Three-dimensional Distribution of the Vitelliform Lesion, Photoreceptors, and Retinal Pigment Epithelium in the Macula of Patients With Best Vitelliform Macular Dystrophy FREE

Best vitelliform macular dystrophy (BVMD) was originally described as an autosomal dominant form of macular degeneration that presents in childhood with a yellow yolk-like or vitelliform lesion in the macula.¹ Fishman and coworkers² found that 76% of patients younger than 40 years of age retain 20/40 visual acuity or better in at least 1 eye, whereas 74% of patients older than 30 years of age have visual acuity of 20/100 or worse in at least 1 eye. In addition to vision loss, patients with BVMD manifest a characteristic abnormal electro-oculogram, with reduced light peak–to–dark trough ratios (Arden ratio, <1.5).³ The light peak is mediated by changing chloride conductance across the basolateral plasma membrane of the retinal pigment epithelium (RPE).⁴

In 1992, Best disease was linked to 11q13 by studying a 5-generation family with 29 affected members,⁵ and the localization was later refined to the pericentromeric region of chromosome 11.⁶ The responsible gene, VMD2 (now known as BEST1), was identified in 1998.^7- 8BEST1 has 11 exons that span 14.1 kilobases and encodes a 585–amino acid protein. To date, nearly 200 disease-causing mutations have been identified in BEST1.^9- 14 Several diseases have been linked to mutations in BEST1, including vitelliform macular dystrophy (both autosomal dominant and recessive), adult-onset foveal macular dystrophy, autosomal dominant vitreoretinochoroidopathy, and, most recently, retinitis pigmentosa.^15- 21

The localization of bestrophin-1 (BEST1), the protein encoded by BEST1, was determined to be in the basolateral plasma membrane of the RPE.²² Physiological studies have provided evidence that human BEST1 functions as a Ca²⁺-sensitive chloride channel^10,23- 25 or plays a role in the regulation of Ca²⁺ channels.^26- 29 Recently, a knock-in mouse model (W93C mutation in BEST1) revealed (1) enhanced accumulation of lipofuscin in the RPE and (2) debris that is thought to be unphagocytosed photoreceptor outer segments and lipofuscin granules in the subretinal space.³⁰

The structural phenotype of the vitelliform lesion has been a subject of debate. Morphological findings described in BVMD donor eyes include (1) the abundant accumulation of lipofuscin in the RPE³¹ (at least associated with some genotypes),³² (2) the mislocalization of the BEST1 protein,^17,33 and (3) photoreceptor degeneration over a morphologically intact RPE layer.^33- 34 None of the histopathologic analyses to date have sampled the vitelliform lesion because the patients who have been examined postmortem had largely progressed beyond this stage.

On the basis of histopathologic findings, researchers have suggested that BVMD affects a more diffuse region than that of the ophthalmoscopically visible lesion. Mullins et al³² suggested that the predilection of the macula for Best disease–related lesions resulted from a relative paucity in normal eyes of wild-type bestrophin in macular locations compared with extramacular locations, suggesting that the entire macula might exhibit photoreceptor or RPE abnormalities rather than being limited to the central macula where lesions are clinically evident. Histopathologic findings from both Weingeist et al³¹ and O’Gorman et al³⁵ suggested the accumulation of lipofuscin granules in the RPE with extrafoveal diffuse macular involvement.

The use of optical coherence tomography (OCT) makes it possible to anatomically examine the vitelliform lesion in vivo. For example, Querques et al³⁶ described the vitelliform lesion “at the level of the “RPE/photoreceptor complex,” and Ferrara et al³⁷ described the vitelliform lesion as existing above the RPE and under the tips of the photoreceptor outer segments. Some of these OCT-guided studies were performed for patients with a clinical diagnosis of BVMD without molecular confirmation of disease. A more precise anatomic understanding of BVMD could provide insight into the pathophysiology of the disease and clinically relevant information for future therapeutic investigation, particularly with respect to the relative effect of the disease on photoreceptors as compared with the RPE. In our study, we investigate the 3-dimensional distribution of the vitelliform lesion and the anatomical condition of the RPE and photoreceptors in the maculas of patients with an array of BEST1 mutations.

Patients were retrospectively recruited on the basis of a clinical diagnosis of BVMD made by a board-certified ophthalmologist and on the basis of molecular confirmation of a mutation in the BEST1 gene; the molecular testing was performed in our laboratory with bidirectional DNA sequencing using ABI 3730 sequencers (Applied Biosystems, Carlsbad, California). Normal controls were matched by age (within 5 years) to the patients with BVMD. These normal controls had normal outer retinas, as determined by a retinal specialist performing indirect ophthalmoscopy, and had no history of retinal disease, diabetes mellitus, or glaucoma.

All patients with BVMD and all control patients underwent OCT imaging using the Spectralis (Heidelberg, Germany) 3-dimensional volume, scan protocol (6.0 × 6.0 × 2.2 mm: 64 [y coordinate], 1048 [x coordinate], and 1024 [z coordinate] voxels, respectively). Color fundus photographs (Zeiss, Dublin, California) and visual acuities were obtained during the same visit. Using our validated, fully 3-dimensional OCT segmentation algorithm, we automatically determined 11 intraretinal surfaces, from the internal limiting membrane (layer 1) to Bruch's membrane (layer 11), in all macular OCT scans from both eyes for all patients (Figure 1).^38- 39

Using our OCT viewing software,⁴⁰ a retina specialist (C.K.) manually excluded a 3-dimensional region encompassing the fovea and, if present, the central Best disease–related lesion from further analysis. Each age-matched case-control pair of patients was then evaluated. For each control patient, a region of exclusion was manually selected to match the volume of the region excluded in the corresponding age-matched patient with BVMD. Measurements of photoreceptor equivalent thickness over the entire macular scan region (excluding the fovea and lesion area as described) were automatically determined, for each A-scan, as the distance in micrometers between surface 8 (photoreceptor inner segment/outer segment junction) and surface 10 (inner surface of the RPE). Measurements of RPE equivalent thickness over the identical scan region were determined, for each A-scan, as the distance in micrometers between surface 10 (inner surface of the RPE) and surface 11 (outer surface of the RPE). The average photoreceptor equivalent thickness and the average RPE equivalent thickness were determined from these surface thicknesses. The mean photoreceptor equivalent thickness and the mean RPE equivalent thickness were calculated by averaging the photoreceptor equivalent thickness and the RPE equivalent thickness of each A-scan, for each eye and for each patient. The photoreceptor equivalent thickness differences and the RPE equivalent thickness differences between patients with BVMD and control patients were then statistically analyzed using a paired t test in Excel (Microsoft, Redmond, Washington) at a significance level of .05. Our study was approved by the institutional review board for human subjects research at the University of Iowa and adhered to the tenets set forth in the Declaration of Helsinki.

A total of 15 patients with BVMD (30 eyes) and 15 control patients (30 eyes) were included. There were a total of 11 mutations in the BEST1 gene among the 15 patients with BVMD (Table). High-definition OCT findings revealed that the vitelliform lesion consists of material located in the subretinal space (Figures 2 and 3). Additionally, deposits of sub-RPE material were noted in the same space in which drusen accumulate, and several patients had a highly reflective dense nodule of material that was located under the RPE. This material is likely to be fibrotic scar tissue, given the similarity of this lesion to fibrovascular pigment epithelial detachments seen in age-related macular degeneration. There is an abrupt angle of the RPE contour due to the strong adherence of RPE cells to their basal lamina. Vitelliform lesions in the subretinal space are shown in Figures 2 and 3. In Figure 2, a vitelliform lesion seen during clinical examination is demonstrated to contain subretinal material on an OCT scan in a 68-year-old woman with a Gln316Pro mutation in BEST1. In this patient, the vitelliform lesion is localized to the subretinal space, with a gradual angle of departure of the reflectivity line that is typically seen with subretinal fluid accumulation. Another example of the subretinal location of the vitelliform lesion is demonstrated in a patient with a Tyr227Asn mutation in BEST1 and 20/25 visual acuity (Figure 3). Atrophic lesions were characterized by disruption of the outer retina and the RPE. This feature is shown in Figure 4, which reveals atrophy of the RPE and loss of normal architecture of the overlying outer retina in a 58-year-old man with a Lys30Arg mutation in BEST1. In another case (ie, a 54-year-old man), multilayered vitelliform material was observed (Figure 5). This patient with a Tyr227Asn mutation in BEST1 showed an extramacular lesion notable for material in multiple layers, including the sub–inner segment/outer segment junction, sub–outer segment tips, and sub-RPE space (Figure 5). In Figure 6, a sub-RPE fibrotic nodule is shown in a 31-year-old woman with a Thr241Asn mutation. The visual acuity in this eye was 20/63.

Automated measurements of photoreceptor equivalent thickness and RPE equivalent thickness could be performed for all patients. The mean photoreceptor equivalent thickness was 28.3 μm for patients with BVMD and 21.8 μm for control patients, an average difference of 6.5 μm (95% CI, −11.12 to −1.83 μm; P < .01). The mean RPE equivalent thickness was 24.5 μm for patients with BVMD and 25.1 μm for control patients, a nonsignificant difference (95% CI, −1.19 to 2.21 μm; P = .53). The measurements of photoreceptor equivalent thickness and RPE equivalent thickness for each patient are shown in Figures 7 and 8.

We further sought to determine how genotype influenced the anatomical features of BVMD in the 15 patients with the disease. Only patients with the Tyr227Asn mutation (2 of the 3) showed extramacular flecks (compared with 0 of the 12 patients with other mutations). Fibrotic nodules were observed in 3 of the 3 patients (and 5 of the 6 eyes) with an Asp302Ala mutation. No other striking genotype-specific structural features were noted.

In our study, we used spectral-domain OCT to characterize the 3-dimensional anatomy of macular lesions in patients with confirmed mutations in the BEST1 gene. The most common OCT-detected phenotypes that we observed were vitelliform material located in the subretinal space, fibrotic nodules under the RPE, and disruption and atrophy of the outer retina and the RPE. We also found that the retina adjacent to these ophthalmoscopically visible lesions was abnormal. Specifically, photoreceptor equivalent thickness was 6.5 μm thicker, on average, in patients with BVMD than in control patients, whereas, on average, the RPE of patients with BVMD was the same thickness as the RPE of control patients. This finding suggests that, although the abnormal protein encoded by BEST1 is expressed in the RPE, its primary anatomical impact is at the photoreceptor level. These data are consistent with histopathologic findings of an attenuated outer retina overlying an intact RPE.³³ The photoreceptor equivalent thickness in patients with BVMD compared with control patients increases with age, perhaps reflecting an age-related accumulation of outer segment debris as was seen in the histopathologic analysis of the knock-in mouse model.³⁰

The diffuse photoreceptor involvement in BVMD that we found is consistent with the well-known abnormality of the electro-oculogram, a large-scale voltage-dependent phenomenon believed to originate from chloride conductance across the basolateral plasma membrane of the RPE. This phenomenon would be difficult to explain by a disease process limited to the macula.^4,41 Arden et al⁴¹ found the electro-oculogram to be normal in patients with localized chorioretinal disease but abnormal when damage is diffuse and affects the majority of the choroid or the RPE. Lending further support to a model of diffuse involvement in BVMD is the immunohistochemical finding that bestrophin (at some level) is expressed to some degree in both the macular RPE and the peripheral RPE.³²

Given prior physiological findings that bestrophin affects the function of the RPE by modulating ion channels, it is likely that the ionic milieu of the subretinal space is altered in BVMD. Our OCT analyses suggest that BVMD may be caused by RPE-mediated changes in the ionic environment of the subretinal space, leading to aberrant interaction between photoreceptors and the RPE, resulting in the accumulation of fluid and outer segment debris in the subretinal space. In normal individuals, the interphotoreceptor matrix is responsible for the tight adhesion of the photoreceptors to the RPE, and structural differences in this matrix have been reported in foveal and extrafoveal locations.⁴² Such regional variations in this matrix may explain the macular location of the vitelliform lesion in BVMD, despite diffuse photoreceptor involvement. Given the light-dependent nature of the electro-oculogram, it is also possible that altered bestrophin function affects other light-dependent events such as the circadian phagocytosis of outer segment tips. If so, the differences between the anatomy of the central macula and the anatomy of the more anterior retina could explain the location of the most characteristic BVMD-related lesion.⁴³

Although it is not possible to draw conclusions regarding a phenotype-genotype correlation given the small numbers of patients with the same mutation, 2 suggestive trends were noted. Two of the 3 patients with a Tyr227Asn mutation had multifocal lesions compared with 0 of 12 patients with other genetic variants, and all 3 patients with the Asp302Ala GAT>GCT mutation revealed a fibrotic nodule at a young age, suggesting that the latter variant may be associated with a more severe fibrosis.

A notable limitation of our study is that it is not longitudinal, and although most patients have been followed for years, typically only 1 spectral-domain OCT image was available per eye. Future studies should include OCT analysis of photoreceptor equivalent thickness and RPE equivalent thickness in individual patients over time to document lesion progression.

In summary, we have provided additional evidence that the vitelliform material described in BVMD is located above the RPE and below the outer segment tips in the classic vitelliform lesion. We also showed that a sub-RPE phenotype exists that may best be described as a fibrotic scar. It is interesting to note that the photoreceptor cells overlying these fibrotic scars can survive for years and support good visual acuity (Table). Lastly, we found that the OCT correlate of photoreceptor outer segments (photoreceptor equivalent thickness) was, on average, 6.5 μm thicker among patients with BMVD than among control patients, whereas the OCT correlate of RPE thickness (RPE equivalent thickness) was not statistically different between patients with BVMD and control patients. Thus, although bestrophin is clearly expressed in the basolateral membrane of the RPE, the greatest anatomical effect of bestrophin dysfunction is a diffuse accumulation of material at the photoreceptor level.