Characterization of Stargardt Disease Using Polarization-Sensitive Optical Coherence Tomography and Fundus Autofluorescence Imaging

Abstract

Purpose.: To identify disease-specific changes in Stargardt disease (STGD) based on imaging with polarization-sensitive spectral-domain optical coherence tomography (PS-OCT) and to compare structural changes with those visible on blue light fundus autofluorescence (FAF) imaging.

Methods.: Twenty-eight eyes of 14 patients diagnosed with STGD were imaged using a novel high-speed, large-field PS-OCT system and FAF (excitation 488 nm, emission > 500 nm). The ophthalmoscopic phenotype was classified into three groups. ABCA4 mutation testing detected 15 STGD alleles, six of which harbor novel mutations.

Results.: STGD phenotype 1 (12 eyes) showed sharply delineated areas of absent RPE signal on RPE segmentation B-scans of PS-OCT correlating with areas of hypofluorescence on FAF. Adjacent areas of irregular fluorescence correlated with an irregular RPE segmentation line with absence of overlaying photoreceptor layers. Eyes characterized on OCT by a gap in the subfoveal outer segment layer (foveal cavitation) showed a normal RPE segmentation line on PS-OCT. Hyperfluorescent flecks on FAF in phenotype 2 STGD (8 eyes) were identified as clusters of depolarizing material at the level of the RPE. Distribution of flecks could be depicted on RPE elevation maps. An increased amount of depolarizing material in the choroid was characteristic for STGD Phenotype 3 (8 eyes).

Conclusions.: PS-OCT together with FAF identified characteristic patterns of changes in different stages of the disease. PS-OCT is a promising new tool for diagnosis and evaluation of future treatment modalities in STGD.

Introduction

Stargardt disease (STGD) is the most common form of juvenile macular dystrophy leading to early visual acuity (VA) loss and central visual field defects. ¹ Autosomal recessive STGD is caused by mutations in the ABCA4 gene, which encodes a transporter protein located in the outer segment disc membranes of the photoreceptors. ² In addition to STGD, ABCA4 gene mutations are associated with several other retinal degeneration phenotypes including cone–rod dystrophy, bulls eye maculopathy, retinitis pigmentosa, and AMD. ^{3 –8} Loss of function due to mutations in the gene results in accumulation of all trans retinal derivatives in the photoreceptor outer segments. Upon phagocytosis of shed photoreceptor outer segments, N-retinylidene-N-retinylethanolamine (A2E) and related bisretinoids are transferred to the RPE and accumulate in lipofuscin granules. ⁹ Histologic examinations of retinas from patients with STGD reported increased lipofuscin accumulation in RPE cells anterior to the equator and markedly enlarged RPE cells filled with lipofuscin granules more posteriorly. Rupture of these cells lead to the end stage of the disease with loss of RPE and photoreceptors. ^10,11  

Clinical findings as well as onset and progression of STGD can be highly variable. ^12,13 On fundoscopy, most patients present with atrophic-appearing lesions of the macula surrounded by yellow flecks to a variable degree. Blue light fundus autofluorescence (FAF) has been shown to facilitate in vivo evaluation of the extent of STGD, since this imaging technique visualizes RPE atrophy as well as increased accumulation of lipofuscin in the RPE. ^{14 –16} Spectral domain optical coherence tomography (SD-OCT) is another useful noninvasive imaging technique allowing for detailed visualization of intraretinal structures, photoreceptor layers, and the RPE. Several studies analyzed the different morphologic aspects of macular changes in STGD including central retinal atrophy and different types of retinal flecks. ^{17 –20} Disorganization or loss of inner and outer segment (IS/OS) junction morphology evaluated by SD-OCT has been shown to correlate with reduced functional parameters such as VA, multifocal ERG, or microperimetry. ²¹ A recent study comparing structural changes on SD-OCT images to those visible on FAF suggested that the structural integrity of the photoreceptors may be affected earlier than changes in the RPE at least as detected by FAF. ²²  

Polarization-sensitive SD-OCT (PS-OCT) is a functional extension of conventional intensity based SD-OCT, combining the advantage of high resolution SD-OCT imaging with a selective identification of the polarization state of backscattered light based on tissue-specific properties. Because melanin containing structures of the RPE change the polarization state in a random fashion, PS-OCT is capable of precisely depicting its integrity and configuration. ^23,24 Detailed imaging of the RPE, in healthy eyes, AMD, and idiopathic juxtafoveal telangiectasia has been described in previous studies from our group. ^{25 –27} Since the RPE plays a major role in the pathophysiology of STGD and other ABCA4-associated retinal dystrophies, a closer evaluation of RPE-specific changes based on imaging with PS-OCT may be of significant clinical value. 

In this study, we imaged eyes with different stages of STGD including a recently described phenotype characterized on OCT by a gap in the subfoveal outer segment layer (foveal cavitation), ²⁸ using a novel high-speed, large-field PS-OCT system, ²⁹ and correlated findings with changes visible on FAF. 

Materials and Methods

Patients

Fourteen consecutive patients with different stages of STGD and mutations in the ABCA4 gene were examined in this prospective cross-sectional observational case series. The protocol was approved by the ethics committee of the Medical University of Vienna and adhered to the tenets of the Declaration of Helsinki. Written informed consent was obtained from all subjects before their participation. 

Clinical Examination

Patients underwent a full ophthalmic examination at the Department of Ophthalmology, Medical University of Vienna, Austria, including slit-lamp examination, applanation tonometry, and dilated fundus examination. Best-corrected visual acuity (BCVA) was measured using Snellen visual charts. The logarithm of the minimum angle of resolution (logMAR) equivalent was used to calculate mean and SD. Patients were classified based on fundoscopic findings according to Fishman et al. ³⁰ Phenotype 1 was characterized by a central, atrophic-appearing macular lesion with or without perifoveal flecks. Phenotype 2 included patients with more numerous flecks, extending anterior to the vascular arcades and nasal to the optic disc. Patients with extensive atrophic-appearing RPE changes and partially resorbed flecks were classified as phenotype 3. 

Electrophysiology

Electrophysiologic assessment was performed using a RETI-port/scan gamma (Roland Consult, Brandenburg, Germany). Full-field ERGs and pattern ERGs incorporated the minimum standards of the International Society for Electrophysiology of Vision (ISCEV) ^31,32 and were recorded using a Dawson-Trick-Litzkow (DTL) fiber inserted into the inferior fornix of the eye. Dark-adapted ERGs were recorded following stimulation with white flashes of intensities of 0.01, 3.0, and 10.0 cd.s.m⁻². Photopic cone system function was assessed after 10 minutes of light adaptation by recording 30 Hz flicker and single flash ERGs (3.0 cd.s.m⁻²; 30 and 2 Hz). Full-field ERG abnormalities are classified into three groups based on the criteria suggested by Lois et al. ³³ Patients with a normal scotopic and photopic ERGs are classified as type 1. Patients with ERG type 2 disease have decreased photopic but healthy scotopic ERGs, and patients with ERG type 3 disease have decreased scotopic and photopic ERGs. 

Blue Light Fundus Autofluorescence Imaging

FAF images were obtained using a confocal scanning laser ophthalmoscope (Spectralis Heidelberg Retina Angiograph [HRA] + OCT; Heidelberg Engineering, Heidelberg, Germany). The device uses blue light with excitation at 488 nm to illuminate the fundus and detects emitted fluorescence signals over 500 nm. Images were recorded after pupillary dilation using a 30° field-of-view mode. To improve the signal-to-noise ratio about 10 images were aligned and a mean image was calculated. Images were recorded in normalized mode; photopigments were not bleached prior to image acquisition. 

Polarization Sensitive SD-OCT Imaging

A novel large-field, high-speed PS-OCT system developed by the Center for Medical Physics and Biomedical Engineering, Medical University of Vienna was used. ²⁹ In contrast to previous PS-OCT systems, ^25,34 the new system provides a denser sampling (up to 1024 × 250 A-scans) over a larger scan field (40° × 40°). The technical principles of PS-OCT are described in detail elsewhere. ^{25,34 –36} In summary, the system enables to measure four parameters simultaneously: the intensity of the backscattered light (as standard SD-OCT imaging), optic axis orientation, retardation, and the degree of polarization uniformity (DOPU). Different retinal layers can be classified into polarization preserving layers (e.g., photoreceptor layers), depolarizing layers (e.g., RPE), and birefringent layers (e.g., retinal nerve fiber layer). Since melanin containing structures within the RPE depolarize backscattered light, ³⁷ PS-OCT is capable of specifically contrasting the RPE. In the case of polarization preserving layers DOPU values close to 1 will be observed, whereas depolarizing structures will exhibit lower DOPU values. The color scale represents DOPU values ranging from 0 (black) to 1 (red) (Fig. 1). Retinal structures displayed in red to orange indicate the polarization-preserving retinal layers, the RPE appears as a depolarizing continuous layer (green to blue colors). The threshold of the DOPU value, below which retinal layers are defined as depolarizing, is DOPU less than 0.75. Areas of DOPU less than 0.75 are selected and used to generate RPE segmentation B-scans, an overlay of the segmented depolarizing material and the intensity image (Fig. 1). 

Figure 1

 

FAF and PS-OCT images of a patient with STGD phenotype 1 (patient number 2): FAF image (A), intensity based B-scans (B1, D1), RPE segmentation B-scans (B2, D2), DOPU images (B3, D3), and depolarizing material thickness map (C). Yellow lines indicate the position of B-scans. Yellow boxes indicate area of magnified images (B1a, B2a, B3a, D1a, D2a, D3a). The diameter of the central area of absent FAF ([A] marked with a white bracket) corresponds to RPE atrophy on the RPE segmentation and DOPU images ([B2, B3] marked with a white double-pointed arrow). Margins of RPE atrophy are well defined on RPE segmentation B-scans ([B2a] indicated with a yellow asterisk). Areas of irregular appearance of RPE ([B2, B3, D2, D3] indicated with white, dashed, double-pointed arrows), showing small focal skip lesions ([B2a, C, D2a] indicated with yellow arrows) correspond in large part with the diameter of abnormal FAF ([A] marked with a white dashed bracket). The intensity based images show extent of transverse loss of the IS/OS junction line of photoreceptors ([B1, B1a, D1, D1a] marked with yellow dashed lines).

Criterion for differentiation between normal and abnormal RPE was discontinuity of the RPE segmentation line as compared with healthy eyes. Additional post processing steps can be carried out for each recorded dataset. In order to depict areas of disturbed RPE patterns, the number of depolarizing pixels is added along each A-scan within the three-dimensional (3D) PS-OCT dataset. The resulting two-dimensional (2D) en face map visualizes the amount of depolarizing material within the retina. In addition to these depolarizing material thickness maps, RPE elevation maps, which allow the detection of irregular RPE shapes, especially drusen or accumulations of depolarizing material can be generated. The algorithm that is used to calculate such RPE elevation maps is described in detail elsewhere. ^25,27 In short, the algorithm works as follows: First, the most depolarizing pixel along each A-scan is detected, which is usually located in the center of the RPE. From these data points, outliers that are located far away from neighboring pixels are removed and the resulting points are fitted with an iterative Savitzky-Golay filter (polynominal order: 3, filter length: 200). The results of this fit yields the approximated normal RPE position. Afterwards, the anterior border of the RPE is found by searching for the first depolarizing pixel (DOPU < 0.75) along each A-scan, starting from the inner retinal layers. From the resulting data points, again outliers are removed and afterwards fitted with the iterative Savitzky-Golay filter (polynominal order: 3, filter length: 6). RPE elevation maps can now be generated by calculating the difference between the estimated normal RPE position and the fitted anterior border of the RPE. 

Correlation of PS-OCT Images With FAF Images

PS-OCT and FAF images were adjusted in size and angle of examination. In order to localize areas of interest in FAF images on corresponding PS-OCT B-Scans, retinal vessels were identified on both FAF images and pseudoscanning laser ophthalmoscope images (depth-integrated intensity images) of the PS-OCT system and used as anatomical landmarks. 

Molecular Genetics

Genomic DNA was extracted from peripheral EDTA-blood by the salting-out method. All 50 exons of the ABAC4 gene were amplified in 25 μL PCR reactions, consisting of 100 ng DNA, 1 μM each primer, 0.2 mM dNTPs, 1.5 mM MgCl₂, (NH₄)₂SO₄-containing amplification buffer, and 0.5 units Taq DNA polymerase (reagents from Fermentas; Thermo Fisher Scientific, Waltham, MA). After cycling (3 minutes 95°C, 38 × [40 seconds 95°C, 40 seconds 60°C, 40 seconds 72°C], 3 minutes 72°C), reaction products were quality controlled by agarose gel electrophoresis and then subjected to standard cycle-sequencing using BigDye terminators and a 3130XL DNA Analyzer (Life Technologies Ltd., Applied Biosystems, Paisley, UK). Sequence trace files were analyzed with the CodonCode Software (CodonCode Corporation, Centerville, MA) using NM_000350.2 as reference sequence. Mutations were evaluated by searching the Human Gene Mutation Database (HGMD; available in the public domain at www.biobase-international.com) and ATP-binding Cassette Transporter Retina (ABCR; available in the public domain at www.retina-international.org) mutation databases. Mutation Taster ³⁸ was used for the prediction of pathogenicity of novel missense mutations. The MaxEntScan modeling was used for predicting the effect of splicing mutations. ³⁹  

Results

Twenty-eight eyes of 14 patients (6 male, 8 female) from 13 different unrelated families with STGD and mutations in the ABCA4 gene were studied (Median age: 32 years, range, 23–52 years). The clinical and genetic findings of patients are summarized in the Table. 

Table

 

Patient Characteristics

Patient Number	Sex	Age	Age of Onset	Visual Acuity RE/LE	Fundus Phenotype	ERG Type	ABCA4 Mutation Allele 1			ABCA4 Mutation Allele 2
							Exon	Position cDNA	Effect on Protein	Exon	Position cDNA	Effect on Protein
1	M	52	19	1.00/1.30	1	2	33	c.4738_4739delTT	p.Leu1580Lysfs*16	46	c.6320G>A	p.Arg2107His
2	F	32	9	1.30/1.00	1	1	19	c.2829delG	p.Pro944Glnfs*6	42	c.5882G>A	p.Gly1961Glu
3	M	29	16	1.30/1.00	1	1	IVS1	c.66+3A>C	/	19	c.2791G>A	p.Val931Met
4	F	32	20	1.00/1.00	1	1	17	c.2588G>C*	p.Gly863Ala*	22	c.3266C>T	p.Thr1089Ile
5	M	28	21	0.52/0.70	1	1	42	c.5882G>A	p.Gly1961Glu	42	c.5882G>A	p.Gly1961Glu
6	F	25	20	1.00/0.80	1	1	13	c.1865delG	p.Ser622Thrfs*27	42	c.5882G>A	p.Gly1961Glu
7	F	32	27	0.05/0.10	2	1	25	c.3626T>C	p.Met1209Thr	33	c.4739T>C	p.Leu1580Ser
8	F	42	17	1.00/1.00	2	1	12	c.1622T>C*	p.Leu541Pro†	42	c.5882G>A	p.Gly1961Glu
9	F	23	23	0.00/0.00	2	1	IVS40	c.5714+5G>A	/	IVS40	c.5714+5G>A	/
10	F	30	16	1.00/1.00	2	1	12	c.1622T>C†	p.Leu541Pro†	19	c.2864A>G	p.Glu955Gly
11	M	45	19	1.30/1.30	3	2	12	c.1622T>C†	p.Leu541Pro†	17	c.2588G>C*	p.Gly863Ala*
12	M	37	14	1.00/1.00	3	2	12	c.1622T>C†	p.Leu541Pro†	19	c.2864A>G	p.Glu955Gly
13	F	27	20	1.00/1.00	3	2	12	c.1622T>C†	p.Leu541Pro†	IVS40	c.5714+5G>A	/
14	M	41	14	2.00/2.00	3	3	IVS13	c.1937+1G>A	/	17	c.2588G>C*	p.Gly863Ala*

 Patient number, sex, age, age of disease onset, visual acuity (logMAR), fundus phenotype (1, STGD phenotype 1; 2, STGD phenotype 2; 3, STGD phenotype 3), ERG type, ABCA4 mutation allele 1 and ABCA4 mutation allele 2; exons and coding DNA (cDNA) positions based on reference sequence NM_000350 (IVS: intervening sequence, intron) are shown. Novel mutations are indicated in bold; patient numbers 10 and 12 are siblings.

*  Linked to c.2828G>A (p.Arg943Glu).

†  Linked to c.3113C>T (p.Ala1038Val).

Stargardt Disease Phenotype 1

Twelve eyes of six patients were classified as STGD phenotype 1 based on fundoscopic findings (central, atrophic-appearing macular lesion with or without perifoveal flecks). 

Intensity B-scans of PS-OCT (as standard SD-OCT imaging), showed focal retinal thinning of retinal layers over a central area of RPE atrophy in four patients (Fig. 1B1). The exact border of RPE atrophy could not be identified reliably on intensity B-scans, since a precise distinction between thinned RPE and Bruch's membrane, which is still visible in areas of RPE loss, was not possible. DOPU B-scans allowed for identification of loss of depolarizing RPE in the eight eyes of these four patients (Figs. 1B3, 1B3a). By thresholding the DOPU image, an overlay of the segmented RPE and the intensity image was generated (RPE segmentation B-scan), precisely delineating the borders of RPE atrophy (Figs. 1B2, 1B2a). RPE transillumination (enhanced penetration depth) visible as hyperreflective area below the absent RPE on intensity B-scans could be observed in six of eight eyes with larger zones of RPE atrophy. RPE segmentation B-scans (Fig. 1B2), DOPU images (Fig. 1B3), and depolarizing material thickness maps (Fig. 1C) revealed significant amounts of depolarizing material in the choriocapillaris and deeper choroidal layers of these eyes. Areas with absent RPE segmentation line (marked with a white double-pointed arrow on Figs. 1B2, 1B3) correlated well with areas of hypofluorescene on FAF (marked with a white bracket on Fig. 1A). Although the margins of RPE atrophy are well defined on DOPU and RPE segmentation B scans, an irregular appearance of the RPE could be identified in areas adjacent to the central atrophy (indicated with dashed double-pointed arrows on Figs. 1B2, 1B3, 1D2, 1D3). Small focal skip lesions could be detected within the DOPU images, RPE segmentation B-scans and depolarizing material thickness maps. Focal skip lesions could be detected only on PS-OCT images because contrast based on intensity B scans is not high enough, and enhanced penetration depth from absent RPE is only visible in larger lesions. These areas corresponded in large part with rings of irregular fluorescence on FAF images (marked with a dashed white bracket on Fig. 1A). Above areas of irregular RPE and focal skip lesions, transverse loss of the IS/OS junction of photoreceptors could be observed on intensity based images (marked with yellow, dashed lines on Figs. 1B1, 1B1a, 1D1, 1D1a). 

Four eyes of two patients were characterized by a subfoveal optical gap (foveal cavitation) with sharp vertical edges on intensity based images (Fig. 2B1). In this area, the IS/OS junction line was absent. RPE segmentation B-scans (Fig. 2B2) and DOPU images (Fig. 2B3) showed no visible irregularities of the RPE beneath the foveal cavitation. FAF showed a small oval-shaped subfoveal area of reduced FAF corresponding to the area of absent photoreceptor layers observed on intensity-based and DOPU images (Fig. 2A). 

Figure 2

 

FAF and PS-OCT images of a patient with an ABCA4-associated phenotype characterized by a foveal cavitation. (patient number 5; Stargardt Phenotype 1): FAF image (A), intensity based B-scan (B1), RPE segmentation B-scan (B2), and DOPU image (B3). Yellow line indicates the position of B-scans. The diameter of the oval-shaped subfoveal area of reduced FAF ([A] marked with a yellow, dashed bracket) corresponds to the extent of transverse loss of the IS/OS junction line of photoreceptors ([B1] marked with a yellow, dashed line) on the intensity-based image. Neither this area, nor the area of increased perifoveal fluorescence is associated with visible RPE changes on RPE segmentation B-scans and DOPU images.

BCVA was reduced in all patients, with a mean of 0.99 logMAR (SD ± 0.23). Five patients showed an ERG type 1 response, one patient showed an ERG type 2 response in both eyes, indicating generalized cone photoreceptor dysfunction. No evident difference between eyes showing ERG type 1 or type 2 responses could be detected on PS-OCT images. 

Molecular analysis of the ABCA4 gene revealed pathogenic mutations in all patients within this group. The four patients showing a central area of RPE atrophy were compound heterozygous for a known missense mutation and for a novel, previously not described, mutation: patient 1 harbored the known c.6320G>A (p.Arg2107His) mutation and a, so far not described, null mutation in exon 33, c.4738_4739delTT (p.Leu1580Lysfs*16), patient 2 the frequent mild missense mutation c.5882G>A (p.Gly1961Glu) and a novel single-base deletion, c.2829delG (p.Pro944Glnfs*6) in exon 19, and patient 3 the known c.2791G>A (p.Val931Met) mutation and a presumably destructive novel splice site mutation within the first intron (c.66+3A>C). Patient 4 harbored a combination of the relatively frequent mild founder allele c.[2588G>C; 2828G>A] (p.[Gly863Ala; Arg943Gln]) and a, so far unknown, missense mutation in exon 22, c.3266C>T (p.Thr1089Ile), which affects a highly conserved threonine residue within the transmembrane helix and is predicted to be deleterious. Both patients characterized by a foveal cavitation carried the frequent mild p.Gly1961Glu mutation, which was homozygous in patient 5 and compound heterozygous with a novel single-base deletion in exon 13, c.1865delG (p.Ser622Thrfs*27) in patient 6. 

Stargardt Disease Phenotype 2

Eight eyes of four patients were classified as STGD phenotype 2, showing numerous flecks, extending anterior to the vascular arcades and nasal to the optic disc. Retinal flecks could be identified on FAF images as hyperautofluoerscent lesions of different size and shape. Correlation of FAF images with PS-OCT images allowed for identification of flecks as accumulations of hyperreflective material at the level of the RPE/Bruch membrane complex on intensity based images. These clusters of hyperreflective material were significantly heterogeneous in size and shape and showed similar depolarizing properties as the RPE on DOPU and RPE segmentation B-scans (Fig. 3). Some of these lesions showed protrusions of depolarizing material expanding further through the interface of the IS/OS junction line, up to the external limiting membrane (Figs. 3G, 3H). Distribution of flecks could be depicted on RPE elevation maps correlating well with FAF images (Figs. 3A, 3B). Four of eight eyes presented a central RPE atrophy with areas of absent and irregular RPE on RPE segmentation B-scans and DOPU images as patients with STGD phenotype 1. 

Figure 3

 

FAF and PS-OCT images of two patients with STGD phenotype 2: FAF image (A), depolarizing material thickness map (B), and RPE segmentation overlay B-scan (C) of patient number 7. Yellow line indicates the position of RPE-segmentation B-scans. Yellow arrows indicate two retinal flecks visible as hyper-autofluorescent lesions on FAF (A), corresponding with accumulations of depolarizing material on RPE elevation map (B), and the RPE segmentation B-scan (C). Yellow boxes indicate area of magnified images of flecks (C1a, C1b). (C2a, C2b) show corresponding DOPU images, (C3a, C3b) corresponding intensity based images. FAF image (F) and two RPE segmentation B-scans (G, H) of patient number 9. Yellow arrows indicate two retinal flecks visible on FAF (F), corresponding with accumulations of depolarizing material at the level of the RPE, expanding through the interface of IS/OS junction line (G) and further through the external limiting membrane (H).Yellow boxes indicate area of magnified images (G1, H1). (G2, H2) show corresponding DOPU images, (G3, H3) intensity-based images.

Foveal sparing in four of eight eyes explained the relatively good mean BCVA of 0.52 logMAR (SD ± 0.48) in this group of patients. All eyes showed an ERG type 1 response indicating normal, generalized rod and cone photoreceptor function. Molecular analysis of the ABCA4 gene revealed known pathogenic mutations in all four patients within this phenotype group (see Table). In one patient, a so far unknown missense mutation in exon 19 was detected, c.2864A>G (p.Glu955Gly), which affects a conserved residue within the ABC transporter 1 domain and is predictably pathogenic. 

Stargardt Disease Phenotype 3

Eight eyes of four patients classified as STGD phenotype 3, showed extensive atrophic-appearing RPE changes and partially resorbed flecks on FAF imaging. Intensity-based B-scans of PS-OCT showed large central areas of RPE atrophy in all patients. DOPU images and RPE segmentation B-scans allowed for identification of loss of depolarizing RPE (Figs. 4C2, 4C3) and precisely delineated the borders of RPE atrophy as in STGD phenotype 1. RPE transillumination visible as hyper reflective area below the absent RPE on intensity-based B-scans could be observed in all eyes (Fig. 4C1). In these areas, RPE segmentation B-scans (Fig. 4C2), DOPU images (Fig. 4C3), and depolarizing material thickness maps (Fig. 4B) revealed significant amounts of depolarizing material in the choriocapillaris and deeper choroidal layers. This is in contrast to findings documented in geographic atrophies secondary to AMD, ²⁴ where considerably fewer depolarizing signals could be segmented in the choroid. Please note the striking difference in the appearance of the two types of atrophies on Figure 4 showing patient number 14 (Figs. 4A–C3), and an exemplary 88-year-old patient with geographic atrophy secondary to AMD (Figs. 4D–F3). Peripapillary sparing was defined as the presence of a uniform autofluorescence pattern in the peripapillary retina on FAF images and could be identified in most patients. In six of eight eyes, PS-OCT imaging revealed preservation of RPE and photoreceptor layers in this area, showing only minor thickness irregularities on RPE segmentation B-scans and DOPU images (Figs. 4C2, 4C3). However, both eyes of patient number 13 showed a markedly reduced RPE signal, multiple skip lesions, and absent photoreceptor layers even in areas corresponding to residual peripapillary sparing on FAF images (Fig. 5). 

Figure 4

 

FAF and PS-OCT images of a 41-year-old patient with STGD phenotype 3 (patient number 14) (A–C3) and an exemplary 88-year-old patient with geographic atrophy secondary to AMD (D–F3): FAF image (A, D), depolarizing material thickness map (B, E), intensity-based images (C1, F1) RPE segmentation B-scan (C2, F2), DOPU image (C3, F3). Yellow line indicates the position of B-scans. Peripapillary sparing is indicated with a yellow asterisk on FAF (A) and RPE segmentation B scans (C2) showing preservation of RPE and photoreceptor layers. The RPE segmentation B-scan, DOPU image, and depolarizing material thickness maps of the patient with STGD show significantly more depolarizing material in the choroid than in the patient with geographic atrophy secondary to AMD.

Figure 5

 

FAF and PS-OCT images of a patient with STGD phenotype 3 (patient number 13) centered on the papilla: FAF image (A), RPE segmentation B-scan (B1), DOPU image (B2). Yellow line indicates the position of B-scans. Residual peripapillary sparing is indicated with a yellow asterisk on FAF (A). The white and black boxes on the RPE segmentation B-scan (B1) and the DOPU image (B2) show markedly reduced RPE signal, multiple skip lesions, and absent photoreceptor layers even in areas corresponding to peripapillary sparing on the FAF image.

BCVA was severely reduced in all patients, with a mean of 1.33 logMAR (SD ± 0.41). Three patients showed an ERG type 2 response indicating generalized cone photoreceptor dysfunction, one patient showed an ERG type 3 response in both eyes, indicating generalized rod and cone photoreceptor dysfunction. No evident difference between eyes showing ERG type 2 or type 3 responses could be identified on PS-OCT images. Molecular analysis of the ABCA4 gene revealed known and fully pathogenic mutations in all four patients also within this phenotype group (see Table). Notably, three of four patients carried the frequent STGD allele c.[1622T>C; 3113C>T] (p.[Leu541Pro; Ala1038Val]), which is generally considered to be associated with more severe disease progression. ⁴⁰ Patient 12 carried this allele together with the same novel missense mutation (p.Glu955Gly) on the second chromosome as his younger sister (patient 10, STGD phenotype 2), which is suggestive of intrafamilial phenotype variation associated with this specific combination of ABCA4 mutations. 

Discussion

The current study is the first to evaluate STGD using a novel large-field, high-speed PS-OCT system in a cross-sectional patient population with well characterized phenotypes and genotypes. PS-OCT-specific alterations were described and correlated with FAF images in the three fundoscopic phenotypes of STGD according to Fishman et al. ³⁰ Intensity-based B-scans and the additional information provided by DOPU images, RPE segmentation B-scans, RPE elevation maps, and depolarizing material thickness maps of PS-OCT have their specific merits in detecting morphologic changes at the level of the RPE and the photoreceptor layers, which are of particular interest in STGD. Intensity-based B-scans as used in conventional SD-OCT imaging have already been shown to facilitate in vivo evaluation of STGD by depicting intraretinal changes including photoreceptor irregularities, configuration of retinal flecks and loss of RPE. ^{17 –20} Erroneous detection of the RPE in SD-OCT imaging has been described in previous studies, ⁴¹ and is related to the limited tissue differentiation of intensity-based images. 

RPE segmentation B-scans and DOPU images of PS-OCT were of particular value for precisely defining the margins of RPE atrophy and to document RPE irregularities such as focal skip lesions, whose identification can be ambiguous on intensity-based SD-OCT images. Identification of the RPE in PS-OCT is based on an intrinsic polarization contrast since melanin containing structures within the RPE depolarize backscattered light. The thickness of the RPE obtained by the segmentation algorithm on RPE segmentation B-scans is larger than the value expected for the healthy RPE based on findings in histology where the RPE is described as a cell monolayer of only a few microns thick. ⁴² The explanation is that the segmented RPE is the result of a convolution of the real tissue thickness with the evaluation window function used for RPE identification. ³⁴  

Irregularities and loss of photoreceptor layers without RPE atrophy could be identified on intensity-based images, particularly in early stages of the disease (Figs. 1, 2). The exact sequence of events in the development of photoreceptor and RPE atrophy in STGD remains controversial. Although the accepted hypothesis was that atrophy of the RPE occurs initially, with secondary photoreceptor degeneration, Gomes et al. ²² reported disorganization of the photoreceptors in regions of preserved RPE based on SD-OCT and FAF findings. This suggests that degeneration of the photoreceptors may occur before that of the RPE. In this study, 12 patients showed RPE irregularities and partly focal skip lesions on RPE segmentation B-scans and DOPU images in regions of disorganization or absence of photoreceptor layers, corresponding in large part with an irregular fluorescence pattern on FAF images. The two patients characterized by a foveal cavitation, possibly representing a transient stage during the disease process, showed a healthy appearing RPE beneath the absent IS/OS junction line. These results suggest that changes in the photoreceptor layer may occur simultaneously with the development of abnormalities in the RPE layer and only in the mildest forms of ABCA4 disease photoreceptor decay seems to precede visible changes of the RPE; it is possible that even in these cases simultaneous damage occurred in the RPE layer and that the resolution of PS-OCT is not high enough to visualize such alterations. 

Focal accumulations of hyperreflective material found on intensity-based images corresponded to retinal flecks on FAF images as also reported in a previous study. ²⁰ These accumulations showed similar depolarizing properties as the RPE on RPE segmentation B-scans and DOPU images. We documented a striking topographic correspondence of these accumulations on RPE elevation maps to hyperautofluorescent flecks on FAF images (Fig. 3). Histologic examinations of eyes affected by STGD reported abnormal accumulation of intracellular material in RPE cells anterior to the equator and single or patches of enlarged RPE cells filled with this material more centrally. Based on histochemical and electron microscopic criteria, the accumulated material has been identified as melanolipofuscin and lipofuscin. ^10,11,43,44 Possibly, these distended RPE cells or extracellular deposits after cell lysis correspond to the yellow flecks observed clinically, ^10,11,43,44 similarly as accumulations of depolarizing material correspond to retinal flecks on FAF images in our study. On the basis of these findings, we suggest that enlarged RPE cells filled with an increased amount of lipofuscin or melanolipofuscin or their content after cell lysis show similar depolarizing properties as normal melanin containing RPE cells. 

Size and appearance of RPE atrophies in STGD phenotype 3 are rather similar to geographic atrophies secondary to AMD on FAF images. Depolarizing material thickness maps revealed the area of geographic atrophies secondary to AMD, which appeared as sharply delineated areas with absent RPE (Schmidt-Erfurth U, et al. IOVS 2012;53:ARVO E-Abstract 3081). Interestingly, RPE segmentation B-scans, DOPU images, and depolarizing material thickness maps revealed considerably more depolarizing material in the choriocapillaris and deeper choroidal layers of some patients with STGD (Fig. 4). Significant amounts of depolarizing material could also be detected in atrophic regions of phenotype 1 (Fig. 1), however, less than in phenotype 3. The histologic nature of these depolarizing signals remains uncertain. Possible explanations include larger numbers of melanocytes in the choroid due to the younger age of STGD patients or accumulation of intracellular lipofuscin/melanolipofuscin from ruptured RPE cells. The latter could explain why flecks and the choroid underneath RPE atrophy share the depolarizing properties. 

A recent study by Giani et al. described “dark atrophy” with indocyanine green angiography (ICGA) in STGD. ⁴⁵ Since this phenomenon is restricted to the area of geographic atrophy in STGD and is much less common in AMD, there may be a relationship between the ICGA-imaged dark atrophy and an increased amount of depolarizing material in the choroid. One explanation regarding the origin of the ICGA-imaged dark atrophy may be that depolarizing material in the choroid obscures ICGA cyanescence. 

A typical feature of STGD is peripapillary sparing, characterized by the absence of flecks and RPE atrophy in the peripapillary area of the retina. This was shown in studies using en face imaging, ^15,46 and a study using SD-OCT to evaluate the integrity of the outer retinal layers and RPE. ⁴⁷ The reasons for this relative sparing in the presence of widespread disease are still controversial. Suggested explanations are less photo-oxidative damage due to a reduced light load on the photoreceptors and RPE in the presence of a thicker peripapillary retinal nerve fiber layer and a more favorable photoreceptor/RPE ratio in the peripapillary area. ⁴⁶ All of our four patients with phenotype 3 STGD disease presented with peripapillary sparing on FAF images to a variable extent. In six eyes of three patients preservation of the RPE and photoreceptor layers could be documented on corresponding RPE segmentation B-scans and DOPU images, showing only minor thickness irregularities (Fig. 4). Patient number 13 presented with a narrow ring of residual peripapillary sparing on FAF images; however, RPE segmentation B-scans and DOPU images showed markedly reduced RPE and photoreceptor layers extending into this area (Fig. 5). Since there is one report showing that a small percentage of patients with STGD develop peripapillary atrophy, ⁴⁸ PS-OCT could be a valuable tool to measure more reliably than other imaging modalities the rate of peripapillary sparing. 

Given the large number of approximately 600 mutations that have been described to be causatively involved in ABCA4-associated disease, prediction of genotype–phenotype relations are inherently complex. Previously suggested genotype–phenotype models for some mutations consider a degree of residual ABCR activity, a degree of misfolding and mislocalization of the protein. This is limited in our patient cohort because six of the 15 STGD1 alleles identified were novel mutations that were not described in other studies so far. Nevertheless, 11 patients carried at least one of the relatively frequent STGD alleles that are considered mild (I: c.[2588G>C; 2828G>A] p.[Gly863Ala; Arg943Glu]; II: c.5714+5G>A; III: c.5882G>A p.Gly1961Glu) or severe (c.[1622T>C; 3113C>T p.[Leu541Pro; Ala1038Val]), respectively. Based on these relatively well-established genotype–phenotype correlations, it is possible to draw the following (cautious) conclusions from our study: STGD phenotype 1 was associated with a missense mutation on one allele and a null mutation on the second allele in three cases. No patient within this group carried a known severe missense mutation, yet three carried a known mild missense mutation. Both patients with the foveal cavitation phenotype carried the mild Gly1961Glu mutation, either homozygous or in combination with a null mutation. On the other hand, STGD phenotype 2 was associated with the presence of the known severe complex mutation Leu541Pro/Ala10398Val in two of four cases, while three of four patients within phenotype 3 carried this mutation. 

Limitations of this study were the relatively small number of patients in each phenotype group, the lack of follow-up data that would allow for monitoring disease progression and that no near-infrared fundus autofluorescence (NI-FAF) imaging was performed. Since PS-OCT and NI-FAF are both detecting melanin related compounds, a comparison of NI-FAF maps with depolarizing material thickness maps would be of interest. 

To conclude, PS-OCT and FAF imaging revealed morphologic changes at the level of photoreceptor layers, the RPE and the choroid allowing for differentiation of characteristic patterns of changes associated with STGD. Data demonstrated the clinical usefulness of PS-OCT to automatically segment the RPE/Bruch's membrane complex, to delineate areas of geographic atrophy and to map the fleck distribution three-dimensionally. PS-OCT provides additional insight in comparison to conventional intensity-based SD-OCT as the depolarizing signals in the choroid, depolarizing properties of retinal flecks, and subtle irregularities of the RPE like skip lesions. The clinical significance of such findings remains open and will be subject of further investigations. Future studies are needed to evaluate the potential of PS-OCT to quantify the different alterations during follow-up. 

Acknowledgments

The authors thank Anna Heger, Simone Stenico, and Iris Schmidt (Neuromuscular Research Department) for their excellent technical support in ABCA4 sequence analysis. 

Disclosure: M. Ritter, None; S. Zotter, Canon, Inc. Tokyo (F); W.M. Schmidt, None; R.E. Bittner, None; G.G. Deak, None; M. Pircher, Canon, Inc. Tokyo (F); S. Sacu, None; C.K. Hitzenberger, Canon, Inc. Tokyo (F); U.M. Schmidt-Erfurth, Canon, Inc. Tokyo (F) 

Footnotes

⁴  See the appendix for the members of the Macula Study Group Vienna

Appendix

The Macula Study Group Vienna 

Department of Ophthalmology, Medical University of Vienna, Vienna, Austria 

Stefan Sacu, MD 

Wolf Buehl, MD 

Roman Dunavoelgyi, MD 

Katharina Eibenberger, MD 

Astrid G. Fuchsjaeger-Mayrl, MD 

Georgios Mylonas, MD 

Sandra Rezar, MD 

Markus Ritter, MD 

Guenther Weigert, MD 

Michael Georgopoulos, MD 

Ursula M. Schmidt-Erfurth, MD