Foveal Sparing in Central Retinal Dystrophies

Abstract

Purpose: To describe foveal sparing (FS) in central retinal dystrophies (RD).

Methods: Participants for this retrospective study were identified from the retinal dystrophy database of the Department of Ophthalmology at Radboud University Medical Center. FS was defined as an intact foveal structure surrounded by at least 180° of chorioretinal atrophy, and a best-corrected visual acuity (BCVA) of <1.0 logMAR (>20/200 Snellen). Eligible eyes were identified using fundus autofluorescence (FAF) images, and FS was confirmed using near-infrared reflectance (NIR) imaging and spectral-domain optical coherence tomography when available. Clinical and demographic data were extracted from medical records. We performed quantification of FS and chorioretinal atrophic areas using semiautomated software on fundus autofluorescence and NIR images. We calculated the chronologic change using eye-wise linear regression.

Results: We identified 36 patients (56 eyes) with FS. RDs included: Stargardt disease (STGD1;20 patients), central areolar choroidal dystrophy (CACD; 7 patients), mitochondrial retinal dystrophy (MRD; 6 patients), pseudo-Stargardt pattern dystrophy (PSPD; 3 patients). Median age at first presentation was 60 (interquartile range [IQR] 54–63) years. Median BCVA at first presentation ranged from 20/25 Snellen in STGD1, to 20/38 Snellen in MRD. Progression of the chorioretinal atrophic area ranged from 0.26 (0.25–0.28) mm/year in PSPD, to 0.14 (0.11–0.22) in CACD. Change in FS area over time was similar between the different dystrophies.

Conclusions: The presence of FS in different RDs suggests a disease-independent mechanism that prolongs the survival of the fovea. The associated preservation of BCVA is important for the individual prognosis and has implications for the design of therapeutic trials for RDs.

Retinal dystrophies (RDs) are among the leading causes for legal blindness in industrial countries.^1–3 Despite differences in underlying mechanisms, the common pathway in these disorders involves a selective atrophy of outer retinal layers, retinal pigment epithelium and associated choroidal layers, functionally paralleled by the development of absolute scotomas.⁴ 

Typically, patches of chorioretinal atrophy initially occur in the parafoveal retina. With spread over time, multifocal atrophic areas coalesce, and new atrophic areas may occur. On clinical examination, the fovea may remain uninvolved from chorioretinal atrophy until late in the course of the disease, when the fovea finally becomes atrophic. This phenomenon is referred to as “foveal sparing,”^5–9 and can be observed in several RDs, such as autosomal recessive Stargardt disease (STGD1),^8,10–13 central areolar choroidal dystrophy (CACD),¹⁴ and maternally inherited diabetes and deafness (MIDD).^15,16 Geographic atrophy and reticular pseudo-drusen in age-related macular degeneration may also show a tendency to spare the fovea for a prolonged period of time.^7,9,17–20 

Although there is no consensus definition of the term “foveal sparing,” it could be distinguished from a perimacular ring scotoma, which is located more peripherally in the macula, resulting in a central visual island frequently observed in retinitis pigmentosa (RP).²¹ Bull's eye maculopathy is another separate entity, where an intact fovea is surrounded by a—usually oval shaped—area of outer retinal atrophy. 

Historically, the term “foveal sparing” was coined by Hart et al.²² to describe the sparing of the fovea from hemifield defects. In the context of macular disorders, Hart et al.²² were the first to describe a “sparing of foveal sensitivity in macular disease,” while Sarks et al.⁷ specified an “enlargement and coalescence of atrophy into an almost complete ring around the fovea.⁷” These perimetric and funduscopic terms were brought together by Sunness et al.²⁰ using the term “foveal sparing”, defined as an intact foveal structure surrounded by outer retinal and RPE atrophy. The atrophy may either arrange as multiple independent spots, or in a ring or horseshoe-like fashion.²⁰ 

The etiology of foveal sparing remains unclear, but it is highly remarkable that such a distinct preservation of the central macular tissue occurs in a variety of heterogenic retinal disorders. This suggests that the underlying mechanism of foveal sparing is, to a certain degree, disease-independent. 

To date, the manifestations of foveal sparing in retinal dystrophies has not been analyzed systematically. Knowledge regarding the natural history of this peculiar phenomenon becomes even more important in the light of emerging therapeutic modalities aimed at prevention or slowing down of progressive chorioretinal atrophy in RDs (e.g., NCT01367444, NCT01736592, NCT01469832, and NCT02402660 on www.clinicaltrials.gov). A better understanding of the disease course of foveal sparing enables better stratification when including patients into a clinical trial and allows for a tailor-made prognosis. In addition, insight in the underlying mechanisms resulting in foveal sparing could open up new avenues for therapeutic approaches aimed at foveal preservation. In this study, we will provide a detailed description of foveal sparing in a large cohort of RD patients. We will analyze commonalities and discriminating features among the different RDs, and we will discuss the consequence for pathophysiological mechanisms regarding foveal sparing. 

Methods

Patient Selection

For this retrospective study, we employed the RD database of the Department of Ophthalmology at Radboud University Medical Center. The 1800 patients in this database have all been diagnosed with some form of inherited retinal disease, the large majority of which have been analyzed genetically. To readily identify patients with foveal sparing, we limited our search to patients with at least one fundus autofluorescence (FAF) imaging investigation. A genetically confirmed clinical diagnosis was also a requirement to be included in the search. In the context of this study, the foveal sparing had to fulfill two criteria. First, the fovea had to be surrounded by at least 180° of chorioretinal atrophy.^6,8,9 Second, a best-corrected visual acuity (BCVA) of ≤1.0 logMAR (≥ 20/200 Snellen) had to be present.⁸ Both eyes of a patient were included unless foveal sparing was only present in one of them. 

The goal of the research is to study chorioretinal atrophy localized to the posterior pole, therefore we excluded generalized chorioretinal atrophy disorders like; retinitis pigmentosa, choroideremia, Bietti crystalline dystrophy, and gyrate atrophy. We excluded retinopathies with solely loss of photoreceptors encircling the fovea, like in Bull's eye maculopathy, because these entities do not fulfill the criteria of chorioretinal loss. 

This retrospective study was approved by the Institutional Ethics Committee at Radboud University Medical Center (Nijmegen, The Netherlands), and was performed in accordance with the Tenets of the Declaration of Helsinki. All patients provided informed consent prior to additional ophthalmologic examinations to complete the clinical assessment and blood collection. 

Clinical Data and Image Acquisition

A detailed medical and ophthalmologic history, including historical BCVA data, age at onset, age at diagnosis, and initial symptoms, was obtained from the medical records. Age at onset was defined as the age at which the patient first experienced visual complaints. In asymptomatic persons and patients where the age at onset could not be determined, the age at disease onset was considered equivalent to the age at diagnosis. 

Standard imaging protocols were followed in the acquisition of retinal images. The FAF and NIR images were acquired using an HRA+OCT device (Spectralis; Heidelberg Engineering, Heidelberg, Germany). The field of view was set to 30° × 30° or 55° × 55° with a minimum resolution of 768 × 768 pixels and centered on the macula. Single images were automatically aligned and averaged to maximize the signal-to-noise ratio using the manufacturer's software (automatic real-time [ART]-mode, Heidelberg Eye Explorer, Heidelberg Engineering). The spectral-domain optical coherence tomography (SD-OCT) scans were performed with the same device, and up to 100 single images were averaged to improve image quality. SD-OCT scans were 6.3 mm horizontal line scans through the fovea and follow-up mode was used for follow-up images. Furthermore, color fundus photographs (CFP) were obtained using a Topcon TRC50IX retinal camera (Topcon Corporation, Tokyo, Japan). 

Image Grading

Qualitative Assessment

Atrophy of retinal pigment epithelium and outer retina is typically associated by a sharply demarcated zone of dramatic reduction of autofluorescence (analogous to the previously employed definition of “definitely decreased autofluorescence”).²³ Where available, SD-OCT was used to confirm foveal outer retinal integrity. If foveal sparing was lost during follow-up, all later visits were excluded from the analysis. 

Qualitative assessment of the available FAF and SD-OCT images was performed to identify characteristic morphologic features and atrophy patterns. In this explorative study, image analysis was performed by one trained reader (ML). As a result of the retrospective nature of this dataset, some of the image grading tasks could not be performed in every eye or for every visit, because image availability varied per patient visit. 

Quantitative Assessment

Measurements of atrophic area and foveal sparing area were performed as previously described,^9,24 using imaging software (RegionFinder, version 2.5.5.0; Heidelberg Engineering). In brief, the dramatic decrease of the FAF signal in GA areas compared with nonatrophic retinal areas is used by the imaging software (Heidelberg Engineering) for the segmentation of atrophy areas. The examiner/operator is required to define the center of each atrophic area. The software's so-called “region-growing algorithm” then identifies the borders of each atrophic area by the sharp change in signal intensity between atrophic and nonatrophic retina. This employed version of the imaging software (Heidelberg Engineering) includes a feature that automatically registers FAF to corresponding NIR images and allows the operator to easily switch from one modality to the other, thereby semi-automating the quantification of RPE atrophy and foveal sparing areas. In case no NIR images were available, measurements were only performed if the atrophy could be clearly discriminated from the physiologically decreased foveal autofluorescence due to the physiological attenuation of the autofluorescence signal in the foveal area by macular pigment. Atrophy measurements were performed in either 30° × 30° or 55° × 55° images within a single eye, as the software does not allow for quantitative comparison between images of different sizes. Only areas of sharply demarcated RPE atrophy, or “definitely decreased autofluorescence” (DDAF), were quantified, and quantification of foveal sparing required an area of DDAF surrounding the fovea by at least 270°.⁹ Therefore, the foveal sparing area was not quantified in eyes with horseshoe-shaped atrophy surrounding the fovea by <270°. In case of incomplete foveal sparing, a constraint was placed by manually drawing a line at the narrowest place of the incomplete part of the residual foveal island. 

Subfoveal retinal thickness (SRT) was assessed in SD-OCT scans, transversing the foveola, as the distance between the internal limiting membrane (ILM) and the outer border of OCT band 4 (corresponding to the RPE/Bruch's membrane complex),²⁵ using a “distance tool” (Heidelberg Eye Explorer; Heidelberg Engineering). A representative example is given in Supplementary Figure 1. Furthermore, the integrity of the retinal bands at the fovea was assessed. 

As a result of the retrospective nature of this study, some of the image grading tasks could not be performed in every eye or for every visit, because image availability varied per patient visit. 

Genetic Analysis

All samples were processed by the Department of Human Genetics of the Radboud University Medical Center in Nijmegen, The Netherlands. Genetic testing was targeted conform the clinical diagnosis. 

Stargardt disease (STGD1): ABCA4 analysis was performed using arrayed primer extension analysis (APEX) microarrays (Asper Biotech, Tartu, Estonia) and Sanger sequencing validation. The presence of two variants in the ABCA4 gene confirmed the diagnosis STGD1. In cases where only one variant in ABCA4 was identified, but the phenotype was characteristic for STGD1, this was deemed sufficient for the diagnosis. The following genetic variants were defined as severe: protein-truncating, canonical splice-site variants, as well as deletions spanning at least one exon. In case of missense variants, severity was predicted by bioinformatic means (SIFT, Polyphen, MutationTaster, and CADD). Cases with a clinical Stargardt-like phenotype and a mutation in the PRPH2 gene, with or without an additional variant in a single ABCA4 allele, were considered having pseudo-Stargardt pattern dystrophy (PSPD).²⁶ 

Central areolar choroidal dystrophy (CACD): analysis of PRPH2 was conducted using Sanger sequencing. One variant in the PRPH2 gene, combined with the typical CACD phenotype, was considered sufficient to establish the diagnosis. 

Mitochondrial retinal dystrophy (MRD): mitochondrial DNA was screened for mutations. The presence of the m.3243A>G mitochondrial mutation, combined with a characteristic clinical picture and visual symptoms, was required for the diagnosis of mitochondrial retinal dystrophy. 

Statistical Analysis

R version 3.3.0 was used to perform statistical analysis. Decimal-scale visual acuity data were converted to logMAR. 

The rate of atrophy progression, and foveal sparing area loss over time, were calculated by eye wise linear regression from square-root transformed values, to reduce the dependency of enlargement rates on baseline lesion size, as previously suggested.²⁷ In this retrospective analysis, data for these parameters were not available for every single visit in every patient. To avoid exclusion of all data from a particular visit in case of a single missing parameter, subsets of visits were analyzed, when necessary. In these cases, cohort sizes and observation intervals were calculated separately. Unless stated differently, all data given represent medians and corresponding interquartile ranges. In this exploratory analysis, statistical tests for significance were not performed. 

Results

A total of 336 patients from the database of 1800 patients met the search criteria: FAF images and a genetically confirmed clinical diagnosis. In this cohort of 336 patients, we identified 36 individuals (56 eyes) with foveal sparing as previously defined. In 20 patients (56%), foveal sparing was present or developed in both eyes. Overall, this cohort included 20 of 176 (11.4%) STGD1 patients (31 eyes), 7 of 103 (6.8%) patients diagnosed with CACD (10 eyes), 6 of 21 (28.6%) MRD cases (10 eyes), 3 of 36 (8.3%) PSPD patients (5 eyes). A comprehensive overview of all patient characteristics is provided in Table 1. Retinal images of foveal sparing in each of the conditions are depicted in Figures 1 through 4. 

Table 1

 

Summary of Patient Characteristics

 

Figure 1

 

Longitudinal retinal imaging of a representative eye with STGD1. Serial fundus autofluorescence images (upper row), and SD-OCT scans centered on the fovea (lower row), of a representative right eye with foveal sparing in a patient with STGD1. Multiple atrophic lesions may be seen that coalesce over time. Cystoid lesions in the inner retina (asterisks) and elevation of the borders of outer retinal atrophy (arrowheads) may be seen. Location of the SD-OCT scan is indicated by the green line.

Figure 2

 

Retinal imaging of two representative CACD cases. FAF, near-infrared reflectance and SD-OCT scans of the left eyes of two patients with CACD. Foveal sparing may demarcate only poorly on FAF. Arrow: center of the foveal pit. Yellow bar: areas of intact retinal pigment epithelium in proximity to the foveal depression. Location of the SD-OCT scan is indicated by the green line.

Figure 3

 

Longitudinal retinal imaging of a representative eye with mitochondrial retinal dystrophy (MRD). The left eye of a patient with MRD showing gradual contraction of the remaining foveal island in 21 months. Fundus autofluorescence (FAF, upper row) and near-infrared reflectance (lower row) demonstrate retinal pigment epithelium atrophy encircling the fovea by 360°. The relatively spared fovea almost exactly centers on the foveal depression, as also shown by the SD-OCT scan (position is indicated by green line).

Figure 4

 

Retinal imaging of two representative PSPD. Fundus autofluorescence (left) and near-infrared reflectance image (right) of two right eyes of patients with PSPD. Multifocal areas of macular atrophy are usually observed in conjunction with more peripheral atrophic spots. Spectral-domain optical coherence tomography (inset, centered on the fovea) discloses outer retinal atrophy and disruption of retinal structure. The yellow bar indicates preservation of the retinal pigment epithelium.

Baseline

There was little variability in the age at first clinical presentation among the different RDs included. Median age at first presentation was 60 (54–63) years. Median values ranged from 56 years in CACD to 62 in PSPD (Table 1). Variability in the duration from experienced symptoms onset to the time where a dystrophy was diagnosed was more pronounced, with median values ranging from 0.8 (0.3–2.3) years in CACD to 15 (12–38) years in PSPD (Table 1). 

At first presentation, median BCVA values were relatively high, ranging from 0.11 (0.09–0.26) logMAR (20/25 [20/24–20/36] Snellen) in STGD1, to 0.28 (0.15–0.30) logMAR (20/38 [20/28–20/39] Snellen) in MRD. 

Qualitative assessment of the available FAF and SD-OCT images revealed characteristic features of foveal sparing morphology and atrophy progression. At baseline (first visit), foveal sparing was present in 52 eyes of 36 patients (bilateral in 15 patients). Thirty-two eyes (62%) presented with a multifocal atrophy pattern (e.g., Fig. 1), 15 eyes (29%) presented with a horseshoe-type pattern (e.g., Fig. 1), and we observed a solitary foveal island in 5 eyes (e.g., Fig. 3). Sixteen eyes did not show sufficient atrophic changes to warrant the diagnosis and in six eyes there may have been foveal sparing, but at the time of investigation the foveal tissue was already lost. We compared both eyes in these 36 patients. We observed a high degree of symmetry between both eyes in individual patients: in MRD and PSPD, foveal sparing could be observed bilaterally in 66.6% of cases. Slightly lower values of 55% and 43% were found in STGD1 and CACD (Table 1). 

Follow-Up

Over time, BCVA remained relatively stable when grouped according to underlying conditions. Decline of BCVA was most noticeable in STGD1 and PSPD, with a loss of 0.11 (0.01–0.13), and 0.13 (0.06–0.15) logMAR, respectively. 

Follow-up was available in 21 patients (58%). Of the 16 eyes without foveal sparing at baseline, five eyes developed foveal sparing over a median period of 3 years (range, 1–4 years). We noticed this development only in STGD1 patients, and in none of the other conditions. Foveal sparing was lost during follow-up in five eyes over a median period of 5 years (range, 3–8 years). In four of these eyes, a multifocal pattern was present at baseline, and foveal sparing was initially not present in one eye. 

In all conditions, we noticed three distinct and consecutive patterns of atrophy, starting with multifocal atrophic lesions immediately adjacent to the fovea, to a horseshoe-type pattern into an isolated foveal island before the occurrence of central atrophy. Figure 5 shows these four stages in the progression of foveal sparing in both eyes of a patient with STGD1. 

Figure 5

 

Foveal sparing atrophy progression pattern. Fundus autofluorescence images of the right (A–C) and left (D–F) eye of a Stargardt patient with asymmetrical foveal sparing. Multifocal areas of RPE atrophy (A) develop and coalesce over time, forming a horseshoe-type atrophy pattern (B). In the last picture (C) the isthmus is severed leaving an isolated foveal island. The left eye reveals a similarly progressive atrophy pattern, albeit in an earlier stage. Figure 5C reprinted from van Huet RAC, Bax NM, Westeneng-Van Haaften SC, et al. Foveal sparing in Stargardt disease. Invest Ophthalmol Vis Sci. 2014;55:7467–7478. © 2014 The Association for Research in Vision and Ophthalmology, Inc. Figure 5E reprinted from Westeneng-van Haaften SC, Boon CJ, Cremers FP, Hoefsloot LH, den Hollander AI, Hoyng CB. Clinical and genetic characteristics of late-onset Stargardt's disease. Ophthalmology. 2012;119:1199–1210. Copyright © 2012 American Academy of Ophthalmology.

Atrophy progression rates varied between 0.26 (0.25–0.28) mm/year in PSPD, to 0.14 (0.11–0.22) mm/year in CACD (Fig. 6A). Loss of surface area of the spared fovea ranged from 0.06 (0.08–0.02) mm/year in STGD1 to 0.10 (0.15–0.10) mm/year in MRD (Fig. 6B), and loss of SRT ranged from 13.84 (19.92–9.47) μm/year in MRD to 0.90 (5.13–1.50) μm/year in CACD. 

Figure 6

 

Change of atrophy area and foveal sparing area over time. Change of atrophy area (A) and foveal sparing area (B) over time as measured in registered fundus autofluorescence images in eyes with distinct retinal dystrophies and foveal sparing. Please note progression rates were square root transformed to reduce the dependency of enlargement rates on baseline lesion size. The horizontal bar inside the boxes represents the median, the hinges correspond to the 25th and 27th percentiles. The upper and the lower whiskers extend to the largest (resp. smallest) value no further than 1.5 * IQR from the hinges.

A summary of all BCVA measurements and quantitative retinal image analyses can be found in Table 2. 

Table 2

 

Summary of Clinical and Retinal Imaging Parameters

 

Genetic Analysis

An overview of all genetic variants is given in Table 3. In only six of twenty STGD1 cases (30%), two ABCA4 variants were detected. The most prevalent variant is c.5461-10T>C (5/26 = 19%). At least one severe ABCA4 variant was detected in 19 cases (95%). In all CACD cases, the same PRPH2 missense variant (p.Arg142Trp) was identified. Likewise, in all MRD cases, the same variant (m.3243A>G) was detected in the mitochondrial DNA. 

Table 3

 

An Overview of All Genetic Variants in This Study Cohort

 

Discussion

Foveal sparing is a regularly encountered phenomenon in a variety of retinal disorders, including RDs like STGD1 and geographic atrophy in age-related macular degeneration. This study analyses the characteristics of foveal sparing in a wide range of retinal dystrophies, including STGD1, CACD, MRD, and PSPD. All the patients were in their fifth decade or older at first presentation, indicating that foveal sparing mainly manifests in RD patients with a relatively low rate of progression, where visual loss occurs later in life. The manifestation of foveal sparing in this heterogenic group of disorders is remarkably similar, in particular with regard to the kinetics of atrophy progression toward the fovea, which varies only slightly between disorder types. In general, RPE atrophy in foveal sparing progresses in a relatively fixed pattern, starting with small, multifocal areas of atrophy that surround the fovea. These, then gradually expand in all directions and coalesce, forming a horseshoe-type atrophy. The bridge of retinal tissue connecting the fovea to the non-atrophic part of the macula gradually narrows until this isthmus disappears, leaving a solitary foveal island. In the final stage this island is lost to atrophy. This sequence of steps is a common finding in these patients and can be of great prognostic help, although the rate of progression may differ. This pattern of RPE atrophy was symmetrical in 80% of the bilateral cases. This indicates, as already suggested by others,²⁸ that the fellow eye cannot automatically serve as control in therapeutic intervention trials for RDs in patients with foveal sparing. 

Although the frequency of foveal sparing in our cohort was highest in the MRD group (29%), this was much lower than the previously reported number of 84% in a 2013 study by de Laat et al.¹⁶ This may be an underestimation on our side due to inclusion bias since this study was not primarily designed to estimate the prevalence of foveal sparing in RDs, including MRD. 

To date, the underlying mechanism of foveal sparing is incompletely understood. Disease independent factors are likely at play in view of the high heterogeneity of the underlying disorders. This notion is supported by the virtually identical kinetics of atrophy progression toward the fovea observed among most conditions (0.06–0.10 mm/year). The presence of multifocal areas of chorioretinal atrophy surrounding the fovea, and very similar square-root transformed atrophy progression rates of 0.116 mm/year were observed in age-related macular degeneration.^9,28 The similarities between the phenomenon of foveal sparing in monogenic and multifactorial disorders further supports the hypothesis of disease-independent mechanisms that shape the foveal sparing phenotype. 

The general susceptibility to retinal degeneration may lie in metabolic differences between regions of the macula. Those differences may be associated with physical characteristics of foveal and peripheral cones; foveal cone outer segments physically resemble those of rods rather than peripheral cones. Furthermore, Müller cells are short and exist in a 1:1 ratio with foveolar cones, but are longer and have a lower ratio extrafoveally.²⁹ 

A number of underlying mechanisms have been proposed involving the rod-derived cone viability factor (RdCVF), variations in macular pigment and peak distribution as well as cone density, increased vulnerability of certain parafoveal photoreceptors, and factors related to RPE and choroid. The RdCVF is secreted from rod photoreceptors and protects cones from degeneration.³⁰ An increased sensitivity of foveal cones to RdCVF, possibly complemented by higher secretion levels of RdCVF, may result in improved central cone survival. 

A second hypothesis involves the macular pigments: lutein, zeaxanthin and meso-zeaxanthin. These carotenoids protect against macular damage through their antioxidant properties and by filtering potential harmful blue light. In eyes with foveal sparing, an uneven distribution of macular pigment might lead to protection of the most central photoreceptors, leaving the parafoveal photorecepotors relativeley unprotected.^31,32 Another explanation may lie in the highly variable peak cone density ranging from 98,200 to 324,100 cones/mm². This remarkable interindividual variability is much less pronounced in the area surrounding the fovea and may contribute to the phenomenon of foveal sparing.³³ In vivo determination of cone density with adaptive optics could solve the role of peak cone density in foveal sparing. Another factor that may be of influence is the increased vulnerability to age and degenerative disease of respectively rod photoreceptors and short wavelength (blue) S-cone photoreceptors.^34,35 Both rods and S-cones are absent in the foveal center and the increased susceptibility to aging and/or disease may explain the relative preservation of the fovea in certain patient. An explanation may also lie in the unfavorably high ratio of rods per RPE cell in the parafoveal retina that could lead to an earlier decompensation of metabolic function promoting perifoveal atrophy.^16,36 Finally, the unique choroidal blood supply to the fovea has been put forward as a factor leading to a local protective effect.^37,38 

A common finding in these foveal sparing patients is late age at which the diagnosis is made. The well-preserved visual acuity leads to patient's delay and at the time of the first ophthalmologic consultation large areas of atrophy are already present. We know from this and other studies that the atrophy progression rate is relatively slow,^39–41 disease-specific changes have therefore been present for years. In the 37 patients with foveal sparing in this study, only one patient experienced a scotoma as initial symptom. The large majority (n = 26; 72%) patients present with loss of visual acuity. This late recognition of patients with foveal sparing RDs narrows the window of opportunity for therapeutic intervention, which is becoming increasingly important as novel therapeutic approaches emerge (NCT01367444, NCT01736592, NCT01469832, and NCT02402660 on www.clinicaltrials.gov). 

In the group of STGD1 patients, the high frequency of self-reported “decreased visual acuity” as initial symptom goes together with a high baseline visual acuity (0.11 [0.09–0.26]) logMAR (approximately 20/25 Snellen). This is likely due to the fact that parafoveal scotomas affect high resolution visual tasks like face recognition and reading, whereas BCVA tests rely on the maximum resolution of the fovea.^7,19,20,42 It is important to be aware that visual acuity tests are an imperfect measurement for macular function—in particular in patients with foveal sparing. 

Preservation of the foveal tissue in the late stage of a RD is beneficial to the patient. The ProgSTAR study recently showed a stable BCVA in the STGD1 cohort with foveal sparing over the course of 3.24 years. In contrast, in the STGD1 cohort without foveal sparing and equivalent baseline visual acuity, an average loss of one line per year could be observed.⁴³ 

Recently, it was shown that discrimination between foveal sparing and nonfoveal sparing STGD1 phenotypes can be made as early as the time the first atrophic lesions become apparent, based on parameters like a late age-at-onset and thinning of the outer nuclear layer and ellipsoid zone.^8,40 Yet, biomarkers that could predict foveal sparing at an even earlier stage are not available. 

With regard to genetic predisposing factors, it has been suggested that the presence of relative mild genetic variants correlate with a less severe phenotype and a generally later age of onset.^11,44 In 14 of the 20 (70%) STGD1 patients with foveal sparing, only one ABCA4 variant could be detected. The percentage of unidentified second mutations is much lower (30%) in patients with a typical Stargardt phenotype with an age-at-onset in the second to third decade, a second mutation has not been identified (Cremers FPM, unpublished observations, 2018). 

Deep-intronic variants were found in several mono-allelic cases.^45–48 Very recently, the hypomorphic intronic variant c.4253+43G>A was found to be associated with late-onset STGD1.⁴⁹ Zernant and colleagues⁵⁰ determined that c.5603A>T (p.Asn1868Ile), previously suspected to be benign because of its high prevalence in the general population (minor allele frequency of 0.07), is disease-causing. This variant was typically found in a compound heterozygous manner with a severe ABCA4 variant in ∼50% of monoallelic cases and in ∼80% of late-onset STGD1 cases.⁵⁰ Runhart et al.⁵¹ in addition found significant differences in age-at-onset between affected siblings and even nonpenetrance in three families with asymptomatic biallelic siblings of affected persons. The penetrance of this variant, when present together with a severe ABCA4 variant in the other gene copy, was estimated to be less than 5%, suggesting a crucial role for genetic or environmental modifiers in STGD1. In the present study, the c.5603A>T variant was found in 3/20 (17%) of STGD1 cases with foveal sparing. In two other STGD1 cases where both alleles were identified, mild variants (p.(Ala1038Val) and p.[Gly863Ala, Gly863del]) were also found, corroborating the hypothesis that late-onset STGD1 can partially be explained by the combination of a severe with a mild ABCA4 variant. 

Therapeutic approaches for RDs are currently emerging. Recently, the Food and Drug Administration (FDA) approved gene therapy for RPE65-associated retinal dystrophy, a form of Leber congenital amaurosis. In addition, therapy trials for STGD1, choroideremia, Usher syndrome type 2A and several more are underway (NCT01367444, NCT01736592, NCT01469832, and NCT02402660 on www.clinicaltrials.gov). To accurately assess efficacy of new therapeutic modalities in clinical trials, careful patient selection and well-thought out outcome measures are essential.⁵² Eyes with foveal sparing could be candidates for inclusion into clinical trials, due to the clearly demarcated, well-measurable areas of progressive RPE atrophy and preserved BCVA, allowing for visual stabilization as one of the outcome measurement. In addition, in about half of the patients with foveal sparing, this phenomenon occurs in both eyes, which opens the opportunity of inter-eye comparison. Finally, these patients may benefit greatly from successful therapeutic intervention, as prolonged preservation of the fovea is paramount to daily functioning, and by extension to quality of life. The use of eyes with foveal sparing for clinical trials inevitably also comes with certain pitfalls. A long follow-up period would be necessary to accurately assess treatment effect if BCVA is used as parameter. And in our cohort, only 43% of the cases has a symmetrical disease progression. In such cases, an alternative outcome measure like atrophy progression, or change in the size of the spared fovea over time, is preferable.⁴¹ 

In summary, the phenomenon of foveal sparing was observed in several distinct retinal dystrophies. All patients presented with late-onset disease, and the pattern of perifoveal RPE atrophy progression was identical in all conditions. Remarkable similarities between foveal sparing pattern in RDs and age-related macular degeneration may suggest disease-independent mechanisms to shape the pattern of foveal sparing. Importantly, the observations made in this study will allow ophthalmologists to provide affected patients with a more accurate prognosis, and the characteristic aspects in terms of visual function loss and atrophy progression should be considered when including foveal-sparing and non-foveal sparing eyes in clinical trials. 

Further research, including histopathology studies and detailed retinal imaging with adaptive optics, is necessary to elucidate the mechanisms involved in foveal sparing. The new insights gleaned from these studies might pave the way for new disease-independent (gene-)therapeutic approaches for prolonged preservation of foveal tissue in degenerative retinal disease. 

Acknowledgments

Supported by the Stichting A.F. Deutman Oogheelkunde Researchfonds, Nijmegen, The Netherlands; The Nederlandse Oogonderzoek Stichting (NOS), Nijmegen, The Netherlands; the Stichting MaculaFonds, The Netherlands; and UitZicht projects (2013–25 and 2014–3). Deutsche Forschungsgemeinschaft, Bonn, Germany, Grant No. FL 658/4–1 and 658/4–2 and LI 284671–1; The Knoop Trust, Oxford, UK. The authors alone are responsible for the content and writing of the paper. 

Disclosure: N.M. Bax, None; D. Valkenburg, None; S. Lambertus, None; B.J. Klevering, None; C.J.F. Boon, None; F.G. Holz, Heidelberg Engineering (F), Optos (F), Zeiss (F), Novartis (F), Bayer Healthcare (F), Genentech (F), Acucela (F), Boehringer Ingelheim (F), Alcon (F), Allergan (F); F.P.M. Cremers, None; M. Fleckenstein, Heidelberg Engineering (F), Optos (F), Novartis (F), Bayer (F), Genentech (F), Roche (F), P; C.B. Hoyng, None; M. Lindner, Heidelberg Engineering (F), Carl Zeiss Meditec (F), Optos (F), Allergan (F), Fresenius Medical Care (F) 

Appendix

Members of the Foveal Sparing Atrophy Study Team (FAST)

Nathalie M. Bax,¹ Dyon Valkenburg,¹ Stanley Lambertus,¹ B. Jeroen Klevering,¹ Camiel J. F. Boon,^2,3 Frank G. Holz,⁴ Frans P.M. Cremers,⁵ Monika Fleckenstein,⁴ Carel B. Hoyng,¹ and Moritz Lindner^4,6,7 

¹Department of Ophthalmology, Donders Institute for Brain, Cognition and Behavior, Radboud University Medical Center, Nijmegen, The Netherlands 

²Department of Ophthalmology, Leiden University Medical Center, Leiden, The Netherlands 

³Department of Ophthalmology, Academic Medical Center, University of Amsterdam, Amsterdam, The Netherlands 

⁴Department of Ophthalmology, University of Bonn, Bonn, Germany 

⁵Department of Human Genetics, Donders Institute for Brain, Cognition and Behavior, Radboud University Medical Center, Nijmegen, The Netherlands 

⁶The Nuffield Laboratory of Ophthalmology, Sleep and Circadian Neuroscience Institute, Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, United Kingdom 

⁷Oxford Eye Hospital, Oxford University Hospitals NHS Foundation Trust, Oxford, United Kingdom