Endothelial failure in the pediatric population can be due to multiple causes that are generally different from those of adulthood. Edematous or vesicular (bullous) keratopathy is rarer after surgery for childhood cataract but may occur after traumatic rupture of the Descemet membrane (DM) by forceps during labor, by exhaustion of the endothelium in congenital glaucoma or by other trauma. Among the dystrophies at this age we can find the hereditary congenital endothelial dystrophy (HCED) and the posterior polymorphic dystrophy (PPD). And there are also cases of endothelial failure of a penetrating keratoplasty (PK).

The general advantages of endothelial lamellar keratoplasty (ELK or DSAEK) over PK add, in pediatric endothelial failure, to avoid certain specific difficulties of this age for the practice of PK, such as the lower scleral stiffness that facilitates the collapse of the eyeball ("vitreous pressure"). On the other hand, many of these eyes are phakic and can have a narrow anterior chamber (AC), which is a challenge for the ELK. The good results of ELK in adults have encouraged it to be practiced in children with increasing frequency^1-6.

SPECIFIC ASPECTS OF THE SURGICAL TECHNIQUE IN THE PAEDIATRIC AGE

The pediatric ELK presents a series of variations regarding surgery in adults:

• Anesthesia. It should be general – except in some very cooperative patients in late pediatric ages – with attention to possible comorbidities.

• Incisions. In phakic patients, we will move them 1 mm towards the upper quadrant – from their usual position at 3 and 9 o’clock – to avoid possible accidental contact with the lens.

• Descemetorhexis. It will not be possible in patients under 12 months. Especially in cases with HCED type 2, the DM is very thickened, and the almost total absence of endothelium will not require descemetorhexis.

• Iridotomy. It is advisable to always perform one at 6 o'clock, since the risk of pupillary blockage is greater.

• Corneal drainage incisions (venting). Although nowadays their use is discussed in adults, they could be useful in children with aphakia and other factors that make it difficult to properly pressurize the eyeball at the end of the intervention.

• Postoperative position. In general, it will not be possible to maintain the supine position and immobility in the early postoperative period, so adhesion of the graft will depend on the pressurization time at the end of the surgery.

• Sutures. Due to the rapid healing at this age, they can be removed between 7 and 30 days.

COMPLICATIONS OF ENDOTHELIAL LAMELLAR KERATOPLASTY IN THE PAEDIATRIC AGE

The possible complications of ELK also present peculiarities in the pediatric group:

• Graft dislocation. Although it is also the most common complication of ELK in the adult, its frequency is higher in children, firstly because of the difficulty or inability to maintain a supine position beyond the anesthesia time.

• Cataracts. Because it is often a matter of phakic subjects, the possible induction of cataract is a complication of greater relevance.

• Primary failure of the graft. It is defined as a graft that does not become functional in the first 2 months after surgery. Apart from the possibility of a non-viable tissue, the greater technical difficulty of ELK in children would explain a greater intraoperative trauma.

• Ocular hypertension and glaucoma. In addition to the cases associated with childhood glaucoma, the risk of pupil blockage by air and of causing an Urrets-Zavalía syndrome would be greater at these ages, due to the lower depth of the AC, the softer and more reactive tissues, with a tendency to the formation of fibrin and adhesions. Therefore, it is important to prevent it with peripheral iridotomy and mydriatics at the end of surgery and be vigilant to detect it early and reduce the size of the air bubble.

• Endothelial rejection. Although generally less frequent than in PK, the most active immune system in children would also increase this risk in ELK.

ADVANTAGES OF ELK VERSUS PK IN THE PAEDIATRIC AGE

In addition to the general advantages of ELK over PK in cases of endothelial failure without stromal opacities, some have special repercussions in pediatric age:

• Intraoperative complications. Lower risk of all those linked to open surgery such as eyeball collapse and even suprachoroidal hemorrhage.

• Astigmatism. For being a practically anastigmatic surgery, the risk of amblyopia is lower, while frequent after PK due to high and irregular astigmatism.

• Resistance to traumatisms. Since the main incision does not usually exceed 4 mm, the eyeball is hardly weakened against the injuries that are not uncommon in these ages.

• Sutures. The possibility of removing them after one week reduces the possibility of many of the complications of PK in this population (infections, epithelial defects, corneal lysis, etc.).

• Visual rehabilitation. The rapid visual recovery and the low variation in refraction contribute to an early visual rehabilitation and reduce the risk of amblyopia.

RESULTS OF ENDOTHELIAL LAMELLAR KERATOPLASTY IN THE PAEDIATRIC AGE

In recent years, several isolated cases and some series of ELK have been reported in pediatric patients. In the most extensive, Madi et al³ studied the results in 19 eyes of 11 pediatric patients (13 HCED, 2 congenital glaucoma, 2 PPD and 2 failed PK). There were 4 graft dislocations that were treated successfully with new air injections. The best corrected visual acuity (BCVA) was >20/40 in 8 of the 13 cases that could be measured. A single episode of rejection was presented 3 months after surgery, which was successfully treated with topical corticosteroids.

Ashar et al⁶ performed ELK in 5 children affected by HCED, who had previously had a PK in the other eye. They present 2 graft dislocations that are treated successfully with a new injection of air. The results regarding spherical defect were not statistically different from those of PK, although the astigmatism was found to be significantly lower and stable. They conclude that the ELK presents as advantages the absence of problems derived from the sutures and a stable refraction faster than the PK. This allows an early treatment of amblyopia and less change of glasses, which entails a better therapeutic compliance. They emphasize that the corneas intervened by ELK do not achieve the transparency of the penetrating ones, with a certain veil (haze) persisting in the corneal stroma. Despite this, BCVA was similar in both groups. Although this opacity probably reduces contrast sensitivity, its ability to induce amblyopia would be more than compensated by early visual rehabilitation and other advantages mentioned above regarding PK.

Aphakia is a difficulty for any technique of endothelial keratoplasty. Once inserted in the anterior chamber (AC), it is necessary to fix the graft with air and therefore the eyes in which the irido-lenticular diaphragm is not preserved have a greater risk of dislocation. Despite this, the advantages of endothelial keratoplasty compared with penetrating keratoplasty (PK) make it preferable also in the most complex situations of endothelial failure. The strategies developed to overcome the difficulties posed by aphakia or the need to exchange the intraocular lens (IOL) seem more feasible with endothelial lamellar keratoplasty (ELK or DSAEK) than with Descemet-endothelial keratoplasty (DEK or DMEK), which defines a preferred field of application of the first.

One of the most frequent causes of endothelial decompensation is complicated cataract surgery. After this, the patient may present with the IOL in the posterior chamber – in the bag or pre-capsular (sulcus) –, an AC IOL, or even without IOL. While a posterior chamber IOL does not usually pose a need for replacement, this issue arises with AC IOLs. In the aphakia it will be necessary to decide if a secondary implant is practiced, either in the same act or in a new procedure.

ENDOTHELIAL FAILURE AND ANTERIOR CHAMBER LENSES

Facing an endothelial decompensation with an AC IOL that requires an ELK (DSAEK), we can choose to explant or not the IOL. The decision must always be individualized, considering the clinical history, the state of the other anterior segment structures, the time elapsed between the IOL implantation and the endothelial decompensation, the visual potential and contralateral eye status and the general state of the patient, their age and visual needs. There are possible arguments in favor and against the explant:

In favor of explant

The disadvantages of an AC IOL include: a) they decrease the volume of the AC^1,2, making difficult the unfolding of the lenticule and its fixation to the cornea; b) they increase the risk of endothelial cell loss^1,3 and endothelial rejection¹. The presence of an AC IOL requires greater manipulation of the graft by the presence of a solid material, which makes it difficult to manage the air bubble and facilitates intermittent contact between the endothelium and the IOL in the postoperative period. This increases the risk of late graft failure⁴. Most studies (table 1) show an increased risk of graft failure in the presence of an AC IOL^1,3,5-8.

Against the explant

Performing this increases the surgical time and a series of risks: associated infection, iatrogenic injury of the iris and ciliary body, hemorrhage or inflammation (iritis), as well as retinal complications – cystic macular edema, retinal detachment, among others. Esquenazi et al⁹ are in favor of respecting the AC IOL. We consider this option only in cases in which the IOL is well centered in an AC of depth greater than 3 mm and the other structures are intact.

ENDOTHELIAL LAMELLAR KERATOPLASTY IN APHAKIA

When proposing an ELK (DSAEK) in an aphakic eye, there is the possibility of implanting a new IOL or not. The situation is analogous when we have decided to explant and replace the present IOL, since we are momentarily creating an aphakic situation. In this case it is recommended to remove the IOL and implant a new one by the smallest incision possible. On the other hand, it should be considered that, in general, corneal transparency will be limited.

Endothelial lamellar keratoplasty in aphakia with IOL reimplantation

The absence of a crystalline lens hinders adhesion of the graft due to the difficulty in maintaining the air bubble and the adequate pressure and involves a risk of dislocation of the disc to the vitreous chamber. Therefore, re-implanting a new IOL supposes an advantage, since it restores at least in part the irido-lenticular diaphragm. Although different alternatives have been proposed to minimize these difficulties^9-16, we do not have comparative studies on which the best option would be, particularly with regard to implanting a new IOL.

In a review, Ghaznawi and Chen¹⁰ conclude that the protocolization of the ELK procedure allows greater safety in complex cases such as aphakia, iridocorneal endothelial syndrome and aniridia. Some groups propose to implant an IOL in the posterior chamber by means of fixation to the sclera with tissue adhesive, which would allow to work safely on the cornea at the same time and with favorable results^11,12. Others defend the use of an IOL fixed to the iris, implanted simultaneously with the ELK². In vitrectomized aphakic patients, inserting an infusion via pars plana would minimize the risk of eyeball collapse, migration of the air bubble and dislocation of the graft to the vitreous chamber¹³. Chiang et al¹⁴ suggest that in these cases the nasal and temporal sclera should be indented with two rods to favor the exit of the fluid from the interface and therefore the adherence of the graft. The use of a special device for the insertion of the graft (EndoGlide) would help to minimize the intraoperative manipulation of the same and the endothelial damage, especially in complex cases of ELK⁴.

Endothelial lamellar keratoplasty in aphakia without new IOL implantation

In the face of an endothelial decompensation and aphakia, we can decide to only treat the cornea, without implanting any IOL in some situations: a) when the purpose is palliative of the discomfort caused by an advanced bullous keratopathy; b) in a high myopic eye where the implant is of scarce refractive relevance; and c) in patients with very low visual potential.

The decision of the keratoplasty technique must also be personalized. Despite the advantages of the ELK, we should not forget that PK still has indications in cases with added anomalies of the anterior segment, especially if there are important anterior synechiae and/or large iris defects, in addition to all cases in which there is a non-reversible opacity (fibrosis) of the corneal stroma.

The difficulty of the ELK in these eyes lies in poor visualization due to corneal opacity and in the absence of the irido-lenticular complex. There is therefore a high risk of intra- and postoperative dislocation of the lenticule to the vitreous chamber. Staining of the lenticule with trypan blue facilitates its visualization, and some authors propose leaving a suture to facilitate its repositioning in case of postoperative dislocation.

IOL OPTIONS FOR SECONDARY IMPLANT OR EXCHANGE ASSOCIATED TO ENDOTHELIAL LAMELLAR KERATOPLASTY

Once we have decided to implant a new IOL or replace the existing one, in view of the need for an ELK, there are several options:

Anterior chamber IOL of angular support

An AC IOL of angular support does not seem the best option when it has often been the main cause of endothelial decompensation that we want to solve, apart from the risk of secondary glaucoma in these patients. They present the inconvenience of needing a 6 mm incision, which must be hermetically closed to withstand the high pressure required at the end of the ELK. However, due to their simplicity of implantation, they could be useful in patients of advanced age and high comorbidity, since they avoid certain risks associated with the greater surgical manipulation that other alternatives carry.

Anterior chamber IOL of iridian support

The IOL anchored to the iris (Artisan type, Ophtec, Groningen, The Netherlands), in its classic form of implantation in the AC, avoids the problems derived from the angular support. However, it has the same disadvantages in terms of the size of the incision and reduction of space in the AC, with greater risk of contact with the graft and, therefore, of endothelial loss, both during the intervention and in the postoperative period¹⁵.

The group of Melles¹⁶ proposes an algorithm of action according to the type of AC IOL and the associated comorbidity. In patients with angular support AC IOL, no significant differences were found in the endothelial loss between the group without explant and the group to which they were exchanged for another one anchored to the iris. They conclude that the main limitation of Artisan is the greater difficulty in the manipulation of the graft and not a greater endothelial loss.

Posterior chamber IOL sutured to iris or fixed to sclera

The IOL implanted in the posterior chamber is the most physiological way of replacing the crystalline lens. In the absence of capsular support, the classic options included suture to the iris or ocular wall ("in sulcus"). More recently, fixation techniques to the sclera without sutures have been developed, through tunneling and the use of tissue adhesive. The implant of an IOL in this position restores the irido-lenticular diaphragm similar to a pseudophakia in the bag and facilitates the simultaneous or subsequent realization of an ELK (Figure 1).

However, all these techniques require a considerable learning curve due to their complexity. Its anatomical results are sometimes unpredictable, with problems such as decentration, discoria and tilting. The latter is less with those fixed to scleral tunnel or iris. In case of defects of the iris or pre-existing discordance, the iris suture of the IOL can solve them simultaneously. They also present associated complications, such as ciliary body hemorrhage or increased risk of retinal detachment with those sutured to the sclera, or macular edema and iris bleeding in those sutured to the latter¹⁷.

IOL of iridian support in posterior chamber

Rijneveld et al¹⁸ were the first to describe the retro-pupillary implantation of an Artisan IOL anchored to the iris, associated with PK. This was subsequently applied to the ELK, where it improves the integrity of the anterior segment without affecting the angle of the AC, with a lower risk of endothelial loss, so it is recommended to perform it at the same time^2,15,19,20. Although this has the advantage of avoiding two interventions (Figures 2 and 3), there is the possibility of doing it in two stages. Performing the secondary IOL or IOL exchange first and later the ELK allows to assure the result of the first one before approaching the second, especially in surgeons who are new to this technique. Practicing them in the reverse order can be advantageous in cases with very poor corneal transparency, to restore this with an ELK before addressing the IOL exchange, although the second phase will involve a certain aggression to the graft.

Currently, we consider the combined surgery of ELK with retro-pupillary Artisan IOL implant the best option for endothelial failure without permanent stromal opacities (Table 2), provided there is sufficient iris support and visibility through the cornea allows it – if necessary by removing the epithelium and superficial opacities, plus the use of glycerin. This technique is simpler and has a lower learning curve than those of scleral fixation or with sutures and provides a faster visual recovery.

IMPACT OF IRIS DEFECTS IN THE PRACTICE OF ENDOTHELIAL KERATOPLASTY

The success of an endothelial transplant, either by endothelial lamellar keratoplasty (ELK or DSAEK) or Descemet-endothelial keratoplasty (DEK or DMEK), requires the attainment of an anterior chamber (AC) filled with air and pressurized at the end of surgery¹. For this, the presence of adequate support by the irido-lenticular or irido-capsular diaphragm is required. In the presence of iris defects, it may be difficult or impossible to confine the bubble in the AC to achieve adhesion of the graft.

The migration of air from the AC backwards usually causes a displacement of the iris root forward – like an iris bombé – that contracts the AC until it becomes minimal or non-existent, with apposition of the iris to the posterior face of the cornea. This situation complicates the placement, centering and fixation of the graft, and can cause it to fail or cause the formation of anterior synechiae and lead to late failure.

The difficulty in retaining the air in the AC will vary, in principle, depending on the degree of the iris defect, the state of the lens capsule and whether an intraocular lens (IOL) and its type are present (see chapter 6.6.2) or if the eye has been vitrectomized (chapter 6.5.3). The possible glaucoma or risk of developing it should also be considered (chapter 6.5.2). To deal with these difficulties, various maneuvers have been proposed, but in any case, these complex situations seem more accessible to the ELK than to a DEK, so they give preference to the first despite the advantages of the second when it is successful.

For some reason, the subsequent migration of the air bubble does not occur in all cases and does not always correlate with the size of the defect or iridectomy. In our experience, there are a considerable number of cases of eyes with ample iridectomies, whether aphakic, with IOLs – either AC or posterior chamber sutured or anchored to the iris – in which no subsequent migration occurs, and the ELK runs without complications (Figure 1). This could be explained by the rigidity of the iris: in cases where it is increased – as in pseudoexfoliation – its resistance to the passage of air through a certain size of orifice would be greater, while the atrophic and flaccid iris would favor this step. The presence of an orthodox pseudophakia (in the bag) helps to avoid such migration in the presence of a wide iridectomy, while cases with AC IOL, sutured or anchored to the iris, are more difficult.

CLINICAL EVIDENCE

There are no specific studies on the success rate of ELK in patients with iridectomies and, given their poorer prognosis, these patients are not usually included in the large series published. On the other hand, it is difficult for the iridectomy to be an independent factor, and most of these eyes are associated with other alterations that influence the prognosis of ELK, such as a trabeculectomy or drainage tube, an aphakia or AC IOL.

In a series of 93 eyes of 85 patients operated on for ELK (DSAEK) at Moorfields Eye Hospital in London (United Kingdom)², a higher dislocation rate was found in 44% who had communication between the AC and the posterior chamber through an iridectomy, with AC IOL or aphakia, compared to 19% that kept the irido-lenticular diaphragm intact. This was incomplete in 25 cases, along with at least one of the risk factors mentioned. They concluded that any communication such as an iridectomy is a risk factor for dislocation and failure of the ELK.

In another series with the first 188 cases of ELK performed at the Bascom Palmer Eye Institute in Miami (Florida, USA)³, the picture of the posterior migration of air in a patient who presented a large inferior iridectomy is described for the first time. Given the initial impossibility of maintaining the air bubble in the AC, 20% sulphur hexafluoride (SF₆) was injected. Although this maneuver should have been useful in the short term, the authors do not clarify the effect that this gas could have on the viability of the graft or the final result of this case.

We have followed two cases of patients with large iridectomies and iris anchoring AC lenses, in which despite the partial closure of the iridectomy with sutures, there was a continuous migration of air to the vitreous chamber with iris bombé formation and flattening of the AC except in the most central area, so that the iris came into contact with the posterior face of the graft on a large part of its surface. Despite this, we pressurized the eye with SF₆ and left the patient in supine position for 3 hours. The graft was adhered the following day in both cases and, in one of them, despite a large inferior displacement of the graft there was a resolution of the edema. The other one did not clear, possibly due to endothelial trauma due to complicated surgery and the apposition of the iris over the lenticule.

MANOEUVRES THAT CAN BE OF UTILITY IN ELK IN PATIENTS WITH IRIS DEFECTS

Although the presence of an iris defect or iridectomy may pose an added difficulty during ELK by subsequent air migration, the visual and anatomical results may be equally good if this complication is avoided. There are several maneuvers that can be useful in this situation.

Check the behavior of the air before introducing the graft

Once the descemetorhexis is done, we inject air as we would at the end of the surgery and observe the behavior of the bubble and the iris. If it is possible to keep the AC deep, filled with air and pressurized, we can continue in the usual way and the probability of success will be high. In a patient with a wide iridectomy this will indicate that it is not necessary to close it, which reduces the manipulation. If, on the contrary, we cannot maintain the AC formed with air and pressurized, we will have to resort to maneuvers to recover the normal separation ("compartmentalization") between both chambers, by closing the iridectomies and implanting or exchanging the IOL.

Closure of iridectomies or iris defects

Many methods for suturing the iris have been described. In general, it is recommended to use dull atraumatic needles (of the "vascular" type) and poorly resorbable materials such as polypropylene (prolene 10-0). The use of sliding knots as recommended by Siepser⁴ favors its placement (Figure 2). The rigidity of the iris and its friability should be assessed, since trying to close a very large defect can lead to tearing and hemorrhaging.

In cases where reconstruction by suture is impossible, there remains the option of plate ring implants or artificial iris (Figure 3). Once the reconstruction has been carried out, we check again the tightness of the irido-lenticular diaphragm.

Implant or IOL exchange in posterior chamber

If the problem persists after closing the iris defect, better separation can be achieved between the cameras with the implant of an IOL in the posterior chamber – either anchored to the iris or the sclera and be secondary in aphakia or by exchange of an AC IOL –. In general, AC IOLs – whether they are angular support or iris fixation – are less effective in separating the compartments of the eye and do not impede subsequent migration of the air or the iris bombé effect.

Final measures

If all of the above is ineffective to achieve bubble retention in AC and its pressurization, we will have to consider the conversion of surgery to PK. However, if the latter is not practicable – due to the lack of full-thickness tissue or having already prepared the lamellar graft – some final measures may be attempted:

• Use SF₆ instead of air.

• Perform corneal drainage incisions to promote disc adhesion (controversial utility).

• In the case of AC IOL anchored to the iris, perfuse continuously with air through an AC maintainer by a paracentesis, while pushing the IOL (and the iris) back with another instrument (a 30 G needle) through another route. This can be maintained for a few minutes until the formed and pressurized bubble achieves adhesion of the graft.

• In vitrectomized eyes, the continuous pneumatization maneuver can be carried out in a pars plana route, with less interference of instruments in the AC.

• Finally, if no maneuver is effective, you can try to leave the eye pressurized even if the iris contacts the lenticule, and leave the patient in the supine position for at least 2 hours.

Until a few years ago, the only option to treat the failure of a penetrating keratoplasty (PK) was to repeat it. This entailed inconveniences such as a prolonged time of visual recovery, an unpredictable refractive result and an increased risk of rejection. In fact, the failure of 1 or several previous PKs was, 10 years ago, the reason for the realization of 18-40% of PK¹. The general advantages of endothelial lamellar keratoplasty (ELK or DSAEK) over PK also make it a preferable procedure in this situation.

INDICATIONS

Given that the majority of PK failures are due to endothelial failure and the stroma is often not irreversibly affected, ELK is currently the best option for its treatment (Figure 1). On the other hand, despite the increasing use of Descemet-endothelial keratoplasty (DEK or DMEK) and its functional advantages in the usual cases, the greater complexity of the situation after PK failure still gives precedence to the ELK.

ELK induces less astigmatism than a new PK and eliminates complications in relation to the scar and sutures, including the least possibility of dehiscence of the surgical wound. In addition, it allows a faster visual recovery, especially when the refractive result of the previous PK was acceptable. On the other hand, ELK presents a lower risk of rejection than a primary PK² and possibly even less compared to a PK re-insertion.

On the other hand, patients with significant fibrosis or other opacities in the stroma besides endothelial failure, with losses of substance (after ulcerations), with high or irregular astigmatism, are bad candidates for ELK on PK. Even in the absence of acute inflammatory processes – like for example the recurrence of a herpetic keratitis – the simple chronicity of an edema due to endothelial failure may end up causing a progressive and irreversible stromal opacity. Nor are good candidates those who never improved their best corrected visual acuity (BCVA) when the PK graft was transparent and those with clinically significant anisometropia¹.

With ELK, in general, the average improvement of the BCVA is 6 lines at 2.3 years³. In our experience, the average BCVA obtained after ELK in a previous PK is slightly lower to that that the patient had while his PK remained transparent. In case of PK failure secondary to a clinically observed rejection, it is recommended to perform ELK on PK at least 6 months after the last episode, in order to achieve an adequate "clean-up" of immunological mediators⁴.

SURGICAL TECHNIQUE

The surgery of the ELK in eyes with failed PK does not differ in essence from the usual one. However, there are some special aspects that are worth remembering.

Descemetorhexis

Some surgeons practice the descemetorhexis just inside the graft-receptor junction of the old PK, taking care not to introduce the dissection instruments on the posterior border of said union^5,6. Others prefer not to perform descemetorhexis at all, to avoid disruption or even dehiscence of the posterior junction between the donor tissue and the recipient PK^3,7. In our experience, in case of performing the descemetorhexis, it is preferable not to remove the sutures of the PK that may remain to avoid weakening or opening the scar of trepanation due to the need to pressurize at the end of the intervention.

It must be borne in mind that the Descemet membrane (DM) is thinner and friable in eyes with failed PK, so its extraction is not as simple as in Fuchs' dystrophy. It is rarely achieved in one piece and will often have to be scraped or even irrigated-aspirated. Since in this situation the endothelium is usually very attenuated or absent and DM is not thickened or with guttae, its removal does not seem mandatory and it has been observed that leaving it does not increase the rate of lamellar graft dislocation³.

Graft size

Many surgeons prefer that the ELK graft be of a size not superior to that of the old PK, to avoid that the posterior projections of the donor-receptor junction of the PK prevent a good local adaptation of the disc and generate spaces through which it would coalesce aqueous humor and could prevent adhesion (Figure 2). This would avoid a possible cause of graft dislocation^5,6. On the other hand, there are authors who place ELK discs with a larger diameter than PK^1,4,7, and have shown that this does not prevent their union, even in the case of irregularity of the posterior corneal surface due to the presence of PK³. The preoperative study by optical coherence tomography (OCT) allows assessing in each case the degree of irregularity present for that cause.

COMPLICATIONS

The rate of complications of ELK on failed PK is greater than for ELK in corneas not previously transplanted^1,7. The most frequent complications in case of ELK over PK include:

• Primary failure of the graft, with rates between 16 and 29%^1,3.

• Rejection of the ELK: 10.5% at 2.3 years of follow-up, according to the Price group³.

• Graft dislocation (Figure 3) has very variable frequencies according to authors, sample size and graft size: 5.9% using grafts smaller than PK⁶; 43% using grafts 0.5 mm greater than PK¹; 14% using grafts of larger diameter than the previous PK⁷; 6.6% using grafts 1 mm greater than the previous PK³; 16.7% using grafts 0.25-0.75 mm larger than the previous PK⁴. Possibly the dislocation is of multifactorial causes, it has to do not so much with the size of the graft but with circumstances as the presence (67%) or not (24%) of a drainage tube (shunt)⁸.

• Decentralization of the graft with respect to PK.

• Intraoperative dehiscence of the scar of the PK, when manipulating its posterior border during the descemetorhexis⁷ (Figure 4, video 6.6.4.1).

• The appearance of leaks (Seidel +) in the postoperative period by 3%⁴.

• Secondary glaucoma, in 26% of cases of ELK in previous PK, as well as cases of glaucoma due to pupillary block⁴.

• Postoperative cystic macular edema in 10% of cases⁴.

SURVIVAL OF ENDOTHELIAL LAMELLAR KERATOPLASTY AFTER FAILED PENETRATING KERATOPLASTY

The survival rate of ELK on previous PK, in an important series of 246 cases, was 89% at one year, 79% at 3 years and 49% at 5 years, which implies an average survival of 80.9% at 17 months⁹. These data should be compared with the survival of the first PK re-grafts in ideal cases (keratoconus) that have been estimated at 98% to 63% per year and from 69% to 28% at 5 years^2,11-13. In addition, starting with a second PK re-transplantation, these percentages decrease drastically.

The existence of a drainage device is an important risk factor for failure of an ELK in previous PK. In its absence, survival improves up to 100% at one year and 96% at 4 years³, while in its presence it was 93% at one year, 74% at 2 years, 44% at 3 years and 22% at 4 years³. This implies a risk of failure 20 times higher in the medium term, and in some studies the survival comes to an end at 5 years¹⁰.

Multivariate statistical studies allow the identification of independent risk factors for an ELK on a PK to fail, such as young age (≤39 years) and episodes of rejection⁹. If the PK had had a rejection episode, it increases by 2 the risk of failure of the ELK⁹. If an episode of rejection of the ELK itself occurs, although initially recovering, the survival rate drops to 70% during the first year while it is 91% in its absence⁹. They also confirm the deleterious effect of the drainage tubes for glaucoma^3,9,10,14. On the other hand, glaucoma alone or a previous trabeculectomy⁹, anterior peripheral synechiae or corneal neovascularization³ would not be risk factors for ELK failure. This implies that alterations of the ocular surface, one of the main causes of PK failure¹⁵, are not a significant problem in endothelial transplantation¹⁶. Glaucoma, also a risk factor in PK¹⁷, would not affect ELK except in the presence of drainage devices.

The number of previous PKs that failed before the ELK is not a factor that statistically affects the prognosis of the latter³, nor is an early dislocation of the lamellar graft³. Although this complication is somewhat more frequent in patients with valves than in those not operated on for glaucoma or with trabeculectomy, this difference is not statistically significant⁹. On the other hand, when ELKs are performed in each center by a single surgeon, this would be a relative protective factor to avoid its failure⁹.

The loss of endothelial cells after this surgery differs among authors. It has been estimated at 57% after 10 months¹. This would not be statistically different from the cell loss in ELK on non-transplanted corneas¹⁹. Price's group finds an endothelial loss of 26% at 6 months and 28% at one year, and it would be statistically not different from what they observe in their primary ELK series³. In conclusion, the published evidence demonstrates the advantages of rescuing a failed PK through an endothelial transplant instead of a new penetrating one.