatically — from about 45% to 16% — after the surgeon stopped using a particular viscoelastic substance during the tissue insertion step, illustrating how refinements in technique drive outcomes.²⁰ One safety concern sometimes raised about combining procedures is retinal detachment, but a large analysis of nearly 2,000 consecutive patients found the two-year retinal detachment rate was under 0.5% for both DMEK alone and triple DMEK, with no meaningful difference between them.²¹

DMEK in Difficult or “Complex” Eyes

Standard DMEK works best in a relatively normal anterior segment — the front compartment of the eye. But many patients who need an endothelial transplant have eyes that are anything but normal. Previous glaucoma surgery, prior vitrectomy (removal of the gel inside the eye), old trauma with iris damage, or a failed earlier transplant can all make the procedure considerably harder. DMEK is still feasible in these situations, but the complication rates are higher and the techniques require adaptation.²²

The core problem in many of these cases is that the air bubble cannot stay where it belongs. Glaucoma drainage devices, surgical holes in the iris, and an absent vitreous gel all create pathways for air to escape from the anterior chamber, which means the graft loses its tamponade and detaches more frequently. Surgeons have developed several workarounds. One approach for eyes with large iris defects involves a safety-net suture threaded behind the graft to prevent it from dislocating backward.²³ Another strategy, sometimes called “Maxi-DMEK,” fills both the front and back chambers of the eye with gas simultaneously, which requires a corneal and vitreoretinal surgeon working together but can achieve graft adhesion even in eyes where standard techniques fail.²⁴

The Surgeon Learning Curve

DMEK has a steeper learning curve than most eye surgeries, and the difference shows up in patient outcomes. A multi-surgeon study comparing early practice (roughly the first 50 cases) against a later period documented a significant drop in average surgery time, from about 59 minutes to 38 minutes for triple DMEK and from about 41 minutes to 27 minutes for DMEK alone. More meaningfully, the rebubble rate fell from about 51% in the early period to about 26% later.² That kind of improvement matters for patients choosing where to have the procedure done. Asking a surgeon about their DMEK volume and rebubble rate is a reasonable way to gauge experience.

Emerging Alternatives That Skip the Donor

An intriguing frontier in treating Fuchs dystrophy dispenses with donor tissue entirely. The procedure, known as Descemet stripping only (DSO) or descemetorhexis without endothelial keratoplasty, involves peeling away the central diseased Descemet membrane and then waiting for the patient’s own peripheral endothelial cells to migrate inward and repopulate the bare area. This approach eliminates the need for donor tissue and the long-term immunosuppressive eye drops that transplant patients use.²⁵ Some surgeons supplement DSO with a topical rho-kinase inhibitor eye drop, which appears to encourage endothelial cell migration and healing.²⁶

DSO is still considered experimental for most patients and works best in milder cases of Fuchs dystrophy where a rim of healthy endothelial cells still exists. The visual recovery can be slower and less predictable than DMEK, and not every eye clears completely. But for the right candidate, it sidesteps donor tissue shortages and the lifelong concern about cell loss in a transplanted graft. Research comparing the two procedures head-to-head is still in early stages, and it will likely take years before DSO’s long-term durability is as well-documented as DMEK’s.