Age-related macular degeneration (AMD) remains a major cause of blindness, with dysfunction and loss of retinal pigment epithelium (RPE) central to disease progression. We engineered an RPE patch comprising a fully differentiated, human embryonic stem cell (hESC)–derived RPE monolayer on a coated, synthetic basement membrane. We delivered the patch, using a purpose-designed microsurgical tool, into the subretinal space of one eye in each of two patients with severe exudative AMD. Primary endpoints were incidence and severity of adverse events and proportion of subjects with improved best-corrected visual acuity of 15 letters or more. We report successful delivery and survival of the RPE patch by biomicroscopy and optical coherence tomography, and a visual acuity gain of 29 and 21 letters in the two patients, respectively, over 12 months. Only local immunosuppression was used long-term. We also present the preclinical surgical, cell safety and tumorigenicity studies leading to trial approval. This work supports the feasibility and safety of hESC-RPE patch transplantation as a regenerative strategy for AMD.


Human ESCs represent a promising source for cellular replacement therapies owing to their availability, pluripotency, and unlimited self-renewal capacity. However, they also carry risks of neoplastic change, uncontrolled proliferation, and differentiation to inappropriate cell types1,2. The eye is advantageous in investigating hESC-based cell therapy as it is accessible and confined, and the transplanted cells can be monitored directly in vivo, with the possibility of being removed or destroyed if there is evidence of neoplastic change3,4. Furthermore, long-term immunosuppression can be delivered locally.

Late AMD is characterized by irreversible cell loss, initially of RPE cells and subsequently of neuroretinal and choroidal cells5, and thus may be amenable to hESC-based cell therapy4. The disease process includes damage to the RPE's specialized basement membrane, Bruch's membrane5. Currently, treatments exist only for the exudative or 'wet' form of AMD. These treatments rely on angiogenesis inhibitors6 or indirect transplantation of an autologous RPE–Bruch's complex (retinal translocation surgery)7. However, the former treatment only suppresses the disease, requiring long-term repeat delivery, and the latter, although restoring the macular anatomy, does not prevent disease recurrence. There is no treatment for atrophic 'dry' AMD, which is characterized by RPE loss and progressive neuroretinal cellular dysfunction.

Suspensions of hESC-derived RPE (hESC-RPE) cells have been transplanted in human subjects with dry AMD and Stargardt's disease, but the extent of cell survival and restoration of vision remains ambiguous8. A recent, single-patient report described transplantation of an autologous induced pluripotent stem cell (iPSC)–derived RPE patch on its own secreted basement membrane9. The iPSC-RPE survived with maintenance, but no improvement, of visual acuity at 12 months. We developed a therapeutic, biocompatible hESC-RPE monolayer on a coated synthetic membrane, herein termed a 'patch', for transplantation in wet and early-stage dry AMD. The choice of membrane material and its preparation, including the human vitronectin coating, has not been described previously to our knowledge. In contrast to RPE suspensions, cells on the patch are delivered fully differentiated, polarized, and with the tight junction barrier formed, that is, in a form close to their native configuration. The synthetic membrane allows the patch to be handled easily and robustly. The main disadvantage of the patch is that it requires a purpose-built delivery tool and a more complicated surgery compared to cell suspensions, and the use of hESCs may require immunosuppression, unlike an autologous cell source. Our delivery tool (Supplementary Fig. 1) confers the benefit of protecting the patch within the tip, thereby minimizing cell loss, cell distribution within the eye, and physical damage to the RPE monolayer.

The clinical trial was designed as a phase 1, open-label, safety and feasibility study of implantation of an hESC-RPE patch in subjects with acute wet AMD and recent rapid vision decline. For safety reasons and to obtain an early efficacy signal, the trial involved patients with severe wet AMD only, although we aim to study the RPE patch in early dry AMD in the future. We reported three serious adverse events to the regulator. These were exposure of the suture of the fluocinolone implant used for immunosuppression, a retinal detachment, and worsening of diabetes following oral prednisolone. All three incidents required readmission to the hospital, with the first two incidents requiring further surgery and the third being treated medically. The three incidents were treated successfully. Both patients achieved an improvement in best-corrected visual acuity (BCVA—vision with an optometrist-determined best glasses-corrected vision) of >15 letters at 12 months after transplantation.


Engineering an RPE patch

Our protocol for manufacturing a clinical-grade RPE patch is summarized in Figure 1. RPE cells were differentiated from the SHEF-1.3 hESC line, which had been derived from the SHEF-1 hESC line. We used a spontaneous differentiation method10, an approach that has been described earlier by others and has been applied to primate11 and human12,13 ESCs. Spontaneous differentiation limits the need for additional factors, facilitating compliance with good manufacturing practice (GMP) guidelines, and with clear advantages over vector-driven methods in terms of the safety and stability of the final RPE. Our differentiation protocol differs from another similar protocol10 in that we used Essential 8 medium rather than human feeder cells during expansion and the early phase of differentiation to avoid exposure to another cell type and to fulfill GMP guidance. While several methods are used to produce hESC-RPE11,13,14,15, because of epigenetic variation from the earliest stages after derivation of hESCs, no method is universally effective15,16. With our protocol, 20–30% of hESC colonies differentiated to RPE (data not shown), which is a similar rate to that of other spontaneous methods11,13 but less efficient than some augmented methods15,16,17.

Figure 1: Generation of hESC-derived RPE for the manufacture of an advanced therapeutic medicinal product (ATMP) to treat AMD.
Figure 1

This figure outlines the chronological process for manufacturing the RPE patch from SHEF-1.3 hESCs. The first four rows show: the stages of differentiation; the coating on the plasticware for each step; the media used at each step; and the tests and checks performed at each step. The bottom row shows brightfield images illustrating each stage: (i) hESC colonies expanded on recombinant human vitronectin (VTN-N); (ii) Spontaneously differentiated RPE cells appear as distinct pigmented foci; (iii) These foci are manually dissected, dissociated, and filtered to achieve a pure single-cell-type RPE population, which is seeded onto plates, where expanding RPE cells establish their classic pigmented, cobblestone morphology; (iv) Fully differentiated RPE-seeded at confluence onto a human-vitronectin-coated PET membrane transwell insert; (v) A therapeutic patch consisting of the RPE monolayer immobilized on the vitronectin-coated PET membrane that has been cut from the transwell; (vi) Following release tests, the advanced therapeutic medicinal product is supplied to the surgical team in a custom-manufactured, single-use sterile container in saline, where it is viable for up to 8 h. PMEL17, premelanosome marker 17; ICC, immunocytochemistry; ISH, in situ hybridization; PET, polyethylene terephthalate; VIR test, visual inspection release test.

Subsequent RPE characterization (Fig. 2) used immunocytochemistry (Fig. 2a), electron microscopy (Fig. 2c), pigment-epithelium-derived factor (PEDF) secretion profile testing (Fig. 2b), and a functional phagocytosis assay (Fig. 2d). During manufacturing, in-process testing, including in situ hybridization with a specific oligonucleotide probe for LIN28A mRNA12, was undertaken to detect hESC impurity at the single-cell level. Differentiated RPE was discarded if any LIN28A-positive cells were detected. The hESC-RPE cells were seeded at confluence onto a human-vitronectin-coated polyester membrane. A 6 × 3 mm (17 mm2) therapeutic element with 100,000 cells (PF-05206388 in the regulatory documentation) was cut to size with a purpose-built punch and loaded into a sealed transport container (Fig. 1). Immediately before release for human transplantation, the cell layer was recharacterized as RPE by means of a visual inspection release test encompassing cell dose, cell identity, and patch coverage checks.

Figure 2: Characterization of hESC-derived RPE.
Figure 2

(a) Confocal brightfield immunofluorescent micrographs depicting staining for typical regional RPE markers as part of the characterization of the monolayer as RPE (each represents a single experiment with no repeats). Scale bars, 25 μm. Predominantly apical PEDF and basal BEST1 expression confirms the polarity of these cells. PMEL17 expression confirms presence of premelanosomes, and ZO1 expression allows easy identification of tight junctions. Absence of Ki67 confirms that cells are not proliferating. Other characteristic RPE markers (CRALBP, MITF, OTX2), crucial for melanogenesis and visual cycle, are also present. Immunocytochemistry was performed once on hESC-RPE cells. BEST1, bestrophin 1; CRALBP, cellular retinyladehyde-binding protein; DAPI, diamidino-2-phenylidole; MITF, microthalmia transcription factor; OTX2, orthodenticle homeobox 2; PEDF, pigment epithelium-derived factor; PMEL17, premelanosome protein 17; ZO1, zonula occludens 1. (b) Quantification of PEDF secretion in spent culture medium during late expansion phase, analyzed using an ELISA assay. The classical RPE PEDF secretion asymptote is visible from around 3 weeks onwards. Each different colored box represents a separate batch of tested cells. (c) Electron micrographs of RPE cells, illustrating the classical ultrastructure associated with normal RPE function including tight junctions (white arrows; i and iii), basal infoldings (black arrows; i and iv), apical microvilli (i–iii), and melanin granules (i–iii), revealing various stages of melanogenesis (ii; stage II/III, III melanosomes and stage IV mature melanosomes). Transmission electron microscopy was performed once on hESC-RPE cells. N, nucleus; AmV, apical microvilli. Images represent a single experiment with no repeats. (d) hESC-RPE cells internalize photoreceptor outer segments (POS), demonstrating hESC-RPE phagocytosis. The extent of phagocytosis was also examined after treatment of the hESC-RPE with MERTK antibody (hESC-RPE + blocker). Data shown are mean internalized POS ± s.e.m. (n = 3 biologically independent cell cultures in each group). Pre-incubation with MERTK antibody had a significant effect on the number of POS ingested by hESC-RPE (P = 0.0000103075196107, two-tailed Student's t-test). This was a single experiment with no replication. Below the graph, confocal micrographs show the internalized POS with a green fluorescent marker (POS) in hESC-RPE and its relative absence in hESC-RPE with MERTK antibody (hESC-RPE + blocker).

Mouse teratoma and in vitro cell-spiking studies

We tested tumorigenicity of the hESC-RPE in NIH III nude mice (Fig. 3a–e). Cell suspensions were used as the mouse eye is too small to administer a patch. Two initial studies conducted with undifferentiated hESCs showed that teratomas could form. In these studies, cell suspensions were injected subretinally, intramuscularly and subcutaneously, with mice followed for 26 weeks. Subsequently, we studied the tumorigenicity of hESC-RPE cells under good laboratory practice (GLP) conditions, including a positive control group injected with undifferentiated hESCs. Given our finding that undifferentiated hESCs formed teratomas, we conducted in vitro cell-spiking studies to examine whether hESCs survive the RPE manufacturing process and culture-seeding conditions.

Figure 3: Preclinical mouse teratoma and pig transplantation studies.
Figure 3

(a) Normal non-injected eye of a mouse (H&E stain). (b) 6-week-old animal injected with hESC-RPE (no teratomas or proliferation were observed in any animals from this group). Section taken 26 weeks after injection (H&E). (c) An example of teratoma formation in an animal injected with undifferentiated hESC. Section taken following euthanization of the animal prior to 26 weeks (H&E). (d) An example of a mesenchymal tumor NOS in an animal injected with undifferentiated hESC. Section taken at termination of the animal prior to 26 weeks (H&E). (e, i) 6-week-old animal injected with hESC-RPE showing persistence of layers of pigmented cells at week 26. (e, ii) 6-week-old animal injected with hESC-RPE and examined with anti-human mitochondria IHC. Immunoreactivity for human mitochondria seen as a brown chromophore (black arrows), demonstrating that the cells represent implanted-hESC-derived RPE. (e, iii) 6-week-old animal injected with hESC-RPE and examined with an isotype control relative to section in e, ii with no positive staining (black arrows) using the same chromophore as in e, ii. (f,g) In vivo, color fundus photographs of pig eyes from the GLP pig study taken 6 weeks after transplantation surgery. (f) The picture shows an eye that has been implanted with a patch consisting of an hESC-derived RPE monolayer on a human-vitronectin-coated polyester membrane. The patch is seen as a consistent and evenly pigmented cellular covering over the entire area. (g) Uncoated patch with no cells, which has been implanted as a control. (h) Cresyl violet stain of pig retina at 6 weeks post-transplantation with a patch of RPE implanted into the subretinal space. INL, inner nuclear layer; ONL, outer nuclear layer. (i) Cresyl violet stain of pig retina at 6 weeks post-transplantation with a patch without cells implanted into the subretinal space. GCL, ganglion cell layer; INL, inner nuclear layer. Note the absence of the ONL. (j) Unstained pig retina 6 weeks post-transplantation with a patch of RPE cells which are pigmented (arrows highlight the position of the polyester membrane). (k) Serial section from the same eye stained with anti-human TRA-1-85 indicates that the cells are human in origin (arrows highlight the position of the polyester membrane).

In the first study, injection of 4.5–8.8 × 104 undifferentiated hESCs into the subretinal space of NIH III mice resulted in localized neoplastic formation in almost half the males injected (12/30) but in very few females (2/30). The tumors showed evidence of pluripotency and appeared to be composed of mesenchymal and epithelial lineages (data not shown).

In the second study of undifferentiated hESCs, neoplastic masses were observed in the injected eye of 12/15 female and 5/5 male mice after cell administration into the subretinal space. Teratomas were observed microscopically in the thigh of all female mice that received 3.6 × 104 or 8.23 × 105 undifferentiated hESCs in BD Matrigel by the intramuscular route, and in the left flank of 2/5 and 1/11 female mice that received 3.6 × 104 or 8.23 × 105 undifferentiated hESCs in BD Matrigel, respectively, by the subcutaneous route. Masses were composed of structures derived from all embryonic germ layers (data not shown).

In the third study, conducted with cells manufactured under GLP conditions, suspensions of hESC-RPE cells, or of undifferentiated hESCs as a positive tumorigenic control, were injected subretinally in NIH III mice. In a total of 80 mice injected with suspensions of hESC-RPE cells, no teratomas were detected (Fig. 3b). Pigmented hESC-RPE cells were observed lining the surface of the retina or lens in the injected eyes of some mice given 6.04 × 104 hESC-RPE cells (Fig. 3e, i,ii,iii) at 26 weeks.

In the same study, administration of undifferentiated hESCs was associated with ocular teratoma formation in four males and five females given 4.28 × 104 cells, all of which were prematurely euthanized (between days 46–62), and in a single male given 4.51 × 103 cells, which survived to the 26 week end point of the experiments. (Fig. 3c). In addition, mesenchymal tumors classified as “Not Otherwise Specified” were present in two males given 4.28 × 104 undifferentiated hESCs (Fig. 3d) and perilenticular mesenchymal hyperplasia in animals of both sexes given 4.51 × 103 cells or 4.28 × 104 cells. All tumors were composed of human cells, as confirmed by immunohistochemistry using an anti-human mitochondrial marker (data not shown). Eyes injected with undifferentiated hESCs also exhibited an increase in the incidence and severity of non-proliferative changes, including lenticular degeneration, posterior synechiae of the iris, and retinal detachment (data not shown). Transplantation of either differentiated or undifferentiated hESCs did not affect body weight.

Given the tumor formation observed with undifferentiated hESCs, we aimed to determine whether hESC-RPE contained any residual pluripotent cells or could support the survival of pluripotent cells, as this would constitute a teratoma risk. Using immunostaining and flow cytometric analysis of cells expressing the pluripotency marker Tra-1-60 (Abcam 16288, Cambridge, UK), we detected no undifferentiated hESCs in dissociated RPE foci before the expansion phase or at 6 to 8 weeks after seeding at the end of the expansion phase (Supplementary Fig. 2). We spiked undifferentiated hESCs into the RPE at 1, 10, 20, or 50% of total cells at the start of the expansion phase. This did not affect the quality of the resulting 6-to-8-week-old cultures at the end of the expansion phase, which showed cobblestone morphology and pigmentation, and stained strongly for PMEL17 (an RPE marker that indicates the presence of premelanosomes). By 2 d post-spiking, viable Tra-1-60-positive cells were no longer detectable by flow cytometry (data not shown). This finding was supported by propidium iodide staining, which showed that 96% of hESCs died when dissociated and seeded in the same manner as RPE prepared for expansion (Supplementary Fig. 3). The cells that survived in some experiments did not have hESC-like morphology and failed to stain for Tra-1-60 or the proliferation marker Ki67, suggesting that they had differentiated or senesced (Supplementary Fig. 3d,e).

Pig single-dose studies

Studies of surgical feasibility and safety of delivery of the RPE patch as well as studies of local and systemic biodistribution and toxicity were carried out in pigs (Fig. 3f–k). Human and pig eyes are similar in size, which allowed for administration of the full-size patch. Two studies were performed; the second used the same clinical surgical technique as in the human clinical trial. Biodistribution of hESC-RPE cells was evaluated in more than ten sites using qPCR detection of multiple human cell markers.

In the first pig study, patch delivery was successful in all 20 animals, as assessed by intraoperative observation of the patch under the neural retina and histological sections showing the patch in the subretinal space (Fig. 3h,i). No retinal detachments were noted in any animal. We found surviving human cells by light microscopy at 2 h, 2 d, and 1, 2, 4, and 6 weeks after implantation (RPE patch: n = 12; control patch without cells: n = 8; figure 3 H-K). Human cell survival was demonstrated despite immunosuppression having been limited to the peri-operative period using oral prednisolone. Surviving cells remained pigmented (Fig. 3f,j). Cells expressed RPE-specific cell markers, showed no proliferation activity, and did not migrate away from the membrane (data not shown). Photoreceptor survival above the patch was observed only in animals that received an RPE patch, as assessed by histology (Fig. 3h). The control membrane without cells did not support photoreceptor survival (Fig. 3i). Microscopically, retinal architecture was intact in most animals receiving human cells. Macrophages were seen and appeared to have phagocytosed some transplanted RPE cells to form large pigmented cells (data not shown). Lymphocytic infiltrate was not observed except in 2 animals at 2 and 6 weeks. Animals that received the control membrane also had macrophage infiltrates but no lymphocytic response (data not shown). These results demonstrated that patch transplantation with hESC-RPE survival was surgically feasible without significant safety issues.

The second pig study differed from the first in that we used the purpose-built surgical tool, which has a protective cradle at the tip from which the patch is mechanically pushed out when the tip is in the subretinal space (Supplementary Fig. 1). The study was undertaken under GLP conditions, a standardized set of requirements to ensure quality and systematic management of the experiment. In this case, it dictated standards for the operating and animal facilities and for the personnel who ran the trial, and defined the conditions and requirements for the keeping and monitoring of animals in the study, including the feeding regimen. Correct surgical delivery of ten RPE patches and ten coated membranes without RPE was achieved in 20 pigs as assessed by clinical examination and histology (Fig. 3f,g). Good cell cover was observed at the 6-week time point in the ten animals that received RPE cells (Fig. 3f). Retinal detachment and rupture of the lens capsule were observed in one male and one female implanted with the patch. There was no evidence of weight loss or early death in any animal. At 6 months after implantation, no hESC-RPE cells were detected at the implantation site or elsewhere in H&E sections of the ten eyes receiving the RPE patch. Anti-human TRA-1-85 immunostaining of the operated eye and optic nerve from these animals was also negative. Microscopic findings of chronic inflammation were seen restricted to the subretinal implantation site, at 6 months in the implanted (left) eye of all animals, and were consistent with the intraocular surgical implantation procedure. The microscopic findings included fibrosis, osseous metaplasia, and small numbers of macrophages and multinucleate giant cells (data not shown). Atrophy of the photoreceptor layer in overlying retina was also present (data not shown). Findings were of similar incidence and severity in both control and cell-treated animals. No positive amplification of any human-specific genes was observed in adrenal, bone marrow (rib and femur), brain, heart, kidneys, liver, lungs, lymph nodes, optic nerve, spleen, or thymus, which were evaluated by qPCR, indicating that the cells did not appear to migrate or survive away from the site of implantation. The second pig study confirmed the surgical feasibility and reliability of the delivery tool and the lack of systemic or local distribution of the hESC-RPE cells.

Clinical trial

Regulatory permission was granted for a phase 1 clinical trial on the basis of the preclinical surgical, safety and tumorigenicity studies reported in this article and published data on the hESC-RPE mono-layer9 (Clinicaltrials.gov: NCT01691261). Permission was granted for ten patients, and we report the primary and secondary outcomes from the first two (Figs. 4,5,6,7). In addition to safety, the trial investigated whether the synthetic membrane would facilitate mechanical delivery of the RPE monolayer, whether the transplanted cells could be sustained long-term with local immunosuppression only, and whether early signals of potential efficacy were evident.

Figure 4: Case 1.
Figure 4

Pre- and post-operative imaging of the cell patch with structural and functional outcomes at 12 months after surgery. (a, i) Pre-operative color fundus photograph demonstrating the extensive sub-RPE and submacular hemorrhage extending under the fovea secondary to wet AMD. (a, ii) Spectral domain OCT (Spectralis, Heidelberg) showing pre-operative section with inset showing the position of the slice. (b, i) Post-operative color fundus photograph showing the patch covered with pigmented cells throughout. The test also shows microperimetry results with two areas of demonstrated sensitivity (Nidek microperimetry) labeled x and y. The circles represent the extent of the visual stimulus demonstrating that it falls completely within the transplanted area. Inset § shows a magnified detail of the patch area with the levels of sensitivity shown by microperimetry in decibels. Inset * shows a close up of the fixation with the Nidek microperimeter fixation outcome superimposed. (b, ii) Spectral domain OCT (Spectralis, Heidelberg) of post-operative section with inset showing the position of the slice. The patch is shown in position with thinned retina over the area of the patch but an intact ellipsoid layer. (c, i,ii) Spectral domain OCT sections through x and y, respectively. The area on the section corresponding to x and y is indicated with a bar. The long red arrow in c, i and c, ii indicates the ellipsoid layer. The long and short green arrows in c, i and c, ii indicate the synthetic membrane and the hESC-RPE, respectively. (d, i) Images from the Rtx1 adaptive optics (AO) fundus camera. (Inset) Color fundus photo with superimposed area of AO imaging. The dotted yellow circle corresponds to the area of positive microperimetry sensitivity marked x throughout this figure. Magnified image of the area indicated by the dotted yellow line, showing bright dots corresponding to cone photoreceptors based on their size, inter-photoreceptor distance, and organization (blue arrows). (d, ii) AO image taken from a normal subject at the same point of the retina as in d, i for comparison. (e, i) Early-phase fundus fluorescein angiogram (FFA) showing early choroidal perfusion in the area of the patch and specifically the two points of demonstrated sensitivity x and y. The dark areas over the transplanted patch, at the nasal end of the patch and the superior border, represent areas of native RPE over and under the patch, masking emission rather than absent choroidal perfusion. (e, ii) Late-phase FFA confirming choroidal perfusion and no leakage under the patch. (e, iii) Scanning laser ophthalmoscope autofluorescence over the patch, including in areas x and y, showing dim but visible autofluorescence.

Figure 5: Case 2.
Figure 5

Pre- and post-operative imaging of the cell patch with structural and functional outcomes at 12 months after surgery. (a, i) Preoperative color fundus photograph of the extensive sub-RPE and submacular hemorrhage extending under the fovea, secondary to wet AMD. (a, ii) Spectral domain OCT (Spectralis, Heidelberg) showing pre-operative section with inset showing the position of the slice. (b, i) 12-month post-operative color fundus photograph showing the patch covered with pigmented cells throughout. There is extensive epiretinal fibrosis associated with the extent of the disease and a retinal detachment that had been successfully reattached. The test also shows microperimetry (Nidek microperimetry) results with two areas of demonstrated sensitivity labeled x and y. Inset § shows a magnified detail of the patch area with the levels of sensitivity shown by microperimetry in decibels. Inset * shows a close-up of the fixation with the Nidek microperimeter fixation outcome superimposed. The circles represent the extent of the visual stimulus, demonstrating that it falls completely within the transplant area. (b, ii) Spectral domain OCT (Spectralis, Heidelberg) showing post-operative section with inset showing the position of the slice. (c, i,ii) Spectral Domain OCT cuts through area x and y, respectively, as indicated. The detail shows the Spectral Domain OCT sections through x and y, respectively. The area on the section corresponding to x and y is indicated with a bar representing the area. The long red arrows indicate the ellipsoid layer. The long and short green arrows indicate the synthetic membrane and the hESC-RPE respectively. (d, i) Images from the Rtx1 AO fundus camera. Inset of color fundus photo with superimposed area of AO imaging. The magnified image of the area indicated shows bright dots corresponding to cone photoreceptors based on their size, inter-photoreceptor distance and organization. (d, ii) An Rtx1 AO fundus camera image taken from a normal subject at the same point of the retina as in d, i for comparison. (e, i) An early-phase fundus fluorescein angiogram (FFA) showing early choroidal perfusion in the area of the patch and particularly the two points of demonstrated sensitivity x and y. The dark areas temporal and superior to the transplanted patch likely represent areas of choriocapillaris loss associated with the primary pathology and surgery. (e, ii) Late-phase FFA confirming choroidal perfusion and no leakage under the patch. (e, iii) Scanning laser ophthalmoscope autofluorescence image over the patch including areas x and y demonstrating autofluorescence over parts of the patch.

Figure 6: Sequences of color photographs of the transplanted patch in cases 1 and 2.
Figure 6

(a) The sequence in patient 1 at 4, 24, and 52 weeks shows the centrifugal expansion of the pigmented areas around the patch and the stability of pigmented areas on the patch. It also shows the regression of the host RPE on the patch, with the original margin indicated by the dotted line. (b) The sequence in patient 2 at 4, 24, and 52 weeks shows the centrifugal expansion of the pigmented areas around the patch and relatively stable areas of pigmentation on the patch. The expansion of pigmented areas in both 6a and b stabilized by week 24.

Figure 7: Best-corrected visual acuity and reading speed in cases 1 and 2 (a) BCVA over 12 months for patient 1.
Figure 7

(b) Reading speed over 12 months for patient 1. (c) BCVA over 12 months for patient 2. (d) Reading speed over 12 months for patient 2.

Using the surgical delivery tool, we placed one RPE patch in the subretinal space, under the fovea, in the affected eye of each patient. Correct placement was confirmed in both patients by stereo-biomicroscopy, fundus photography, and spectral domain optical coherence tomography (SD-OCT) (Figs. 4b, i,ii and 5b, i,ii).

In patient 1, OCT and native-level autofluorescence immediately after surgery showed a 'double thickness' of RPE in one area nasal to the fovea, indicating overlap of native RPE and the patch (Fig. 4b,c,e). In all other areas in patient 1, and in all areas in patient 2, a single layer of RPE on OCT suggested there was no residual native RPE over the patch (Figs. 4b, ii and 5b, ii). In both patients, hESC–RPE was present over the full area of the patch at 12 months as evidenced by dark pigmented cells covering the patch, although unevenly, photos of the fundus, and a hyper-reflective monolayer on the patch seen by SD-OCT (Figs. 4b, i,ii and 5b, i,ii and 6a, b). In both patients, the patches showed uneven autofluorescence (Figs. 4e, iii and 5e, iii), which suggests functioning RPE phagocytosis18,19. Also, visible in both patients were darker, pigmented areas continuous with the patch, which may represent RPE cell migration off the patch onto adjacent RPE-deficient areas. These areas spread from the patch edge outward over the first 6 months after surgery before stabilization (Fig. 6a,b). The areas were contiguous with the patch RPE signal on OCT and were absent in areas where the native RPE layer persisted. There was no evidence of neoplastic transformation either on regular review of the fundus by an ocular oncologist or by serial ocular ultrasound.

In addition to RPE survival, we studied visual recovery in both cases. Testing was always carried out by an independent qualified observer. The Early Treatment Diabetic Retinopathy Study (ETDRS) letter chart was used to define BCVA, which improved over 12 months from 10 to 39 and from 8 to 29 letters, in patients 1 and 2, respectively (Fig. 7a,c). Microperimetry, a test of perception of microlocalized light stimuli, showed visual fixation at the center of the patch and vision over the patch in both patients at 12 months (Figs. 4b,i and 5b,). Figures 4b,i and 5b,i, show sample areas of visual sensitivity localized totally within the patch. Reading speed improved from 1.7 to 82.8 and from 0 to 47.8 words/min in patients 1 and 2, respectively, over 12 months on the University of Minnesota MNRead test (Fig. 7b,d), an improvement and final level not found in the Submacular Surgery Trial20. Pelli–Robson contrast sensitivity scores (Log) improved from 0.45 to 1.35 in patient 1, and 0 to 1.05 in patient 2 over 12 months. At each point that showed microperimetry sensitivity over hESC-RPE, we observed choroidal filling by angiography (Figs. 4e, i,ii and 5e, i,ii); RPE-autofluorescence (Figs. 4e, iii and 5b, iii) and presence of the ellipsoid layer (indicative of preserved photoreceptors) by SD-OCT (Figs. 4c, i,ii and 5c, i,ii). We note that it was not possible to ascertain whether the ellipsoid zone was present pre-operatively, owing to the poor detail in the pre-operative OCT scans (Figs. 4a, ii and 5a, ii). Rtx1 adaptive optics camera images showed survival of cone photoreceptors in the areas corresponding to areas of sensitivity on microperimetry (Figs. 4d, i and 5d, i).

We reported three serious adverse events that were unrelated to the RPE patch. The first was exposure of the suture of the fluocinolone implant in patient 1, which required conjunctival revision surgery. The other two, both in patient 2, were a worsening of diabetes following oral prednisolone, which was treated medically, and a retinal detachment. The retinal detachment was an asymptomatic, infero-temporal, proliferative vitreoretinopathy (PVR)-associated traction retinal detachment under silicone, which did not extend past the inferior arcade and thus did not affect the implant. It was observed at the 8-week follow-up, with the retina having been completely attached at the 4-week check. It was treated with a single surgery with peeling of the PVR membranes, inferior retinectomy of the peripheral retina (180 degrees), and laser to the retinectomy edge. The silicone oil was retained. The retina was attached at the end of the surgery and has remained attached after subsequent surgery to remove the silicone oil. There was a residual epiretinal band over the posterior pole with some focal macular traction (Figs. 5b, i and 6b), which was not treated.

Ocular pressures were never raised in either patient. No changes of concern were noted in the liver and renal function tests and by liver ultrasound in either patient. On full-field electro-retinography (ERG) recording, there was evidence of a mild but consistent reduction in photoreceptor function at 6 months in both patients with additional consequent electro-oculography (EOG) reduction (in the eyes operated on). The reduction in photoreceptor function persisted in patient 1 but recovered in patient 2 by 12 months.


The results presented here provide an early indication of the safety and feasibility of manufacturing an hESC-RPE monolayer on a synthetic basement membrane and delivering the patch into the subretinal space as a potential treatment for AMD. Our data suggest early efficacy, stability, and safety of the RPE patch for up to 12 months in two patients with severe vision loss from very severe wet AMD.

hESC-RPE on a membrane shows optimized differentiation, polarization, viability, and maturation of the monolayer at the time of delivery that contrasts favorably with delivery of cell suspensions, where the cells are necessarily not in a monolayer and thereby not polarized or fully differentiated. Proper orientation is readily confirmed by the color difference between the white membrane and the pigmented RPE. Cells delivered in suspension may be lost due to reflux through the retinotomy, with potential vitreous seeding, and the cells undergo shear stress and damage when ejected through the delivery cannula21. Furthermore, cells in suspension are required to adhere to and form a monolayer on a damaged native Bruch's membrane, which leads to poorer cell survival and widespread apoptosis22. Previous work showed poor differentiation using suspensions and that RPE mono-layers on membranes appear superior23.

Therapeutic human RPE patch transplantation has been reported using a harvested autologous RPE–Bruch's membrane–choroid patch from the same eye24,25,26 and an induced pluripotent stem cell (iPSC)-derived RPE patch9. The main advantages of our system over these are mechanical ease of handling due to the rigidity of the synthetic membrane. The tool we developed also allowed consistent patch delivery with a small localized retinal detachment over the macula, whereas with the intraoperative harvested autologous technique, half of the entire retina must be reflected to ensure consistent delivery without RPE damage27,28. The availability of an off-the-shelf patch would be especially advantageous and critical in cases of severe wet AMD with sudden vision loss, as described in this study. Treatment is required rapidly29, and an autologous iPSC patch could not be prepared in a suitable time frame for transplantation. It is also possible that the hESC-RPE patch may alter the natural history of the disease as the cells are zero years old, rather than the 60-plus years of the patients, and are not genetically or environmentally predisposed to develop AMD; however, only further investigation can address this. The main disadvantage of our technology is the need for immunosuppression, although for the two cases reported here we have demonstrated that only local immunosuppression was necessary for long-term hESC-RPE survival.

We present the preclinical safety and tumorigenicity studies that supported regulatory approval of our clinical study. The literature suggests that adult human RPE cells30, terminally differentiated hESC-RPE31, and spontaneously immortalized RPE cells32 are non-proliferative cell types and appear to lack the potential to form teratomas. The GLP study of tumorigenicity in NIH III mice showed that the hESC-RPE was not associated with tumor formation or other notable proliferative changes. Injected hESC-RPE cells survived for the full 26 weeks in some animals. Transplantation of undifferentiated hESCs under the same conditions was associated with teratoma formation as well as unclassified mesenchymal tumors, peri-lenticular mesenchymal hyperplasia, and an increase in the incidence and severity of degenerative changes in the treated eye.

Owing to tumor formation by undifferentiated hESCs, a major safety concern was the potential for survival and persistence of undifferentiated hESCs through the manufacturing process. In-process testing for undifferentiated cells is essential in the manufacture of any cell product from pluripotent cells for human transplantation. Spiking studies and single-cell labeling studies demonstrated no detectable undifferentiated cells in the final product. Even when we contaminated primary foci of hESC-RPE cells with up to 50% undifferentiated hESCs, no pluripotent hESC cells were present by the end of the expansion phase. Furthermore, hESCs were not viable when dissociated and seeded into RPE expansion conditions. Thus, we demonstrated that undifferentiated hESCs were not detectable at stage 4 of the production process, and that the RPE differentiation medium does not support the survival of hESCs.

Our preclinical studies in pigs investigated surgical feasibility, biodistribution, and toxicity. We showed consistent facilitated mechanical delivery of the RPE patch in all 20 pigs operated on in the GLP final study using our purpose-built tool. Implantation of the control membrane without cells led predictably to a foreign body reaction, but this was minimal when RPE cells covered the membrane. The presence of RPE was also associated with persistence of the native photoreceptor layer and less retinal atrophy than membrane alone in both pig and clinical studies.

A qPCR analysis of systemic biodistribution in pigs at 26 weeks after implantation of one clinical-sized graft (100,000 hESC-RPE cells) showed no evidence that cells migrated or survived away from the site of administration. The lack of distribution of the cells is consistent with previous studies on ocular administration of RPE cells32. This is also supported by our NIH III mouse study, in which teratomas from undifferentiated hESCs were found only in the eye, where the cells had been administered, and not in tissues distal to the site, suggesting that undifferentiated hESCs do not migrate or do not survive away from the site of implantation.

In the GLP pig study with no immunosuppression, no definitive hESC-RPE cells were identified at 26 weeks, whereas persistent hESC-RPE cells were found at 6 weeks in the earlier pig studies (in which some animals were immunosuppressed perioperatively), and at 26 weeks in the NIH III mouse teratoma studies (immune-deficient animals). Histology of the implanted patch at 6 weeks from animals in the first pig study showed persistence of human RPE and support of normal retinal architecture relative to animals receiving the membrane alone. Microscopic findings consistent with a localized chronic inflammatory reaction around the polyester membrane were present in animals from both groups in the GLP pig study at 6 months, in the absence of any RPE cell cover. There was no difference in the incidence or severity of the inflammatory reaction in either group.

In the clinical trial, the transplanted RPE patch survived, as demonstrated by a clear RPE signal on OCT for 12 months and the visible persistence of a pigmented cellular monolayer, although some of these cells may have been pigmented macrophages. There was evidence of early autofluorescence in both patients 1 and 2, suggesting that RPE phagocytosis has commenced18,19. The retina over the patch was thinned in patient 1 but not in patient 2, which may reflect a difference in pre-operative disease or level of microtrauma from manipulation during surgery. It is possible that the decrease in central pigmentation in patient 1 represented cell loss from delayed rejection. Both patients retained features of normal architecture and visible areas of the ellipsoid zone. Furthermore, there was clear evidence of retinal function over the patch, as demonstrated by fixation microperimetry, which showed focal sensitivity, and increased visual acuity and reading speed, all of which were sustained or improved over 12 months post-transplant. However, in the absence of a control, proving the improvement was due to the transplant is not possible. The improvement in reading exceeded that of similar cases in the Submacular Surgery Trial20. Visual function remained variable across the transplanted area, with poor visual function and thinning centrally in patient 1, which we feel may reflect intra-operative surgical trauma.

While histological evidence of hESC-RPE survival was not possible, the extensive number of structural and functional features discussed above support the conclusion that the hESC-RPE survived. Furthermore, the presence of transplanted hESC-RPE cells immediately adjacent to the neuroretina suggest that they are associated with retinal function over the patch. We did not intentionally remove the subretinal choroidal neovascular membrane, but cannot exclude that it was removed inadvertently. Therefore, the effect of the original choroidal neovascular membrane on function remains ambiguous. However, the co-localization of choroidal perfusion, survival of hESC-RPE, retinal sensitivity, and presence of photoreceptors strongly support the conclusion that the visual improvement and stability was associated with the transplanted RPE patch.

Retinal detachment was the most severe clinical complication seen. The GLP pig studies had two retinal detachments in 20 animals, which is similar to the rates reported for early human autologous transplantation and translocation surgeries33,34. The retinal detachment in patient 2 of the clinical trial was likely a PVR-associated detachment that occurred between 4 and 8 weeks. It was treated with a single operation using standard techniques. The retinal detachment did not extend far enough to reach the patch. However, despite the delivery of RPE cells on a patch and in a protected delivery device, both of which minimize the shedding of RPE cells into the vitreous, we cannot be certain that no hESC-RPE cells were released nor that they did not contribute to the risk of developing PVR. The true surgical risk can be assessed accurately only with a larger series of patients.

In patient 1, there was an area of native RPE over the patch that was clearly delineated and separate from the large area over the patch where the native RPE was lost and only hESC-RPE was present. Notably, the patient fixated not over native RPE but over the RPE patch. Subjectively, patient 1 acknowledged that her vision improved relative to the pre-operative state and that she could see letters and read directly with the central vision; however, she described troublesome distortion, similar to that experienced by patients who have had previous retinal detachment or wet AMD. The second patient described his vision as continuously improving but also being “dimmer” than before the onset of the disease, which is consistent with his longer-standing disease at presentation and, likely, more damaged neuroretina.

Reporting these first two cases at 12 months is valuable because the functional improvement, robust imaging data, lack of major safety concerns, and demonstration of sufficiency of local immunosuppression represent steps in support of hESC-based regenerative therapy for AMD and other diseases of the eye. We show that differentiation of hESCs into a therapeutic cell with delivery, survival, and rescue of vision in very severe disease is feasible. Although 12 months is sufficient to begin to describe cell survival and clinical outcomes, it is early in terms of safety monitoring, especially for late teratoma formation. The patients will be followed for five years after surgery. These two early cases are also instructive as they show an encouraging outcome despite very advanced disease, which increases the complexity of surgery and involves more-damaged neuroretina. Additionally, there was no evidence of recurrence of the neovascular membrane and no need to administer angiogenesis inhibitors to either patient at 12 months, although with only two cases it is difficult to attribute this to the transplant.

The role of immune privilege in the subretinal space remains ambiguous, and the immunological effect of the surgery and patch transplant is unknown. Stability of the transplant was achieved with immunosuppression consisting of perioperative oral prednisolone and long-term intraocular steroid implants. For patient 2, who has type II diabetes, there was a period of poor blood sugar control with the need to add insulin, due to the systemic steroid use. Given the concerns about long-term systemic immunosuppression, a notable finding of this study is that the transplanted hESC-RPE cells survived at least 12 months with only local immunosuppression. Although long-term, local immunosuppression can be provided in the eye without systemic side effects, the possibility of long-term ocular morbidity remains. In our two patients there was no associated intraocular pressure rise or need for pressure-reducing medication.

Stem-cell-based tissue transplantation is a potentially effective treatment strategy for neurodegenerative or other diseases with irreversible cell loss. Here we addressed challenges related to engineering, manufacturing, and delivering a clinical-grade hESC-RPE patch, leading to stabilization and improvement of vision for at least 12 months in two subjects with severe vision loss from AMD. These findings support further investigation of our approach as an alternative treatment strategy for AMD.


Ethics statement.

All animal (mouse and pig) procedures were conducted in accordance with the provisions of the United Kingdom Animals Scientific Procedures Act (ASPA) 1986. The animal experimentation under the ASPA is overseen by the Home Office, UK government, in England.

The mouse teratoma studies were carried out by the Independent Contract Research (CRO) Organization Huntington Life Sciences (HLS), now Envigo (Alconbury/Huntingdon, Cambridgeshire, UK). (HLS Animal Project - Licence number PCD 70/8702).

The pig surgery was carried out under GLP conditions at the Northwick Park animal experimental surgery facility (Northwick Park and St Mark's Hospital, Harrow, Middlesex, UK). Pigs were cared for in an independent GLP registered facility including throughout the 6-month's follow-up period. The animal reporting including organ sampling was carried out independently by HLS.

Pfizer Inc. sponsored and was the sole funding source of the clinical trial. Two subjects out of a planned 10 were enrolled before Pfizer suspended funding for strategic commercial reasons unrelated to this trial. The cells for transplantation were prepared to the state of confluent RPE at Roslin Cells Ltd. (Edinburgh, United Kingdom,) before transport to London. The preparation and cutting of the patch was completed at Cells for Sight (Institute of Ophthalmology, Bath St., London, UK).

Approval of the clinical trial was granted by the UK Medicines and Health Products Regulatory Authority (MHRA), the Gene Therapy Advisory Committee (GTAC), the Moorfields Research Governance Committee and the London–West London & GTAC Research Ethics Committee (Clinicaltrials.gov: NCT01691261). The study complied with Good Clinical Practice guidelines according to the European Clinical Trials Directive (Directive 2001 EU/20/EC), the Declaration of Helsinki and has an independent External Data Monitoring Committee (E-DMC). The data monitoring committee had three representatives who were retinal surgeons, with two having specialty expertise in ocular oncology and the third in ocular immunology. Informed consent was obtained from each patient. The study compliance with protocol was reviewed regularly, and none of the recorded protocol deviations had an impact on subject safety or study integrity. The authors vouch for the accuracy and completeness of the data and data analyses. The datasets generated by HLS and Aros are being transferred to the academic authors.


The only statistical analysis reported was for the phagocytosis study (Fig. 2d). Data shown are mean ± s.e.m., n = 3 biologically independent cell cultures in each group. As the n is less than 10, individual data points are indicated. Analysis was by two-tailed Student's t-test with P = 0.0000103075196107. This was a single experiment with no replication.

Preparation of hESC-RPE (GMP conditions and facility).

The original cell source, the SHEF-1 hESC (NIBSC - UK Stem Cell Bank - http://www.nibsc.org/ukstemcellbank) line was expanded according to GMP guidelines at the Stem Cell Derivation Facility, Centre for Stem Cell Biology (CSCB), University of Sheffield35. The SHEF-1 and 1.3 cell lines were authenticated by a short-term repeat analysis fingerprint, which was performed by University of Wisconsin Hospital and Clinics. The analysis showed an identical STR genotype profile across eight human STR loci for both SHEF-1 and SHEF-1.3. The SHEF-1.3 cell line was derived from the expanded SHEF-1 hESC cells under the same GMP conditions at the same facility. The passage and expansion of SHEF-1.3 hESCs was carried out in recombinant human vitronectin (VTN-N) (Life Technologies)–coated culture vessels. Medium was removed from cultures and hESCs were washed with PBS−/− (without Ca++ and Mg++), incubated with EDTA (Sigma) until the cell colonies begin to detach. Cells were resuspended in Essential 8 (E8) medium (Life Technologies) and seeded onto VTN-N-coated culture vessels. hESCs were replenished with E8 medium daily, and passaged every 3–4 d depending on visual observations of colony morphology and size.

hESC differentiation to RPE was carried out in flasks coated with plasma-derived vitronectin (Amsbio). Sets underwent regular media replenishment with E8 medium for up to 14 d until hESC colonies were 80–100% confluent and then transitioned to TLP medium consisting of 389 ml of Knockout DMEM, 1 ml of 2-mercaptoethanol, 5 ml of NEAA (non-essential amino acids), 100 ml of knockout serum and 5 ml of L-glutamine per 500 mL (Invitrogen – LifeTechnologies; Thermo Fisher Scientific). Cells were maintained on a twice-weekly media replenishment regimen. RPE cells appear in culture as distinct pigmented foci; visible to the naked eye, which continue to expand in diameter and can be maintained in this culture system for up to 22 weeks.

RPE foci were manually isolated using sterile microblades (Interfocus) and collected in TLP medium. Pooled foci were washed with PBS−/− and incubated at 37 ± 2 °C with Accutase (Sigma) until a cell suspension is observed. The suspension was then passed through a 70-μm cell strainer (Corning), washed in TLP medium and counted. The cells were plated on CELLstart-coated (Life Technologies) 48-well plates at 4.8 × 104 cells/well. Cells were maintained on a twice-weekly media replenishment regimen of TLP medium until they formed a confluent, pigmented cell sheet with cobblestone morphology, generally for a minimum of 5 weeks.

To ensure safety and purity of the RPE, it was essential to exclude the possibility of undifferentiated hESCs being administered to a patient. In situ hybridization was used to assess hESC impurity. Samples of differentiated cells were prepared as a monolayer on microscope slides. Specific oligo probes against LIN28A mRNA were used to evaluate the presence of hESC in the RPE population. LIN28A-stained cells were identified using a nuclear stain. If any LIN28A positive cells were detected, the RPE batch would be rejected. Differentiated RPE was further assessed using immunocytochemistry against the melanosome-specific protein PMEL17 (Dako), an RPE cellular marker. PMEL17 positively stained cells were counted and expressed as a percentage of total cells. Upon testing of cells on representative cell patches, the RPE purity ranged from 99.8 to 100% on positive staining of PMEL17 and no LIN28A-positive cells were detected.

RPE cells were assessed using a light microscope for pigmentation, cobblestone morphology, health and signs of contamination and processed further only if they passed this visual check. Media was removed, cells were washed with PBS−/−, and incubated at 37 ± 2 °C with Accutase for 1–2 h until a cell suspension was observed. The suspension was passed through a 70-μm cell strainer, washed in TLP medium and then counted. Cells were subsequently seeded (1.16 × 105 cells/well) into custom manufactured transwells, with polyester membranes coated in plasma-derived human vitronectin. The membrane was 10-μm thick polyethylene terephthalate (PET), with a 0.4-μm pore size at a density of 1 × 108 pores/cm2 (Sterlitech, Kent, Washington, USA). TLP medium was added to the outer well containing the insert, and the plate then incubated at 37 ± 2 °C. The RPE cells were maintained with twice weekly TLP media replenishment until required for drug product manufacture.

On the day of surgery, the transwell was removed from the culture plate and placed directly onto the cutting device. The patch is cut from the membrane and is then placed into the storage solution (0.9% sodium chloride) within the storage container that is then sealed (Fig. 1a) The patch is assessed visually through the clear lid of the storage container for integrity, pigmented cell coverage and viability. The patch was transported out of the GMP facility and transferred to the operating theatre.

Patch cell count.

The theoretical cell count on each patch was calculated using an observed cell diameter of 14 μm and this gives 100,000 cells per patch. To complement this, two patches from preclinical production batches had the cells dissociated and counted and gave counts of 110,000 and 120,000 cells total (i.e., a cell diameter of 12–13 μm). The theoretical cell count of 100,000 was within the confidence intervals from these two counts, and so was selected.


Immunostaining was performed as described previously17. Briefly, cells were fixed in cold 4% paraformaldehyde (PFA) in 0.1 M phosphate buffer and then blocked and incubated with an appropriate combination of the following primary antibodies: TRA-1-60 (Life Technologies); Ki67 (VectorLabs); CRALBP (Thermo Scientific); PMEL17 (Dako); MERTK (Abcam); ZO-1 (Life Technologies); OTX2 (Millipore); Bestrophin (Millipore) and PEDF (Millipore). They were then incubated in the appropriate secondary antibodies including AlexaFluor 647 conjugated donkey anti-mouse Ig-G (Life Technologies), 594 conjugated donkey anti-rabbit Ig-G (Life Technologies) and Dylight 488 conjugated donkey anti-mouse Ig-M (Startech). Nuclei were stained using Hoechst 33342 (Life Technologies).

Transmission electron microscopy (TEM).

RPE cells immobilized on a membrane were fixed in Karnovsky's fixative and prepared for TEM as described previously17.

PEDF assay.

PEDF assay on spent media was completed using a developmental enzyme-linked immunosorbent assay (ELISA; Meso Scale Discovery Platform S600, Mesoscale, USA) according to manufacturer's instructions.

Phagocytosis assay method.

This was based on a previously described method36. Purification of photoreceptor outer segments (POS) was performed as described previously37. The POS were labeled as described previously38. The labeled POS were seeded onto hESC-RPE monolayer patches grown on plasma-derived human vitronectin at 1 × 107 POS/ml. In one condition the cells were pre-incubated with 1:50 anti MERTK antibody (ab52968, Abcam) to block phagocytosis for 1 h. The cells were incubated at 37 °C in 5% CO2 for 6 h. The samples were fixed and stained as described previously32. To quantify phagocytosis, three distinct areas per patch of hESC-RPE or hESC-RPE with blocker (MERTK antibody) (n = 3) were imaged blind on the Zeiss 710 confocal. The total number of internalized fluorescent POS were counted in a 135 μm × 135 μm area, using the orthogonal view function.

Mouse teratoma studies.

The cells tested were undifferentiated SHEF-1.3 pluripotent hESCs (hereafter 'hESCs'), and SHEF-1.3 hESC-derived RPE ('hESC-RPE'). The cells were grown at a Pfizer facility before transfer to HLS. They were injected into the eyes of NIH III immune-deficient mice (strain Crl:NIH-LystbgFoxn1nuBtkxid, 6-7 weeks old; Charles River, UK) at HLS. The injections were carried out by A.Vu. and A.A. from University College London. Histological and pathological analyses were carried out by HLS, which provided a report on the study.

The second study was conducted as there were fewer tumors in female animals in the first study. It was concluded that this was most likely linked to a decline in cell suspension quality for dosing of the females, rather than an actual sex difference. The second study focused on female animals. In addition, we studied the relative rate of teratoma formation following subretinal, subcutaneous or intramuscular injection of undifferentiated hESCs in female NIH-III mice. In the second study five males were also dosed with undifferentiated hESCs subretinally.

In the third study, under GLP conditions, NIH III mice were administered suspensions of hESC-RPE cells or suspensions of undifferentiated hESCs as a positive tumorigenic control, with all injections being delivered subretinally. Other controls carried out in all the mouse teratoma studies are listed in the tables below. Animals were followed for up to 26 weeks, allowing for the assessment of tumors associated with the administration of the test cells, prior to the animals having an excessive burden of spontaneous tumors. The numbers of animals injected, their sex and the site injected are listed in the following table.

First study: 2-month pilot study of the teratoma formation potential of hESCs in NIH III immune-deficient mice

Test conditionInjected number/sex animals
1. Untreated control5/sex(10 total)
2. DMEM control – subretinal5/sex(10 total)
3. hESCs at 45,000 M/44,000 F cells – subretinal15/sex(30 total)
4. hESCs at 88,000 M/68,000 F cells – subretinal15/sex(30 total)
5. Vehicle – IM5/sex(10 total)
6. hESCs at 88,000 M / 68,000 F cells – IM5/sex(10 total)

Second study: 2-month pilot study of the teratoma formation potential of hESCs in female NIH III immune-deficient mice

Test conditionInjected number/sex animals
1. DMEM – subretinal5F(5 total)
2. hESCs at 35,600 cells – subretinal5M/15F(20 total)
3. Matrigel – IM5 F(5 total)
4. HESCS at 35,500 cells – IM10F(10 total)
5. hESCs at 822,500 cells – IM10F(10 total)
6. Matrigel – SC5F(5 total)
7. hESCs at 35,500 cells – SC8F(8 total)
8. hESCs at 822,500 cells – SC12F(12 total)

Third study (GLP conditions): teratoma formation potential of hESC-RPE in NIH III immune deficient mice for 26 weeks following intraocular dosing

Test conditionInjected number/sex animals
1. KO DMEM control10/sex(20 total)
2. mTeSR control10/sex(20 total)
3. hESC-RPE at 5,340 cells in KO DMEM20/sex(40 total)
4. hESC-RPE at 60,400 cells in KO DMEM20/sex(40 total)
5. hESCs at 4510 cellsa in mTeSR19 M/20 F(29 total)
6. hESCs at 42,800 cellsa in mTeSR20 M/19 F(29 total)
  1. DMEM = Dulbecco's Modified Eagle Medium; KO DMEM = knockout DMEM; IM = intramuscular; SC = subcutaneous. M = male; F = female.
  2. aRepresents the average number of cells administered to the group.

Necropsy and histology.

At the end of the study animals were subject to a detailed necropsy, and macroscopic examinations were performed on all treated and control animals. The following tissues were processed for histological examination: abnormal masses, epididymides, eyes, heart, kidneys, liver, lymph nodes (mesenteric, submandibular), optic nerves, ovaries, spleen, sternum with bone marrow, testes. Tissues and organs were collected and fixed in 10% neutral buffer except the eyes and optic nerves, which were fixed in Davidson's fluid. Following histological processing and review, any teratomas or abnormal masses were subject to immunohistochemistry (IHC; anti-human mitochondrial marker staining), using a validated method to confirm whether these were of human or mouse origin.

Pig transplantation feasibility studies.

The analysis of the biodistribution data in pig tissues (looking for evidence of human cells in different organs) was done using qPCR by the independent CRO Aros (Aarhus, Denmark) with reports provided.

Surgical safety, successful delivery of the RPE patch, local and systemic biodistribution and toxicology were examined in these studies. The pig was selected since the size of the pig eye allowed for the administration of the intended full-size patch and the study of surgical feasibility, biocompatibility and biodistribution. All pig studies described were performed by the same surgeon who operated in the two cases described in the clinical trial (L.d.C.). In the second pig study the same clinical surgical technique was used as in the human trial.

First pig study.

In this study, hESC-RPE cells were seeded onto a BD Matrigel (BD Biosciences, Erembodegem, Belgium)-coated polyester membrane and manually implanted as a 1x3-mm membrane into the subretinal space of pig eyes via 20G pars plana vitrectomy, bleb detachment with retinotomy followed by air tamponade. Twenty animals underwent surgery and 12 of these received hESC-RPE cells on the membrane, 2 at each time point, with the remainder receiving a membrane without cells, 1 at each time point except 6 weeks where there were 3 animals. Six of the animals received immunosuppression with oral cyclosporine (650 mg for up to 2 weeks following surgery). Animals were observed, euthanized and enucleated at 2 h, 2 d, 1, 2, 4, and 6 weeks. Eyes were examined by light and electron microscopy.

Second pig study.

Safety/biodistribution study of therapeutic patch in pigs for 26 weeks

Table 4: Following surgical (intraocular), subretinal implantation


The feasibility of the surgical procedure and the general safety of PF-05206388 was assessed in studies performed in the Large White/Landrace hybrid pigs (in all cases).

In the GLP pig study, a group of 5 male and 5 female pigs were administered a patch at a dose of 1 graft in the left eye (approximately 100,000 hESC-RPE cells on a vitronectin-coated polyester membrane) and observed for 26 weeks (6 months). A similarly constituted control group of 5 male and 5 female animals received the vitronectin-coated polyester membrane only. The membrane with or without cells was placed into the subretinal space in all 20 pigs using the same tool and surgical technique as in the clinical trial. Animals received prednisolone (1 mg/kg) orally 7 d before surgery, which continued for 14 d post operatively, intravitreal 4 mg triamcinolone acetonide (TCA) during the surgery, sub-Tenon TCA 40 mg at the end of surgery, and Carprofen (4 mg/kg) was given orally postoperative on day 2 of the study. Eye drops of 0.1% dexamethasone, 0.5% hypromellose and 0.35% neomycin were administered to all the pigs 4 times per day for 17 d post-surgery.

Systemic distribution of cells.

The distribution of the hESC-RPE cells was evaluated as part of the pivotal 26-week toxicity study in pigs. The biodistribution was evaluated using qPCR techniques and the human cell markers huGAPD, huPEDF, HSSRY, huPOU5F1, and huOTX1/2 and the pig marker, pigACTB. A total of 60 samples from each animal (5/sex/group) that included the following: adrenal, bone marrow (rib and femur), brain (7 sections), heart (5 sections), kidneys (5 sections), liver (6 sections), lungs (5 sections), lymph nodes (4 nodes, total 8 sections), optic nerve (left and right, proximal and distal to eye), spleen (5 sections), thymus (3 sections) were evaluated. To avoid a high number of false positives, positives in at least two individual human markers in the same sample were required. No animals that received the matrix only had an increase in amplifying human transcripts.

Cell spiking studies.

The effect of deliberately spiking hESCs into RPE was assessed. Different proportions (0, 1, 10, 20, 50 or 100%) of hESCs were mixed with dissociated RPE foci (P0 cells), seeded into 96-well plates at 38,000 cells/cm2 and cultured for 6 weeks (P1 cells) using standard RPE culture conditions. Feeder-free hESC were used for spiking to eliminate counting inaccuracies due to the presence of fibroblast feeder cells. Plates were fixed and stained for TRA-1-60 or PMEL17 and the corresponding no primary antibody controls (n = 4 plates, 1 well/plate stained with each antibody/control). A single-cell suspension of hESCs was also seeded onto Matrigel in mTeSR1 media with ROCK inhibitor 2 d before fixing as a positive control for TRA-1-60 staining.

Flow cytometry.

Cells were washed with PBS and treated with Accutase to produce a single-cell suspension, before pelleting and resuspended in 3% BSA/PBS to 1 million/ml. 100,000 cells per sample were incubated on ice for 20 min with IgM-PE isotype control (eBioscience 12-4752- 82), TRA-1-60-PE (eBioscience 12-8863-82, PE) or Tra-1-81-PE (eBioscience 12-8883-80) at 1/20 dilution. For propidium iodide (PI) staining, 100 μl 1.5 μM PI (Sigma) was added to 100 μl cells and incubated for 10-15min on ice. Samples were pelleted and resuspended in 150 μl PBS alone (for PI stained samples) or PBS containing 50 nM TOPRO3 (Invitrogen), a dead cell stain, for IgM and TRA-1-60 samples. Samples were run on the Accuri Cflow flow cytometer. Debris, TOPRO3 positive dead cells (except when deliberately analyzing dead populations) and cell doublets were excluded from the analysis.


For live PI staining, 100 μl 1.5 μM PI (Sigma) in PBS was added to wells containing 100 μl cells/media. After 10 min PI/media was aspirated and replaced with 100 μl 1:10,000 HOESCHT in PBS. Plates were imaged live on the Image Express Micro platform (Molecular Devices Corporation). For immunostaining, cells were fixed with 4% PFA for 20 min, washed with PBS−/− and blocked with 5% NDS in 0.3% triton-X/PBS for 1 h. Primary antibody in 1% NDS/TX/PBS was then applied for a further hour at room temperature. The antibodies used were TRA-1-60 at 1:100 (Abcam 16288, mouse), PMEL17 at 1:25 (DAKO HMB45, mouse monoclonal) or 1:500 Ki67 (Vector K451, rabbit). Wells were washed three times with 1% NDS/TX/PBS and secondary antibody, at 1:200 in 1% NDS/TX/PBS applied for 1 h at room temperature. For TRA-1-60 Jackson anti-mouse IgM (715-485-020) was used, for PMEL17 Invitrogen Alexa Fluor 488 anti-mouse and for Ki67 Alexa Fluor 488 antirabbit was used. Wells were then washed three times with PBS−/− and 1:10,000 HOESCHT applied. These were imaged on the Image Express Micro platform (Molecular Devices Corporation); automated image analysis was performed using MetaExpress software. The best settings for this analysis were determined both by visual observation of the plates and by choosing.

The introducer tool.

The patch delivery tool was custom designed and manufactured with the aim of providing protected delivery of the therapeutic patch until its placement into the subretinal space. The device consists of a handle containing a mechanism to advance a flexible rod that passes from the wheel, through the shaft and to the tip where the rod physically pushes the patch out of the tip of the device (Supplementary Fig. 1.). The rod is advanced by virtue of the wheel on the handle that the surgeon rolls forward to advance the patch out of the device and into position. The patch was loaded manually into the tip by drawing it into place with microforceps.

Clinical trial.

Subjects in the trial underwent extensive screening and follow-up investigations including (i) blood sampling for: hematology: hemoglobin, hematocrit, RBC, platelet, WBC count and differential; Chemistry: urea, creatinine, glucose, calcium, sodium, potassium, chloride, total CO2, AST, ALT, total bilirubin, ALP, uric acid, albumin, total protein; HIV, HepB and HepC testing; (ii) urinalysis for pH, glucose, protein, blood, ketones, nitrites, leukocyte esterase, microscopy and a drug screen.

A physical examination, including head, ears, eyes, nose, mouth, skin, heart and lung, lymph nodes, gastrointestinal, musculoskeletal, and neurological systems, blood pressure, pulse rate and Electrocardiograms, was also performed.

Primary endpoints.

  1. Incidence and severity of adverse events.

  2. Change from baseline in Early Treatment Diabetic Retinopathy (ETDRS) BCVA; proportion of subjects with an improvement of 15 letters or more at Week 24.

Secondary endpoints.

  1. Change in baseline in ETDRS BCVA. Proportion of subjects with an improvement of 15 letters or more.

  2. Mean change of BCVA from baseline by study visit.

  3. Position of PF-05206388 by serial biomicroscopic evaluation.

  4. Position and presence of pigmented RPE cells by serial fundus photography.

  5. Mean change from baseline in contrast sensitivity by Pelli Robson test.

  6. Change in liver and renal function by blood tests and liver ultrasound.

  7. Change in leakage or perfusion in normal fundal vasculature and presence of abnormal vasculature by fundus fluorescein angiography.

  8. Change in central 30 degree of visual function by Humphrey Field test.

  9. Change in thickness of RPE layer by B-mode orbital ultrasound.

Inclusion criteria.

  1. Male and /or post-menopausal (defined as at least 12 consecutive months with no menses without an alternative medical cause) female subjects aged 60 years or above.

  2. Diagnosis of wet age-related macular degeneration (AMD) with evidence of good foveal fixation in the study eye plus at least one of:

    • History of sudden decline in vision within 6 weeks of the time of surgery (Day 1) associated with evidence of an RPE tear extending under the fovea.

    • History of sudden decline in vision within 6 weeks of the time of surgery (Day 1) associated with a submacular hemorrhage with greatest dimension of 6 disc diameters.

    • Evidence of failure to respond to anti-VEGF treatments as defined by declining vision ≥25 ETDRS letters within 20 weeks prior to surgery (Day 1) despite at least 3 injections of anti-VEGF.

  3. BCVA value recorded of 6/36 Snellen equivalent or worse in the study eye during or within 14 d of the screening period.

  4. An informed consent document signed and dated by the subject or a legal representative.

  5. Subjects who are willing and able to comply with all scheduled visits, treatment plan, laboratory tests, and other study procedures.

Exclusion criteria.

  1. Evidence or history of clinically significant hematological, renal, endocrine, pulmonary, gastrointestinal, cardiovascular, hepatic, psychiatric, neurologic, allergic disease (including drug allergies, but excluding untreated, asymptomatic, seasonal allergies at time of dosing) or other severe acute or chronic medical or surgical condition or laboratory abnormality that may increase the risk associated with study participation or investigational product administration or may interfere with the interpretation of study results and, in the judgment of the investigator, would make the subject inappropriate for entry into this study.

  2. Pregnant females; breastfeeding females; and females of childbearing potential.

  3. Positive urine drug screen for illicit drugs or drugs of abuse.

  4. History of regular alcohol consumption exceeding 14 units/week for females or 21 units/week for males within 6 months of screening.

  5. Treatment with an investigational drug within 30 d (or as determined by the local requirement, whichever is longer) or 5 half-lives preceding the first dose of study medication.

  6. Blood donation of >1 pint (500 mL) within 56 d prior to dosing.

  7. Subject unwilling or unable to comply with the Lifestyle Guidelines described in this protocol.

  8. Contraindication to general anesthesia (as determined by the attending anesthetist).

  9. Previous significant retinal disease in the study eye (other than AMD), as determined by the investigator.

  10. Current or previous significant other ocular disease in the study eye, as determined by the investigator.

  11. Any ocular disease in the study eye that might alter the ocular media and reduce the posterior segment view, as determined by the investigator.

  12. Any previous retinal surgery in the study eye.

  13. Contraindication to prednisolone, triamcinolone, cefuroxime or fluocinolone acetonide (specifically increased intra-ocular pressure or glaucoma).

  14. History of sensitivity to heparin or heparin-induced thrombocytopenia.

  15. Use of anti-VEGF therapy (e.g., Lucentis) within 7 d of surgery in the study eye.

  16. Participation in another ongoing interventional clinical study.

  17. Limited use of non-prescription medications that are not believed to affect subject safety or the overall results of the study may be permitted on a case-by-case basis following approval by the sponsor.

  18. Subjects who are investigational site staff members or relatives of those site staff members or subjects who are Pfizer employees directly involved in the conduct of the study.

Visit schedule.

Subjects attended the hospital on an out-patient basis for up to 3 d during the screening interval (Day –21 to Day 0). Eligible subjects attended the hospital on an out-patient basis the day before surgery (Day 0) for pre-surgery assessments. Insertion of PF-05206388 (the hESC-RPE patch) occurred the following day (Day 1). Subjects were discharged from the hospital the day after surgery (Day 2). There was a second operative procedure performed at the Week 10 visit (Visit 9) whereby the silicone oil, which was inserted at the time of PF-05206388 insertion, was removed from the eye. Subjects with significant cataracts noted at screening were permitted to have the cataracts removed at the time of this second surgical procedure. For the first procedure, subjects received up to 1 mg/kg oral prednisolone (up to 60 mg maximum) daily commenced 2–4 d pre-operatively and continued for at least 2 weeks followed by a tailing off over at least one further week. Subjects also received single doses of sub-Tenon triamcinolone acetonide 40 mg and subconjunctival cefuroxime (250 mg in 1 mL) after patch insertion. For the second procedure, subjects received up to 1 mg/kg oral prednisolone (up to 60 mg maximum) daily, commenced 7 d pre-operatively and continued for at least 2 weeks followed by a tailing off over at least one further week. Subjects also received an intravitreal implant of fluocinolone acetonide either 0.19 or 0.59 mg as an anti-inflammatory agent and subconjunctival cefuroxime (250 mg in 1 mL) post-surgery.

The first surgical procedure was carried out on day 1, subjects were discharged from hospital on day 2 and attended for post-surgical evaluations at week 1, week 2, week 4, week 8, week 12, week 16, week 24, week 36 and week 52.

Surgical procedures.

Transplantation operation: patients underwent standard general anesthesia. Routine sterile povidone-iodine 5% preparation of the eye with and draping of the face was completed. The eyelids were separated with a disposable Barraquer speculum. For patient 1 a cataract was removed and a single focus, one-piece, acrylic lens for emmetopria was inserted, at the time of the transplant surgery, while patient 2 was already pseudophakic. A 270 degree peritomy with a standard 23-gauge infusion and two further ports were inserted. Pars plana vitrectomy with induction of PVD when not present was completed with subsequent 360 degree Argon laser. A macular retinal detachment was created with a 38-gauge de Juan dual bore cannula, the retinotomy was enlarged with microscissors and the subretinal space irrigated to remove all blood and fibrin. The temporal port was widened for introduction of the delivery device and insertion of the patch into the subretinal space. The macular was reattached with heavy liquid and the retinotomy lasered. A routine search of the retinal periphery was completed in both cases with no iatrogenic breaks detected. Air/fluid followed by air/silicone oil exchange was carried out. The ports were sutured and the infusion removed. Sub-Tenon triamcinolone acetonide 40 mg and subconjunctival cefuroxime (250 mg/1 ml) were injected followed by sub-Tenon (0.5% bupivacaine).

Removal of silicone oil: silicone oil was removed under sub-Tenon local anaesthesia. A 23-gauge infusion was inserted in the infero-temporal quadrant and a second port created to aspirate the oil. The retinal periphery was examined for breaks. The ports were closed and subconjunctival antibiotics were given. For patient 2 an operation to reattach a detached retina, consisting of epiretinal membrane peeling, inferior 180-degree retinectomy and laser, was carried out before the oil removal.

Ocular tests and examinations.

Biomicroscopic evaluation and intra-ocular pressure measurement were carried out using a Haag-Streit 900 slit-lamp (Koeniz, Switzerland) and Goldmann applanation tonometer.

Fundus photography, angiography.

Color, red-free fundus photographs and standard fluorescein and indocyanine green angiography (50°) were acquired through dilated pupils using the Topcon TRC Retinal Camera (Oakland, NJ, USA).

Spectral-domain OCT scan and cSLO reflectance imaging.

Volume scan and EDI volume scan, 20° × 20°, 49 sections, ART 12, High Speed scans and 30°, 55° short-wave reflectance autofluorescence images were acquired through dilated pupils using the Spectralis OCT (Heidelberg Engineering GmbH, Heidelberg, Germany).

Adaptive optics.

AO images were acquired through dilated pupils using the ImagineEyes Camera (Orsey, France). Alignment and data collection were repeated for several retinal locations.

For calculation of cellular density, a freely available image-processing program (ImageJ, National Institutes of Health, Bethesda, Maryland) was used to manually identify the cones in each subject′s retinal image.

Pattern ERG, full-field ERG and EOG.

Pattern ERG, full-field ERG and EOG were performed according the ISCEV standards. Dark-adapted ERGs were recorded to flashes of 0.01 and 10.0 cd.s.m-2 (DA0.01; DA10.0); light-adapted responses were recorded to flashes of 3.0 cd.s.m-2 (30 Hz and 2 Hz).

Automated perimetry.

Visual fields were assessed using Humphrey Field Analyzer (Carl Zeiss Meditec, AG, Jena, Germany). Central 24-2 Threshold tests were performed using Goldmann III white stimulus at background luminance of 31.5ASB, in SITA-FAST strategy.


A customized rectangular grid of 9-23 loci covering the area of the RPE patch was used in Nidek MP1 Microperimeter (NAVIS software 1.7.2, Nidec Technologies, Padova, Italy). Stimuli were set to Goldmann III-V size, white color, 200-ms duration and 0-20 dB intensity with background luminance at 1.27 cd/m2. A single red-cross 2–3° target was used for 30 s fixation stability assessment. A 4-2 staircase strategy was employed to assess the retinal sensitivity through dilated pupils after 15-min dark adaptation.

ETDRS/logmar BCVA—distant and near, including contrast sensitivity (CS).

Refracted BCVA was tested by independent trained optometrists using the Early Treatment of Diabetic Retinopathy Study (ETDRS) charts and contrast sensitivity tested by the Pelli–Robson chart at 1 m.

Minnesota reading test (MNRead).

The MNRead sentences were used to assess the visual processing capabilities and eye-movement control required for normal text reading. Each sentence contains 60 characters (including a space between each word and at the end of each line) printed as three lines with even left and right margins. The vocabulary used in the sentences is selected from high-frequency words that appear in second- and third-grade reading material. The charts contain sentences with 19 different print sizes. From the recommended viewing distance of 40 cm (16 in), the print size ranges from +1.3 to –0.5 logMAR (Snellen equivalents: 20/400–20/6).

B-mode orbital ultrasound.

Ultrasound scans were acquired through sterile gel, using the ACUSON Sequoia512 system (Siemens Healthcare, USA).

Ocular oncologist review.

Specialist Ocular oncologists reviewed the patients with biomicroscopic evaluation and review of images from the ultrasound, fundus photography and fundus angiography.

Data validation.

This manuscript is generated from an unlocked database and includes some data that Pfizer is not responsible for validating/storing.

Life Sciences Reporting Summary.

Further information on experimental design is available in the Life Sciences Reporting Summary.

Data availability.

The majority of data generated or analyzed during this study are included in this published article and its supplementary information. Any further data concerning the current study are available from the corresponding author on reasonable request. There are no accession codes, unique identifiers or web links to publicly available data sets related to this study.

Additional information

Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.


  1. 1.

    Human embryonic stem cells: early hints on safety and efficacy. Lancet 379, 689–690 (2012).

  2. 2.

    et al. Development of human embryonic stem cell therapies for age-related macular degeneration. Trends Neurosci. 36, 385–395 (2013).

  3. 3.

    et al. Stem cell based therapies for age-related macular degeneration: The promises and the challenges. Prog. Retin. Eye Res. 48, 1–39 (2015).

  4. 4.

    et al. Developing cellular therapies for retinal degenerative diseases. Invest. Ophthalmol. Vis. Sci. 55, 1191–1202 (2014).

  5. 5.

    & Understanding age-related macular degeneration (AMD): relationships between the photoreceptor/retinal pigment epithelium/Bruch's membrane/choriocapillaris complex. Mol. Aspects Med. 33, 295–317 (2012).

  6. 6.

    et al. Ranibizumab for neovascular age-related macular degeneration. N. Engl. J. Med. 355, 1419–1431 (2006).

  7. 7.

    et al. Adaptive optics imaging shows rescue of macula cone photoreceptors. Ophthalmology 121, 430–431.e3 (2014).

  8. 8.

    et al. Human embryonic stem cell-derived retinal pigment epithelium in patients with age-related macular degeneration and Stargardt's macular dystrophy: follow-up of two open-label phase 1/2 studies. Lancet 385, 509–516 (2015).

  9. 9.

    et al. Autologous induced stem-cell-derived retinal cells for macular degeneration. N. Engl. J. Med. 376, 1038–1046 (2017).

  10. 10.

    et al. Elucidating the phenomenon of HESC-derived RPE: anatomy of cell genesis, expansion and retinal transplantation. Exp. Neurol. 214, 347–361 (2008).

  11. 11.

    et al. In vitro and in vivo characterization of pigment epithelial cells differentiated from primate embryonic stem cells. Invest. Ophthalmol. Vis. Sci. 45, 1020–1025 (2004).

  12. 12.

    International Stem Cell Initiative. et al. Characterization of human embryonic stem cell lines by the International Stem Cell Initiative. Nat. Biotechnol. 25, 803–816 (2007).

  13. 13.

    et al. Derivation and comparative assessment of retinal pigment epithelium from human embryonic stem cells using transcriptomics. Cloning Stem Cells 6, 217–245 (2004).

  14. 14.

    et al. Directed differentiation of human embryonic stem cells into functional retinal pigment epithelium cells. Cell Stem Cell 5, 396–408 (2009).

  15. 15.

    , & The derivation of clinical-grade human embryonic stem cell lines. FEBS Lett. 580, 2875–2878 (2006).

  16. 16.

    et al. Restriction landmark genome scanning identifies culture-induced DNA methylation instability in the human embryonic stem cell epigenome. Hum. Mol. Genet. 16, 1253–1268 (2007).

  17. 17.

    et al. Human embryonic stem cell-derived cells rescue visual function in dystrophic RCS rats. Cloning Stem Cells 8, 189–199 (2006).

  18. 18.

    & Fundus autofluorescence and RPE lipofuscin in age-related macular degeneration. J. Clin. Med. 3, 1302–1321 (2014).

  19. 19.

    , , & (eds.) The Atlas of Fundus Autofluorescence Imaging (Springer, 2007).

  20. 20.

    et al. Submacular Surgery Trials (SST) Research Group. Surgery for hemorrhagic choroidal neovascular lesions of age-related macular degeneration: ophthalmic findings: SST report no. 13. Ophthalmology 111, 1993–2006 (2004).

  21. 21.

    , & The effect of injection using narrow-bore needles on mammalian cells: administration and formulation considerations for cell therapies. J. Pharm. Pharmacol. 67, 640–650 (2015).

  22. 22.

    , & Reengineering of aged Bruch's membrane to enhance retinal pigment epithelium repopulation. Invest. Ophthalmol. Vis. Sci. 45, 3337–3348 (2004).

  23. 23.

    et al. Subretinal implantation of retinal pigment epithelial cells derived from human embryonic stem cells: improved survival when implanted as a monolayer. Invest. Ophthalmol. Vis. Sci. 54, 5087–5096 (2013).

  24. 24.

    et al. Retinal pigment epithelium translocation after choroidal neovascular membrane removal in age-related macular degeneration. Ophthalmology 109, 1492–1498 (2002).

  25. 25.

    , , , & A free retinal pigment epithelium-choroid graft in patients with exudative age-related macular degeneration: results up to 7 years. Am. J. Ophthalmol. 153, 120–7 e2 (2012).

  26. 26.

    et al. Long-term visual and microperimetry outcomes following autologous retinal pigment epithelium choroid graft for neovascular age-related macular degeneration. Clin. Experiment. Ophthalmol. 37, 275–285 (2009).

  27. 27.

    , , , & Retinal pigment epithelium-choroid graft with a peripheral retinotomy for exudative age-related macular degeneration: long-term outcome. Retina doi: 10.1097/IAE.0000000000001945 (2017).

  28. 28.

    , , & Surgery for CNV and autologous choroidal RPE patch transplantation: exposing the submacular space. Graefes Arch. Clin. Exp. Ophthalmol. 248, 37–47 (2010).

  29. 29.

    et al. New algorithm for assessing patient suitability for macular translocation surgery. Clin. Experiment. Ophthalmol. 35, 448–457 (2007).

  30. 30.

    , & Long-term outcome of RPE allografts in non-immunosuppressed patients with AMD. Eur. J. Ophthalmol. 9, 217–30 (1999).

  31. 31.

    et al. Morphological and functional rescue in RCS rats after RPE cell line transplantation at a later stage of degeneration. Invest. Ophthalmol. Vis. Sci. 49, 416–421 (2008).

  32. 32.

    et al. Long-term safety and function of RPE from human embryonic stem cells in preclinical models of macular degeneration. Stem Cells 27, 2126–2135 (2009).

  33. 33.

    & Macular translocation with 360 degree retinotomy for management of age-related macular degeneration with subfoveal choroidal neovascularization. Am. J. Ophthalmol. 134, 560–565 (2002).

  34. 34.

    et al. A comparison of macular translocation with patch graft in neovascular age-related macular degeneration. Invest. Ophthalmol. Vis. Sci. 50, 1848–1855 (2009).

  35. 35.

    , & Derivation of GMP raw materials for use in regenerative medicine: hESC-based therapies, progress toward clinical application. Clin. Pharmacol. Ther. 82, 448–452 (2007).

  36. 36.

    et al. Molecular characterization and functional analysis of phagocytosis by human embryonic stem cell-derived RPE cells using a novel human retinal assay. Mol. Vis. 15, 283–295 (2009).

  37. 37.

    & Analysis of photoreceptor outer segment phagocytosis by RPE cells in culture. Methods Mol. Biol. 935, 285–295 (2013).

  38. 38.

    et al. Rescue of the MERTK phagocytic defect in a human iPSC disease model using translational read-through inducing drugs. Sci. Rep. 7, 51 (2017).

Download references


We acknowledge H. Moore, Stem Cell Derivation Facility, Centre for Stem Cell Biology (CSCB), University of Sheffield for derivation of the original SHEF-1 hESC line and P. Keane and M. Cheetham for comments on the paper. We thank R. McKernan for support and input throughout the project. L.d.C. and P.J.C. received the following grants and donations and would like to acknowledge that they were used to fund the studies reported in this article: Anonymous Donor, USA, Establishment of The London Project to Cure Blindness - Donation. Lincy Foundation, USA, The London Project To Cure Blindness: Funding Towards The Production Of A Cell Based Therapy For Late Stage Age-Related Macular Degeneration - P12761. Macular Disease Society Studentship – Donation. MRC, Stem Cell Based Treatment Strategy For Age-Related Macular Degeneration (AMD) - G1000730. CIRM (California Institute of Regenerative Medicine) LA1_C2-02086. Pfizer Inc, The Development Plan For A Phase I/IIa Clinical Trial Implanting HESC Derived RPE for AMD - PF-05406388. Moorfields Biomedical Research Centre, National Institute for Health Research (NIHR) - BRC2_011. The Michael Uren Foundation R170010A.

Author information


  1. The London Project to Cure Blindness, ORBIT, Institute of Ophthalmology, University College London (UCL), London, UK.

    • Lyndon da Cruz
    • , Kate Fynes
    • , Odysseas Georgiadis
    • , Yvonne H Luo
    • , Ahmad Ahmado
    • , Britta Nommiste
    • , Shazeen M Hasan
    • , Sakina B Gooljar
    • , Amanda-Jayne F Carr
    • , Anthony Vugler
    • , Conor M Ramsden
    •  & Peter J Coffey
  2. NIHR Biomedical Research Centre at Moorfields Eye Hospital NHS Foundation Trust, UCL Institute of Ophthalmology, London, UK.

    • Lyndon da Cruz
    • , Odysseas Georgiadis
    • , Yvonne H Luo
    • , Adnan Tufail
    • , Anthony G Robson
    • , Graham E Holder
    • , Mandeep S Sagoo
    •  & Peter J Coffey
  3. Moorfields Eye Hospital NHS Foundation Trust, London, UK.

    • Lyndon da Cruz
    • , Odysseas Georgiadis
    • , Yvonne H Luo
    • , Conor M Ramsden
    • , Adnan Tufail
    • , Anthony G Robson
    • , Graham E Holder
    •  & Mandeep S Sagoo
  4. Wellcome/EPSRC Centre for Interventional & Surgical Sciences (WEISS), Charles Bell House, London, UK.

    • Lyndon da Cruz
  5. Pfizer, Granta Park, Cambridge, UK.

    • Julie Kerby
    • , Magda Bictash
    • , Mike Fenster
    • , Juliette Steer
    • , Tricia Harbinson
    • , Anna Wilbrey
    • , Gang Feng
    • , Mark Whitlock
    • , Peter T Loudon
    •  & Paul Whiting
  6. Cell and Gene Therapy Catapult, London, UK.

    • Julie Kerby
  7. Cells for Sight, Transplantation & Research Program, UCL Institute of Ophthalmology, London, UK.

    • Amanda Vernon
    •  & Julie T Daniels
  8. UCL Institute of Neurology, Queen Square, London, UK.

    • Paul Whiting
  9. Center for Stem Cell Biology and Engineering, NRI, UC, Santa Barbara, California, USA.

    • Peter J Coffey


  1. Search for Lyndon da Cruz in:

  2. Search for Kate Fynes in:

  3. Search for Odysseas Georgiadis in:

  4. Search for Julie Kerby in:

  5. Search for Yvonne H Luo in:

  6. Search for Ahmad Ahmado in:

  7. Search for Amanda Vernon in:

  8. Search for Julie T Daniels in:

  9. Search for Britta Nommiste in:

  10. Search for Shazeen M Hasan in:

  11. Search for Sakina B Gooljar in:

  12. Search for Amanda-Jayne F Carr in:

  13. Search for Anthony Vugler in:

  14. Search for Conor M Ramsden in:

  15. Search for Magda Bictash in:

  16. Search for Mike Fenster in:

  17. Search for Juliette Steer in:

  18. Search for Tricia Harbinson in:

  19. Search for Anna Wilbrey in:

  20. Search for Adnan Tufail in:

  21. Search for Gang Feng in:

  22. Search for Mark Whitlock in:

  23. Search for Anthony G Robson in:

  24. Search for Graham E Holder in:

  25. Search for Mandeep S Sagoo in:

  26. Search for Peter T Loudon in:

  27. Search for Paul Whiting in:

  28. Search for Peter J Coffey in:


L.d.C., P.T.L., P.W., and P.J.C. designed all of the animal studies and the clinical study, developed the methodology for these studies, collected the data, performed the analysis, and wrote the manuscript. L.d.C. performed the pig and human surgery. K.F., J.K., A.A., A.Ve., J.T.D., B.N., S.M.H., S.B.G., A.-J.F.C., A.Vu., C.M.R., M.B., M.F., J.S., T.H., and A.W. developed, isolated, and prepared the hESC-RPE and performed the engineering of the hESC-RPE patch; and assisted in designing and conducting the mouse and pig studies, collecting the data, performing the analysis, and writing the manuscript. A.A. and A.Vu. performed the mouse surgery. O.G., Y.H.L., A.A., A.T., G.F., M.W., A.G.R., G.E.H. and M.S.S. assisted in designing the clinical study, developing the methodology, collecting the data, performing the analysis, and writing the manuscript.

Competing interests

J.K., M.B., M.F., J.S., T.H., G.F., M.W., P.T.L., and P.W. were all employees of Pfizer during the period of this clinical trial. This study was sponsored by Pfizer Inc. L.d.C. and P.J.C. are named on two patents lodged by University College London Business. They are Patent Application No. PCT/GB2009/000917 (for the patch) and International Patent Application No. PCT/GB2011/051262 (for the surgical tool).

Corresponding author

Correspondence to Lyndon da Cruz.

Integrated supplementary information

Supplementary information

About this article

Publication history