Epidemiology

In indtrialized countries the end-stage of age-related maculopathy (ARM) is the principal cause of legal blindness in persons older than 60 years.^1–3 Age-related macular degeneration (AMD) occurs in an atrophic (one-third) and an exudative form (two-thirds). While we currently have an effective treatment to manage exudative AMD, there is no equally effective treatment in sight for geographic atrophy (GA). This situation is likely to increase the relative proportion of blindness through GA compared with exudative AMD in the future. Increasing life expectancy will account for a rise in the absolute number of GA. (Epidemiology is discussed further in Chapter 63, Epidemiology and risk factors for age-related macular degeneration.)

Alternative treatments for AMD

Surgical treatment

As an alternative treatment, five surgical treatment modalities have been described for exudative AMD and to a lesser extent for dry AMD.

Both macular translocation surgery and the transplantation of a free autologous graft of RPE and choroid will probably achieve a better VA than no intervention. Though macular translocation surgery might produce better outcomes,²⁷ it has a greater risk of complications and there is a physical limit to the amount of rotation possible (even after 360° retinectomy), which makes it less suited to large CNV.

This chapter focuses on the transplantation of free autologous grafts of retinal pigment epithelium and choroid. It will discuss its history, surgery, published results in patients with exudative and atrophic AMD, and conclude with potential developments.

Rationale for reconstitution of retinal pigment epithelium

The spectacular functional restoration achieved in some patients with exudative age-related macular degeneration after macular translocation by rotation proved the principle of reopposing a reversibly compromised fovea to a fresh undersurface of functioning RPE cells.^10,28 Other key discoveries in the concept of restoring the underlayer of the macula have included:

Consequently, many different surgical approaches to recreate a functioning RPE underlayer of the macula have been tried. These approaches may broadly be defined by several complementary techniques: autografts versus allografts; cells in suspension versus cell sheets or patches; RPE versus iris pigment epithelium (IPE) cells (Table 121.1).

Table 121.1 Reconstitution of RPE: cell suspension or cell sheets; allograft or autograft; RPE or IPE

*Flapover technique; ^†Partial thickness graft; ^‡Geographic atrophy.

RPE, retinal pigment epithelium; IPE, iris pigment epithelium.

Transplantation of a full-thickness patch from the midperiphery

With the current lack of demonstrable presence or function of RPE or IPE suspension transplants^35,37,38 the authors decided to pursue the use of a sheet of autologous RPE on its own substratum. Peyman reported a case history on the use of a full-thickness flap with a pedicle. Follow-up was 6 months and stabilization of vision at 20/400 was reported.¹⁹ Aylward, in nine patients, used a full-thickness patch, cut out from a location adjacent to the removed subfoveal membrane. In four patients, some function on microperimetry could be shown over the patch; but the preoperative vision was too low to assess vision in detail using the techniques available at the time.²⁰ Fibrosis of the patch, however, developed in the second year of follow-up in most patients and previously documented function on the graft had disappeared.⁵¹ In Aylward’s patients the grafted paramacular choriocapillary appeared sclerotic and damaged by the surgery and we speculated that it was therefore less likely to be successfully revascularized. Van Meurs and Van Den Biesen²¹ sought to improve on Aylward’s technique by harvesting a relatively healthy midperipheral full-thickness RPE and choroid patch with the advantage of having easy access to cut out the patch and direct control of bleeding from the donor site. Surgeons in Cologne, Kiel, Liverpool, Vienna, Verona, Beijing, London, and Rotterdam have published further reports, some with a different surgical approach with a flapover technique to expose the subretinal space. Surgeons have also reported the use of a split-thickness graft, removing most of the choroid with an excimer laser³⁹ or with a spatula,⁴⁰ or using a spontaneously occurring free RPE sheet in patients with a pigment epithelium detachment.⁵² The last approach is particularly interesting in view of the recent development of free RPE sheets, without choriocapillary, grown from embryonic stem cells (see below).

Surgery

Keyhole approach

After the induction of a posterior vitreous detachment, a complete vitrectomy is performed (see video under http://www.eyemoviepedia.com/videos/3971697493/153). The macular retina can be separated from the RPE and/or blood and/or CNV membrane by injecting a balanced salt solution into the subretinal space through a 28-gauge subretinal cannula. A paramacular temporal retinotomy is made, through which the CNV membrane and/or subretinal hemorrhage is removed from the subretinal space with Thomas subretinal forceps. Remaining hemorrhage is flushed out by irrigation with a 135° bent 32-gauge needle connected by tubing to a syringe filled with BSS and operated by an assistant.

Initially, a small retinotomy was made in the raphe; in later surgeries a radial cut was made along the raphe, from the retinotomy further peripheral. However, a vertical retinotomy temporal to the macula (Fig. 121.1) was found to work best for two reasons: (1) less tendency to enlarge in the direction of the fovea and (2) easier for later introduction of the flat graft (Francois Devin, Marseille, pers. comm.). After circular heavy diathermy or laser in the midperiphery at the 6 or 12 o’clock position, vitreous scissors are used to cut a full-thickness graft of retina, RPE, Bruch’s membrane, and choroid of approximately 2.5–3 × 2.5–4 mm. Initially, the graft was loaded onto an aspiration-reflux spatula. Subsequently the technique was changed so that the graft could be grasped from the choroidal side by fine forceps. The retina is removed from the graft just before it is repositioned under the macula through the existing paramacular retinotomy (Fig 121.2). Perfluorocarbon liquid is injected over the macula to hold the graft in position during retraction of the instrument. A vibration device attached to the forceps facilitates the release.⁵³ The midperipheral donor site is then surrounded with laser photocoagulation, followed by silicone oil tamponade.

Fig. 121.1 A vertical retinectomy (as suggested by F. Devin) is currently preferred because this retinotomy has less tendency to enlarge towards the fovea and the graft is inserted more easily.

Fig. 121.2 Technique of a free retinal pigment epithelium–choroid transplantation using the keyhole technique. PFCL, perfluorocarbon liquid.

In a second procedure, approximately 3 months later, the silicone oil can be removed and the inner limiting membrane (ILM) peeled to avoid the risk of macular pucker. In phakic patients, lensectomy and intraocular lens (IOL) insertion can be performed during the first or second procedure.

Flapover approach

To obviate the need for a parafoveal and midperipheral retinotomy, to enable better control of choroidal bleeding and to enable the surgeon to fashion the size of the graft to the RPE defect a technique from macular translocation by retinal rotation can be used.⁵⁵

A temporal retinal detachment is created by subretinal infusion of balanced salt solution (BSS) through a 41-gauge needle, followed by a retinectomy near the ora serrata in the temporal 13 clock hours with subsequent folding over of the retina over the disk to expose the temporal subretinal space. While working under BSS for membrane removal and later under PFCL for preparation of the graft more direct control of obtaining a correctly sized patch is possible. It may be the approach of choice in large lesions and allows direct control of bleeding from the stalk of the CNV (Fig. 121.3). This technique is currently favored by more surgeons and the best results so far reported are by Pertile in Verona.⁴⁵ A specific concern here is that the detached temporal retina may incarcerate in the nasal sclerotomy. Measures to prevent this occurring are: the meticulous removal of vitreous base remnants, a low infusion pressure, and a retinectomy that should not be extended too far to the nasal sclerectomy. When incarceration occurs, you can pull the incarcerated retina out of the sclerotomy with a ring forceps, after closing the infusion (Gracia Pertile, May 2011, pers. comm.).

Fig. 121.3 Technique of RPE choroid graft transplantation under direct view after flapping over the temporal retina.

Results in exudative AMD

Reports of free RPE grafts generally have been reported on small groups of patients and have had a follow-up of only 6 months to 3 or 4 years (Table 121.1). MacLaren and Aylward have presented patients with a 5- and 6-year follow-up, but for only four of nine operated patients.⁵⁶

A recent report followed a cohort of 132 consecutive patients (134 eyes) who underwent RPE–choroid graft surgery; in some patients VA and/or microperimetry results of 4–7 years after surgery were documented (Fig. 121.4).⁵⁷ In several patients, clear evidence of retained retinal function on the graft proved in principle the longevity of this technique and at timepoints not yet reached with anti-VEGF treatments (Figs 121.5, 121.6).

Fig. 121.4 Visual acuity (VA) before surgery (preoperative) and each year up to 7 years thereafter. Three groups are depicted. Groups 1 and 2 comprise percentages of patients with a VA of >20/200 or ≤20/200. The third group comprises percentages of patients whose last measured VA had been >20/200 before they were lost to follow-up (LFO), and patients who had not able to reach their next yearly follow-up visit yet, while VA was >20/200 at last measurement (LFO while >20/200). The data of this third group have been carried forward in this figure.

Fig. 121.5 A 78-year-old male. Preoperative visual acuity (VA) 20/100, minimally classic choroidal neovascularization. NIDEK MP-1, 5 years postoperatively, with VA 5 20/40.

Fig. 121.6 An 80-year-old woman. Preoperative visual acuity (VA) 20/320, >50% hemorrhagic lesion. NIDEK MP-1, 4 years postoperatively, with VA 20/50.

The number of patients with preserved macular function remains modest, however, due to several factors: pre-existing and surgical damage to the retina; ischemic and surgical trauma to the RPE–choroid graft; revascularization delay and potential revascularization injury of the graft. Comparison with the results of the submacular surgery trials, however, suggested that it is the RPE graft rather than simply the removal of the submacular choroidal neovascular scar and blood, that was responsible for the preservation of reasonably good macular function in some patients.²⁶

Recently, it was shown that spectral domain OCT, a non-invasive technique, allows the monitoring of revascularization of a RPE–choroid graft (Fig. 121.7).²⁶

Fig. 121.7 A 79-year-old woman with 2-month-old submacular hemorrhage, after six anti-VEGF injections. (A) Preoperative: spectral domain optical coherence tomography (SD-OCT). (B) One day postoperative: SD-OCT demonstrating small and clear vessel lumina in the graft. (C) Six days postoperative: larger diameter of vessel lumina and a gray shading of the lumen suggest connection of afferent and efferent vessels with flow.

Spectral domain OCT may also prove to be an important tool to select patients with a potentially functional retina, by selecting patients with preserved outer retina structures.

RPE transplantation in dry AMD

Full macular translocation is usually complicated by rapid onset of RPE atrophy at the new location of the macula.⁵⁸ Two prospective non-randomized studies are available looking at the prognosis after patch translocation in GA.^41,44 A follow-up of more than 2 years is required to be able to discern a therapeutic effect compared to the slow progressive natural course.⁴⁴

Surgical aspects in dry AMD

In patients with dry AMD, several surgical steps are slightly different from those for exudative AMD (see video under http://www.eyemoviepedia.com/videos/4166636950/67). After a three-port pars plana vitrectomy, balanced salt solution was injected subretinally temporal and adjacent to the macula. In almost all eyes, the macula did not detach, but only the retina around the macula. Additional mechanical separation of the macula proved necessary. We used an angulated infusion needle as spatula and fluid jet. At the same time Bruch’s membrane was scratched to create breaks in Bruch’s membrane allowing choroidal vessels to connect to the transplant later on, as graft non-perfusion was observed in our first patients (Cologne, Rotterdam) (Fig. 121.8) when this step was omitted. Access to the subretinal space was through a temporal retinotomy. The excision site of the graft was chosen in the inferior retinal periphery after demarcation by laser coagulation. After removal of the overlaying retina, the graft was mobilized from the sclera, grasped with angulated forceps from the choroidal side and inserted under the macula through the temporal retinotomy.⁵⁹ Heavy silicone (Densiron, Fluron, Ulm) was used as vitreous tamponade for 3 months.

Fig. 121.8 One of the first patients with dry AMD and a free retinal pigment epithelium–choroid graft, in whom Bruch’s membrane was left untouched and no revascularization occurred.

Functional outcome of ten patients: prior to surgery, seven of the ten patients were able to read compared with five patients being able to read 2–3 years postoperatively. Microperimetry showed central fixation in five eyes prior to surgery and in two eyes 2–3 years after surgery. In all but one eye revascularization of the graft was confirmed by ICG angiography beginning as early as 3 weeks after translocation. Autofluorescence was present in all eyes throughout the follow-up period.

Recurrent CNV appeared in three eyes at the edge of the graft, not affecting the fovea. Macular pucker was noted in three eyes. PVR retinal detachment outside the posterior pole occurred in three eyes.

Conclusions for patient benefit in geographic atrophy

In terms of timing of surgery, the insidious onset of changes in the choroid and RPE in atrophic AMD with subsequent atrophic changes in the overlying retina compare unfavorably with the relatively acute-onset changes in RPE and choroid in patients with exudative AMD, with more often some degree of preservation of the overlying retina.

None of our patients with atrophic AMD improved in visual acuity. Stabilization or further visual loss is a likely long-term outcome. Retinal atrophy could not be reversed by RPE–choroidal grafting. Excentric preoperative fixation will persist. Tight adhesion of central retina to the choroid is characteristic to GA. This explains a higher risk of surgical damage to photoreceptors compared to exudative AMD, in addition to pre-existing retinal atrophy as part of dry AMD. Only a very small selection of patients with preserved central fixation and an intact junction of inner/outer segment photoreceptor junction on OCT may profit from grafting, at a considerable risk, however, of losing central vision.

Of interest, we observed no recurrence of RPE atrophy on the graft, as opposed to its rapid recurrence after full macular translocation.

RPE–choroid translocation and future stem cell treatments for AMD

A stem cell is a dividing self-renewing cell also capable of differentiating into a variety of adult differentiated cell types to renew various tissues. While it has always been appreciated that embryonic stem cells are pluripotent, in that they give rise to a whole embryo and any one of the approximately 220 different cell types that define the adult mammalian organism, it was not until the cloning of Dolly the sheep in 1997 that it became apparent that adult differentiated cells could also have a similar potential. In the case of Dolly, the DNA from a mammary gland cell taken from her genetic mother was extracted and injected into a newly fertilized oocyte, from which the original gamete-fused nucleus was removed.⁶⁰ This process, known as somatic cell nuclear transfer (SCNT), led to the development of a clone of embryonic stem cells which was subsequently able to differentiate into the sheep, which was born after implantation into a surrogate mother.

The fact that the sheep was normal proved the principle that any adult somatic cell could potentially be reprogrammed by undefined factors within a newly fertilized oocyte cytoplasm to become an embryonic stem cell and this was subsequently confirmed in primates.⁶¹ Somatic cell nuclear transfer requires a newly fertilized oocyte and so brings with it some ethical considerations and practical limitations. However, a decade after the successful cloning of Dolly the sheep it was discovered that adult differentiated cells could be reprogrammed without the oocyte by manipulating the histones which are the protein switches that inactivate specific genes, particularly histone H3, or by direct methylation of the DNA promoter sequences.^62,63 Both of these approaches switch specific genes on and off and effectively reprogram the cell to become, for instance, a stem cell.⁶⁴ This research identified the transcription factors that facilitative reprogramming of skin cells that were then reimplanted to become a fully differentiated mouse embryo and the reprogramming part (without implantation) has subsequently been achieved in human cells.⁶⁵ Hence, current talk about stem cells is not discussing fetal cells or even cells modified via the oocyte, it is believed that induced stem cells may hold the key to regeneration of diseased tissues in future. These could be obtained potentially from any patient and manipulated into the desired cell type prior to transplantation – referred to as induced pluripotent (iPS) stem cells.

It should be immediately apparent at this stage that since an iPS cell could potentially give rise to a normal mammalian organism, it should by definition be able to give rise to any of the cell types present in that organism. This includes of course the RPE, photoreceptors, retinal ganglion cells and other components of the visual system. This raises the possibility that we may be able to take adult skin cells from a patient with AMD and reprogram these into RPE cells prior to transplantation. This sounds like an exciting concept, but we need to consider some of the observations learnt through two decades of AMD surgery research. We know from the studies mentioned above that suspensions of healthy RPE cells isolated from the retinal periphery in AMD patients are unable to reform a physiological monolayer underneath the macula. This may in part be due to the fact that these cells need to be fused with the underlying Bruch’s membrane and the latter is almost certainly diseased and compromised in the later stages of AMD.⁶⁶ The choriocapillaris which gives rise to Bruch’s membrane may be derived from the mesoderm germ layer whereas the RPE originates from the neuroectoderm.⁶⁷ This is highly relevant, because in studies to date, that have derived RPE cells from both ES cells and IPS cells, it has become apparent that the RPE cell differentiation does not include the creation of a choriocapillaris or Bruch’s membrane.^68,69 The key challenge of the application of iPS-derived RPE cells will be to generate these cells not as a suspension, but fused to an underlying substrate of Bruch’s membrane, or similar, that can facilitate polarization and hence balanced homeostasis of RPE cells once transplanted into the subretinal space.⁷⁰

It can therefore be seen that the techniques described above, particularly in relation to RPE–choroid patch graft surgery, are highly relevant to any future stem cell approach because it is almost certain that a very similar surgical technique would be employed to deliver reengineered tissue into the subretinal space. In cases where the retinal degeneration has progressed to the point of photoreceptor loss, any stem cell strategy to transplant photoreceptors back into the fovea would need to be combined with restoration of the underlying RPE, because otherwise the transplanted cells would not survive.⁷¹ It may therefore be possible to use iPS-cell-derived RPE engineered on a suitable Bruch’s substrate via a standard surgical procedure to restore submacular RPE as the first step, followed by reintroduction of photoreceptors via an alternative stem cell approach at a later stage. It is however worth noting that the RPE–choroid patch graft results, described in detail in the section above, suggest that there is no intrinsic problem in using autologous cells harvested from the retinal equator for subfoveal RPE replacement. This would also solve the Bruch’s membrane substrate problem, because RPE cells would remain adherent to the choriocapillaris throughout the transplantation procedure. While there is no doubt that photoreceptors would almost certainly need to be derived from an alternative possibly stem cell source, the idea of using stem cells or iPS cells to replace the RPE is less clear, because the problem of recreating the Bruch’s membrane substrate still remains (Fig. 121.9). Furthermore, the concept of introducing genetically modified and potentially teratogenic cells brings with it additional risks that would delay clinical application,⁷³ whereas autologous grafting techniques are already established in the clinical arena.

Fig. 121.9 Ultrastructure of a full-thickness RPE–choroid graft. Scanning electron micrograph showing the regular alignment of retinal pigment epithelium (RPE), choriocapillaris and the continuous Bruch’s membrane between (arrowheads). The challenge of any stem cell approach will be to recreate a structure similar to this, particularly as autologous tissue can be harvested with minimal risk from the anterior region of the eye and supports foveal vision up to at least 20/30.

(Reproduced with permission from MacLaren RE, Pearson RA. Stem cell therapy and the retina. Eye 2007;21:1352–9.⁷²)

Recently, a Phase I clinical trial has been approved by the FDA to use a stem cell approach to treat 12 patients with Stargardt disease.⁷⁴ This concept has many challenges, not least of which because many patients with Stargardt disease lose vision prior to disruption of the underlying RPE, presumably due to photoreceptor dysfunction.⁷⁵ Similarly, it is not clear if these transplanted RPE cells have any capacity to adhere to or polarize on the underlying Bruch’s membrane in these patients. Nevertheless, it is likely that observations on the behavior of these embryonic stem-cell-derived RPE cells will provide useful information on the viability or not of this as a potential future treatment. Moreover many of the technical challenges relating to surgery with these cells will be overcome, at least in part, as a result of what we have learnt to date with the surgical approaches using autologous RPE–choroid grafts.