Figure 1. Appearance of central atrophy and pisiform flecks in Stargardt disease on color fundus photography and autofluorescence imaging of patient 1. A, Color fundus photograph demonstrating typical yellow fleck lesions throughout macula and a central area of retinal and retinal pigment epithelial atrophy. B, The same fundus imaged with fundus autofluorescence imaging with excitation at 488 nm demonstrating a defined area of sharply demarcated hypoautofluorescence, indicating an area of retinal pigment epithelial loss. Additional areas of mottled hypoautofluorescence in parafovea were observed adjacent to the central area of atrophy. Hyperautofluorescent flecks in concentric patterns are found extending out from the central macula. C, The same fundus imaged with near-infrared autofluorescence imaging with excitation at 795 nm demonstrating more diffuse and widespread abnormalities than seen on color fundus photography or fundus autofluorescence. There are proportionally more hypoautofluorescent flecks than hyperautofluorescent flecks.
Figure 2. Longitudinal changes in autofluorescence associated with macular atrophy in Stargardt disease in patient 1 showing fundus autofluorescence (FAF) and near-infrared autofluorescence (NIA) in the right (A) and left (B) eyes. Areas of clinically evident macular atrophy have a corresponding area of markedly decreased autofluorescence on FAF imaging; atrophic regions are outlined and can be observed to enlarge by contiguous spread with time. The areas of hypoautofluorescence on NIA imaging were larger than the corresponding areas of clinical atrophy and extended beyond the borders of the outlined atrophic areas.
Figure 3. Longitudinal changes in fundus autofluorescence of fleck lesions in Stargardt disease captured using 488-nm excitation, with autofluorescence images of patient 2 (A), patient 3 (B), and patient 4 (C) with Stargardt disease. Numbers 1, 2, and 3 indicate separate time points during follow-up. The color composites superimpose all 3 images as color-coded layers (blue indicates time point 1; green, time point 2; and red, time point 3). A, Spread of hyperautofluorescent fleck lesions in a centrifugal wave centered on the fovea. The color composite demonstrates the progressive outward spread. B, Hyperautofluorescent flecks can be seen extending contiguously as trails in a center-to-peripheral direction. Insets, Magnified views of 1 representative fleck (in boxes); the fleck can be observed to progress with a hyperautofluorescent leading edge, leaving behind a hypoautofluorescent trailing edge. The color composite demonstrates fleck progression. C, A similar example of contiguous fleck progression. Inset, Magnified view of 1 fleck (in box) as it extended contiguously over time.
Figure 4. Longitudinal changes in near-infrared autofluorescence of fleck lesions in Stargardt disease captured using 795-nm excitation in the right (A) and left (B) eyes of patient 2. The near-infrared autofluorescent patterns occur symmetrically between the eyes. Diffuse areas of hypoautofluorescence with few hyperautofluorescent flecks were observed mostly near the outer border of the affected area. These areas of hypoautofluorescence demonstrate outward expansion from the center to the periphery of the macula. Boxes indicate areas of hyperautofluorescence observed to decrease as a function of time.
Figure 5. Correlative analysis of fundus autofluorescence (FAF) and near-infrared autofluorescence (NIA) imaging of fleck lesions in patient 2 (A), patient 3 (B), and patient 1 (C) with Stargardt disease. Correlation between modalities shows that flecklike lesions are matched between FAF and NIA images. Color composites (right) superimpose FAF (green) and NIA (red) images. Orange indicates a superimposition of FAF and NIA hyperautofluorescent signals, where a subset of lesions was found to be hyperautofluorescent on both FAF and NIA (solid boxes). Dashed boxes indicate a subset of lesions found to be hyperautofluorescent on FAF but hypoautofluorescent on NIA, which appear green in the color composite. No lesions appeared hyperautofluorescent on NIA but hypoautofluorescent on FAF. A few lesions appearing as red on the color composite were strongly hyperautofluorescent on NIA and close to background autofluorescence on FAF.
Figure 6. Graphical model representing the time-dependent changes in autofluorescence on fundus autofluorescence (FAF) and near-infrared autofluorescence (NIA) imaging in fleck lesions of Stargardt disease.
Cukras CA, Wong WT, Caruso R, Cunningham D, Zein W, Sieving PA. Centrifugal Expansion of Fundus Autofluorescence Patterns in Stargardt Disease Over Time. Arch Ophthalmol. 2012;130(2):171-179. doi:10.1001/archophthalmol.2011.332
Author Affiliations: Division of Epidemiology and Clinical Research (Dr Cukras), Ophthalmic Genetics and Visual Function Branch (Drs Cukras, Caruso, Zein, and Sieving), Office of the Scientific Director (Dr Wong), and Office of the Clinical Director (Ms Cunningham), National Eye Institute, National Institutes of Health, Bethesda, Maryland.
Objective To study the longitudinal changes in autofluorescence in Stargardt disease to reveal aspects of disease progression not previously evident. Changes in autofluorescence reflect changing fluorophore compositions of lipofuscin and melanin in retinal pigment epithelial cells, which has been hypothesized to contribute to Stargardt disease pathogenesis.
Methods We examined the temporospatial patterns of fundus autofluorescence with excitation at both 488 nm (standard fundus autofluorescence) and 795 nm (near-infrared autofluorescence) in a longitudinal case series involving 8 eyes of 4 patients (range of follow-up, 11-57 months; mean, 39 months). Image processing was performed to analyze spatial and temporal cross-modality associations.
Results Longitudinal fundus autofluorescence imaging of fleck lesions revealed hyperautofluorescent lesions that extended in a centrifugal direction from the fovea with time. Patterns of spread were nonrandom and followed a radial path that left behind a trail of diminishing autofluorescence. Longitudinal near-infrared autofluorescence imaging also demonstrated centrifugal lesion spread but with fewer hyperautofluorescent lesions, suggestive of more transient hyperautofluorescence and more rapid decay at longer wavelengths. Fundus autofluorescence and near-infrared autofluorescence abnormalities were spatially correlated with each other, and together they reflect systematic progressions in fleck distribution and fluorophore composition occurring during the natural history of the disease.
Conclusions Stargardt disease fleck lesions do not evolve randomly in location but instead follow consistent patterns of radial expansion and a systematic decay of autofluorescence that reflect changing lipofuscin and melanin compositions in retinal pigment epithelial cells. These progressive foveal-to-peripheral changes are helpful in elucidating molecular and cellular mechanisms underlying Stargardt disease and may constitute potential outcome measures in clinical trials.
Clinical hallmarks of Stargardt disease include the accumulation of yellow flecks in the retina in the earlier phase1 and the onset of central retinal and retinal pigment epithelial (RPE) atrophy and central vision loss in the later phase.2 Histopathological studies indicate that Stargardt disease–related fleck lesions are composed mainly of lipofuscin,3- 5 a by-product of the visual cycle that consists of multiple fluorophores, including A2E.6The accumulation of lipofuscin in the retina occurs normally with aging7 but has an accelerated course in certain retinal diseases such as Stargardt disease.8
The identification of ABCA4 as a causative gene for Stargardt disease9 has led to the appreciation that a lack of ABCA4 function results in the accumulation of lipofuscin and related compounds such as A2E10 in photoreceptors and ultimately in RPE cells.11 The way accumulation of lipofuscin, A2E, and other bisretinoid adducts subsequently leads to RPE and photoreceptor cell death in Stargardt disease is currently being investigated. While these pathways are not completely elucidated, research has demonstrated that lipofuscin-related compounds can act as photosensitizers, increasing the production of reactive oxygen species12,13 that through the chemical modification of proteins and DNA could induce RPE cell death.
The accumulation of lipofuscin in the RPE cell may also detrimentally alter the composition and distribution of other intracellular components such as melanin. Melanin, a compound important in RPE homeostasis7 and present intracellularly as melanin granules, is concentrated in RPE cells at the fovea, where RPE cell height is increased and more melanin granules per RPE cell are found.14 With aging, melanin, unlike lipofuscin, decreases in the RPE; melanin granules in RPE cells decrease in number as those containing lipofuscin accumulate.7These trends have led to the hypothesis that melanin in RPE cells may regulate lipofuscin accumulation and/or the photo-oxidation of lipofuscin-related compounds.8,15 How these melanin-lipofuscin interactions figure in the accelerated course of lipofuscin accumulation in Stargardt disease has not been fully explored.
The fluorescent properties of both lipofuscin and melanin enable them to be quantified in the eye noninvasively by in vivo autofluorescence imaging. Delori et al16 described a detailed method for measuring lipofuscin in the human fundus in vivo. This method was used to demonstrate that patients with Stargardt disease display increases in fundus autofluorescence (FAF) with spectral properties consistent with lipofuscin accumulation.17 Other reports have also described FAF findings in Stargardt disease, reporting characteristic patterns of increased fluorescence in early disease and decreased autofluorescence in cases of central atrophy.18- 22 In addition to standard FAF imaging, which uses excitation wavelengths in the blue part of the visible spectrum and detects lipofuscin autofluorescence, autofluorescence imaging using longer or near-infrared wavelengths—termed near-infrared autofluorescence (NIA) imaging—likely detects ocular melanin autofluorescence.14,23 Imaging of NIA has also been used in imaging Stargardt disease.18,24
While it has been previously hypothesized that changing lipofuscin and melanin content in RPE cells may compose part of the pathobiology of Stargardt disease, the longitudinal study of autofluorescence patterns in Stargardt disease, which may reveal changing patterns of lipofuscin and melanin in the fundus, has not previously been characterized to our knowledge. In this study, we used both FAF and NIA imaging to follow the progression of flecklike changes and central atrophy in Stargardt disease and identified a remarkable pattern of centrifugal progression of patterns of fleck creation and dissolution in the 8 eyes of 4 patients whom we have followed up for 11 to 57 months. This progression is best appreciated when viewed as a video. This novel observation regarding the natural history of Stargardt disease may reflect underlying retinal gradients including the distribution of cone photoreceptors, RPE cell density, and intracellular melanin and macular pigment, which are highest at the fovea and decrease with increasing distance from the fovea. These new observations provide features that may be useful in the development of new quantitative outcome measures in clinical trials. Additionally, the dynamic patterns of changes of FAF and NIA reflect cumulative alterations in lipofuscin and melanin content of RPE cells that can assist in elucidating intracellular processes underlying disease pathophysiology.
Eight eyes of 4 unrelated patients with a clinical diagnosis of Stargardt disease were selected as a subset of a larger study based on the presence of longitudinal FAF data during at least 3 time points. Study patients were examined in the eye clinic of the Ophthalmic Genetics and Visual Function Branch at the National Eye Institute, National Institutes of Health, Bethesda, Maryland, between January 2004 and December 2009. Informed patient consent and local institutional review board approval were obtained for this retrospective study, which was conducted in accordance with the ethical standards of the 1964 Declaration of Helsinki. The clinical diagnosis was established based on patient and family history, ophthalmoscopy findings, visual field testing findings, and/or full-field or multifocal electroretinography according to the guidelines of the International Society for Clinical Electrophysiology of Vision. Ages of the patients at the first visit ranged from 9 to 44 years. The results of ABCA4 genetic testing were available for 2 patients. The length of follow-up ranged from 11 to 57 months (mean, 39 months). The general demographic and baseline information of these patients is summarized in the Table.
Color fundus photography (CFP) was performed using a standard digital imaging system (OIS, Sacramento, California). Imaging of FAF was performed with a confocal laser scanning ophthalmoscope (HRA2; Heidelberg Engineering, Heidelberg, Germany) using an excitation wavelength of 488 nm and a barrier filter at 500 nm. Imaging of NIA was performed with the same instrument using an excitation wavelength of 795 nm with a band-pass filter with a cutoff at 810 nm. Images were captured at a rate of 8.8 frames/second, and a total of 15 images were averaged to generate the final autofluorescence image. Images with motion artifacts were excluded from the averaging process.
Longitudinal analysis of fundus images was performed by aligning consecutive images to create a video file using ImageJ version v1.441i software (National Institutes of Health). Automated and manual alignment of images was performed using stationary landmarks such as the optic nerve or retinal vessels. Correlative analyses between imaging modalities were performed by importing aligned images into Photoshop CS4 version 11.0.1 (Adobe, San Jose, California) as semiopaque images layered serially on top of each other. Images were colorized according to time point or modality to facilitate correlation and comparison.
Two main defining clinical characteristics of the disease, central atrophy and yellow subretinal pisiform flecks, typically characterized on CFP, could be correlated with features on FAF and NIA imaging. Central atrophy, seen as a circumscribed patch of depigmented RPE on CFP (Figure 1A), corresponded to a clearly demarcated area of uniform decreased autofluorescence on FAF imaging (Figure 1B) that was surrounded by a border of patchy, mottled hypoautofluorescence. On NIA imaging (Figure 1C), this area of central atrophy also demonstrated decreased autofluorescence; however, the borders of this zone were not well defined and extended beyond the area of central atrophy as visualized on CFP and FAF imaging.
Pisiform flecks, typically characterized as yellow lesions on CFP, can also be visualized on both FAF and NIA imaging. On FAF imaging, the flecks corresponded to areas where the autofluorescence signal differed from the overall background level. Most flecks seen on FAF imaging had increased autofluorescence (hyperautofluorescence) that exceeded background levels by varying degrees (Figure 1B); some of the flecks were also surrounded by a ring of decreased autofluorescence. A minority of flecks were primarily hypoautofluorescent. Conversely, on NIA imaging (Figure 1C), most flecks were hypoautofluorescent, with relatively few flecks appearing hyperautofluorescent. These few hyperautofluorescent flecks were sometimes surrounded by a ring of hypoautofluorescence.
These observations indicate that the clinical hallmarks of Stargardt disease were associated with abnormal autofluorescence properties that varied in intensity and extent depending on the wavelength of the illumination used. They also reveal that flecklike lesions, appearing mostly homogeneous on CFP, may be composed of different fluorophores present at varying levels.
We followed changes in the appearance of central atrophy in study eyes on autofluorescence imaging. Figure 2 displays sequential FAF images of fellow eyes from a patient with Stargardt disease with bilateral central atrophy. On sequential FAF imaging, the sharply demarcated area of central atrophy in both eyes gradually enlarged by expansion into contiguous areas. In the left eye, the emergence of a new island of atrophy was also observed (Figure 2B). New areas of atrophy appeared to emerge and extend selectively into locations demonstrating mottled hypoautofluorescence, akin to the pattern of progression of geographic atrophy in age-related macular degeneration.25 On the NIA image (Figure 2), the area of hypoautofluorescence contained but extended beyond the borders of central atrophy as seen on CFP and FAF imaging.
While pisiform flecks have been thought to emerge and increase in number during the natural history of Stargardt disease, the details of their progression with time have not been previously examined using autofluorescence imaging to our knowledge. Study eyes were imaged longitudinally using FAF imaging, and resulting images were spatially aligned to follow progressive changes in fleck distribution. Examination of all study eyes revealed a number of characteristic progression features. The number of flecks located in the macula increased radially as a function of time. In addition, fleck accumulation exhibited a pattern of centrifugal addition of new flecks beginning from the fovea and extending toward the outer edges of the macula. The extent and rate of this centrifugal extension varied considerably between patients. Figure 3A shows a study eye in which fleck distribution spread in a circular wave centered on the fovea; the superimposition of images of flecks colorized as a function of time illustrates the centrifugal stepwise extension of the wave. Figure 3B and C illustrate study eyes with fewer flecks at baseline in which the pattern of outward spread of fleck lesions proceeded at a slower rate. In these examples, the pisiform flecks extended in a radial direction following a continuous path (Figure 3B and C). Estimates for the average rates of centrifugal extensions of flecks ranged from 14 to 58 μm per month. The dynamic nature of fleck progression can be most clearly depicted by viewing consecutive FAF images as a video of sequential temporal images aligned spatially and stacked on top of each other (video 1, video 2, and video 3).
The natural history of fleck progression is also characterized by progressive changes on the level of FAF signal associated with flecks. Newer flecks appear hyperautofluorescent on FAF imaging, while more posteriorly located older flecks become progressively more hypoautofluorescent with time. In Figure 3A, fleck progression can be visualized as a hyperautofluorescent wave moving outward, leaving behind a wake of hypoautofluorescent flecks. In Figure 3B and C, flecks extend radially with a hyperautofluorescent leading edge, leaving behind a hypoautofluorescent trailing edge.
Progressive fleck-associated changes are also observed on NIA imaging. Longitudinal imaging of study eyes revealed a similar pattern of centrifugal spread of autofluorescence alterations from the fovea (Figure 4). The overall pattern on NIA imaging appeared as a progressive spread of predominantly hypoautofluorescent lesions. A few lesions demonstrated increased autofluorescence, and these tended to be distributed near the periphery of the overall pattern of altered autofluorescence. As was noted on FAF imaging, flecks that appeared hyperautofluorescent on NIA imaging were observed to become hypoautofluorescent with time (Figure 4A).
Spatial correlation of fleck-associated abnormal autofluorescence on FAF and NIA imaging was performed by superimposing color-coded FAF and NIA images from the same time point. We observed that fleck-associated changes from background fluorescence were correlated across the 2 modalities. Most hyperautofluorescent flecks seen on NIA imaging (coded in red) corresponded with hyperautofluorescent flecks seen on FAF imaging (coded in green), resulting in orange-colored lesions on image superimposition (Figure 5). However, many more hyperautofluorescent flecks on FAF corresponded with hypoautofluorescent flecks on NIA, resulting in a superimposed image that is primarily green (Figure 5). This indicated that while early hyperautofluorescent flecks became hypoautofluorescent over time on both FAF and NIA imaging, fleck-associated hyperautofluorescence on NIA imaging may precede or be concurrent with that on FAF imaging, while hyperautofluorescence on FAF imaging is unlikely to precede that on NIA imaging.
Descriptions of the clinical course of Stargardt disease, studied primarily from clinical examination and CFP, have depicted a process of increasing fleck accumulation and macular atrophy. Herein, we have followed the natural history of the disease with autofluorescence imaging captured using 2 ranges of excitation and emission wavelengths and performed longitudinal and cross-modality analyses of the data.
Our observations reveal that the spatial accumulation of flecks in Stargardt disease in the macula did not develop randomly across the macula but progressed spatially from the fovea in an intriguing radial pattern. This pattern indicates that the earliest events leading to fleck formation occur close to the fovea and that foveal-to-peripheral gradients in the retina—such as the distribution cone photoreceptors,26 RPE cells,27 and intraretinal concentrations of melanin14 and macular pigment28—may contribute to fleck progression. One possibility is that the high density of cones at the fovea may specify it as the initial locus where a limiting factor (eg, ability to sustain metabolic activity) is exceeded, which leads to the formation of by-products that negatively influence neighboring cones. Our analysis also reveals that fleck lesions in some cases spread contiguously, suggesting that intercellular communications may play a role in disease progression whereby affected cells induce pathological alterations in directly adjacent cells, thus driving a continuous spatial spread of fleck lesions.
Another interesting observation in the natural history of fleck lesions is the varying nature of the fleck-associated autofluorescent signal over time. Fleck lesions on FAF imaging were observed in general to increase in hyperautofluorescence, reach a peak of hyperautofluorescence, then decrease subsequently in autofluorescence to near-background levels, and eventually become hypoautofluorescent relative to background—a pattern that has been similarly reported in the progression of geographic atrophy in eyes with age-related macular degeneration.25,29 In Stargardt disease, this progression of fluorescent changes in individual flecks can be most dramatically visualized in the supplementary videos (video 1, video 2, and video 3).
Changes in fleck autofluorescence were also apparent on NIA imaging, but these appeared to progress on a different temporal scale compared with FAF imaging. This pattern suggests that while flecks on NIA imaging undergo a similar decay of autofluorescent signal such as that seen on FAF imaging, the rate of decay into hypofluorescence may occur more quickly in NIA than FAF imaging. These longitudinal observations may be summarized graphically in a model timeline of changes on the 2 different imaging modalities (Figure 6).
Several studies, including one performed by Delori et al16 more than 15 years ago, have identified lipofuscin in the retina as a primary fluorophore excited at 488 nm.17 More recent studies of the distribution of autofluorescent patterns and spectral properties of the measured fluorescence suggest that melanin may be a key fluorophore in the fundus at longer wavelengths, contributing to the NIA pattern.7,14,17,23,30 These longitudinal observations may relate to how lipofuscin and melanin composition and distribution may change in vivo within diseased RPE cells. These changing compositions may be associated with cellular changes observed in histopathological study of RPE cells in Stargardt disease where, apart from RPE cell atrophy located near the central macula in some eyes,3- 5 substantial cellular changes were also observed in nearby macula. In intact RPE cells, lipofuscin granules have been observed to accumulate, displacing intracellular melanin granules. Histopathologically, lipofuscin content in affected RPE cells has been observed to further increase, followed by fusion between lipofuscin and melanin granules3 and later leading to the eventual loss of most of the melanin granules.4 The accumulation of lipofuscin displacing melanin granules apically, combined with the generation of intermediates such as melanolipofuscin and oxidized melanin, may underlie the increases in autofluorescence on NIA imaging.31 Later, continued accumulation of lipofuscin with loss of melanin granules may be reflected in the increase in lipofuscin-associated FAF signal and the concurrent decrease in melanin-associated NIA signal. With time, RPE cells become engorged with lipofuscin granules and finally undergo deterioration with the breakdown of cellular membranes and the ultimate lysis and death of the RPE cell, which may culminate in loss of autofluorescence on both FAF and NIA imaging and the emergence of clinically evident atrophy. It is notable that frank clinical atrophy in Stargardt disease and the concurrent loss of all autofluorescence signals typically occur in the region of decreased FAF signal as a final event following earlier fleck-related autofluorescence changes in the same area. As the hypofluorescence on FAF imaging of these atrophic areas is quite distinct, FAF has been used to detect areas of atrophy in an attempt to properly measure them.32 Interestingly, confluent areas of hypoautofluorescence or absent FAF correlate with loss of photoreceptor inner segment–outer segment reflectivity and RPE on optical coherence tomography.33 The status of photoreceptors overlying areas of abnormal or patchy FAF is less clear. New technologies such as adaptive optics will likely be important in linking these fluorescent changes directly to photoreceptor loss,34 and early studies suggest that increased FAF is correlated with increased cone photoreceptor spacing.34
Identifying a temporospatial pattern of change in Stargardt disease may provide insights into the underlying disease pathogenesis and may be helpful to clinicians in following its course. These patterns of progression, when quantified, can also be useful in the development of new outcome measurements for clinical trials testing novel therapies for Stargardt disease. Because these changing autofluorescence patterns may reflect intracellular events in RPE cells, they may be helpful in gauging the biological effect of potential therapies and in interpreting treatment effects. Longitudinal cross-modality autofluorescence imaging may constitute a useful ancillary evaluation technique for the clinical follow-up and clinical trial study of Stargardt disease.
Correspondence: Catherine A. Cukras, MD, PhD, Division of Epidemiology and Clinical Research and Ophthalmic Genetics and Visual Function Branch, National Eye Institute, National Institutes of Health, Bethesda, MD 20892 (firstname.lastname@example.org).
Submitted for Publication: April 12, 2011; final revision received July 18, 2011; accepted August 1, 2011.
Published Online: October 10, 2011. doi:10.1001/archophthalmol.2011.332
Author Contributions: Dr Cukras had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Financial Disclosure: None reported.
Funding/Support: This work was supported by the National Eye Institute Intramural Research Program.