Iris Pigment Epithelial Translocation in the Treatment of Exudative Macular Degeneration: Title and subTitle BreakA 3-Year Follow-up FREE

Age-related macular degeneration (AMD) is the leading cause of legal blindness among elderly persons in Western societies.^{1 - 3} Approximately 10% to 20% of patients with AMD develop the exudative form of this disease that is characterized by choroidal neovascularization (CNV).³ Growth of new vessels from the choroid through the Bruch membrane and into the subretinal space causes exudation and hemorrhagic lesions with progressive visual loss, eventually resulting in the formation of disciform scars with loss of photoreceptors. In addition to AMD, several other diseases, including dominant drusen, presumed ocular histoplasmosis syndrome, idiopathic CNV, and myopic maculopathy, can be associated with CNV resulting in visual loss.

Currently, the only treatments for CNV that have been proven effective are laser photocoagulation^{4 - 5} for extrafoveal-juxtafoveal classic CNV and photodynamic therapy especially for subfoveal predominantly classic CNV.⁶ However, even when these treatments are successful, re-treatments are required in a significant number of cases because of recurrence or persistence of the CNV after photocoagulation and reperfusion after photodynamic therapy.^{6 - 8} Surgical extraction of subfoveal neovascular membranes in AMD transforms the exudative process into a dry and, therefore, more stable lesion, but as the underlying retinal pigment epithelium (RPE) and parts of the Bruch membrane are likely removed or damaged, visual outcome has been disappointing in patients with AMD. Furthermore, several studies have shown progression of atrophy after membrane extraction and recurrence of the CNV in 30% to 40% of patients.^{9 - 18} Transplantation of RPE cells to the subretinal space has been examined in many experimental and clinical studies as a means of restoring normal subretinal conditions.^{19 - 21} However, transplantation of heterologous adult and fetal RPE cells into human subjects has not been successful.^{20 - 21} Subsequently, transplantation of autologous RPE cells has been suggested^{22 - 25} ; and a few studies have shown the feasibility of transplanting autologous RPE cells. However, harvesting of autologous RPE cells is associated with complications such as proliferative vitreoretinopathy, and to date functional results have not been encouraging. Because autologous RPE cells are difficult to harvest and because autologous iris pigment epithelial (IPE) cells can be easily obtained, it has been suggested that IPE cells may substitute for RPE cells for transplantation subretinally to replace missing or damaged RPE cells. During the last few years, several investigators have demonstrated that RPE and IPE cells have many important properties in common, such as pigmentation, cellular morphologic features, and formation of tight junctions.^{10 ,26 - 31} It has been shown that human, porcine, and rat IPE cells grown in vitro acquire the ability to phagocyte photoreceptor outer segments, which is a specific property of RPE cells in situ.^{32 - 34} In rabbits autologous IPE cells can be transplanted to the subretinal space, where they form a monolayer on top of the original RPE, phagocyte photoreceptor outer segments, develop microvilli, establish contact with the photoreceptor outer segments, and show no evidence of rejection during a 20-week follow-up.^{35 - 36} Recent studies have demonstrated that bovine IPE cells express messenger RNA for proteins involved in retinol metabolism, namely, RPE65 (an RPE gene), CRALBP (cellular retinaldehyde-binding protein) , and 11-cis-RDH (11-cis–retinol dehydrogenase). However, immunohistochemistry did not reveal the presence of these proteins in IPE cells in situ.³⁷ Since adult IPE and RPE cells retain the ability to transdifferentiate into other cell types such as lens epithelial cells^{38 - 39} and neural retinal cells,^{40 - 42} it can be postulated that IPE cells transplanted into the subretinal space would transdifferentiate into RPE cells or RPE-like cell types.

Previously we published the 1-year results of our pilot study on the effects of surgical removal of CNV combined with transplantation of autologous IPE cells to the subretinal space in 20 patients.⁴³ We report the functional and anatomical outcome of the same patient group 3 years after treatment.

Between June 23, 1998, and November 9, 1998, 20 patients with exudative maculopathy underwent surgery for removal of subretinal neovascular membranes followed by transplantation of autologous IPE cells to the subretinal space. Criteria of inclusion consisted of the following: (1) CNV that involved the geometric center of the foveal avascular zone, (2) best-corrected visual acuity (VA) of 20/50 or less in the eligible eye, and (3) decreased VA within the last 3 months. Informed consent indicating that the procedure was a novel experimental approach was obtained from each patient. The alternatives, potential benefits, risks, and complications of this surgical treatment were explained to the patients in detail before recruitment. Ocular criteria for exclusion consisted of disciform lesions, pigment epithelial tears, and eyes with recurrent CNV after laser photocoagulation or conventional submacular surgery. Patients with a history of heart attack, stroke, and presenile dementia were also excluded.

To establish a baseline, all patients underwent a comprehensive ophthalmologic examination preoperatively, including anterior segment biomicroscopy, intraocular pressure measurement, and fundus examination with slitlamp biomicroscopy. Best-corrected distance VA was measured by a certified visual examiner (S.A.) with the use of the Early Treatment Diabetic Retinopathy Study (ETDRS) chart according to a fixed protocol. For each vision test, both eyes of the patient were tested with different charts. Color photographs of the macula and the optic disc of each eye were taken; fluorescein and indocyanine green angiograms were performed by a certified photographer. The size of CNV was measured in late-phase fluorescein angiograms; measurements were taken in triplicate and the CNV size is given as disc area in accord with the Macular Photocoagulation Study Group.⁴ The size of CNV was measured in 10 of 20 patients; for the remaining 10 patients either the angiograms were unavailable or the angiograms did not allow for accurate measurement because of cloudy media or quality of the angiogram. Microperimetry of the study eye was performed using scanning laser ophthalmoscopy (SLO II; Rodenstock, Munich, Germany). Patients had a complete medical history, physical examination, electrocardiogram, chest radiogram, and blood test to screen for major organ pathologic conditions.

A large iridectomy was performed on the eye undergoing vitrectomy through a peripheral corneal incision. The tissue was placed in a glass well with a drop of balanced salt solution (BSS; Alcon Pharma GmbH, Freiburg, Germany) and the IPE cells were mechanically isolated from the underlying stroma without the use of enzymes to avoid possible damage and alteration of the cells. For injection the cells were suspended in 20 μL of balanced salt solution.

In all patients a 3-port pars plana vitrectomy was performed. A small retinotomy was made superior or temporal to the fovea; the fibrovascular membrane was seized with forceps and extracted slowly through the retinotomy. If necessary, the bleb retinal detachment was enlarged by injecting a stream of balanced salt solution. The IPE cells (approximately 12 000 cells per eye) were injected into the subretinal space using a Hamilton syringe fitted with a glass pipette tip. The syringe, which had been coated with autologous serum prior to filling it with IPE cells, was introduced and positioned just over the retinotomy. After the cells were injected, a fluid-air or a fluid-gas exchange was performed. In one case cataract surgery prior to iridectomy and vitrectomy was performed. Choroidal neovascular extraction and IPE transplantation was performed by 3 different surgeons (P.W., P.E., and K.U.B.-S.) following identical surgical protocol and technique.

All patients were scheduled for examination at 3 weeks and at 3, 6, 12, and 36 months. The 36-month examination took place between 36 and 48 months. Postoperative ophthalmic examinations included all the parameters of the preoperative examination. Three-year follow-up was completed in 17 patients; 3 patients were unavailable for follow-up at 3 years—2 patients died of diseases unrelated to the treatment and 1 patient was unable to attend the scheduled visit because of severe heart disease. All 3 patients lost to follow-up belonged to the group with classic CNV secondary to AMD.

The primary efficacy outcome was change in VA after 3 years (logarithm of the minimum angle of resolution [logMAR] and Snellen equivalent) of the operated-on eye compared with baseline VA. Data were described by mean (SD) and frequencies, as well as corresponding 95% confidence intervals. Detailed comparisons were performed using 95% confidence intervals corresponding to the least significant difference t test. Effects are considered statistically significant if the P value of the corresponding test drops below .05. To study the possible bias resulting from the missing observations at the 3-year follow-up visit, 2 populations were investigated. First, the data from the full analysis set with complete observations (n = 17) were evaluated. Next, data from the full analysis set using missing observation substitution (last observation carried forward) (n = 20) were evaluated. Computations were performed using SAS software (SAS Institute Inc, Cary, NC) under MacOS (Apple Computers, Inc, Cupertino, Calif).

Submacular surgery with CNV extraction was performed in 20 eyes of 20 patients; autologous IPE cells were transplanted in 19 of 20 patients. In 1 patient the retinal bleb could not be enlarged sufficiently to inject the cell suspension. The series consisted of 17 female and 3 male patients who had a median age of 67.9 years (age range, 33-85 years). Follow-up ranged from 36 to 48 months (mean, 40.2 months). Baseline characteristics and VA data of these patients are summarized in the Table. Preoperative best-corrected VA ranged from 20/1000 (logMAR, 1.7) to 20/125 (logMAR, 0.8), with a mean of 20/320 (logMAR, 1.2). Preoperative fluorescein angiography showed classic subfoveal CNV due to AMD in 9 eyes and occult subfoveal CNV in 7 eyes. The eyes of 4 younger patients showed either idiopathic CNV or CNV associated with dominant drusen or pathologic myopia. Choroidal neovascularization size ranged from 0.8 to 10.7 disc diameters, with a median of 4.1 disc diameters (Table). Functional outcome did not correlate with lesion size in these patients.

Three years after IPE transplantation, best-corrected VA ranged from hand movements (logMAR, 1.8) to 20/100 (logMAR, 0.7), with a mean of 20/320 (logMAR, 1.2). Compared with preoperative VA, 1 patient showed improvement of 3 lines or more on the ETDRS chart, 13 patients retained stable VA with a change of less than 3 lines, and 3 patients showed decreased VA of 3 lines or more (Table). Of the 3 patients who showed decreased VA, 1 patient had proliferative vitreoretinopathy necessitating 2 revisional vitrectomies.

In fundus photographs transplanted cells appeared as dark spots in the area of the transplantation that appeared to remain stable in size and shape throughout the observation period (Figure). Throughout this 36 to 48 months’ follow-up period, the adjacent retina did not show evidence of immunologic rejection, such as inflammatory reactions with destruction of the transplanted cells or macular edema. Postoperative angiograms did not show signs of recurrence of the CNV at any time in any of the patients (Figure). None of the patients showed central fixation or responses over the transplanted areas using macular perimetry 3 years after submacular surgery.

All 19 phakic patients developed cataract on the operated-on eye and underwent cataract surgery; in 1 case phacoemulsification was combined with the initial vitrectomy and IPE translocation. Two patients experienced significant complications, namely, retinal detachment and proliferative vitreoretinopathy, during the first 2 months postoperatively and underwent revision vitrectomy with silicone oil tamponade at 3 months. The eye that could not be successfully injected with IPE cells developed a macular pucker, and that patient underwent revision vitrectomy with epiretinal membrane peeling 5 months after the initial surgery.

The main criterion of analysis was comparison of change in VA (logMAR) after 3 years vs baseline. Mean change in VA (logMAR) considering the 17 patients who completed the 36 months’ follow-up was not significant; mean change in all 20 patients substituting missing values by last observation carried forward showed the same tendency. In a 3-factor repeated-measures analysis of variance model with all factors treated as continuous, VA showed no dependency from age. However, logMAR values increased linearly, depending on baseline values (P<.001), and slightly over time.

The development of neovascularization may lead to irreversible damage to the RPE and photoreceptors, and removal of the membranes is associated with the traumatic loss of the RPE cell layer in AMD. Photodynamic therapy as the current standard in the treatment of at least predominantly classic subfoveal CNV in AMD can stabilize VA in most patients without the risk of surgery associated complications and adverse effects. However, only a few patients with exudative maculopathy are eligible for the treatment with photodynamic therapy.^{6 ,44 - 45} Furthermore, a considerable number of re-treatments are necessary because of recanalization or recurrence of the CNV. Surgical removal of the neovascular complex is technically feasible in almost all patients with exudative AMD; but the procedure is usually associated with traumatic loss of the RPE cell layer in AMD.^{46 - 47} Most studies that investigated the effect of submacular surgery in exudative AMD have concluded that VA in the best case can be stabilized or slightly improved after membrane extraction. To improve postoperative VA, membrane extraction should be accompanied by restoration of the diseased and surgically damaged Bruch membrane–RPE complex.^{14 - 15}

Gouras et al⁴⁸ postulated that RPE cell translocation to the area of RPE loss would reestablish functional and morphologic integrity of the retina-photoreceptor complex thus improving or restoring vision. Algvere et al^{19 - 20} transplanted human fetal RPE cells in patients with nonexudative and exudative AMD; however, the transplanted cells did not survive and proliferate, due to rejection of the allograft, insufficient attachment to the Bruch membrane, or other unknown causes. Algvere et al also demonstrated that transplantation of homologous adult human RPE cells to the subretinal space is feasible and that in patients with an intact blood-retinal barrier, namely, patients with dry AMD, the transplanted RPE cells survive for up to 12 months without adversely affecting the photoreceptors. To prevent rejection it would be necessary to transplant autologous RPE cells or to immune suppress the host after transplantation of allogenic RPE cells. The feasibility of isolating and transplanting autologous RPE cells at the time of CNV excision has been demonstrated by Binder et al²⁴ in a trial with 20 patients, who have been followed up for 12 months. Recently van Meurs et al⁴⁹ published the results of a series of 8 patients treated with a modified technique of autologous RPE cell translocation. However, the routine isolation and transplantation of viable autologous RPE cells at the time of CNV excision is challenging. Full-thickness transplants that were taken from areas adjacent to the CNV showed sequestration after a longer observation period.^{22 ,25} Furthermore, proliferative viteroretinopathy rates, as one of the main surgery-associated complications, have been relatively high in these first case series.^{24 ,49} Because autologous RPE cells are difficult to obtain in a sufficient number of viable cells and possibly carry the same defect as the subfoveal diseased RPE, it may be preferable to transplant autologous IPE cells that hypothetically could acquire RPE functions in the subretinal space. The hypothesis that IPE cells may acquire RPE functions in the subretinal space is supported by experiments in which IPE cells were transplanted as a suspension to the subretinal space of rabbits and of Royal College of Surgeons (RCS) rats. In a 20-week study, IPE cells transplanted to the subretinal space of rabbits survived without aid of immunosuppressive agents, formed a single layer of cells between the RPE and the photoreceptor cells layer, developed microvilli, attached to the retina and to the RPE layer, and demonstrated phagocytosis of photoreceptor outer segments without disturbing the morphology of the photoreceptor layer.³⁴ In the RCS rat, which is a model of AMD, transplanted IPE cells demonstrated the same characteristics and in addition prevented photoreceptor degeneration.^{32 ,50 - 51} Several studies examined the feasibility of transplanting autologous IPE cells to the subretinal space.^{43 ,52 - 55}

In this study, 1 of 17 patients who were examined 3 years or longer after autologous IPE transplantation showed significant improvement, and 13 patients showed stabilization of VA, a visual outcome similar to that reported after classic membrane extraction alone.^{12 - 16} Since visual loss after membrane extraction is in part caused by the destruction of photoreceptors, visual recovery after epithelial cell translocation could be expected only in areas where photoreceptors have not been damaged. Patients enrolled in this study had advanced neovascular macular degeneration with relatively poor VA at baseline; damage to the RPE and consequently to the photoreceptors was likely extensive. Furthermore, mean lesion size in our series was relatively large and greater compared to results of the VIP/TAP or the SST studies.^{45 ,56} In contrast to photodynamic therapy studies that showed better results in patients with smaller CNV size, the result of this study showed no effect of CNV size on visual outcome. Similarly, the results of the SST study showed no influence of CNV size on visual outcome.⁵⁷

Although good visual recovery was unexpected, the contribution of the transplanted autologous IPE cells to visual outcome in these patients is unclear. Iris pigment epithelia cell transplantation might result in better visual outcome in patients with advanced dry AMD with geographic atrophy, and hold promise for this subgroup of patients with AMD for whom therapeutic options are limited. In addition, the development of techniques that are less damaging to the RPE-retina complex followed by IPE transplantation should result in better visual outcome and perhaps restoration of reading VA. Even though IPE cells transplanted as a suspension assume proper morphology in rabbits, in humans proper morphology, and thus acquisition of functionality, of the transplanted cells may require transplantation of sheets of cells with apical and basal domains already established.³⁵ The assessment of cell fate in the subretinal space may be visualized in future studies with optical coherence tomography. Of particular interest is the observation that even after 3 years vascular membranes did not recur in any of the patients considering the high rate of recurrence when subfoveal neovascular membranes are removed without transplantation of cells.^{12 - 15} Similar observations were also reported by Binder et al²⁴ for patients who had undergone translocation of autologous RPE cells. If these observations can be confirmed, it might suggest that transplanted IPE or RPE cells may produce substances that inhibit the formation of blood vessels.

To our knowledge, in this study we provide for the first time long-term results after CNV extraction and transplantation of autologous IPE cells. The observations that autologous IPE cells are well tolerated for a period of 36 months or longer, and that no recurrence of neovascularization was observed in any of the patients are important. These findings suggest that with additional research, autologous IPE translocation combined with surgical excision of the CNV may become a suitable treatment option for patients with exudative AMD and may hold promise for the treatment of advanced dry AMD.

Correspondence: Gabriele Thumann, MD, Center of Ophthalmology, University of Cologne, Joseph-Stelzmann-Strasse 9, D-50935 Cologne, Germany (G.Thumann@uni-koeln.de).

Submitted for Publication: February 27, 2005; accepted March 9, 2005.