The basic technique of endothelial lamellar keratoplasty (ELK) by peeling the Descemet membrane (DM) from the recipient and mechanized cutting of the donor, which corresponds to the English acronym DSAEK, consists of a series of steps that are summarized in Figure 1^1-3.

PREPARATION OF THE DONOR TISSUE

Lamellar cut with microkeratome

The first part of the preparation, if it has not already been done in the eye bank ("pre-cut" tissue) consists of a lamellar cut with a microkeratome. It is necessary to have a corneoscleral segment of sufficient size to mount it in the artificial anterior chamber (AC). Depending on the model, the minimum total diameter should be between 15 and 17 mm. The cornea should have a "white to white" diameter of at least 11 mm. The most widespread system is ALTK from Moria SA (Anthony, France) (Figure 2), but there are others (see chapter 2.4) such as the Amadeus II from Ziemer (Switzerland) (Figure 3, video 6.4.1.1), or the Horizon, which uses single-use material – both artificial AC and heads – and pressurization with air (Figure 4).

Once the corneo-sclera has been mounted on the artificial AC and suitably pressurized, the microkeratome is activated, and a complete lamellar cut is obtained, with a free cap anterior stroma. Normally we remove the epithelium previously, except in corneas of less than 550 μm, which allows to gain about 50 μm of depth and achieve a finer bed. We seek to obtain a bed of between 80 and 130 μm – although up to 200 μm is considered acceptable – because it does not have much influence on the visual result. The thickness of the cut is variable in practice, which may be due, in addition to the different thickness of the donor corneas, to the use of thick gauge heads, whose variability in the cut is intrinsically greater than with the finer ones. In addition, the cadaverous edema causes that the measured thicknesses are greater than those that the tissue will have when it recovers its normal hydration.

With the ALTK system, many authors use the 300 μm head^1-3, although we have good results with the 400 μm one. To obtain discs between 70 and 120 μm, we have developed a nomogram for Amadeus II based on the principle that, the lower the speed of passage, the thicker the sheet obtained will be (Table 1)⁴. This allows to obtain consistent results in corneas between 475 and 650 μm. When they are thicker we perform a double step, first with a 140 μm head and then another in the opposite direction according to the same nomogram and the thickness that was left after the first cut.

Trepanation of the donor disk

Once the lamellar cut of the donor has been made, we proceed to trepanation (this can be done immediately or wait after the descemetorhexis of the recipient). We usually use the Barron's trepan-punch (Katena Products, Denville NJ, USA). The donor corneoscleral segment is placed at the base with the endothelial side facing upwards, with the anterior cap replaced in its bed with its edges marked with marker points, which allows to verify that it is well centered. The die fits in the guides and is applied under pressure (Figure 5); the anterior corneal cap serves as a shock absorber for the tissue with the endothelium.

We obtain a disk with posterior stroma, DM and endothelium. We check its thickness and configuration by lifting it carefully with a forceps (Figure 6), avoiding folds that would damage the endothelial cells, and leaving it on the same base, covered with balanced saline solution (BSS).

The choice of graft diameter should be decided for each receiver. The most common are disks of between 8 and 8.5 mm (and up to 9 mm)⁵. The larger, the greater the endothelial population we will transplant. However, in eyes with small corneas and narrow AC, using a large disk in excess can cause complications and even graft failure (Figure 7).

PREPARATION OF THE RECIPIENT

Marking of the cornea and incisions

The procedure in the recipient (video 6.4.1.2) starts with a superficial mark – on the epithelium or in the Bowman if it has been removed – centered on the pupil and with the same die of the donor tinted or another marker of equal diameter to the graft. We usually use retrobulbar anesthesia, although it is possible to do it with topical⁶. We perform three paracenteses to manipulate the graft with the least trauma.

Usually we make a 4 mm main corneal incision, although this may vary depending on the insertion system we use. It is important to build it carefully, to facilitate tightness and avoid hypotonia in the immediate postoperative period, one of the main reasons for graft dislocation. A small incision induces less astigmatism but causes greater donor tissue damage by compression when inserted into AC⁷.

Dissection by peeling the DM-endothelium complex (descemetorhexis)

The peeling of the DM-endothelium complex or descemetorhexis is the common maneuver that defines the set of techniques of modern endothelial keratoplasty⁸. To do this, we can reform the AC with a cohesive viscoelastic and keep the pupil dilated. This allows us to work with a good background reflex (Figure 8), which is common when cataract surgery is associated. If we prefer to avoid the viscoelastic, we reform the AC with air or install a continuous infusion with an AC maintainer of 23 or 25G (Figure 9), with the pupil in miosis.

The maneuver begins by "marking" or cutting the DM in a circle according to the surface line and detaching it progressively with an inverse Sinskey's hook instrument (e.g., K3-5002, Katena) or others designed for that purpose. We separate the DM from the periphery with a Melles’ scraper (50.212-D, DORC, Zuidland, The Netherlands), with a John's forceps (AE-4962, Asico LLC, Westmont, IL) or with a capsulorhexis forceps, until it is removed. We recommend scraping the posterior stroma about 1.5 mm inside the entire edge of the descemetorhexis, to expose the fibers of the posterior stroma and improve the adhesion of the donor disc⁹.

Four punctures may be added for drainage through the cornea inside the circle, although its usefulness has been questioned. Some surgeons perform a peripheral iridotomy at 6 o’clock – after instilling 1% acetylcholine to close the pupil – in order to obviate the risk of pupillary block in the early postoperative period. In case of having used viscoelastic, we proceed to irrigate-aspirate it completely.

INSERTING THE GRAFT IN THE ANTERIOR CHAMBER

The way to insert the donor disk in the most used CA during the years of development of the technique was by means of tweezers, folded like a Mexican taco. Later, a series of devices were devised for this purpose, whether they are the type of an injector that pushes the tissue inside the AC (push out) or as a glide that slips into the main incision while the tissue is pulled from an opposite incision (pull through).

Insertion with forceps

We apply a small amount of viscoelastic on the endothelial side of the donor disc. With a capsulorhexis forceps we fold the endothelium on the endothelium, so that 40% of it remains above the remaining 60%,⁹ we take it with a pair of Ogawa’s insertion forceps (ref. 20003, Moria) or Katena’s (K5-8010) and we introduce it in the AC.

This maneuver must be performed with decision, since the AC without viscoelastic or pressurized collapses easily and damages the graft (Figure 10). Once inserted in the AC, we deepen this with BSS through one of the paracenteses, which allows the disc to unfold progressively.

Insertion with a glide

One of the most used sliders for endothelial grafting is Busin's (19098, Moria). It requires a main incision of 4.5 mm and a paracentesis at 180°, with the AC maintainer of 25 G through another paracentesis. We place the donor disk inside the slider with the endothelium upwards and without using viscoelastic (Figure 11a).

With the Busin’s forceps (ref: 20004, Moria) or a coaxial capsulorhexis forceps, we pull the edges of the disk until it appears at the end of the instrument (Figure 11b), which we invert and approach to the main incision. We introduce the same forceps through the opposite incision, crossing the AC until its end appears through the main incision; we take with it the edge of the disc and introduce it in the AC (Figure 12).

The pressure of the AC maintainer helps to unfold the disc. We have designed a 23G forceps with a slightly larger curvature than Busin’s and finished with two small 0.12 mm teeth (Figure 13) to prevent the tissue from slipping off (Asico AE - 4219)¹⁰.

Other devices for the introduction of the donor disk

The ideal insertion technique would be one that minimizes folds, compression and friction of the tissue when passing through the incision, and therefore the trauma to the endothelium of the graft. To this end, multiple instruments have been designed (table 2) that try to avoid the disadvantages of more traditional instruments such as forceps. However, with an injector there is also the possibility that the tissue is compressed or folded inside. Therefore, the following evidences should be considered:

a) The endothelial loss is inversely related to the size of the incision. Regardless of the technique used, the 3 mm incisions produce more endothelial trauma than the 5 mm incisions. The greatest damage with these occurs in the periphery and may not be identified in the initial postoperative period but may contribute to late graft failure⁷.

b) To avoid direct contact or overlap between different areas of the endothelium in a standard donor disk (8 mm in diameter and 200 μm in thickness), the internal diameter of an injector must be at least 3 mm¹¹.

UNFOLDING AND FIXING THE GRAFT WITH AIR

Once the donor disk is inserted, it is necessary to unfold it, apply it with precision on the bed of the descemetorhexis and fix it in a stable way. For the deployment, with the technique with forceps, we inject an air bubble under the disk through a paracentesis or through the main incision (Figure 14a).

With the slider technique, the tissue usually floats in the AC since it is maintained by the continuous infusion. To complete the filling with air we can use a 27 G cannula after removing the AC maintainer or changing the infusion line from BSS to air (Figure 14b). We will thus maintain the AC formed with air in a controlled manner the rest of the intervention.

To center the disc in the stromal bed we can use the same hook-type instrument used in the descemetorhexis or forceps from its edge. The tightness can often be achieved with a single suture in the main incision, but you should not hesitate to place the necessary to ensure it and avoid hypotonia in the early postoperative period. If the eye is hypotonic, it may be necessary to inject BSS before leaving the AC completely filled with air, if this alone is not enough to obtain the proper pressure. To measure it, an intraoperative applanation tonometer type Barraquer is useful, with marks of 20 and 30 mmHg.

In order to ensure that there is no fluid in the interface – and to complete the centering – we can perform massage maneuvers depressing the cornea from the center to the edges with a roller or a blunt spatula, if necessary protecting the epithelium with viscoelastic. If this has been removed, we apply a therapeutic contact lens.

THE IMMEDIATE POSTOPERATIVE PERIOD

After about 10-15 minutes in the surgery room, we extract 40% of the air from the AC to prevent it from causing hypertension due to pupillary block. In the presence of a peripheral iridotomy and inducing mydriasis it is possible to leave the air longer. To ensure disc adhesion, the patient must remain in supine and relatively immobile for at least 2-3 hours, after which we perform a slit lamp examination. If the disc is in position, the patient can begin to mobilize with caution – avoiding rubbing the eyes – and start postoperative treatment (see chapter 6.7.2).

Although Descemet-endothelial keratoplasty (DMEK) has become our routine procedure for uncomplicated cases of corneal edema, we continue to use DSAEK or automated endothelial keratoplasty with Descemet membrane (DM) peeling as the procedure for more complex cases, as in eyes with intraocular lens (IOL) in the anterior chamber (AC), trabeculectomy or drainage tube, previous vitrectomy, aphakia and large iris defects. The technique described below has been minimally modified for these complex cases, and with it we have obtained a very low rate of dislocations (2%) and rejection of the graft (less than 1%) in more than 14 years of use and more than 2,000 cases^1-3.

PREPARATION OF THE DONOR'S LENTICULE

Lamellar cutting with microkeratome

The most common microkeratome for DSAEK surgery is ALTK (Moria LLP, Pennsylvania). The artificial AC is filled with the Optisol-GS preservation liquid itself (Bausch & Lomb, Rochester, NY). The corneoscleral segment of the donor is covered with a thin layer of Healon on the endothelium, placed on the AC plunger and oriented according to the longest diameter of the cornea in the horizontal meridian. Once the tissue is in position, the plunger is raised until it is firmly secured. With a syringe filled with Optisol, the pressure in the AC is increased to >65 mmHg and the stopcock is closed to stabilize it. With a Merocel sponge, the epithelium is removed.

The guide ring for the microkeratome can be adjusted in height to obtain the desired diameter for tissue resection. Since the DSAEK requires the largest possible diameter, of about 10 mm, said ring is placed in the lowest possible position without the use of a spacer. The microkeratome head is usually 300 μm and the resection diameter is at least 9.5 mm. For corneas over 580 μm, a 350 μm head can be used. However, thicker cuts increase the variability and risk of perforating the bed. After checking the high pressure with a gravity applanation tonometer or by the touch of a finger, the microkeratome head is mounted in the guide ring and passed over the donor cornea in about 4 to 5 seconds (Figure 1). This cuts a complete sheet of anterior tissue that will remain inside the head.

Disassembly of the corneoscleral segment

After drying the residual bed and checking its smoothness and diameter, its edge is marked to help with the centering of the trepanation. The cut anterior cap is repositioned, using the marks as reference. At its center point another mark is placed and we wait a moment for it to adhere to the bed. The donor tissue is carefully removed from the artificial AC, preventing a collapse that could damage the endothelium. With hummingbird forceps we push the edge of the sclera backwards to detach it from the metal cover. The plunger is lowered very gently and, as necessary, Optisol is injected to prevent the collapse of the artificial AC. The cover is removed leaving the tissue on the plunger with the AC formed. We lift the edges of the sclera gently in each quadrant to detach it from the plunger until air enters the AC. As the tissue rises, the air slowly pushes the Healon, which spills down the opposite edge forming a cohesive bolus between the endothelium and the plunger.

Trepanation from endothelial side

Once the corneoscleral segment has been freed, it is irrigated gently on both sides with Optisol, to remove the remains of Healon. Excess fluid on the endothelium is absorbed with a sponge at the scleral edge, away from the endothelium. We place it on the trepanation block with the endothelial side facing up, centered with the help of the central mark we made on the anterior cap. I use a Barron’s die (Katena, Denville, NJ) of the same diameter intended for descemetorhexis. After cutting with the trephine, the tissue is left covered with a thin layer of Optisol until it is inserted.

PREPARATION OF THE RECIPIENT

Incisions and surface marking

In general, I make a temporary scleral incision for the DSAEK, which provides the greatest manual access and visibility. Previously I practice two paracenteses in the limbus on both sides, separated from each other for about 5 clock hours, with a 1 mm diamond knife (Figure 3). These must be very beveled to achieve the proper hermeticity.

Through one of them, AC is modified with Healon cohesive viscoelastic (Abbot Medical Optics, Santa Ana, CA). Dispersive viscoelastic (such as Viscoat) should be completely avoided in this type of surgery, as they can be trapped in the interface and prevent adhesion of the donor disc.

A circular marker of 8.0, 8.5 or 9.0 mm in diameter is applied on the epithelial surface, depending on the size of the cornea and the preference of the surgeon. It should be remembered that the difficulty to deploy and position a graft increases with its size, significantly between 8.0 and 8.5 mm. The circular mark should never overlap the inner edge of the incisions and be painted with gentian violet after checking its correct centering.

With scissors, a temporal limb peritomy of the conjunctiva is performed, of about 3 hours of arc. After cauterizing the scleral bed, an incision of 5.0 mm in length is made approximately 0.5 to 1 mm behind the corneal limbus and concentric, with a tri-face diamond knife at 350 μm depth. A deeper initial incision results in less beveled wound closure and increased risk of perforation. With a semilunar knife, the corneoscleral tunnel is dissected, at 75-85% depth and up to 1 mm inside the limbus, without yet penetrating the AC. Incisions <5 mm are associated with endothelial loss, higher dislocation and rejection rates, and offer no topographic advantage.

Descemetorhexis

For the DM peeling, I use a simple inverted Sinskey-Terry’s hook with a blunt tip (Bausch & Lomb Surgical, St. Louis, MO), which is introduced in the anterior chamber through one of the paracenteses until it reaches the endothelium at the height of a distal point of the superficial circular mark. The blunt tip easily punctures DM – thickened in Fuchs' dystrophy – but not stromal lamellae, which allows DM only to be dissected without altering the stroma, creating a very smooth bed for grafting (Figure 5a). This maneuver extends along an almost perfect circle that follows the superficial mark slightly inside, so that we will have a certain overlap of the DM of the recipient with the edge of the donor tissue. The recipient bed is thus delineated for endothelial transplantation.

Once the perimeter mark of the bed has been completed, the same hook is used to separate the DM, pulling its edges in 360° (Figure 5b) until it is free. To open the tunneled main incision to the AC, I prefer a 2.8 mm diamond knife for cataract surgery – although others may be used – trying to make it at least 0.5 mm outside the circular mark. To extract the DM and send it to pathology, we use capsular forceps or another type of soft forceps.

Scraping of the peripheral stroma

In order to favor the adhesion of the graft, I consider it fundamental to scrape the periphery of the recipient's bed, to expose the stromal fibrils and to make the surface rough. The scanning electron microscopy shows that, after removing the DM, a smooth like a crystal bed remains. There is a total absence of cut stromal fibrils that are seen after manual lamellar dissection (DLEK). I believe that these fibrils contribute significantly to the adhesion of the donor disc, and their absence after peeling may explain in part why there are more dislocations in the DSAEK than in a DLEK.

For this maneuver I designed a special instrument, the Terry’s Scraper (Bausch & Lomb), similar to an inverted Sinskey but with a hardened and wide tip (1 mm). This is introduced through the main incision in order to scrape 1.0 mm from the periphery of the bed in 360°, taking care to leave intact the central area of ​​5 or 6 mm in diameter. The creation of stromal fibrils is visually verified by their whitish appearance and should not be subtle: it is better to scrape excessively than not enough.

Removal of viscoelastic

After scraping the bed, the main incision is extended to 5.0 mm from the external entrance (Figure 7a) and temporarily closed with a vicryl 10-0 suture. Irrigation-aspiration of the viscoelastic is performed (Figure 7b). It is important that this removal be exhaustive, because any remaining Healon in the AC can prevent the adhesion of the donor disk. The pupil contracts with acetylcholine (Miochol) until it is <4 mm, for safety for the lens during insertion and unfolding of the tissue. The eye is left somewhat hypotonic and the preparation of the donor is completed.

INSERTION AND FIXATION OF THE GRAFT

Insertion technique with Charlie II forceps

After removing the vicryl suture, the AC is reformed with BSS. The donor disk is prepared at the base of the trephine with a drop of Healon on its endothelial surface (Figure 8a) and bent in the form of an asymmetric Mexican taco at 60:40% (Figure 8b). The fold is held along the stromal side with the Charlie II (Bausch & Lomb) forceps. These are thin and toothless, have a separation of 150 μm in the distal tips and a stop that prevents the crushing of the donor tissue. The polished surface of the tips allows them to release the tissue after insertion. The insertion maneuver in the AC must be with decision, with the 60% portion facing forward (the recipient's bed), the 40% backward (the iris) and the taco opening to the left of the surgeon (Figure 9a). With these forceps you can also center the disc by gently pushing along the sides of the stroma, or by massaging the outside of the cornea.

Deployment of the graft

To deploy the donor tissue, the AC is first reformed. In order to guarantee its stability, we can close the scleral wound beforehand with 3 sutures of nylon 10-0. Through the paracentesis on the right, a 27G cannula is inserted with the tip over the iris. BSS is injected gently to deepen the AC, which is usually enough for the disk to begin to unfold. If necessary, we can push the IOL back a bit with the cannula itself. If the tissue is not deployed at all, you can insist from the paracentesis on the left or use the BSS to move the Healon from the endothelial side and thus help the deployment. Since the disk was folded asymmetrically, it will invariably unfold in the correct orientation. Ideally, it should be opened enough so that the edge of its posterior 40% rests on the iris at an angle of about 80°. Once this is achieved, a bubble of air is injected very gently, by the paracentesis on the left, into the fold of the disc until it is fully deployed on the recipient bed (Figure 9b).

Centering and fixation of the graft

Once the deployment is completed, if the disc was not perfectly centered, we can try the corneal massage-sweep: with the AC filled with 70% air, the eye is rotated in the direction in which we want to move the tissue – e.g., if it is off-center down, we will make the eye look up. This maneuver places the air bubble in the angle behind the tissue and allows it to move "downhill" by depressing and sweeping the cornea with a limbus-to-limbus "Cindy Sweeper” (Bausch & Lomb) in the direction of that we want the tissue to move (Fig. 10). Sometimes this is achieved by applying a small jolt with the sweeper.

If the sweep fails to center the disc, we can manipulate it directly, either on the endothelial side or the stromal side. For the former, the Sinskey-Terry’s inverted hook is applied over the peripheral endothelium and the disc is moved to the desired position (Figure 9c). Although this maneuver causes some endothelial damage at the point of peripheral contact, we have not detected that the endothelial count at 6 months is worse than after one after a penetrating keratoplasty. This should only be done with the AC partially filled with air; if it were completely full we would cause stretch marks in the tissue and more damage. The positioning can also be carried out from the stromal side of the interface, with a 30 G needle through the main incision and with the AC completely filled with air.

Drainage of the interface

Once the tissue is properly centered, it is critical to ensure that there is no liquid left in the central or peripheral interface, which may interfere with or delay adhesion. For this we can use Price's "stump" technique: with the AC completely filled with air, beyond the edges of the graft, an intraocular pressure (IOP) >40 mmHg, the center of the cornea is strongly compressed with the roll, with movements from the center to the periphery in each quadrant, so that the liquid passes to the AC. By using only this maneuver (without drainage incisions) we have achieved a low rate of graft detachments (2%) in routine DSAEK.

With the disk well centered and the fluid from the interface removed, we leave the AC with air between 20 and 30 mmHg and wait about 10 minutes with the light of the microscope turned off. Although the choice of this time is arbitrary, we believe that it allows the tissue to reach body temperature and that the endothelial pump starts functioning, as well as a cohesive interaction of the stromal fibrils from the periphery of the recipient with the surface of the donor. We take advantage to instill mydriatics (cyclopentolate 1% and phenylephrine 2.5%) to prevent the bubble that we will leave in the AC – from 6 to 8 mm – to cause a pupillary block, which has not been presented in any of our first 500 cases of endothelial keratoplasty.

Extraction of air

After the waiting time with the AC pressurized with air, we proceed to remove it completely to check that it has not been trapped behind the iris and that the AC is deepened well and the IOP is normal. This is done by exchange with BSS with a 27G cannula through a paracentesis. We orient the patient's face in the opposite direction so that the paracentesis is "uphill”, and the air goes out as we inject BSS. If air is felt in the pupillary area we can suck it directly there. We pressurize the eye with BSS until a normal IOP and only then reinject the 6-8 mm bubble that we want to leave in the postoperative period (Figure 11). We verify that it moves freely and does not adhere to the pupillary ridge.

We verify that the incision is sealed, and the knots are buried on the scleral side. The conjunctiva is closed with suture or cautery. We routinely place a 24-hour collagen shield on the cornea, soaked in antibiotics (moxifloxacin) and steroids (dexamethasone), as well as an occlusive dressing and a shield, taking care not to apply pressure. The patient is taken to the recovery room, where he is kept in the supine position facing the ceiling, for at least one hour and, if possible, until he fully recovers from the anesthesia, when he is discharged with instructions to go home and remain in bed with the nose pointing at the ceiling as long as possible, getting up only for meals and to go to the bathroom, for a maximum of 20 minutes, until you see him again at the clinic the next morning.

The femtosecond laser assisted procedure (FSL) offers significant benefits in keratoplasty. The laser cuts the corneal tissue in a precise, reliable and reproducible way. Both the donor button as the recipient cornea can be modelled with advanced designs, beyond the reach of the most skilled human hand or any other mechanical device. With the FSL, all types of cuts can be designed for both penetrating keratoplasty (PK) and deep anterior lamellar (DALK), with better adjustment of the edges of the Bowman between donor and recipient, reduction of irregular astigmatism and ocular surface problems^1-4. In principle, all the benefits of FSL should also be applicable to endothelial keratoplasty^5-9. But reality is more complex than it may seem.

FEMTOSECOND LASER AND DESCEMETO-ENDOTHELIAL KERATOPLASTY

The two main endothelial transplant procedures are endothelial lamellar keratoplasty (ELK) by Descemet’s membrane (DM) peeling and mechanized donor cutting (DSAEK), and Descemet-endothelial keratoplasty (DMEK).

A FSL-assisted predescemetic lamellar cut would be technically possible using a method similar to that of the "big bubble". The expansion of the gases after repeated impacts with the proper configuration and depth can release the predescemetic layer of Dua together with the DM and the endothelium. This has been achieved in small areas (Figure 1) and in theory the FSL could be calibrated to create more extensive separations at this level. Unfortunately, the photo-disruption caused by the laser generates an ion plasma and gas containing significant concentrations of reactive oxygen species at each impact on the stroma¹⁰. Its probable toxicity to endothelial cells makes a cut with FSL in its proximity impracticable, as estimated, at less than 100 μm¹¹. Therefore, it does not appear that FSL can become a practical way to cleanly separate DM-endothelium.

FEMTOSECOND LASER AND ENDOTHELIAL LAMELLAR KERATOPLASTY

Many surgeons consider ELK (DSAEK) a valid option for endothelial failures, especially in complex situations. Since in this technique the tissue commonly carries 100 μm or more of stroma, it would be possible to prepare it with the help of FSL without causing endothelial toxicity. The possible advantages over microkeratomes are summarized in table 1^1,12,13.

In the first experiences of endothelial keratoplasty assisted by FSL in rabbits, both the donor button and the receptor stroma were pre-cut. The lamellar cut was made at the level of the middle stroma of both corneas, where the collagen density is higher, and it was found that the healing was not significantly different from what is observed in a LASIK¹⁴. With the introduction of the DM peeling that made the cut unnecessary in the receiver, the use of FSL to create lenticules for ELK (DSAEK) showed a very high reproducibility¹³. However, the first clinical series showed poor visual results compared to mechanical procedures or PK^12,15. This was explained by two problems: the interface between the donor and the recipient was rough and irregular^16,17, and the thickness of the lenticule was not regular, creating a wavy posterior cornea.

The technique of the double anterior layer

To even the button interface and obtain smooth lenticules with FSL, a double layer method was designed (Figure 2)¹⁶. It consists of first carving a sheet of the anterior cornea, removing it and then carving a thinner one in the bed to leave a posterior layer of 100-150 μm according to OCT.

In a first experiment to minimize the energy applied to the tissue,¹⁶ cultivated human corneas were prepared, in order to reproduce the conditions of the clinical procedure¹⁸. An FSL of 60 kHz (Intralase, AMO) was used, with the lowest energy parameters to achieve the lamellar cuts. A LASIK flap of 100 μm thickness served as control. Resistance was evaluated to separate the two stroma sheets. With the appropriate energy level (1.0 μJ for the second cut), impact size and distance between impacts (4:4), the dissection was easy and reproducible (Figure 3) and the scanning electron microscope (SEM) showed regular button surfaces, similar to LASIK flaps (Figure 4).

Despite the anatomical improvement, clinical results with lenticules pre-cut with FSL using the double layer technique remain limited. When comparing this (2L) with the previous single lamellar cutting technique with double pass of the FSL (2P), we obtained better visual acuity (VA) with the first one, but in the last control the differences were scarce (Figure 5)¹⁹ and they could not compete with the mechanical ELK/DSAEK¹⁵.

The posterior coupling technique

The attention was focused on the irregularities of the thickness of the graft, visible in the patient transplanted with both slit lamp and OCT²⁰. While the frontal part of the cornea is mainly responsible for its optical power, the posterior layers can cause a critical irregular astigmatism. Due to the preservation method, the cornea for pre-cut with FSL presents an irregular thickness due to the edema, which also creates a greater swelling of the posterior laminae. When the FSL is coupled by the epithelial side, the cut takes as a reference the plane of Bowman's layer and will be perfectly parallel to it but not to the endothelium. Employing a coupling on the posterior side should, therefore, achieve a better parallelism with the endothelial plane.

Indeed, the experiences through posterior coupling of FSL, on the endothelium protected by a thin layer of viscoelastic substance, were able to create perfectly flat lenticules in a reproducible manner (Figure 6)²¹. And despite a certain traumatic reduction, the endothelial population remained functional²². However, despite achieving good transparency with this technique, the visual results remained unchanged, below those of the mechanical procedures (unpublished data).

Effects of FSL on the cutting surface

Some patients with poor outcome after ELK assisted with FSL, or secondary endothelial failure or rejection, were re-operated by PK, which allowed to study the failed lamellar graft interface with the recipient stroma. Interestingly, said interface was regular, thin and smooth. Given this, a possible explanation for the limited VA would be that the FSL action on the stroma and the healing process interfere with its transparency. To investigate this, we created with FSL – with the recommended parameters – endothelial discs with the double layer technique in 3 cultured human corneas, previously desedematized²³. Optical histology revealed an amorphous band of 1-5 μm at the edges of the lamellar cut (Figure 7).

With the SEM at high magnification, the surfaces showed diffuse lamellae like condensed collagen. We hypothesized that said amorphous layer would generate an unusual healing pattern in the grafted patients. By confocal and live microscopy, we observed a high density of keratocytes in the stroma of the recipient on both sides of the lamellar dissection, while the interface appeared acellular and amorphous (Figure 8).

The study of 5 failed buttons after ELK with FSL and 6 after mechanical technique (DSAEK) – re-operated by PK – showed that, while the former remained strongly attached to the recipient stroma, some of the latter partially detached²⁴. The histology revealed, on the entire graft side of the interface in cases assisted with FSL, reorganization of the collagen lamellae and significant densification of the collagen fibrils, with an amorphous layer >10 μm thick in 2/3 of the cases (Figure 9).

By means of transmission electron microscopy, approximately 100 μm of the length of the interface was studied in each sample. In all recipients and mechanically carved grafts, the collagen lamellae could be arranged normally or irregularly, but in no case disorganized fibrils were observed. These, on the other hand, were found systematically on the side of the donor carved with FSL. In these, the disorganized collagen layer was at least 1 μm thick at 96% of the interface length studied. This would confirm that FSL cutting produces changes in posterior corneal stromal collagen which, together with scarring, results in a dense and strong layer at the interface that, while strengthening the graft union, probably degrades visual quality and explains the worst results with this technique.

In conclusion, although assistance with FSL could offer significant advantages over mechanical procedures in keratoplasty, cutting with FSL for ELK remains a challenge. The healing of collagen does not seem to behave in the same way in the anterior and posterior stroma of the cornea after a cut with FSL. With this, the strong healing pattern offers advantages in both DALK and PK. However, it is likely to participate in the limitation of visual results in the ELK. New parameters, procedure remodeling, cutting designs, built-in imaging devices, additional navigation systems, or a new laser wavelength are possible options to address current problems, before being accepted as a routine procedure.