It is important to know the possible complications during the practice of endothelial lamellar keratoplasty (ELK or DSAEK), to prevent them and know how to treat them when they occur. We can describe them according to the moment they occur.

COMPLICATIONS DURING THE PREPARATION OF THE GRAFT

Obtaining the graft for ELK is a very technical phase of the intervention and possibly its safety improves when it is partly entrusted to specialized technicians of the Eye Bank. Even so, complications can occur.

Superficial perforation (buttonhole)

Even with the most advanced technology, the microkeratome can sometimes perform a lamellar cut that becomes too superficial in the center and produces a buttonhole in the anterior cornea (video 6.7.1.1). This prevents the tissue from being used for ELK, but the Descemet membrane (DM) could be dissected with the endothelium for a Descemet-endothelial keratoplasty (DEK or DMEK).

Deep perforation (of the bed)

If we force the cut to obtain a very thin back sheet, we can get to perforate with the microkeratome the central area of ​​the corneal bed. To avoid this, the nomograms of each microkeratome must be followed¹, according to the pachymetry and the speed of movement of the blade (see chapter 6.4.1). The pressure in the artificial anterior chamber (AC) is important so that the cut is reproducible. For it to remain stable, it is recommended to keep the infusion open during cutting. Otherwise, the compression produced by the microkeratome in its path can cause a large increase in pressure that would cause the rupture of the corneal bed.

Eccentric trepanation

The lamellar cut usually has a central area of ​​about 9-10 mm in which the thickness of the back layer is the desired one. When the graft is trepanned from the endothelial surface, a good centering within that diameter is crucial, beyond which we will invade the uncut cornea and a very thick sector of the disc edge will remain in the shape of an arc (Figures 1 and 2).

To avoid this, it is useful to put marks with a surgical marker on the edges of the cut with the microkeratome, which will serve as a reference for centering the base of the trephine-punch, so that it acts inside the marked circular area.

DIFFICULTIES AND COMPLICATIONS DURING THE DESCEMETORHEXIS

The resection of DM-endothelium is a unique maneuver, which defines modern endothelial keratoplasty and raises a series of possible difficulties and specific complications.

Limited visibility

Corneal edema or superficial opacities can prevent the correct visualization of DM. In these cases, the epithelium is usually very altered, and it is preferable to remove it. If this is not enough because the stromal edema is very opaque, we can reduce it with the application of a few drops of glycerin. Lighting is also a factor to consider; we can improve it with a surgical slit lamp or place a fiber optic light (Chandelier) in the AC through one of the paracentesis².

Collapse of the anterior chamber

When the descemetorhexis is performed without viscoelastic or continuous infusion (under aqueous humor or air), it is easy for the AC to collapse when introducing the instruments, which makes maneuvers difficult. The ability to retain an air bubble during this phase will depend on a very beveled construction of the paracentesis and the ability of the surgeon not to deform them. We can hydrate them with BSS in an attempt to make them more watertight when passing a fine instrument like a reverse Sinskey’s hook. It will often be necessary to reform the AC several times, all of which is avoided with an AC maintainer (with BSS or air), which also allows to regulate the intraocular pressure (IOP). If viscoelastic is used, it must be of cohesive type to facilitate its later removal. It is also useful to decrease the vitreous pressure by intravenous infusion of 20% mannitol just before surgery, to limit the volume of peri-or retrobulbar anesthesia, and to apply a Honnan-type balloon for a minimum of 10 minutes.

Hypertension due to posterior air migration

When intraocular air is used, a collapse of AC with sudden hypertension may appear. This is due to a migration of air behind the iris, which is pushed forward and closes the angle. In this situation it is not necessary to try to deepen the AC by injecting more air, since – in case of achieving it – this worsens it. We must introduce a cannula slowly to reach the pupil – or an iridectomy if there is one – and aspirate the air that has passed behind (video 6.7.1.2). If the hypertension is extreme and this maneuver is impossible, we will resort to air drainage via pars plana with a 27 or 30 G needle.

Ocular hypotension

Aphakic intraoperative ocular hypotension may occur in aphakic patients with localized choroidal detachment ("hack sign"). We will proceed to reform the eyeball with BSS until firm pressure is obtained. In these cases, it is preferable to place – even from the start – a continuous infusion via pars plana, which allows the IOP to remain stable.

Incomplete descemetorhexis

Once the descemetorhexis has been performed and the DM removed, it is convenient to check its extension and integrity. When the DM is thin or is very attached to the stroma, it is often not possible to dissect it in one piece and the extraction will be by pieces. In these cases, there may be adherent DM islets (Figure 3). To discard them, it is advisable to fill the AC with air and pass an endothelial scraper, since leaving them may hinder graft adhesion.

Posterior migration of Descemet’s fragments

The withdrawal of DM can be especially difficult and patchy in cases of aphakic edematous keratopathy without Fuchs’ dystrophy – with fine and very "adherent" DM. In the absence of crystalline or IOL, the fall of fragments of DM to the posterior segment has been observed, although its clinical significance is not clear.

COMPLICATIONS DURING SIMULTANEOUS CATARACT SURGERY

Limited visibility

During the practice of cataract surgery in a patient with corneal edema, poor corneal transparency can hinder maneuvers, especially capsulorhexis. It is therefore recommended to dye the capsule with trypan blue, as well as the other maneuvers mentioned above regarding the descemetorhexis to improve visibility. The aid by means of optical fiber can be done from the AC, although its placement behind the lens has also been described, via pars plana³.

Capsular break

Cataract surgery through a cornea with poor transparency increases the risk of complications such as rupture of the posterior capsule, an additional reason to try to obtain the best possible visibility conditions and to proceed with caution.

COMPLICATIONS DURING THE INSERTION OF THE GRAFT

The insertion of the graft is a critical phase of the intervention, where complications can occur with great repercussion on the result.

Displacement or expulsion of the graft by the infusion

The pull-through insertion technique requires placing an infusion line to deepen the AC. To prevent that flow from displacing or expelling the graft, it is important to follow a precise sequence of activation and closure of the infusion. It should not be opened until the moment when the forceps are inserted through the paracentesis and should be closed as soon as the tip peeks through the main incision. It does not matter that the AC collapses at that moment. If we leave it open, the BSS flow through the incision can displace the graft when we approach it (video 6.7.1.3). We will only open it again when the graft is already held firmly by the forceps in the mouth of the incision, before inserting it by traction in the AC (video 6.7.1.4). We will close it again as soon as the graft is in and before releasing it, with the forceps about to exit through the paracentesis (video 6.7.1.5). Otherwise, the force of the infusion can displace, reverse or even expel the graft through the main incision. When an infusion injector is used, it must also be closed just before its tip has just been removed from the AC.

Passing of the graft behind the iris

By inserting the graft through injectors, it can happen that it passes behind the iris. Inserting the tip of the injector into the AC until beyond the pupil prevents this. To extract a disc from behind the iris to the AC, air can be used, avoiding as much as possible the viscoelastic.

Graft fall to the vitreous chamber

The manipulation of a posterior lamellar graft in the AC of an aphakic eye with a large pupil exposes us to the fact that the disc can fall to the vitreous chamber during the intervention. If the iris allows it, we must close the pupil – be it medically or with sutures – before introducing the graft. In patients with aniridia, the fall can occur even in the presence of an IOL sutured to the sclera, by the peripheral space. The use of a continuous infusion of air through the pars plana avoids this complication. If the mydriasis is paralytic, it is not possible to correct it and the eye will continue aphakic, it is advisable to fix the graft with a 10/0 nylon suture that prevents its fall in case of detachment in the postoperative period (Figure 4).

Iris drag when inserting the graft

Inserting the disc into a collapsed AC may result in an entrainment of the iris with the instruments and even cause hemorrhage or iridodialysis. This can occur in an insert with forceps, and especially when introducing an injector that requires to advance through the AC. If it has coaxial infusion (e.g. Endosaver, Sightlife, Winston-Salem, NC, USA) it is important to control it with an adjustable pressure system and pedal opening. This allows to keep the AC formed during the insertion of the tip with the right pressure (20-30 mmHg) and to increase it only when we want to deploy the graft. This rarely comes out spontaneously – without using the plunger mechanism – before reaching 70-90 mmHg of infusion, especially if you take care that, when mounted on the tab of the injector, it is dry and the disc does not slide easily. You can also use infusion by another route (AC maintainer), so that the formation of the AC does not depend on the injector.

Inversion of the graft

When the graft is inserted into the AC, it may happen that it flips, with the endothelial side facing forward. If this goes unnoticed, it can ruin an otherwise impeccable surgery. It is recommended to record all interventions to be able to review them in case of doubt. If forceps are used, to avoid turning, the graft should not be released until it begins to open, either spontaneously or with the help of small blows on the cornea. If an injector is used, it is important that the deployment be slow to see that the orientation is correct. For greater safety, a mark can be made unambiguously ("S", "F", etc.) with gentian violet on the stromal side of the disk edge. For this purpose, specific instruments are available (Figure 5).

Eccentric placement of the graft

After deploying the graft, it must be centered before fixing it. Several techniques can be used for this purpose (see chapter 6.4.1). A discrete decentering may not have consequences, but occasionally an area of ​​the peripheral cornea with persistent edema or the eccentric edge may create an iris synechia.

GENERAL INTRAOPERATIVE COMPLICATIONS

Some complications can occur during any phase of the operation:

Hemorrhage in the anterior chamber

AC maneuvers can produce hemorrhages, especially when performing an iridotomy, when releasing anterior synechiae, or with any manipulation that tracts the iris, as described when inserting the graft. If the bleeding does not stop spontaneously – which usually happens – it may be helpful to fill the AC with air and wait a few minutes. If a bleeding point is identified, fine-tip coaxial diathermy 23G can be applied. If clots of a certain importance remain, they can be removed by irrigation-aspiration once the hemostasis has been obtained, taking care not to reactivate the bleeding. In patients with known hemorrhagic risk – de-coagulated, hypertensive, etc. – it can help to place them in anti-Trendelenburg and avoid maneuvers that can traumatize the iris, as well as hypotonia.

Vitreous in anterior chamber

In the presence of defects in the irido-capsular diaphragm – posterior capsulotomy, large iridectomies, zonular defects – we can find vitreous in AC, either pre-existing or provoked. This will hinder maneuvers, especially graft placement, adhesion and viability. Given the suspicion of vitreous in AC, it is necessary to eliminate it by means of anterior vitrectomy – assisted with triamcinolone – before inserting the disc.

Dislocation of the intraocular lens

In cases of ELK in pseudophakic patients with large capsulotomy or pseudophakodonesis, high air pressures in AC should be avoided, during surgery or at the end of surgery, since we could cause a dislocation of the IOL to the vitreous chamber through an enlarged capsulotomy or by zonular disinsertion⁴.

Patients undergoing endothelial lamellar keratoplasty (ELK or DSAEK) need a protocolized postoperative follow-up. Despite the lower frequency of rejections compared to penetrating keratoplasty (PK), these are possible, and it is necessary to warn the patients to go to the emergency room when there is an episode of blurred vision or any other sign of inflammation¹.

FACTORS THAT INFLUENCE POSTOPERATIVE EVOLUTION

After an ELK, the survival, the probability of complications and the speed of visual recovery will be different according to:

The etiology of endothelial failure. In Fuchs’ dystrophy, the postoperative evolution and visual acuity (VA) recovery is better than in postsurgical corneal decompensation without previous dystrophy². Cases after failed PK have a worse prognosis.

The status of the stroma. A very opaque stroma contraindicates an ELK, although sometimes it is difficult to judge how much transparency it can recover. In any case, the VA gain is greater and faster when the stroma is still transparent (Figure 1).

Glaucoma and other associated pathologies. Glaucoma and in particular the presence of a drainage tube in the anterior chamber (AC), as well as any inflammatory pathology, are associated with a higher probability of rejection and lower graft survival³.

Complexity of surgery. Surgical trauma and endothelial loss will have been greater in more complex interventions: aphakic, vitrectomized eyes, when there is communication between the AC and the vitreous or when the surgery is associated with a secondary IOL implant, either fixed to the iris or the sclera (Figure 2).

Quality of the donor endothelium. Good cellularity of the donor endothelium will provide a faster and longer lasting recovery.

POSTOPERATIVE PERIODS

In the postoperative follow-up of an ELK we can differentiate several periods:

The first week

In the first days of the postoperative period, the most frequent complications are graft dislocation, pupil air block and primary graft failure. If the air takes longer than usual to reabsorb and iridotomy has not been done, mydriasis should be maintained. As in any intraocular surgery, monitor the signs of a possible infection. The progressive disappearance of corneal edema gives us an idea of ​​the functionality of the implanted endothelium (Figure 3).

The first 3 months

In the first days the edema must have disappeared if the endothelial functionality is good. The first endothelial count is performed, monitoring of the IOP, exploration of the posterior pole. If it has not been possible to do it before, macular edema should be ruled out in patients with postoperative edematous (bullous) keratopathy. If the VA has already recovered enough, we can prescribe the refraction it needs. Recovery may, however, be slower over months, depending on the condition of the stroma and the condition of the macula.

We must maintain vigilance over a possible infection, typically fungal, which may start as punctate opacities at the interface even 2 months after the intervention and with hardly any previous signs of inflammation (Figure 4).

From the 4^th month onwards

Endothelium transplantation requires periodic controls of endothelial cell density, even more than in PK. Two evolution patterns have been observed. The most common is that there is an initial loss, consistent with the surgical trauma, followed by a little more physiological decrease (Figure 5). In some cases, however, sustained endothelial loss is maintained at rates much higher than the physiological one (Figure 6). This type of accelerated descent persists even in the 8 years that we have been able to follow-up until now. In the absence of objective signs of rejection, it may be a form of chronic rejection or other mechanism that causes an increase in cell loss.

Acute rejection is rare after ELK. We have only seen it in 1 case of more than 300. The florid form of rejection that can be seen in a PK would be here almost impossible due to several factors – the smaller amount of tissue transplanted, its isolation in the AC without contacting the limbus or the ocular surface, the absence of donor dendritic cells, etc. – that increase the immunological privilege of these transplants. However, it is possible that graft antigens end up coming into contact with the recipient's immune system slower and this results in a more temperate response^4-7. In the episodes that have been observed, "keratic" endothelial precipitates appear, similar to those of PK rejection, accompanied by a thickening of the graft and the recipient cornea. In any case it is necessary to treat these episodes as soon as possible and establish a closer and longer surveillance of these patients.

POSTOPERATIVE TREATMENT

During the first week, topical antibiotic treatment with broad-spectrum antibiotic eye drops (e.g. 0.5% moxifloxacin or 0.3% tobramycin every 4 hours) is instituted. The anti-inflammatory treatment includes topical corticosteroids at full dose (e.g. prednisolone 1% every 2 h) during the first 2 weeks, followed by a descending pattern until the 5^th month at 1 drop/day, which can be maintained for life or withdraw later if it is appreciated that the endothelium is stable for more than a year. On the other hand, if the cell loss is accelerated, it is recommended to maintain higher doses (2-3 drops/day), always monitoring the intraocular pressure (IOP). Antibiotic combinations with corticosteroid should be avoided since, to avoid the creation of resistance, the former should never be progressively reduced or maintained prophylactically for more than 10 days.

If the epithelium has been removed, it is preferable to place a therapeutic contact lens and use eye drops without preservatives for the time it takes to grow again until completely stabilized. It is not uncommon for these patients to have difficulty re-epithelializing; in this case, one should not hesitate to apply all the necessary measures, such as autologous plasma derivatives, etc. In all cases it is advisable to keep the ocular surface well lubricated, always with products without preservatives.

During the first month, a systemic corticosteroid (prednisone 1 mg/kg/day in 2 doses) is also administered with a progressive withdrawal schedule starting from the first week, and under gastric protection (omeprazole 20 mg/day).

PROTOCOL OF REVISIONS/MEASURES TO BE TAKEN

The calendar of explorations and possible measures to be taken in the usual postoperative period after an ELK can be as follows (table 1):

First day

Perform biomicroscopy (BMC) and IOP control by digital palpation. Check the adherence of the graft, if necessary by means of optical coherence tomography (OCT) (Figure 7). The main situations that will require action include:

Pupillary block. The way to solve it is by the immediate drainage of part of the air. This can be done in the office, after instilling anesthetic eye drops and povidone iodine 0.5%. With a 30 G needle attached to a 2 ml syringe with BSS, we open one of the paracenteses minimally. We remove the needle as soon as some air comes out – it does not interest that it completely empties. The ideal is to empty something more than half the AC. If this is done as a rule on the same day of the intervention, pupillary blocks are completely avoided. If the iris has not detached from the cornea, you can reject it backwards with the needle and reform the AC with BSS – for which an assistant will be useful –. Blocking is especially frequent in ELK in phakic eyes, where it is preferable to remove all the air on the day of surgery and maintain mydriasis (Figure 8).

Detachment or dislocation of the graft. It will be necessary to reform the air bubble, which can also be done at the slit lamp. In this case the 30 G needle, bent to generate a convexity, is attached to a syringe with air. Instead of entering through a lateral paracentesis – which would cause loss of AC – the lower corneal limbus is punctured (at 5 o'clock if we are right-handed), we position the lenticule and introduce the air, also with help and making sure the tone is normal. The patient is then placed in a supine position on a stretcher for at least 1 hour, after which we will check the position of the graft and the IOP.

First week

VA with pinhole, BMC of anterior and posterior pole and measurement of IOP. Corrections should not be applied because of the increased thickness. If VA is lower than what would correspond to the condition of the cornea, assess macula with OCT. Perform corneal topography-tomography (Scheimpflug or OCT). If corneal transparency is not enough, these scans are postponed to the next revision.

First month

VA and refraction, BMC, IOP and evaluation of the endothelium by specular or confocal microscopy. The suture is removed if applicable.

Third month

VA and refraction, BMC, IOP and endothelial count. Prescription glasses are prescribed if applicable.

First year and subsequent annual reviews

VA and refraction, BMC, IOP and endothelial count.

This calendar will be modified whenever the evolution is not favorable or if there are other associated pathologies.

Endothelial lamellar keratoplasty (ELK or DSAEK) is not a surgery without complications, despite the good results it offers. The surgeon should know them, prevent them as much as possible and be prepared to solve them when they occur. Well managed, many of them have little impact on the visual result although they may require a small intervention (bubble reformation). However, the potentially devastating ones (infections, glaucoma) are not lacking. We can consider two main groups according to whether they relate to the graft or not.

POSTOPERATIVE COMPLICATIONS RELATED TO THE GRAFT

Decentration of the graft

A lack of transient adhesion during the early postoperative period can cause the disc to slip and to find it stuck but off center within a few hours in the first control. This may not require any action if the displacement is discrete, but if it is important it will be better to re-center it immediately (Figure 1).

Detachment or dislocation of the graft

Probably the most frequent complication of ELK is detachment or dislocation due to lack or loss of adherence between the donor disc and the recipient (Figure 2). The published series present rates as variable as 82% to 0% (it would be an average of 14%)¹, which seems to indicate that it is due to variable aspects of the surgery and the experience of the surgeon. The most commonly involved factor is hypotonia: it is more frequent in all situations in which adequate pressurization is not achieved or it is not maintained long enough in the early postoperative period, as in aphakia, presence of drainage, vitrectomized, etc.

Any more or less elastic and curved membrane tissue – as is the case of the cornea – responds to the pressure applied with an increase in tissue tension and therefore its surfaces are smoothed. When tension decreases, it is mainly the concave surface that tends to wrinkle – as with the Schepens’ sign of retinal detachment. The folds – or microfolds – by hypotonia in the posterior layer of the recipient cornea probably act as channels, that the convex and smoother surface of the graft fails to seal, and they allow fluid to enter the interface, which prevents adhesion.

The time in which the patient is usually left with the anterior chamber (AC) filled with air and in the supine position varies considerably among surgeons. The fact that some authorities in the field use relatively short times (see chapter 6.4.2), suggests that a good adhesion does not require a very long time but the right conditions. This could explain the success of maneuvers such as continuous air pressurization (pneumatization) by pars plana in aphakic patients during the operation.

A common mistake is to assume that if it is completely filled with air, the pressurization will automatically be adequate. Given the small volume of the AC with respect to the total of the eyeball, in a hypotonic eye it is easy that despite filling the AC completely with air – even rejecting the iris backwards – a normal IOP may not be achieved. This will require reforming the eyeball with liquid (BSS), either from the AC or by pars plana. The critical importance of IOP in the adhesion phase of the graft makes its intraoperative measurement of utmost importance. For this purpose, gravity applanation tonometers (Maklakov-Barraquer) with adequate reference marks (20-30 mmHg) exist.

Graft dislocation has been related to the use of viscoelastic and it is understandable that if this is retained at the interface it will prevent adhesion. Therefore, there is a consensus to prohibit dispersive ones due to their difficulty of removal. The question is more controversial in the case of cohesive viscoelastics, since many surgeons use them without apparent impact on the adhesion of the graft, provided they are removed exhaustively before the insertion of the disc in the AC.

The attitude towards a detachment of the graft will depend on whether it is partial or total. To specify this, it is often necessary to perform an OCT (see Figure 7 of chapter 6.5.3 and Figure 2 of chapter 6.6.4). When it is partial, especially if the detached area is small, you can wait until spontaneous adhesion occurs. The endothelium does not suffer from the fact that it is detached as photoreceptors of a detached retina do, but it can be traumatized by friction with the iris or an AC IOL. If the detachment of the disc is total or almost total, we will proceed to the reformation of the bubble with air or sulphur hexafluoride (SF₆). When the graft dislocation is repeated several times despite the bubble reformations, it may be necessary to suture the graft or proceed with a PK.

Primary failure of the graft

It is considered a primary failure of the graft when this does not become functional and there is no improvement in the edema after the first days or weeks of the postoperative period (Figure 3). This can happen with the disc well attached, but an unsuccessful graft can be manifested by an inability to adhere or a detachment that appears after a few days.

Given the current quality controls in eye banks, the viability of the tissue supplied can be considered generally guaranteed. Therefore, the primary cause of primary failure is excessive trauma to the graft during the intervention – including the lamellar cutting phase – and its frequency is inversely related to the surgeon's experience.

Endothelial rejection

The specific immune reaction of the host against donor endothelial antigens can in principle occur in all cases in which a homograft of this cell layer is performed. Although the data published on rejection in ELK present remarkable variability – between 0% and 45% depending on the series –, the resulting average, around 10%¹, is clearly lower than that of the PK. However, there is no shortage of those who have not found statistically significant differences between both techniques, in cases of secondary endothelial failure².

Endothelial rejection in a patient with ELK is not always accompanied by the typical clinical process after PK – corneal edema, intraocular inflammation, endothelial deposits, Khodadoust’s line – and often one or several of them are missing (Figures 4 and 5). It should be suspected in a patient with good previous results and cell counting, who presents increased pachymetry or edema, even without visible deposits. These cases usually respond well to intensive treatment with topical corticosteroids and only rarely require systemic treatment.

Alterations of the Interface

Sometimes we find alterations or opacities in the graft-receptor interface, which can explain a limited visual recovery. When a certain diffuse veil is seen, we suspect anomalous healing by an inconsistent stromal bed, perhaps related to the way of obtaining the disc (e.g. manual or automated microkeratome), which can explain the variability in the visual results of the published series.

It is in this plane that foreign materials can be trapped as hematic remains – usually peripheral and that tend to be reabsorbed with time –, remnants of the receptor’s Descemet’s membrane (DM) (see Figure 2, Chapter 6.7.1)⁶ or iris pigment. Diffuse reticular opacities have been described at the interface (see chapter 6.9.4) that could be viscoelastic remnants. Epithelial invasion by superficial cell implantation during surgery has rarely been observed^3-5. These can be localized (Figure 6), but if they extend they lead to graft failure. Remember that this is also where infections usually appear due to graft contamination (see below).

Medium-long term endothelial cell loss

Endothelial cell loss is a problem of great importance for any type of endothelial transplantation, since it determines its long-term survival. The reduction occurs initially due to surgical trauma and then progressively, apart from possible sudden decreases, due to episodes of rejection or other trauma. The problem, unresolved, is that of progressive loss in the absence of obvious rejection and its cause. In the important series of Price’s group⁷, the initial loss (the first year) was greater for ELK than for PK, but then the slope of loss softens and by the third year the curve intersects with that of PK, which continues its descent at a faster pace (Figure 7). According to this, the viability of ELK graft is in the long run greater than in PK.

The remarkable variability between different individuals, however, does make us suspect that multiple factors participate. When the rate of progressive endothelial loss is high, it could be a subclinical form of chronic rejection and a long-term topical corticoid treatment should be considered (see chapter 6.7.2).

Other less frequent complications

Among the less frequent incidents that can lead to a poor functional result, we find the presence of folds in the graft (Figure 8), very thick discs that do not allow a sufficient transparency of the stroma (Figure 9), and persistent epithelial defects. In a patient with previous LASIK surgery, its lamellar interface may be opened with fluid accumulation in it (Figure 10)⁸.

GENERAL POSTOPERATIVE COMPLICATIONS

After an ELK, any of the known complications in intraocular surgery can occur. The most important ones include:

Infections

The appearance of corneal opacities in the postoperative period should make us consider a possible bacterial, viral or fungal infection. The latter have turned out to be the most specific of ELK, probably due to tissue contamination during conservation or preparation. It usually occurs relatively late and with few inflammatory signs, such as whitish opacities at the interface (see chapter 6.9.6). Although they are rare, their results can be devastating, especially if they extend to the inside of the eye (Figure 11)^9,10. As in deep anterior lamellar keratoplasty (DALK; see chapter 5.6.4), the efficacy of antifungals is very limited in this location. But unlike an anterior keratoplasty, in ELK we do not have integrity of the eyeball and a simple replacement of the endothelial graft is exposed to spread of the infection. In our opinion, the safest treatment, provided that the infection is limited to the interface, is a PK with trepanation in block of the cornea and the disc, under intense antifungal/antibiotic coverage. If there are signs of intraocular invasion it should be treated as an endophthalmitis.

Ocular hypertension and glaucoma

The elevation of IOP after ELK raises questions similar to those that occur in other corneal transplants. It should be differentiated between those due to a response to corticosteroids, those related to glaucoma or a pre-existing tendency to glaucoma (pseudo-exfoliation, etc.), and those that can be attributed to the technique itself – such as a pupillary block or the formation of peripheral anterior synechia – (see chapter 6.5.2).

To prevent pupillary block, some surgeons perform an inferior iridotomy, although this would not be necessary if the air is removed largely at the end of the procedure. However, prolonged intraoperative hypertension greater than 35-40 mmHg can cause iris ischemia and lead to an Urrets-Zavalía’s syndrome with irreversible paralytic mydriasis (see chapter 6.9.5), an additional reason for systematically performing intraoperative applanation tonometry.

Complications related to the IOL

The state and characteristics of a pre-existing IOL can give rise to complications related to intraoperative air pressure, such as its dislocation to the vitreous chamber¹¹, and opacification when they are hydrophilic (see chapters 6.5.1 and 6.7.1). Some of these problems may not be apparent until the postoperative period.

Vitreoretinal complications

The appearance of vitreoretinal complications after an ELK is very variable according to the series, but in general not very frequent. Cystic macular edema was estimated at 4.5%¹. Retinal or choroidal detachment seems to be more related to pre-existing conditions such as aphakia, high myopia or glaucoma than to the technique itself, and in general its rate does not exceed 1%. However, they could be related to repeated collapses of the eyeball during surgical maneuvers. Performing vitreoretinal surgery after an ELK does not appear to be a threat to graft fixation, but endothelial losses of 11% have been observed in the short term¹³.

Despite its advantages and lower risks with regard to penetrating keratoplasty (PK), a part of the endothelial lamellar keratoplasties (ELK or DSAEK) end up failing. Due to its thin graft nature, only adhered to the receiving cornea without a lateral scar, ELK disc replacement is an attractive option due to its simplicity. However, there is no abundance of published data on this topic, so it is difficult to establish criteria for action, when and how-to re-transplant, and what is the prognosis that the patient can expect.

INDICATION

The re-transplantation rate in ELK varies between 3.5% and 15%. The percentage tends to exceed 10% in novice surgeons^1-6. The indication of graft replacement depends on multiple factors and therefore must be individualized and discussed with each patient. Although visual acuity (VA) is important, its value does not on its own determine the need for reoperation. Sometimes patient satisfaction depends on other aspects such as the elimination of photophobia and discomfort related to epithelial edema.

REPEAT THE ELK OR PROCEED WITH A PK?

The advantages of ELK over PK make it a principle that it is preferable to repeat it before going to a PK. In the case of a re-graft, the inherent advantages of ELK (structural, refractive, faster visual recovery) are added to the lower risk of rejection, which increases more in a second PK. However, in the series of 119 unsuccessful ELK studied by Li and Wilhelmus, in 57% an ELK was repeated and in the remaining 43% a PK, with similar results².

There may be situations that make a PK preferable, from the preference of the surgeon for being more familiar with it, including the availability of the instruments, or the fact that the failure has been associated with specific problems of the ELK, such as repeated disk dislocations (see Figure 3, chapter 6.6.4). In the face of an infection of the interface, PK may be almost the only option – if we still have time to prevent its intraocular spread –. Other times they can be factors of the patient, such as an inability to maintain supine decubitus, desire a different technique after repeated problems with an ELK that ends up failing, or have had a good experience with a previous PK in the other eye. Again, the decision must be individualized.

AT WHAT MOMENT SHOULD WE RE-TRASPLANT?

Once the replacement of the ELK has been decided, the ideal time to carry it out is also subject to some controversy. The published periods vary between 2 weeks and 18 months. Faced with an early failure, delaying the surgery allows to check whether the endothelial dysfunction is transient or not, or even to see if it is possible to replace the damaged donor endothelium by migration of the recipient's cells. When the failure occurs beyond the first few weeks, these arguments lose their meaning.

Supporters of early re-transplantation offer two main reasons⁷: a) to minimize the immunization of the recipient with respect to the transplanted tissue, and b) to prevent the progression of edematous (bullous) keratopathy, which can lead to stroma opacification due to fibrosis. The progressive damage to keratocytes would affect the remodeling of the extracellular matrix and the maintenance of corneal transparency, as well as the trophism of the endothelial cells.

SURGICAL TECHNIQUE

The procedure to replace an ELK is basically the same as in primary surgery, except for the removal of the failed graft (video 6.7.4.1). Under peribulbar anesthesia plus sedation, a small incision is made through which we introduce an inverted Sinskey’s hook – or rotator –, with which we will carefully remove the margins of the graft. In general, the greatest adhesion of the disk is in its perimeter and, once it has detached, the adhesions of the rest of the interface are generally weaker, where a fan-sweeping maneuver will suffice (Figure 1). Once released, the lenticule is removed through the same incision (Figure 2). The remaining bed is prepared for the insertion of a new lamellar graft, and then we proceed in the same way as in the primary ELK, with the carving, insertion, centering and fixation with air of the new disc (Figure 3). The postoperative measures are also similar to those after the primary ELK.

PROGNOSIS

The rate of intraoperative complications, risk of rejection and postoperative IOP elevation are comparable in primary ELK and ELK replacement, so the management of both situations is similar.

It has been suggested that the results after an ELK re-graft depend on the condition of the failed cornea. Thus, Kim et al³ obtained VA> 0.5 in 59% of 20 cases of ELK re-transplantation, a rate that reached 75% of 37 patients in Price’s group¹. This difference has been explained by the different indications of re-transplantation in the two studies: in the first, it was the objective decompensation of the graft, and in the second the unsatisfactory VA due to folds, irregularities and/or opacities in the graft. Therefore, visual results after graft replacement would be better if performed when the patient begins to have discomfort – possibly earlier – than when waiting for obvious decompensation. In the second series, 97% of the patients obtained a significant improvement of the VA. It is, however, a relatively short series, so further studies will be necessary to confirm this hypothesis.

Our failure rate in ELK, at the Hospital de Cruces in Bilbao between 2011 and 2013, was 17.7% (11 cases). The main cause would be iatrogenesis linked to the learning curve. The causes of failure are detailed in table 1. We have not found a statistical association between the use of cultivated or cold-preserved corneas and graft failure (p = 0.282). The endothelial density prior to the surgery of the failed corneas ranged between 2,364 and 3,021 cells/mm². The graft was replaced in 7 cases, 6 of them successful and one failed for the second time. In another case we opted for a PK with good results. The surgical technique was chosen in each case after discussing the surgeon's options with the patients.