Deep Scleral Exposure: A Degenerative Outcome of End-Stage Stargardt Disease

Abstract

PURPOSE:

To describe a distinct phenotypic outcome of outer retinal degeneration in a cohort of genetically confirmed patients with recessive Stargardt disease (STGD1).

DESIGN:

Retrospective case series.

METHODS:

Twelve patients, who were clinically diagnosed with STGD1 and exhibited a unique degenerative phenotype, were included in the study. Two disease-causing mutations were found in all patients by direct sequencing of the ABCA4 gene. Clinical characterization of patients were defined on fundus photographs, autofluorescence images (488-nm and 532-nm excitation), spectral-domain optical coherence tomography (SD-OCT), and full-field electroretinogram (ffERG) testing.

RESULTS:

Mean age at initial presentation was 67.8 years and reported age of symptomatic onset was 14.1 years (mean disease duration [ 53.8 years). Best-corrected visual acuity ranged from 20/400 to hand motion. All patients exhibited advanced degeneration across the posterior pole resulting in a reflectively pale, blonde fundus owing to unobstructed exposure of the underlying sclera. SD-OCT revealed complete loss of the outer retinal bands (external limiting membrane, ellipsoid zone, interdigitation zone, and retinal pigment epithelium) and choroidal layers. Scotopic and photopic waveforms on ffERG were nonrecordable or severely attenuated in 8 patients who were tested.

CONCLUSIONS:

Widespread scleral exposure is a clinical outcome in a subset of STGD1 following a long duration of disease progression (~50 years). The blonde fundus in such cases may exhibit phenotypic overlap and shared therapeutic implications with other aggressive chorioretinal dystrophies such as end-stage choroideremia, gyrate atrophy, or RPE65-Leber congenital amaurosis.

AUTOSOMAL RECESSIVE STARGARDT DISEASE (STGD1; MIM #248200) is the most common inherited retinal dystrophy, responsible for mostly adolescent-onset progressive central vision loss.¹ The causal gene, the photoreceptor-specific ATP-binding cassette transporter, ABCA4, was identified in 1997²; since then >1000 disease-associated variants have been reported.³ The disease phenotypes, resulting from biallelic mutations in ABCA4, vary extensively and sometimes exhibit phenotypic overlap with conditions caused by mutations in other genes. For example, bull’s-eye and occult maculopathy are well-described early clinical abnormalities detected in patients harboring the c.5882G>A (p.Gly1961Glu) mutation in ABCA4,^4,5 as well as other conditions ranging from maculopathies caused by mutations in CRX^6,7 and PROM1,⁸ central areolar choroidal dystrophy (RDS/PRPH2),^8–10 RP1L1-occult macular dystrophy,^11,12 and achromatopsia (CNGA3,¹³ CNGB3,¹⁴ GNAT2,¹⁵ PDE6C,¹⁶ PDE6H,¹⁷ ATF6¹⁸) to drug-induced toxicities (chloroquine and hydroxychloroquine).¹⁹

Clinical precision decreases with disease progression as the manifestation of pathognomonic features, such as peripapillary sparing²⁰ and the appearance of pisciform flecks,^21,22 become indiscernible from gradual deterioration of retinal tissue. These features are subsequently replaced by the appearance of bone-spicule pigment deposition, vessel attenuation, optic disc pallor, and generalized attenuation of cone and rod function, which reflect characteristics of panretinal diseases such as retinitis pigmentosa.^23,24 Comprehensively characterizing the expansive clinical presentation of a disease improves diagnostic accuracy in the clinic and provides invaluable scientific insight into disease etiology and natural history. Furthermore, detailed clinical characterization of STGD1 guides and facilitates the effective design of interventional trials, some of which are currently ongoing, including, gene therapy (NCT01345006) and the slowing of A2E formation by the oral ingestion of deuterated vitamin A (NCT02402660).

The current study describes a phenotypic outcome of advanced degeneration in the natural history of STGD1. Clinical documentation consists of multimodal retinal imaging in a study cohort with a mean disease duration of over 50 years.

METHODS

PATIENTS:

All study procedures were defined under protocol #AAAI9906 approved by the Institutional Review

Board at Columbia University Medical Center. The study adhered to tenets set out in the Declaration of Helsinki. A retrospective review of 300 patients with a clinical diagnosis and genetic confirmation (at least 2 disease-causing mutations in the ABCA4 gene) of STGD1 was conducted at the Department of Ophthalmology, Columbia University. Patients identified and selected for the study exhibited widespread chorioretinal degeneration of the posterior pole resulting in visibility of the underlying sclera. Patients with degeneration of the outer retina but not the choroid (ie, presence of continuously intact choroidal vessels) were not included in the characterization of this phenotype. Assessment of scleral visibility was made on digital color fundus photographs (50-degree field).

CLINICAL EXAMINATION AND CHARACTERIZATION:

Each patient underwent a complete ophthalmic examination by a retinal physician (S.H.T.), which included a slit-lamp and dilated funduscopy examination, best-corrected visual acuity (BCVA; Snellen), color fundus photography, fundus autofluorescence (AF, 488 nm, 532 nm, and 787 nm), spectral-domain optical coherence tomography (SD-OCT) scanning and full-field electroretinogram (ffERG) testing. Imaging across all modalities was conducted following pupil dilation (>7 mm) with tropicamide (1%) and phenylephrine hydrochloride (2.5%). Fundus autofluorescence (488 nm) images and 9 mm horizontal foveal SD-OCT scans were acquired with the Spectralis HRA+OCT (Heidelberg Engineering, Heidelberg, Germany). Ultra-widefield autofluorescence images were acquired with an Optos 200 Tx (Optos PLC, Dunfermline, United Kingdom). The ffERGs were recorded with silver-impregnated fiber electrodes (DTL; Diagnosys LLC, Littleton, Massachusetts, USA) on the Espion Visual Electrophysiology System (Diagnosys LLC) in accordance with International Society for Clinical Electrophysiology of Vision (ISCEV) standards.²⁵

MOLECULAR CHARACTERIZATION:

Screening of the ABCA4 gene was performed by next-generation sequencing (NGS) as previously described.^26,27 All detected possibly disease-associated variants were confirmed by Sanger sequencing and analyzed with the Alamut software (http://www.interactive-biosoftware.com). Segregation of the new variants with the disease was analyzed in families if family members were available. Functional annotation of variants was determined using computational software including Annovar using pathogenicity scores of MCAP, REVEL, Eigen, CADD, DANN, and SPIDEX.^26–32 As a general guideline, pathogenic consequences are predicted for variants with scores over 0.025 for MCAP, 0.5 for REVEL, 0.5 for Eigen, 20 for CADD, 0.97 for DANN, and more than −2 or less than 2 for SPIDEX (psi z-score).^28–34 The allele frequencies of all variants were compared with The Genome Aggregation Database (gnomAD) (http://gnomad.broadinstitute.org/gene/ENSG00000198691; accessed January 2018).

RESULTS

ALL PATIENTS (N = 12) INCLUDED IN THE STUDY PRESENTED with a long history of retinal degeneration. Mean cohort age at presentation was 67.8 years (range, 48–85 years). The reported age of symptomatic onset varied from 5 to 29 years of age (mean = 14.1 years), giving the study cohort an average disease duration of 53.8 years. Visual acuities were not correctable outside 20/400 to hand motion (HM) in all patients. Table 1 further summarizes demographic, clinical, and genetic characteristics. Funduscopic examinations in each patient were remarkable for features consistent with advanced chorioretinal degeneration including optic disc pallor, attenuation of the retinal vasculature, and dark pigment migration in the macula and periphery. Wide-spread loss of retinal pigment epithelium (RPE) and choroidal vessels was observed in the posterior pole of all patients, resulting in exposure of underlying scleral tissue (Figure 1). The fundus at this stage was highly reflective, exhibiting a blonde hue and an irregularly tessellated appearance in certain regions. These profound areas of degeneration extended into the far periphery, in some cases, beyond which residual retinal tissue and large choroidal vessels became visible. Widespread nummular and bone-spicule (nonparavascular) pigment deposition was noted in all patients (Figure 1, blue arrowheads).

TABLE 1.

Demographic, Clinical, and Genetic Characteristics of Patients in the Scleral Exposure Stage of Stargardt Disease

						BCVA		ffERG
Patient	Age (y)	AO (y)	DD (y)	Sex	Race	OD	OS	Scotopic	Max	Flicker	Photopic	Follow-up Duration (y)
1	85	20	65	M	White	CF	CF	NR	NR	NR	NR	10
2	68	29	39	M	White	20/400	20/400	↓↓	↓↓	↓↓	↓↓	10
3	74	14	60	M	White	20/400	20/400					1
4	55	9	46	F	White	20/400	20/400	NR	↓↓	NR	NR	10
5	65	8	57	F	White	CF	CF	NR	NR	NR	NR	6
6	48	5	43	F	White	20/400	20/400	NR	NR	NR	NR	6
7	75	25	50	M	White	CF	CF	↓↓	↓↓	↓↓	↓↓	4
8	62	14	48	M	White	CF	20/400	NR	NR	NR	NR	N/A
9	72	17	55	M	White	HM	HM	NR	NR	NR	NR	2
10	68	18	50	F	White	20/400	20/400					2
11a	70	5	65	F	White	CF	20/400					5
12a	72	5	67	M	White	CF	CF					5

y = years; AO = age of onset; CF = counting fingers; DD = disease duration; ffERG = full-field electroretinogram; HM = hand motion; N/A = not available; NR = nonrecordable; ↓↓ = severely attenuated.

^aPatients 11 and 12 are siblings.

FIGURE 1.

Fundus montage and photographs of Stargardt disease patients exhibited advanced retinal degenerative features such as optic disc pallor, retinal vasculature attenuation, and pigment deposition (blue arrowheads). The fundus of each patient, following a prolonged duration of disease progression, is characterized by a reflectively pale, blonde appearance owing to exposure of the underlying sclera.

Areas of degeneration seen on fundus examination corresponded to homogenously hypoautofluorescent lesions with well-delineated, scalloped edges, continuous across a large area (Figure 2, Top) or in a diffuse pattern of numerous, round coalescing foci (Figure 2, Bottom) on AF (488-nm and 532-nm) imaging. Nonatrophic areas adjacent to degeneration were heterogeneously atrophic, exhibiting a punctate appearance of alternating hyper-and hypoautofluorescent flecks reaching postequatorial regions. Macular SD-OCT scans showed extensive retinal thinning and a visible loss of the characteristic foveal contour in all cases except Patient 6 (Figure 3). A residual presence of the retinal nerve fiber layer (RNFL), ganglion layer, and plexiform layer was noted in all patients. A closer examination was required to detect the intermittent presence of, albeit considerably thinned, remnants of the outer nuclear layer (ONL). The most notable finding on SD-OCT was the complete absence of all hyperreflective outer retinal bands that are attributable to photoreceptor inner/outer segments, ellipsoid zone (EZ) and interdigitation zone (IZ), and RPE, as well as the underlying choroid, resulting in increased signal transmission into the sclera throughout the length of the B-scan (Figure 3, red arrows). Longitudinal assessment was possible in all except Patient 8, through follow-up visits ranging between 1 and 10 years after initial presentation. Further lesion growth and the appearance of peripheral degeneration were noted at subsequent visits; however, the pale regions of visible sclera remained effectively unchanged over time (ie, no increase or loss of pigment deposition) (Figure 4). ffERG testing revealed virtually nonrecordable cone and rod responses in 6 out of 8 patients tested, while severely attenuated cone and rod responses were measured in Patient 2 and Patient 7 (Table 1).

FIGURE 2.

Ultra-widefield fundus autofluorescence (AF) imaging (488-nm and 532-nm) of patients in the scleral exposure stage of Stargardt disease. Regions of atrophy are hypoautofluorescent with well-defined edges, covering the posterior pole as either (Top) a continuous region or (Bottom, left and right) multiple, coalescing foci. An extensive heterogeneity in AF attributable to confluent accumulations of flecks is visible in the mid to far periphery. The optic disc in each fundus image is outlined (dotted white line).

FIGURE 3.

Spectral-domain optical coherence tomography (SD-OCT, center column) with corresponding fundus regions (left column) of patients in the scleral exposure stage of Stargardt disease. Horizontal SD-OCT scans through the fovea reveal a complete loss of the reflective outer retinal bands, external limiting membrane, ellipsoid zone, interdigitation zone, and retinal pigment epithelium, as well as choroidal layers. Hypertransmission of the OCT signal is visible throughout the scan, revealing the underlying sclera (red arrowhead) in maginfied SD-OCT insets (right column). A central region of residual choroid (yellow arrow) was observed in Patient 6.

FIGURE 4.

Longitudinal documentation of Stargardt disease patients exhibiting widespread, scleral-deep chorioretinal degeneration. No detectable changes in pigment deposition or further vascular deterioration were noted after 5–10 years in Patient 4 (Top rows) and Patient 1 (Bottom rows), respectively, indicating the complete absence of chorioretinal tissue.

Complete sequencing of the ABCA4 gene identified at least 2 (expected) disease-causing variants in all patients (Table 2). Variants found in the cohort included 17 missense variants, of which 4 were complex alleles, including c.[2588G>C; 5603A>T] (p.[Gly863Ala; Asn1868Ile]),³⁵ c.[1253T>C; 5603A>T] (p.[Phe418Ser; Asn1868Ile]), and c.[4594G>A; 5603A>T] (p.[Asp1532Asn; Asn1868Ile]) twice. All missense variants were predicted to be ‘‘deleterious’’ by SIFT (score = 0) and MutationTaster (p = 1) and pathogenic by M-CAP (scores = 0.391–0.791), REVEL (scores = 0.77–0.98), and CADD (scores = 27–42). Other variants included a stop-gain, c.6088C>T (p.Arg2030*) variant in Patient 3 and 8 noncoding variants: 5 variants in canonical splice site sequences (c.4773+3A>G, c.2160+1G>C, c.768G>T, c.3050+5G>A, and c.5714+5A>G) and 3 deepintronicvariants(c.302+68C>T, c.4539+2028C>T and c.4539+2001G>A) that likely have a negative effect on exon splicing.³⁶ The sibling pair Patient 11 and Patient 12 harbored a deep intronic variant, c.4539+2028C>T, and a deletion/insertion, c.6148–698_c.6670del/insTGTGCACCTCCCTAG, described in a previous report.³⁷ Patient 5 harbors another deep intronic variant, c.4539+2001G>A.

TABLE 2.

Summary and Pathogenicity Analysis of ABCA4 (NM_000350.2) Variants in the Study Cohort

Patient	cDNA Variant	Protein Variant	Type	Coding Effect	M-CAP	REVEL	Eigen	CADD13	DANN
1	c.3322C>T	p.(R1108C)	Substitution	Missense	0.797	0.89	0.81	35	1.00
	c.4139C>T	p.(P1380L)	Substitution	Missense	0.391	0.87	0.70	28	1.00
2	c.4139C>T	p.(P1380L)	Substitution	Missense	0.391	0.87	0.70	28	1.00
	c.5714+5G>A	p.(?)	Substitution	?	-	-	-	-	-
3	c.2588G>Ca	p.(G863A)	Substitution	Missense	-	0.80	0.58	27	1.00
	c.5603A>Ta	p.(N1868I)	Substitution	Missense	-	0.40	0.03	26	0.92
	c.6088C>T	p.(R2030*)	Substitution	Nonsense	-	-	0.54	42	1.00
4	c.3322C>T	p.(R1108C)	Substitution	Missense	0.797	0.89	0.81	35	1.00
	c.1253T>C	p.(F418S)	Substitution	Missense	0.582	0.93	0.81	29	1.00
5	c.161G>A	p.(C54Y)	Substitution	Missense	-	0.98	0.87	29	1.00
	c.2160+1G>C	p.(?)	Substitution	?	-	-	-	-	-
6	c.768G>T	p.(?)	Substitution	?	-	-	-	-	-
	c.4539+2001G>A	p.(?)	Substitution	?	-	-	-	-	-
7	c.3050+5G>A	p.(?)	Substitution	?	-	-	-	-	-
	c.4594G>Aa	p.(D1532N)	Substitution	Missense	0.722	0.77	0.80	28	1.00
	c.5603A>Ta	p.(N1868I)	Substitution	Missense	-	0.40	0.03	26	0.92
8	c.3056C>T	p.(T1019M)	Substitution	Missense	0.611	0.96	1.10	33	1.00
	c.3056C>T	p.(T1019M)	Substitution	Missense	0.611	0.96	1.10	33	1.00
9	c.161G>A	p.(C54Y)	Substitution	Missense	-	0.98	0.87	29	1.00
	c.4773+3A>G	p.(?)	Substitution	?	-	-	-	-	-
10	c.4139C>T	p.(P1380L)	Substitution	Missense	0.391	0.87	0.70	28	1.00
	c.4594G>Aa	p.(D1532N)	Substitution	Missense	0.722	0.77	0.80	28	1.00
	c.5603A>Ta	p.(N1868I)	Substitution	Missense	-	0.40	0.03	26	0.92
11	c.302+68C>Ta	p.(?)	Substitution	?	-	-	-	-	-
	c.4539+2028C>Ta	p.(?)	Substitution	?	-	-	-	-	-
	c.6148–698_ c.6670del - 4770 bp del	p.(?)	Deletion	?	-	-	-	-	-
12	c.302+68C>Ta	p.(?)	Substitution	?	-	-	-	-	-
	c.4539+2028C>Ta	p.(?)	Substitution	?	-	-	-	-	-
	c.6148–698_ c.6670del - 4770 bp del	p.(?)	Deletion	?	-	-	-	-	-

^aVariants in cis; predicted pathogenicity: M-CAP (>0.025); REVEL (>0.5), Eigen (>0.5), CADD13 (>20), DANN (>0.97).

DISCUSSION

THE CURRENT STUDY DESCRIBES AN END-STAGE PHENOTYPE in a genetically confirmed cohort (n = 12) of STGD1 patients with disease duration of >50 years. All patients presented with advanced disease characteristics such as optic disc pallor, retinal vasculature attenuation, and wide-spread pigment deposition. Poor visual acuity (20/400 to HM) in the cohort was attributable to widespread degeneration across the posterior pole with near-complete loss of the outer retinal layers, as evidenced by the visible absence of the hyperreflective RPE, IZ, EZ, and ELM bands on SD-OCT, and choroidal vasculature across the posterior pole. Consequently, the fundus in these patients exhibited a reflectively pale, blonde hue resulting from an unobstructed view of the underlying sclera. The absence of observable changes within these regions (eg, further pigment accumulation) over many years after initial examination reflects the complete loss of chorioretinal tissue and is indicative of the end of a longstanding degenerative process in the retina.

The finding of sclera-deep degeneration broadens the list of differential diagnoses to conditions that are characterized by aggressive chorioretinal deterioration and rod-cone dystrophies such as choroideremia (CHM, MIM #303100),^38,39 gyrate atrophy of the choroid and retina (OAT, MIM #258870),⁴⁰ Leber congenital amaurosis (RPE65, MIM #613794),⁴¹ and clinically advanced cases of C2orf71-related retinopathy.⁴² Ocular history and ffERG testing would most effectively differentiate STGD1 from these other listed conditions, as patients affected with the latter generally report progressive visual field constriction and nyctalopia (rod-cone attenuation on ffERG) as opposed to early central vision loss and a cone-rod dysfunction on ffERG. Anatomically, the current study cohort bears the most clinical resemblance to patients with end-stage choroideremia, who exhibit the similar pale fundus appearance from exposed scleral reflectance. The precise pathophysiology of choroideremia remains largely unknown despite promising advances in adeno-associated viral (AAV) gene therapy (NCT01461213).^43,44 The protein product of CHM, Rab escort protein-1 (REP1), has been localized exclusively to rods,⁴⁵ although numerous studies have presented histopathologic and adaptive optics–scanning light ophthalmoscopy data supporting RPE as the primary site of degeneration.^46–52 The outcome of partial or complete deterioration of the outer retina and underlying choroid owing to severely incapacitated RPE is therefore a conceivable eventuality given the physiological interdependence of adjacent layers in this part of the retina.^53–55 Likewise, such a pathway can be recapitulated by ABCA4 dysfunction in which the formation of bisretinoid fluorophores in phagocytized photoreceptor outer segments perpetuates the rapid demise of lipofuscin-laden RPE.^6,56–59

Consistent with other rod-cone degenerative conditions, patients with choroideremia experience early, progressive night blindness but retain visual acuity until relatively late in the disease.⁴⁹ A recent study of 56 consecutive patients at Oxford Eye Hospital found the median survival (Kaplan-Meier analysis) of 20/20 acuity, bilaterally, to be 39 years,⁶⁰ whereas the patients in the present STGD1 cohort all experienced a comparatively earlier age of central visual acuity loss (mean 14.1 years). This difference in ocular history can thus serve as a diagnostic aid in situations of phenotypic overlap between the 2 conditions. A further point of distinction between STGD1 and choroideremia is the frequency at which this phenotypic outcome presents to the clinic. While scleral visibility within areas of degeneration is routinely observed in choroideremia,^38,39 only 12 out of a database of 300 (~4%) STGD1 patients presented with this finding. It is unclear whether specific genotypes are an underlying factor behind this clinical outcome. Most ABCA4 variants in these patients are either deleterious or very severe. Each patient would be appropriately diagnosed as cone-rod dystrophy with a group 3 ffERG classification⁶¹; however, their overall disease trajectory is not as severe as in patients with the rapid-onset chorioretinopathy sub-phenotype who are exclusively biallelic null cases (ie, completely lack functional ABCA4 protein) and report symptomatic onset within the first decade of life. The most significant underlying factor could be the long duration of disease progression (>50 years) in these patients, who comprise a STGD1 demographic (>60 years of age) that is poorly characterized in the current literature.

The outcome of complete chorioretinal degeneration has serious implications for therapeutic approaches to end-stage STGD1. The pathophysiology of ABCA4 dysfunction will require replacement of both photoreceptors and RPE to restore function in the retina; however, the absence of a native choroid, as described in this study, presents an added obstacle involving tissue revascularization. Macular translocation of RPE-choroid grafts has been demonstrated in patients with age-related macular degeneration, with considerable success.^62–67 Such an approach, coupled with gene therapy, if the tissue source is autologous, may be a feasible strategy for STGD1 if the neurosensory retina can also be incorporated. Further applications of gene therapy include imparting photosensitivity to secondary neurons in the inner retina using optogenetic therapy.⁶⁸ Preliminary efficacy of delivering channelrhopsin-2 (ChR2), a light-gated cation channel isolated from the green alga Chlamydomonas reinhardtii, to patients with advanced retinitis pigmentosa is currently being evaluated (NCT02556736). Visual restoration can also be artificially achieved by the implantation of a bionic prosthesis, although as with the current potential of optogenetic therapy, the promise of high-resolution vision remains a work in progress. Nevertheless, long-term safety and improved visual function with the Argus II Retinal Prosthesis System was reported in a group of 30 subjects, which included a patient with choroideremia (NCT00407602).⁶⁹

In summary, an end-stage sub-phenotype of genetically confirmed STGD1 characterized by complete loss of the outer retina and choroid, resulting in widespread scleral exposure, is associated with long disease duration (>50 years) in older patients. This clinical stage exhibits significant phenotypic overlap with aggressive chorioretinal dystrophies such as choroideremia, but can be distinguished, in addition to genetic screening, by an ocular history of central vision loss and a cone-rod pattern of functional attenuation on ffERG.