Assessment of 180° Rotation of the Choroid as a Novel Surgical Treatment for Age-Related Macular Degeneration

Abstract

Purpose.: Our objective was to examine the feasibility of rotating choriocapillaris, Bruch's membrane (BM), and retinal pigment epithelium (RPE) through 180° on a vascular pedicle and to assess revascularization and tissue preservation postoperatively. Such an approach could be used in the treatment of age-related macular degeneration where there is focal disease at the macula with healthy tissues located peripherally.

Methods.: Successful surgery was performed in six rhesus macaque monkeys, which have a very similar choroidal blood supply to humans. After inducing a retinal detachment, the recurrent branch of the long posterior ciliary artery was used as a pedicle around which a graft stretching to the temporal equator was rotated. Retina was reattached over the rotated graft and eyes were followed up for up to 6 months with repeated angiography and optical coherence tomography (OCT). The morphology of retinal cells and BM were assessed by immunohistochemistry and electron microscopy.

Results.: Revascularization of the choroid was limited, with reestablishment of drainage to the vortex veins seen in only one case. There was a secondary loss of the RPE and outer retina evident on histological analysis three months after surgery. The underlying BM however remained intact.

Conclusions.: Pedicled choroidal rotation surgery is technically feasible in vivo with intraoperative control of bleeding. However, lack of graft revascularization with the technique in its current form leads to neuroretinal and RPE tissue loss, and graft shrinkage. We found no evidence that rotational grafts are likely to improve the outcomes presently achieved with free graft techniques.

Introduction

Age-related macular degeneration (AMD) is the leading cause of blindness in the developed world.^1,2 It is predicted to have increasing socioeconomic impact with proportionately more people affected owing to the increase in life expectancy.³ Central to the pathophysiology of AMD is accumulated damage to Bruch's membrane (BM), attributable in part to a chronic autoimmune process that results in complement deposition and secondary dysfunction of the overlying RPE.^4–6 The loss of RPE leads to a chronic secondary degeneration of overlying photoreceptors, culminating in loss of central vision (dry AMD).⁷ In some patients, complexes of abnormal choroidal new vessels (CNV), accompanied by fibroblasts, perforate the compromised BM disrupting the RPE and retina leading to cell death and acute visual loss (wet AMD).^8–10 CNV may be treated with repeated injections of VEGF inhibitors,^11–13 but there is currently no effective treatment for chronic dry AMD. Because in most cases the underlying chronic AMD disease process continues even with anti-VEGF treatments, the search continues for more effective restorative therapies. 

AMD is a focal disease that develops in the central macula but spares the retinal periphery. Surgical approaches for AMD aim to reoppose foveal photoreceptors with healthy extrafoveal RPE-choroid before the photoreceptors undergo irreversible damage. This not only prolongs photoreceptor survival, but can also promote recovery of damaged photoreceptors, as evidenced by regeneration of outer segments.¹⁴ Current clinical techniques can be broadly divided into those that move the foveal photoreceptors away from the focal RPE disease by detachment and rotation of the neurosensory retina (macular translocation), or those that replace diseased RPE by grafting tissue harvested from healthier extramacular regions (RPE-choroid grafts). Macula rotation can provide good visual acuity results but torsion of the retinal image causes diplopia if vision in the fellow eye is good, and there is a limit to the angle of rotation possible. This makes it unsuitable for the first eye affected or for larger lesions.^15–19 RPE-choroid grafts can be used on first eyes, as the foveal photoreceptors are not displaced, and also have the potential for good improvements in vision.^14,20–32 Free RPE-BM-choriocapillaris patch grafts allow larger subfoveal defects to be treated than has proved possible with either translocation surgery or submacular rotation of pedicled grafts. There is, however, an early period of ischemia before revascularization of free grafts,^31,32 which is a possible cause of further compromise to the overlying photoreceptors. 

The complexity and complications associated with current surgical procedures has limited their application. At present, surgery is largely restricted to patients with specific mechanical conditions, such as RPE tears³³ or large submacular haemorrhages,²⁵ which have a poor prognosis with anti-VEGF therapies. Surgical approaches have, however, recently regained renewed interest owing to the possibility of using embryonic stem (ES) or induced pluripotent stem (iPS) cells to replace compromised RPE cells in all types of AMD. Such cells can be induced to form an RPE monolayer, which has the capacity to phagocytize photoreceptor outer segments when transplanted into a rodent model of RPE disease.^34–36 However in human AMD, as opposed to the rodent disease models, there is accumulated damage to the underlying BM which compromises transplanted RPE cell survival. The clinical translation of stem cell–related technologies in AMD is therefore dependent on developing a means of repairing or replacing the damaged subfoveal BM.³⁷  

Against this background, we therefore sought to develop a novel surgical technique to incorporate some of the advantages of macular rotation and RPE-choroid graft surgery, but which might also provide a means of replacing submacular BM necessary for future ES/iPS cell approaches and reduce the potential for early ischemia. The technique involves rotating a full-thickness graft of choroid, BM, and RPE on a vascular pedicle such that healthy extramacular tissues are repositioned under the fovea. In contrast to free RPE-choroid grafts, the aim was to maintain the arterial blood supply and evaluate whether vascular flow in the graft could be reestablished with the peripheral choroid. Potentially such an approach would allow diseased submacular tissues in the early stages of AMD to be replaced with healthy tissue rotated from the periphery before the onset of significant photoreceptor degeneration. We describe the results of the surgery in the macaque monkey, as it has a very similar choroidal blood supply to humans and similar retinal structure, including vascular arcades and a fovea. 

Materials and Methods

The choriocapillaris temporal to the macula and extending as far as the equator of the eye is supplied by a recurrent branch of the temporal long posterior ciliary artery (LPCA) (Figs. 1A, 1B).^38–40 This recurrent branch of the LPCA pierces the sclera through an arc-shaped window and travels a few millimeters across the suprachoroidal space before supplying the lateral choroid. The temporal point of entry of this artery can be seen by ophthalmoscopy in patients whose choroid has undergone atrophy in diseases such as choroideremia (Figs. 1C, 1D). This position and the relatively long suprachoroidal course makes it an ideal arterial pedicle on which to center a rotational graft spanning between the optic nerve head and temporal ora serrata. It should be noted, however, that owing to the unique blood supply of the choroid, venous drainage follows a different course horizontally across the choroid to the vortex veins at the equator and hence the pedicle in this case would contain only the arterial vessel. 

Figure 1.

 

Rotation of a full-thickness RPE-BM-choriocapillaris graft on a vascular pedicle. The choroidal blood supply of a human left eye is shown in scanning electron microscopic images of methyl methacrylate microvascular casts ([A], and enlargement of boxed area [B]). Images courtesy of Jane Olver. (A) Vortex veins are highlighted with red arrows and the lateral and medial long posterior ciliary arteries (LPCA) with white arrows, which originate either side of the optic nerve (ON). Choriocapilliaris temporal to the macula is supplied by a recurrent branch of the lateral LPCA (red and white arrowheads in [B] respectively). This recurrent branch transverses the sclera to enter the suprachoroidal space temporal to the macula and can be seen in fundus photographs from a patient with choroideremia in which the overlying RPE has degenerated ([C] and corresponding enlargement [D]; vessel course highlighted with white arrowheads and red arrow, respectively). Full-thickness RPE-BM-choriocapilliaris grafts were rotated on this blood supply temporal to the macula. Thus, in an eye with a subfoveal hemorrhage ([E], retina in blue) the temporal retina would first be folded back nasally to access the subretinal tissues, and a graft fashioned around the vascular pedicle to include the subfoveal tissues ([F], graft indicated with white circle, vascular pedicle in red). This disc of tissue is then rotated on its vascular pedicle (G), such that damaged subfoveal tissues are replaced with healthy peripheral tissues, and the overlying retina replaced (H).

Animals and Anesthesia

Surgery was performed in adult female Macaca mulatta monkeys. The animals were cared for in accordance with the ARVO statement for the use of animals in ophthalmic and vision research and the UK Home Office Animals (Scientific Procedures) Act 1986. All procedures were carried out in accordance with the European Communities Council Directive 1986 (86/609/EEC), the US National Institutes of Health Guidelines for the Care and Use of Animals for Experimental Procedures, and the UK Animals Scientific Procedures Act, and were approved by the local ethical review committee at University College London's Institute of Ophthalmology. Surgery was performed on right eyes only and animals were followed for up to 6 months postoperatively. Before surgery and postoperative assessments, pupillary dilation was achieved with 1% tropicamide and 2.5% phenylephrine hydrochloride administered topically. Anesthesia was induced with a ketamine/medetomidine mixture (6.4 mg/kg ketamine and 0.08 mg/kg medetomidine, intramuscularly [IM]) and intravenous access established following atropine administration (atropine sulfate 0.04 mg/kg IM). Short-duration anesthesia (up to 30 minutes for ocular imaging) was maintained with a single administration of IM ketamine/medetomidine alone. For surgical recovery procedures, anesthesia was maintained for up to 4 hours using isoflurane in oxygen and the level of anesthesia was deepened as required during retinal surgery and particularly during surgery to the choriocapillaris. Animals were intubated and the blood pressure, electrocardiogram, end-tidal CO₂, and peripheral oxygenation were continually monitored. A thermostatically controlled heat blanket was used to maintain body temperature at 37°C and intravenous fluids were given routinely. 

Surgical Technique

Surgery was performed under sterile conditions using standard 20-gauge vitrectomy instrumentation, a Zeiss operating microscope (Zeiss, Oberkochen, Germany), and BIOM wide-angle viewing system (Oculus, Wetzlar, Germany). A xenon chandelier light source was also used during manipulation of the choroid. The surgery consisted of a vitreolensectomy, followed by detachment of the temporal half of the retina to access the subretinal tissues. A full-thickness graft of RPE, BM, and choroid was circumscribed with laser, dissected free, and rotated by 180° under heavy liquid. The recurrent branch of the LPCA was identified beneath the choroid during the dissection and preserved. The retina was then reattached so that the fovea lay over RPE-choroid rotated from the extreme temporal periphery and the vitreous cavity was filled with silicone oil before recovery. The steps are discussed in further detail as follows and are illustrated in Figure 2. 

Figure 2.

 

Choroidal surgery in the Macaque monkey. Following a lensectomy and vitrectomy, a retinal detachment was induced through injections of fluid in to the subretinal space (A). The peripheral retina was treated with endolaser, dissected (B) and folded nasally over the optic disc (C). To control hemorrhage, the edges of the RPE-BM-choroidal graft were lasered (C) and then cut (D) and rotated around the deep vascular blood supply of the recurrent branch of the LPCA (E). At the end of surgery, the retina was folded back over the rotated graft (F) and kept in place with further peripheral laser and silicone oil tamponade.

1.  
   Before surgery, the periorbital regions were thoroughly cleaned with povidine iodine and sterile surgical drapes were applied. A pediatric lid speculum was applied to the right eye. Marcaine was injected into Tenon's space to partially proptose the eye and facilitate access (as the supraorbital brow of the Macaque is prominent compared with that in humans). Three sclerotomies were made 2 mm posterior to the limbus. The inferotemporal port was used to fix a perfusion cannula, which was secured in place with 7/0 vicryl.
2.  
   Pars plana lensectomy to render the eye aphakic was performed using a fragmatome and infusion line connected to a 20-gauge needle.
3.  
   A pars plana vitrectomy with excision of the posterior hyaloid face, core vitreous, and anterior vitreous was performed. Posterior vitreous detachment was induced through the combination of vacuum aspiration and with the use of a pick where necessary.
4.  
   Temporal retinal detachment was induced through subretinal injection of Ca²⁺ and Mg²⁺-free BSS Plus (Alcon Laboratories, Fort Worth, TX) using a 41-gauge cannula.
5.  
   A 180° temporal retinotomy was created with vertical cutting scissors following demarcation with endolaser. The detached retina was folded nasally by pulling on the anterior vitreous and then held in place with perfluorocarbon (Bausch & Lomb, Kingston upon Thames, UK, model VRL 110).
6.  
   The site of the choroidal graft was first delineated with endolaser and then cut with vertical cutting scissors. Grafts extend from midway between the optic disc and fovea to close to the temporal ora serrata, and so measured up to 8 disc diameters across their largest dimension. In the macaque, the macular vessels extend farther anteriorly than in humans (i.e., the latter have more anterior growth of the eye). Hence, although the grafts may appear small relative to the posterior vessels, the temporal margin is in fact very close to the ora serrata and the equivalent-sized graft relative to the eye in humans would represent rotation of almost the entire posterior choroid. Infusion fluid containing 0.01% adrenaline was used to control bleeding. Once free of surrounding choroid, grafts could easily be rotated around the deep arterial pedicle by up to 180° and positioned in contact with the temporal border of the uncut peripapillary choroid.
7.  
   The neurosensory retina was rolled back to its original position through aspiration of heavy liquids overlying the choroidal graft in the posterior pole of the eye and then injection over the optic nerve to gradually fold the retina back. At completion of the procedure, endophotocoagulation was carried out using an illuminated laser pipe to the site of the retinotomy. A direct perfluorocarbon silicone oil (1300 centistokes, Bausch & Lomb) exchange was performed using the viscous fluid injection pump and passive aspiration using a 20G Charles flute. The silicone oil was left in situ throughout the postoperative period.
8.  
   Before recovery, sclera and conjunctival wounds were closed with 7/0 and 8/0 vicryl sutures respectively. Subconjunctival cefuroxime and Depo-Medrone were administered to the operated eye and animals were given subcutaneous injections of antibiotic and nonsteroidal anti-inflammatory medication. Antibiotic and nonsteroidal anti-inflammatory medications (meloxicam) were administered orally for a minimum period of 5 days following the procedure.

In Vivo Assessments

Confocal scanning laser ophthalmoscopy (cSLO) was performed under anesthesia with the Heidelberg Spectralis hardware and software (Heidelberg Engineering, Dossenheim, Germany). Autofluorescence images were obtained from surgical and control eyes. Angiography was performed with fluorescein and indocyanine dyes injected intravenously. Spectral domain optical coherence tomography (OCT) images were also obtained using the Spectralis system. All in vivo assessments were performed with similar settings to those used in patients. 

Tissue Processing

The animals were killed by an overdose of intravenous pentobarbital while under deep anesthesia, and the control and procedured eyes enucleated. For light microscopy, eyes were fixed with PBS (pH 7.4) 4% paraformaldehyde overnight, cryoprotected and embedded in optimal cutting temperature compound (OCT) (R A Lamb, East Sussex, UK) with care to maintain track of the orientation. Frozen specimens were stored at −20°C and sectioned on a cryostat at 12 to 20 μm. Retinal sections were blocked in 0.01M PBS with serum and 0.1 % Triton-X 100 for 1 hour, then incubated with 1:400 primary antibody at 4°C overnight. After rinsing with PBS, sections were incubated with 1:1000 secondary antibody for 2 hours at 4°C, rinsed, and counterstained with Hoechst 33342 before mounting. The following antibodies were used: rabbit anti-red/green opsin (Millipore, Watford, UK), rabbit anti-PKCα (Epitomics, Burlingame, CA), rabbit anti-RPE65 (kind gift from Rosalie Couch), rabbit anti-glial fibrillary acidic protein (GFAP) (Abcam, Cambridge, UK), lectin peanut agglutinin (PNA)-AlexaFluor 488 (Molecular Probes; Invitrogen, Grand Island, NY), and lectin RCA-fluorescein (Vector Labs, Peterborough, UK), with Alexa-tagged secondary antibodies (Molecular Probes, Invitrogen) as appropriate. PAS (ClinTech, Guildford, UK) and Alcian blue (Merck, Darmstadt, Germany) staining was performed according to manufacturer's guidelines. 

Electron Microscopy

For scanning electron microscopy, tissue was fixed with 3% glutaraldehyde and 1% paraformaldehyde in 0.08M sodium cacodylate-HCl buffer (pH 7.4). Fixed tissue was then rinsed in PBS and immersed in 1% aqueous osmium tetroxide solution for 2 hours at room temperature. Following osmication, samples for resin embedment were rinsed in distilled water (three times over 10 minutes) and dehydrated by passage through an ascending alcohol series comprising single immersions in 50%, 70%, and 90% ethanol and then three immersions in absolute ethanol. Each step lasted 15 minutes and was carried out at room temperature. Samples were then loaded into a BalTec CPD 030 dryer (Leica Microsystems, Milton Keynes, UK), critical point dried, mounted, and coated with 1.5 nm of platinum using a Cressington 308R coater (Cressington Scientific Instruments, Watford, UK) before imaging in a Zeiss Sigma field emission scanning electron microscope (Carl Zeiss NTS, Cambridge, UK) at 5 kV. 

Results

Surgery

Surgery was performed in eight eyes of eight animals in total. In six animals, choroidal flaps were successfully created and rotated on their arterial pedicle without intraoperative complications (Fig. 2 and Supplementary Movie). Long-term monitoring of the grafts (see later in this article) was undertaken for four of these six animals, as despite successful surgery, two were excluded from follow-up because of a latent undetected renal problem. In two cases, free grafts were translocated but without preservation of the arterial pedicle. Ca²⁺ and Mg²⁺-free BSS Plus (BSS Plus part I; Alcon, Hemel Hempstead, UK) was used during induction of the temporal retinal detachments. This solution has previously been found to reduce the strength of attachment between RPE cell microvilli and photoreceptor outer segments in rabbit eyes, and may thus facilitate the creation of a less-traumatic iatrogenic retinal detachment.^41,42 Three or more such subretinal infusions were required in each case to detach the temporal retina, which was then grasped by the anterior vitreous frill and peeled back up to the optic nerve to expose the submacular choroid. During the period of follow-up, there was evidence of epiretinal membrane formation but no instance of postoperative retinal detachment. 

Angiography

At 4 and 12 weeks postoperatively, the grafts were assessed with fluorescein and indocyanine green (ICG) angiography. Persistent graft revascularization was, however, seen only in one of the four eyes that had undergone graft translocation with the arterial pedicle intact (Fig. 3). In this eye, two directions of arterial flow were seen arising from the LPCA branch at the center of the graft: one draining to the peripheral choroid superiorly and the other extending temporally. The “clockwise” course of the blood flow was in keeping with the original direction of rotation of the temporal wedge of choroid supplied by the recurrent branch of the LPCA pedicle. Despite clear signs of early revascularization, by week 12 the blood flow through the graft had reduced considerably and only the central area of arterial flow extending superiorly was still evident (Fig. 3). No significant arteriovenous outflow from the graft was seen in any of the other three operated cases with pedicles. Similarly, no significant arteriovenous outflow was seen in the two eyes in which the choroidal graft had no arterial pedicle, and these eyes were not examined further. 

Figure 3.

 

Fundus infrared imaging (A, D) and ICG angiography in the same operated eye at week 4 (A–C) and week 12 (D–F) postoperatively. In the infrared reflectance images, the peripheral margins of the RPE layer of the graft are shown with black arrowheads. (B) and (E) are early images following ICG injection, whereas (C) and (F) were taken after several minutes. At week 4, two areas of arterial flow are seen within the graft, extending radially from the region of the vascular pedicle toward the graft margins. By week 12, flow was reduced and evident only in a single region extending superiorly from the vascular pedicle.

Histology

Histology at 4 to 6 months allowed a more detailed analysis of the retinal changes resulting from choroidal graft surgery. The region overlying the center of the graft was compared with the unoperated nasal retina, which acted as an internal control for the effects of inflammation and silicone oil tamponade. A section from the extreme temporal periphery overlying unoperated choroid acted as a control for the effects of retinal detachment and heavy liquid contact with photoreceptors during the surgical procedure (Fig. 4). The unoperated eye was also used as a further control for the effects of complex surgery on the eye. 

Figure 4.

 

Immunohistochemistry from control and operated eyes. The operated eyes include representative sections from the nasal retina (internal control), extreme temporal retina over nongrafted choroid (control for retinal detachment and heavy liquid), in addition to the retina overlying the graft. Staining for GFAP shows widespread Müller cell activation, including nasal retina (A–D). Bipolar cell marker PKC alpha shows preservation of these cells despite loss of photoreceptors overlying the graft (E–H). Cones survive in all regions (I–L) but only in the fovea in the retina overlying the graft as evidenced by cone sheath staining with PNA. Cone opsin staining is reduced in the temporal retina, which may reflect changes induced by retinal detachment /heavy liquid (K), but is present although mislocalized in the fovea overlying the graft (L). Blood vessels stained with the RCA lectin (M–P) are also clearly seen in the choroidal graft (P). RPE visual cycle function as determined by RPE65 activity (Q–S) is not seen in the pigmented RPE-like cells overlying the graft (T). Scale bar = 50 microns.

In general, loss of the outer nuclear layer (ONL) was observed in regions overlying the center of the graft, but ONL preservation was seen in the temporal periphery (Fig. 4). This confirmed that photoreceptor loss was most likely attributable to the specific effects of the lack of blood flow through the graft, rather than the retinal detachment and/or use of heavy liquid during surgery. Müller cells were activated throughout the operated retina including the nasal region as evidenced by widespread increase in GFAP (Figs. 4A–D). The presence of PKC alpha staining confirmed that the residual retinal layer seen in the grafted regions was predominantly composed of preserved bipolar cells (Figs. 4E–H). Cones could be clearly identified in the temporal retina through cone opsin staining and preservation of cone sheaths with PNA (Figs. 4I–L). Here opsin staining was reduced in the extreme temporal retina (Fig. 4K), which may reflect some damaging effects of heavy liquid or the creation of retinal detachment, as this region was reattached to nonrotated RPE-choroid. Most regions of the retina overlying the graft were devoid of cone staining, although a few surviving cones could be identified in the foveal region in one retina, but with short outer segments and mislocalized opsin (Fig. 4L). Lectin staining with RCA was used to identify blood vessel morphology at the microscopic level (Figs. 4M–P). Despite the angiographic findings, this showed preserved blood vessels throughout the graft, although without the characteristic overlying layer of RPE that was seen in nonoperated regions (Fig. 4). Although pigmented cells could be identified in the grafted area, these did not express RPE65, which is in contrast to other regions of the retina and suggests that any RPE cells overlying the rotated graft had become severely de-differentiated as result of the surgery (Figs. 4Q–T). 

Because there was preserved microvasculature in the rotated choroidal graft but no obvious overlying RPE, it was of interest to study the intervening BM in more detail. Using PAS staining with Alcian blue, the BM could be clearly identified in all operated and control regions (Fig. 5). In regions within the graft, the BM staining could frequently be seen without overlying RPE (Fig. 5D), in other regions the overlying RPE formed a continuous monolayer (Fig. 5E). In some areas discontinuous RPE cells could be seen but linked by a continuous underlying BM (Fig. 5F). This confirmed the preservation of BM and also identified that some RPE cells had survived, despite losing RPE65 immunoreactivity. 

Figure 5.

 

Specific analysis of BM with PAS and Alcian blue staining shows control regions (A–C) and the retina overlying the graft (D–F). The preserved BM is marked by white arrowheads. RPE cells are lost in some regions overlying the graft (D), whereas a continuous monolayer was seen elsewhere (E). Isolated RPE cells separated by gaps could also be identified on continuous BM in some regions (F). ONL, outer nuclear layer; OS, outer segments; RPE, retinal pigment epithelium. Scale bar (A) = 20 μm.

To examine the overall structure of the preserved BM in more detail, scanning electron microscopy was used to view the retinal surface of the choroidal graft in one animal (Fig. 6). This showed generalized loss of RPE cells but good preservation of BM across the entire graft. The position of the tight junctions that would have previously anchored RPE cells could be identified as hexagonal outlines on the otherwise smooth BM surface (Fig. 6C). 

Figure 6.

 

Scanning electron microscopy of a graft 4 months after surgery. The position of the rotated graft at the end of surgery is shown (A). This can be compared directly to the fixed eye cast with the retina removed used for scanning (B) and the regions selected for analysis. The residual BM (C) has polygonal shapes representative of the attachments of the previous RPE cells. This morphology matches the size and shape of RPE cells, which can be seen for comparison on the nasal area (D). ON, optic nerve. Scale bars: (B) = 200 μm, (C, D) = 20 μm.

Discussion

This study demonstrates that despite the high blood flow within the choroid,⁴³ it is technically feasible to create and manipulate full-thickness retina-RPE-choroidal grafts on their deep arterial blood supply in vivo. The results showed, however, that revascularization of the grafts postoperatively was poor, resulting in degeneration of the overlying retina and RPE, but relative preservation of BM. 

The finding that BM was a relatively robust structure and could be transplanted from the periphery is of potential relevance for stem cell approaches toward RPE replacement. Unless transplanted replacement RPE cells are able to bind to a suitable substrate, their survival and differentiation is short-lived,⁴⁴ which is a hurdle that will need to be overcome before clinical trials in this field of research.³⁷ BM with its pentilaminar structure, specific permeability properties, and an embryological origin involving two germ layers is a complex structure to regenerate from stem cells in vitro. Our results, however, suggest that autologous BM harvested from the temporal retina may provide a natural nonsynthetic alternative, and thus facilitate the application of stem cell approaches to RPE replacement. A further implication of our results is that it is almost certain that an RPE monolayer would not survive if the underlying choroidal blood flow is compromised. Hence, for conditions such as dry atrophic AMD, restoration of adequate choroidal blood flow should be a prerequisite for successful RPE replacement. 

The key question from the results, however, is why the choroid does not become revascularized in the way one might expect, given that it was rotated around an arterial pedicle and left in contact with the residual choroid at the circumscribed margin. It is notable that the early revascularization became less at later stages. Normally one would expect the revascularization of transplanted tissue to improve over time as capillaries open up and create new anastomoses. The behavior of the choroid in this regard is surprising, particularly as the tissue itself has an intrinsic ability to regenerate new vessels in response to ischemia, as evidenced by the exudative form of AMD. In the experiments we have performed, it may be that the natural elasticity of the choroid is causing peripheral contraction toward the central arterial pedicle, which is in conflict with any peripheral attempts to regenerate bridging vessels so as to reestablish the venous outflow at the graft margin. Clearly, the biphasic arterial blood flow through the vascular pedicle alone was insufficient to maintain adequate choroidal graft perfusion. 

A previous clinical study identified that smaller pedicled flaps of RPE-choroid could be mobilized in the submacular space and initially maintained blood supply, but later this was lost.^21,26 In large series of free RPE-choroid grafts harvested from the periphery, most grafts are found to revascularize.^{24–25,31–32,45} This revascularization is predominantly horizontal, extending from unprocedured choriocapillaris, and is sufficient to sustain grafts transplanted into areas of chorioretinal atrophy.⁴⁶ With spectral domain OCT and angiography, afferent and then efferent vessels can be seen to form between free grafts and neighboring unoperated choroid, resulting in changes in graft thickness, in the days following surgery.⁴⁵ It is therefore likely that a high degree of contact between the graft and nonprocedured choroid is required to promote revascularization. The elastic nature of the choroid means there is a tendency toward graft shrinkage. Of note, though, all grafts in our series had at least 180° of peripheral contact with the nasal choroid at the end of the procedure and this was maintained throughout the follow-up period. 

Alternatively, we speculate that there may have been adequate oxygenation in the subretinal space and as a result insufficient ischemic drive for revascularization. The mechanism of RPE cell death could be attributable to disruption of the normal vascular trafficking of other nutrients or minerals critical for RPE function. Death of RPE cells leads to secondary atrophy of the underlying choroid, as evidenced by the clinical phenotype of choroideremia and also the recently described dominant mutations in RPE65.⁴⁷ Loss of the RPE cell function is also known to lead to secondary degeneration of photoreceptors. Hence, one possible explanation for the observed choroidal shrinkage and ONL atrophy is that it is attributable to primary dysfunction of the RPE cells overlying the graft. This is supported by our observation of preserved RPE cells without enzymatic activity. Similarly, the survival of foveal cones overlying the center of the graft may be because these cells can use Müller cells rather than the RPE for metabolism of the visual cycle.⁴⁸ Our study differs from the clinical studies in that most patients who previously underwent RPE-choroid graft surgery would have had advanced neovascular AMD with an abundance of endogenous VEGF perioperatively. It is notable that free RPE-choroid grafts placed over atrophic regions in dry AMD, which do not have active CNV, do not revascularize unless the underlying BM is incised to promote blood vessel ingrowth.⁴⁶  

The technique in its current form demonstrated that the highly vascular choroid can be manipulated and rotated in vivo around an arterial pedicle, but provided no evidence that rotational grafts are likely to improve the outcomes presently achieved with free graft techniques. 

Supplementary Materials

MOV S1 

Acknowledgments

The authors thank Jane Olver for the electron microscopy images in Figure 1, Graham Nunn for technical assistance with vitreoretinal surgical equipment, Steven Hughes for histology advice, and Rosalie Crouch (Medical University of South Carolina) for the anti-RPE65 antibody. 

Footnotes

 Supported primarily by the Medical Research Council UK (G0601588 awarded to REM, GWA, and AMS), assisted by the National Institute for Health Research Biomedical Research Centre, the Royal College of Surgeons of Edinburgh, the Special Trustees of Moorfields Eye Hospital, the Oxford Stem Cell Institute, The Health Foundation and Medical Research Council Grant G0701535.

Footnotes

 Disclosure: E. Lee, None; M.S. Singh, None; H.E. Jones, None; B. Ahmed, None; I.M. Andolina, None; J.T.C. Clements, None; V. Luong, None; P.M. Munro, None; M.P. Lawton, None; K.L. Grieve, None; G.W. Aylward, None; A.M. Sillito, None; R.E. MacLaren, None