Clinical and Molecular Characteristics of Childhood-Onset Stargardt Disease

1Stargardt macular dystrophy (STGD) is the most common form of juvenile-onset macular degeneration; it is inherited as an autosomal-recessive trait and caused by mutations in the ABCA4 gene.– Most cases present with central visual loss in early teenage years and ophthalmoscopy classically reveals macular atrophy with yellowish-white flecks at the posterior pole at the level of the retinal pigment epithelium (RPE).

2A large number of studies have described wide phenotypic variability and variable severity in ABCA4-associated retinopathy. The various phenotypes encompass macular atrophy without flecks, bull's-eye maculopathy, fundus flavimaculatus (retinal flecks without macular atrophy), a foveal sparing phenotype, cone-rod dystrophy, and “retinitis pigmentosa.”– There is also considerable allelic heterogeneity, with >700 variants in ABCA4 having been reported to date.,,–

3Patients with childhood-onset STGD tend to develop early severe visual acuity (VA) loss, markedly compromised retinal function on electroretinography with generalized rod and cone system dysfunction, and rapid enlargement of RPE atrophy and progressive loss of retinal function.,,,, Patients with adult-onset disease are more likely to retain useful VA for longer and show milder retinal dysfunction at diagnosis.,,,, There have been no previous studies specifically describing the clinical findings in a large cohort of molecularly confirmed STGD patients presenting and examined in childhood; the majority of previous reports relate to clinical features of patients examined in adulthood, some of whom may have had childhood-onset disease.

4The purpose of this study was to describe the detailed clinical and molecular genetic findings of a large cohort of patients from a single center with childhood-onset STGD examined before 17 years of age.

5Forty-two patients diagnosed with STGD at <17 years of age, between January 2001 and January 2012, were ascertained from the pediatric inherited retinal disease clinics at Moorfields Eye Hospital. Two subjects have been described in a previous case report. Blood samples were collected and genomic DNA extracted from peripheral blood leukocytes after obtaining informed consent. The protocol of the study adhered to the provisions of the Declaration of Helsinki and was approved by the local Ethics Committee of Moorfields Eye Hospital.

6A detailed medical history was obtained and a full ophthalmologic examination performed. The age of onset was defined as either the age at which visual loss was first noted by the patient or, in the “asymptomatic” patients, when an abnormal retinal appearance was first detected. The duration of disease was calculated as the difference between age at onset and age at most recent examination in childhood. The follow-up data were obtained before the age of 17 years.

7Clinical evaluation included best-corrected VA, dilated ophthalmoscopy, color fundus photography, fundus autofluorescence imaging (AF), spectral-domain optical coherence tomography (SD-OCT), and electrophysiologic assessment. Best-corrected Snellen VA was converted to equivalent logarithm of the minimum angle of resolution (logMAR) VA. Follow-up data of logMAR VA, color fundus photography, and AF imaging were compared with those at baseline.

8Color fundus photography was performed with a TRC-50IA Retinal Fundus Camera (Topcon, Tokyo, Japan). Patients were divided into 1 of 6 fundus appearance groups based on the presence and location of central (macular) RPE atrophy and yellowish-white flecks (Table 1).

9Autofluorescence images before 2009 were obtained with an HRA 2 (Heidelberg Engineering, Heidelberg, Germany; excitation light, 488 nm, barrier filter, 500 nm; field of view, 30×30°); imaging after 2009 was undertaken using the Spectralis with viewing module version 5.1.2.0 (Heidelberg Engineering; excitation light, 488 nm; barrier filter, 500 nm; fields of view, 30×30° and 55×55°). Patients were classified into 1 of 3 AF patterns, as previously described (Table 1).,

10Spectral domain OCT was undertaken with the Spectralis with viewing module version 5.1.2.0. The HEYEX software interface (version 1.6.2.0; Heidelberg Engineering) was used for retinal thickness measurements., Central foveal thickness was defined as the distance between the inner retinal surface and the inner border of the RPE.

11Electrophysiologic assessment included full-field electroretinogram (ERG), and pattern ERG, recorded with gold foil electrodes. Protocols incorporated the recommendations of the International Society for Clinical Electrophysiology of Vision., Full-field ERGs were used to assess generalized rod and cone system function and included (i) dark-adapted dim flash 0.01 cd·s·m−2 (DA 0.01), (ii) dark-adapted bright flash 11.0 cd·s·m−2 (DA 11.0), (iii) light-adapted 3.0 cd·s·m−2 30 Hz flicker ERG (LA 3.0 30 Hz), and (iv) light-adapted 3.0 cd·s·m−2 at 2 Hz (LA 3.0). The pattern ERG P50 component was used to assess macular function. All the components of the ERG and the pattern ERG P50 component were examined to classify patients into 1 of the 3 previously described electrophysiologic groups (Table 1).,

12Blood samples were collected in EDTA tubes and DNA was extracted with a Nucleon Genomic DNA extraction kit (BACC2; Tepnel Life Sciences, West Lothian, UK). All 50 exons and exon–intron boundaries of the ABCA4 gene were amplified using Illumina Truseq Custom Amplicon protocol (Illumina, San Diego, CA), followed by sequencing on Illumina MiSeq platform., The next-generation sequencing reads were analyzed and compared with the reference genome GRCh37/hg19, using the variant discovery software NextGENe (SoftGenetics LLC, State College, PA). All detected possibly disease-associated variants were confirmed by Sanger sequencing.,

13All the missense variants identified were analyzed using 2 software prediction programs: SIFT (Sorting Intolerant from Tolerant; available from www.sift.jcvi.org/; accessed November 1, 2013), and PolyPhen2 (available from www.genetics.bwh.harvard.edu/pph/index.html; accessed November 1, 2013). Predicted effects on splicing of all the missense and intronic variants were assessed with the Human Splicing finder program version 2.4.1 (available from www.umd.be/HSF/; accessed November 1, 2013). The allele frequency of all variants was estimated by reference to the Exome Variant Server (NHLBI Exome Sequencing Project, Seattle, WA; available from www.snp.gs.washington.edu/EVS/; accessed November 1, 2013).

14Patients harboring ≥2 mutations were classified into 3 genotype groups based on mutation type: Group A included patients with ≥2 definitely or likely deleterious (severe) variants; group B included patients with 1 deleterious variant and ≥1 missense or in-frame insertion/deletion variants; and group C included individuals with ≥2 missense or in-frame insertion/deletion variants (Table 1). One disease-associated intronic change of unknown effect was treated as a deleterious allele owing to the associated severe clinical phenotype previously reported., It should be noted, however, that assigning severity (e.g., a deleterious effect) to a mutation was not always straightforward, especially for missense alleles and some variants in splice sites.

15To investigate differences between the patients with childhood-onset STGD and those with adult-onset STGD, clinical and molecular genetic data of patients with adult-onset STGD ascertained at Moorfields Eye Hospital were reviewed. The comparison group consisted of all patients who had adult-onset STGD (older than 17 years), and who had ≥2 disease-causing ABCA4 variants.

16Statistical analysis was performed using commercially available software: Excel Tokei 2010 (Social Survey Research Information Co., Ltd., Tokyo, Japan). The eye used for analysis was selected according to the Random Integer Generator (available from www.random.org/). The Mann–Whitney U test was applied to investigate the differences between the 2 groups (childhood-onset STGD vs adult-onset STGD) in terms of logMAR VA, and central foveal thickness. The chi square statistic was applied to investigate the association between selected categorical variables of childhood-onset and adult-onset disease, including fundus appearance, flecks (macular, peripheral, and no flecks), presence of pigmentation, AF pattern, electrophysiologic group, and genotype group. P <0.05 was considered to indicate statistical significance.

17Forty-two unrelated patients with childhood-onset STGD were ascertained; the clinical findings are summarized in Table 2 (available at www.aaojournal.org). There were 22 female and 20 male patients. Eight (19%) were from consanguineous families. The median age of onset was 8.5 years (range, 3-16), and the median age at baseline examination was 12.0 years (range, 7–16). The median logMAR VA at baseline in all 42 patients was 0.74 in the right eye and 0.74 in the left eye (range, 0.10–1.30 and 0.12–1.40, respectively). The mean duration of disease at baseline was 2.0 years (range, 0–9). Follow-up data were available for logMAR VA, fundus photography, and AF imaging, in 24, 14, and 11 patients, respectively. The detailed changes in these parameters during follow-up are presented in Table 3 (available at www.aaojournal.org).

18The median logMAR VA at baseline in the 24 patients that were monitored was 0.75 in the right eye and 0.75 in the left eye (range, 0.10–1.30 and 0.12–1.30, respectively); the median logMAR VA at follow-up was 1.00 in the right and 1.00 in the left eye (range, 0.05–1.40 and 0.20–1.60, respectively) at a median age of 15.0 years (range, 12–16). Fifteen patients (15/42; 36%) had logMAR or ≤1.0 VA in the better eye at baseline. Thirteen of 24 patients (54%) with available follow-up data had logMAR VA of ≤1.0 in the better eye at follow-up (range, 11–16). Follow-up data were available in 14 of 27 patients with VA better than logMAR 1.0 in the better eye at baseline; 6 (43%) had logMAR of ≤1.0 VA in the better eye at follow-up (range, 13–15).

19Color fundus photographs, AF images, and SD-OCT images of 5 representative cases are shown in Figure 1. Baseline color fundus photographs were obtained in 39 patients (Table 2). Among the 39 patients, there was 1 (2.5%) with a grade 1 fundus appearance at baseline, 1 (2.5%) with grade 2, 11 (28%) with grade 3a, and 26 (67%) with grade 3b. There were no patients with a grade 3c or grade 4 fundus appearance. Central atrophy was present in 37 of the 39 patients (95%) at baseline; flecks were detected at the macula in 4 of the 39 patients (10%) and in the periphery in 23 (59%), with no visible flecks in 12 individuals (31%; Table 2). Retinal pigmentation was present in 2 of the 39 patients (5%; patients 24 and 34).

20Serial color fundus photographs were available in 14 patients (Table 3), 3 of whom showed a fundus grade transition. Macular flecks, which were not present at baseline, developed in 2 subjects (patients 7 and 13) and macular and peripheral flecks became visible in 1 individual (patient 26). Color fundus photographs and AF images of 4 representative cases who developed flecks during the follow-up interval are shown in Figure 2.

21Patients 17 and 18 had fine dots at the central macula surrounded by numerous peripheral flecks, classified into fundus grade 3b (patient 17; Fig 3). Clinical and molecular genetic data of these 2 patients have been previously described. Only 1 patient had asymmetric fundus findings, with a central atrophic-appearing lesion with peripheral flecks extending anterior to the vascular arcades in the right eye, and macular atrophy with flecks, subretinal fibrosis, and hyperpigmentation at the level of RPE in the left eye (patient 29; Fig 3).

22We obtained AF images for 32 patients at baseline (Table 2). There were 10 of the 32 patients (31%) with type 1 AF pattern, 22 (69%) with type 2 AF, and no subjects with type 3 AF. Serial AF images were obtained in 11 patients during the follow-up interval (Table 3); no patient demonstrated an AF grouping transition.

23We obtained SD-OCT images for 21 patients at baseline (Table 2). Outer retinal disruption at the fovea was present in all 21 patients. The median central foveal thickness of the right and left eyes was 60.0 and 55.0 μm, respectively (range, 33–138 and 35–140, respectively). Eighteen of the 21 patients (86%) had severe foveal thinning in both eyes (<100 μm).

24Electrophysiologic assessment was performed in 25 patients at baseline (Table 2). Nine of the 25 patients (36%) were in ERG group 1 (isolated macular dysfunction),1 (4%) was in ERG group 2,and 15 (60%) were in ERG group 3 (generalized cone and rod dysfunction).

25Detailed molecular genetic results including in silico analysis to assist in the prediction of pathogenicity of the variants are shown in Table 4 (available at www.aaojournal.org). Forty-six ABCA4 variants were identified: 27 missense, 7 splice-site alterations, 7 nonsense, 3 frameshifts, 1 in-frame duplication, and 1 definitely disease-associated intronic variant for which the exact pathogenic mechanism is not known. Thirteen novel definitely or highly likely disease-causing variants were identified: p.Gln8fs, p.Cys519*, p.Asp586Gly, p.Arg587Lys, p.Glu905fs, p.Tyr1027*, p.Met1066-Arg, p.Arg1097*, p.Thr1721fs, p.Tyr1770Asp, p.Ala1739dup, p.Ser2072Asn, and c.6817-2A>C (Table 4). Four homozygous variants (p.Glu905fs, p.Glu1022Lys, p.Tyr1027*, and c.64719+1G>A) were identified in patients from consanguineous families and the other 42 variants were detected in heterozygous state. Four of 8 patients from consanguineous families had homozygous variants (patients 3, 5, 6, and 28), 2 had compound heterozygous variants (patients 2 and 17), and 2 had no variants identified (patients 1 and 4).

26At least 1 disease-causing ABCA4 variant was detected in 38 of the 42 patients (90%); of these, ≥2 variants were identified in 34 (81%) and 1 variant in 4 (9.5%; Tables 2 and 4). Only 4 of the 42 individuals (9.5%) had no variants identified. The 34 patients harboring ≥2 disease-causing variants were classified based on the number and mutation type (with suggested severity) into 3 genotype subgroups: 7 patients (21%) in genotype group A, 15 (44%) in group B, and 12 (35%) in group C (Table 2).

27Sixty-four patients with adult-onset STGD harboring ≥2 disease-causing ABCA4 variants were reviewed. The clinical and molecular genetic data were compared between 34 patients with childhood-onset STGD harboring ≥2 disease-causing ABCA4 variants and the aforementioned 64 patients with adult-onset STGD (Table 5, available at www.aaojournal.org; Fig 4).

28There were significant differences in terms of fundus appearance classification (chi-square = 23.2; P = 0.001), presence of pigmentation (chi-square = 14.9; P = 0.000), genotype group classification (chi-square = 7.3; P = 0.003), and central foveal thickness in the selected eye (P = 0.012; Table 5, available at www.aaojournal.org; Figs 4 and 5, available at www.aaojournal.org); with childhood-onset STGD being associated with less retinal pigmentation, a greater proportion of patients harboring deleterious alleles, and a thinner central fovea. No differences were identified in terms of location of flecks (chi-square = 4.0; P = 0.136), AF pattern (chi-square = 5.6; P = 0.061), electrophysiologic group (chi-square = 3.8; P = 0.148), or logMAR VA in the selected eye (P = 0.781). However, a greater proportion of patients with childhood-onset STGD were in ERG group 3 (10/18; 56%) compared with adult-onset STGD (18/59 [31%]; Table 5; Fig 4), but the difference, although showing a strong trend, did not attain significance (Fig 4).

29This manuscript reports a series of childhood-onset patients with molecularly confirmed STGD, and compares the genetic, clinical, and electrophysiologic data with those in an adult-onset group.

30The classical phenotype of STGD is characterized by the presence of yellowish-white fundus flecks and macular atrophy, but the fundus appearance can be variable.,,, Fishman described 4 groups based on fundus appearance and electrophysiologic findings; the author did not distinguish between childhood-onset and adult-onset disease. In addition, the classification did not fully encompass the range of phenotypes present in childhood-onset disease and thus was modified for the present study (Table 1). Most children in this study had the classical fundus appearance of STGD with macular atrophy and macular and/or peripheral flecks, but one third of children had no visible flecks at presentation. Subsequent development of flecks was observed during the follow-up interval in 3 of these 12 patients (Fig 2). Similar development of macular/peripheral flecks over time have also been described in a young adult patient with STGD.

31There were no children with paracentral atrophy without central atrophy (observed in the foveal sparing phenotype, a milder phenotype seen in a minority of patients with STGD).,,, This observation is in keeping with previous reports that patients with a foveal sparing phenotype typically present in later adult life., The subset with a foveal-sparing phenotype show relatively preserved foveal structure, which results in a relatively wide CFT range in the adult-onset STGD group (Fig 5).

32Marked disruption of foveal outer retinal structure was present on SD-OCT in all children imaged, indicating that changes in foveal structure occur early in the disease process. Visual loss may precede ophthalmoscopic abnormalities in childhood-onset STGD and this may lead to nonorganic visual loss being considered. In such cases, SD-OCT imaging and/or electrophysiologic assessment will avoid misdiagnosis. The early foveal involvement in STGD without flecks, or other AF imaging evidence of increased levels of lipofuscin in the RPE, lend support to the hypothesis that A2E, which is elevated in STGD, may be directly toxic to cone photoreceptors.,

33Of the 24 patients, 9 (36%) were in ERG group 1, 1 (4%) in ERG group 2, and 15 (60%) in ERG group 3. A greater proportion of patients were in group 3 compared with the cohort with adult-onset disease, indicating that childhood-onset STGD is more likely to be associated with generalized retinal dysfunction. This is further evidence for childhood-onset STGD having a more severe retinal phenotype.,,

34Twenty-two patients (58%) had ≥1 deleterious variant and 7 subjects (18%) had 2 deleterious variants, which was significantly higher than observed in the adult-onset cohort (45% and 5%, respectively). The 5 patients (71%) with available ERGs in genotype group A (harboring 2 deleterious variants) all had generalized rod and cone system dysfunction (ERG group 3). These findings when taken together suggest that patients harboring deleterious ABCA4 variants are more likely to have an earlier presentation (childhood) and a more severe functional phenotype.

35There are potential limitations of this study, including the definition of age of onset and choosing to classify childhood-onset as before the age of 17. The age of onset was defined as either the age at first symptom or the age when a retinal abnormality was first detected in “asymptomatic” patients. These 2 groups (symptomatic and asymptomatic) may have different clinical characteristics, including the symptomatic patients would be expected to have foveal involvement and thereby reduced VA. However, the vast majority of children were symptomatic in our cohort. It is also possible that dividing patients by age 17 may potentially introduce a selection bias.

36This study specifically addresses, for the first time, the clinical features and molecular genetic findings of childhood-onset STGD in a substantial group of patients. Childhood-onset disease is associated with more severe VA loss from the early stages of disease. The classical flecks are not always present at diagnosis, but can appear later in the course of disease. Generalized cone and rod system dysfunction is more common than in adult-onset disease, in keeping with a more severe phenotype. Two or more disease-causing variants were detected in >80% of children and a higher proportion of definitely or possibly deleterious variants were demonstrated compared with adult-onset STGD, which is likely to underlie the earlier onset and more severe phenotype in childhood. The rapid deterioration of function in childhood-onset disease suggests that the investigation of novel therapies in this age group is more likely to lead to timely recognition of any treatment effect compared with adults with more slowly progressive disease.