The pathogenetic mechanism underlying retinal degeneration in choroideremia is not clearly understood, but may be due to a deficiency in the prenylation of multiple Rab proteins.48
Some histological reports of eyes with advanced CHM have shown lack of OS phagocytosis by the RPE,19
while others suggested secondary loss of RPE and photoreceptors caused by choroidal atrophy.20
The eyes from a 30-year-old man with CHM showed independent degeneration of the choriocapillaris, RPE, and photoreceptors, but also reported significantly better visual function than would be predicted by the retained RPE and choriocapillaris, perhaps suggesting that RPE or choroidal atrophy precedes photoreceptor degeneration.14
While some OCT studies of CHM patients show early photoreceptor IS/OS and ONL degeneration in the presence of normal21
or attenuated RPE,49
others demonstrate preservation of the ONL and OS in areas with attenuated or absent RPE.50
Based on histological studies of female carriers, authors have concluded that the primary defect was at the level of RPE,15
or the retina, RPE, and choroid simultaneously.45
Given the lack of clarity regarding the initial site of disease in the published literature, we examined retinal structure at high resolution in living eyes with CHM in an effort to shed light on the pathogenesis of this disease. We observed characteristic FAF findings in all carriers similar to the pattern reported in animal models6
and in prior studies of human CHM carriers.51
It has been proposed that the observed hyperautofluorescence results from rod loss; the outer segments of degenerating rods are digested by the RPE, causing increased lipofuscin accumulation and hyperautofluorescence.6
Hypoautofluorescence, in contrast, occurs secondary to loss of RPE cells and thinning of the RPE layer.6
Patches of hypoautofluoresence in the two asymptomatic carriers who had normal ERG findings and full visual fields correlated with patches of hypopigmentation throughout the fundus. Prior work has reported fundus hypopigmentation, RPE mottling, and reduced dark-adapted ERGs as the first manifestations of complete deletion of the CHM
gene in a 4-year-old boy, with dark-adapted cone sensitivity abnormalities in his obligate carrier mother.50
Since our asymptomatic carriers had normal dark-adapted ERGs but showed hypopigmented fundus findings and attenuation of band 3 on SD-OCT that correlated with areas of hypoautofluoresence, degeneration of the contact cylinder between the RPE apical processes and the external portion of the cone outer segments, possibly including the cone outer segment tips,42
may be the first manifestation of CHM in these carriers.
Although AOSLO images were not acquired in areas of abnormal autofluorescence, we observed abnormal cone structure in areas where the band 3 was disrupted in the presence of an intact band 2 on SD-OCT. In the present study all subjects except for two carriers and one affected male with early disease showed attenuation of band 3 with an overlying intact band 2. Others have interpreted this finding to indicate that RPE cells degenerate prior to photoreceptor degeneration; CHM eyes showed abnormally thin RPE underlying normal OS and ONL layers, in contrast to the pattern seen in patients with RP, where thinning occurs first in the OS, then in the ONL with relative preservation of the RPE.49
These observations may suggest that the RPE is the primary layer affected by CHM.
However, some studies have defined the fourth outer retinal band as follows: As illustrated in Supplementary Figure S1 (see Supplementary Material and Supplementary Fig. S1, http://www.iovs.org/content/54/2/950/suppl/DC1
), band 4a represents a combination of rod outer segment tips and apical RPE, and 4b corresponds to the region of basal RPE and Bruch's membrane, at extrafoveal locations.41
In our images of CHM patients, the delineation between sublayers 4a and 4b was ambiguous in most scans, but was more evident in scans acquired using the Bioptigen system (see Supplementary Material and Supplementary Fig. S1, http://www.iovs.org/content/54/2/950/suppl/DC1
, bottom panels; A, A, C). In A, A, and C, regions of degeneration showed preservation of band 2 in the absence of band 3 with preservation of band 4, perhaps indicating degeneration of the interface between cone outer segments and apical RPE processes. Regions with loss of band 3 in the presence of preserved bands 2 and 4 may represent selective cone photoreceptor outer segment damage in regions with intact RPE cells, which would be perfectly consistent with the apparently normal RPE autofluorescence at those regions as shown in A.
However, AOSLO images at regions where the band 3 was attenuated but band 2 was intact showed disrupted cone photoreceptors in an affected male (, insets D2 and C1) and increased cone spacing in two symptomatic carriers (, insets A1, A3, and B1). This finding suggests cone photoreceptors were abnormal in areas with disruption of the contact cylinder between the RPE apical processes and the cone outer segment tips, in the presence of preserved inner segment ellipsoid (ISe) bands on SD-OCT.42
High-resolution images of the cone photoreceptor mosaic may provide insight into the relationship between photoreceptor and RPE cell death, and suggest that the contact cylinder between the RPE apical processes and the cone photoreceptor outer segments degenerates even in the presence of an apparently intact ISe band observed on SD-OCT. The attenuation or loss of this layer could be due to loss of the RPE apical processes, the cone photoreceptor outer segments, or the interface between these structures.
The SD-OCT bands 4a and 4b are complex and represent interdigitations between rod and cone outer segments and the apical microvilli of the RPE,41
which are not well resolved or discriminated using standard, commercially available SD-OCT systems including the systems used in the present manuscript. It is possible that higher resolution OCT systems,52
such as adaptive optics OCT,53,54
could distinguish between cone and rod outer segment tip loss and RPE damage. Since disruption or loss of the OS/RPE, cone outer segment tip, and rod outer segment tip bands may represent primary degeneration of photoreceptors or simultaneous photoreceptor and RPE cell degeneration, the ambiguity in our images highlights areas for future investigation using AO-OCT to probe these interfaces in choroideremia patients, which may provide additional insight into the primary site of degeneration in choroideremia.
In other male patients, areas of normal cone spacing were seen 2 to 4 degrees peripheral to the fovea while cone spacing was increased near the fovea. Unlike patients with primary cone degeneration, in which central cone spacing is increased even in early stages—or retinitis pigmentosa (RP), in which cones are well-preserved centrally but become increasingly sparse with eccentricity at the edges of scotomas23
—the present manuscript demonstrates abnormal cones at the fovea in affected males with CHM, while cone spacing Z-
scores were more normal at the edges of the preserved retinal regions (C, D). Possible explanations for this finding include the following. First, cone spacing may appear more normal at the edge of degeneration because the imaging properties of degenerating cones change. The degenerating photoreceptors were observed to form ORT, which have been reported in imaging studies of CHM,43,44
Bietti crystalline retinopathy,43,56
retinal degeneration associated with a mitochondrial DNA mutation (A3242G), central serous retinopathy, age-related macular degeneration, and other diseases such as pattern dystrophy associated with choroidal neovascularization.43
ORT typically occur in areas with disruption of outer retinal architecture and relative preservation of the photoreceptor layer with preserved ISe bands, often overlying RPE damage or at the margin between preserved and absent RPE and photoreceptor layers.43
Zweifel and colleagues have proposed that ORT may represent a final common pathway in a variety of retinal degenerative conditions that are initiated by loss of interdigitation of the outer segments with RPE or degeneration of the RPE, followed by disruption of attachment to neighboring neural elements such as Müller cells with outward folding of the photoreceptor layer until opposite sides of the fold establish contact and form new lateral connections, reconstituting the IS/OS junction and forming a tubular structure.43
The occurrence of ORT in CHM may indicate primary degeneration of RPE cells with secondary effects on photoreceptors, and may affect cone spacing measures as the outer retina forms tubules adjacent to regions of atrophy. In addition, the interlaminar bridges first reported by Jacobson and colleagues21
may contribute to ORT formation, and similarly may affect the imaging properties of cones near the advancing margins of degeneration. The interlaminar bridges have been attributed to hyperplastic Müller cells in response to retinal degeneration, with altered optical properties that affect their imaging characteristics21
; a similar phenomenon may account for the apparent decreases in cone spacing observed in the present manuscript at the edges of degeneration. Second, cones may be more abnormal at the fovea than at the edge of atrophy. This is unlikely for 2 reasons: first, automated perimetry showed foveal sensitivity was better preserved, while sensitivity was more abnormal at retinal locations with increased eccentricity; and second, SDOCT images showed that the outer retinal layers were better preserved near the fovea than in extrafoveal locations; but it is possible that the cones are structurally more intact, although less normal functionally, at the margins of degeneration. Even if cone spacing is more normal at the peripheral edges of the AOSLO montages, the cones at the margins of degeneration are likely not normal; cones in these regions appeared crenelated and shrunken in comparison with normal cone mosaics (, insets A3, B3, C1, and D1). Third, rods may be present at the edge of atrophy. Since rods are smaller than cones, quantitative measures of photoreceptor spacing that include rods will produce a lower mean spacing than ones that include only cones. However, cone spacing measures by Fourier spectrum analysis of AOSLO images showed similar results to cone spacing obtained by manually selecting cones, suggesting rods were not included in our manual cone spacing measures.
In primary photoreceptor degenerations such as RP, cones are better preserved at the fovea and cone spacing increases near the edge of the remaining photoreceptor mosaic, beyond which RPE cell mosaics may be seen.23
RPE mosaics were not observed peripherally in the CHM patients in this study, perhaps because RPE cells degenerate primarily or simultaneously with photoreceptors. Alternatively, CHM may affect the imaging properties of RPE cells, which may require normal melanin content and distribution to be visualized; RPE cells in CHM carriers have abnormal melanin granule distribution16
and lack basal microvilli and infoldings.45
The present manuscript also demonstrates distinct information that can be acquired using different AO imaging modalities (AOSLO and flood-illuminated AO). At the fovea of a young CHM patient with well-preserved visual acuity and cone spacing, we observed low-frequency structures simultaneously with overlying cones using an AO flood-illuminated system. The low-frequency structures were consistent in size with RPE cells at the location imaged, and were associated with a similar number of overlying photoreceptors as have been associated with RPE cells in histological studies.57–61
Gao and Hollyfield59
reported RPE cell center-to-center spacing ranging from 12.45 μm at the fovea to 13.87 μm at 120 μm from the foveal center, and reported a density of 22 cones/RPE cell at the fovea; Watzke and associates reported center-to-center spacing of 14 μm within 250 um of the foveal center,61
and Dorey and associates reported RPE cell sizes of 10.12 μm58
; however, these measures from histological studies of excised human ocular tissues may reflect artifact introduced by histological processing. Roorda and colleagues observed RPE cells directly in AOSLO images after extensive cone photoreceptor loss with cell spacing of 14.85 to 15.2 μm in locations ranging from 100 to 750 μm.47
Morgan and colleagues62
reported RPE cell nearest neighbor distances ranging from 10.8+/−1.7 μm at 5 degrees superior to the fovea using AOSLO autofluorescence imaging in normal humans, but no locations comparable with those in the present image were studied. The low-frequency structures in the present study measured 12.23 μm at a distance of 130 μm from the fovea, which is comparable in size with other reports. The structures observed in the present study have not been observed with the same imaging system in normal eyes or other eyes with retinal degeneration, and were not visible on the AOSLO image of the same location due to confocality of the SLO system; light not originating from the focal plane of the retina was excluded through the use of a pinhole conjugate to the retinal focal plane, thus increasing the contrast of the final image.63
However, scattered light from the fundus that comes from the RPE is usually masked by the light that comes from the overlying photoreceptor mosaic, and therefore, it is unusual to visualize RPE cells at a region where cone mosaics are intact.47
The findings in suggest that RPE cells are more visible in CHM patients than in normal eyes when imaged using flood illuminated adaptive optics systems with a near-infrared light source. The mechanism for this enhanced visibility may be due to abnormalities of melanin deposition in RPE cells with mutations in the CHM gene. Alternatively, enhanced RPE cell visibility could be due to selective abnormalities at the level of cone inner segments which cause a relative reduction of wave-guided light originating from the cones without reducing backscattered light originating from the RPE.
In summary, high-resolution retinal imaging demonstrated abnormalities of FAF, retinal layer morphology, and cone morphology and spacing in CHM patients and carriers with a spectrum of clinical characteristics. Our high-resolution images of cone structure demonstrate a pattern of photoreceptor degeneration with ORT, microcystic inner retinal edema, and cone loss at the fovea with smaller, atrophic-appearing cones at the edge of degeneration which has not been observed in patients with primary photoreceptor degenerations.23,32
Cone abnormalities in regions where SD-OCT band 3 showed attenuation in regions in which band 2 was intact suggest early involvement of RPE cells, likely in association with simultaneous photoreceptor cell degeneration because cone spacing was abnormal in many regions in which band 2 was intact. Longitudinal studies using high-resolution retinal imaging techniques may provide further insight into the ways cones and RPE cells are affected by CHM
mutations. Furthermore, in vivo surrogate markers of photoreceptor structure such as those presented in the current manuscript should prove useful in evaluating the safety and efficacy of experimental therapies for CHM, including gene replacement trials which are underway.