A total of 244 candidate variants were detected by exome sequencing

At least 1 disease-causing ABCA4 variant was identified in 38 patients (90%), including 13 novel variants; ≥2 variants were identified in 34 patients (81%). Patients with childhood-onset STGD more frequently harbored 2 deleterious variants (18% vs 5%) compared with patients with adult-onset STGD.

Patients and Methods The protocol of the study adhered to the provisions of the Declaration of Helsinki. After informed consent was obtained, blood samples were taken and molecular analysis on the ABCA4 gene was performed as described by Maugeri et al. 14 The charts of patients with ABCA4 mutations who originally had received diagnoses of isolated or autosomal recessive CRD were reviewed. All patients originated from the University Medical Centre Nijmegen (Nijmegen, The Netherlands) and the University of Heidelberg (Heidelberg, Germany). In this study the diagnosis of CRD was based on the following criteria: initial symptoms of blurred central vision without a history of night blindness, impairment of color vision, and fundoscopic evidence of maculopathy without or with mild peripheral retinopathy. 3 4 5 7 8 In patients with recordable ERGs a cone–rod pattern of degeneration had to be present (i.e., the photopic b-wave impairment had to be greater than or equal to the scotopic b-wave amplitude impairment). Patients 9250 and 13163, who had nonrecordable ERGs, were included because their histories and clinical features were similar to those of other patients with cone–rod degeneration and they were believed to represent advanced cases of CRD. In addition to an ophthalmic examination, Goldmann kinetic perimetry routinely was performed using III-4-e and I-4-e isopters. Color vision was tested with the Ishihara and Panel D15 tests, except in patients 9369, 9378, and 10125, who were tested under conditions described earlier. 18 Because these patients were examined in two different clinics and ERGs were recorded over a long period, the methods, instrumentation, and analysis techniques of the electroretinography varied. The ERGs in patients 9369, 9378, 10125, and 11872 were performed as described by Thijssen et al. 19 The ERG method used in patients 9370, 9553, 9633, and 13163 was described by Alexandridis and Krastel. 20 The ERGs of the remaining patients (9250, 9371, and 9650) are of a more recent date and were performed according to International Society for Clinical Electrophysiology of Vision (ISCEV) standards. 21 Fundus photographs were taken in most patients and some of the patients (9650, 9369, 9378, and 10125) also underwent fluorescein angiography.  Results The characteristics of 12 patients with ABCA4-associated retinal dystrophy resembling CRD are summarized in Table 1 . Most did not have affected family members, and therefore their retinal dystrophies could not be classified as autosomal dominant, autosomal recessive or X-linked. Four patients reported a brother or sister with subnormal vision. In view of the reputedly normal visual acuity of the parents and the molecular defects, the inheritance pattern of the gene defects in these patients (individuals 9303, 9369, 9553, and 13163) was classified as autosomal recessive.  The visual acuity of the patients did not exceed 20/200 and, on average, was much lower. With the exception of patient 9553, the age of onset was at or before the age of 12, and in each of the patients, blurred vision was the initial symptom. Night blindness did not occur except in patients 9378 and 10125, in the final stages of retinal degeneration. Evidence of maculopathy in the form of bull’s eye maculopathy or pigmentary changes was present in all the patients reported in this study (Fig. 1A) . The functional equivalent of the mainly centrally located retinal disease was a central scotoma, varying from 8° to more than 40°. In all but one patient, the scotoma was absolute. Only in patient 9378 was the central scotoma relative and surrounded by absolute scotomas. Fundoscopic evidence of early peripheral involvement of the retina was mild, and only in the later stages of the disease did peripheral changes characteristic of RP, such as narrowing of retinal vessels and bone spicula, occur in patients 9369 (Fig. 1B) and 10125. Similarly, mild constriction of the visual fields occurred only in two patients (9650 and 10125) and only in the advanced state. Color vision was tested in 10 patients. Six demonstrated a red–green defect, and in two of these (patients 11872 and 10125), it was accompanied by a blue-yellow defect. In the remaining four patients, color vision was so severely disturbed that the exact type of impairment could not be assessed.  The ERG recordings demonstrated degeneration of both rods and cones. When ERG responses could be elicited, the cones appeared to be affected as much as the rod photoreceptors and, in most of the patients, even more severely. The ERG responses in five patients progressively deteriorated until no photopic and scotopic responses could be recorded. In these patients, with exception of patients 9250 and 13163, ERG recordings of an earlier date were used in Table 1 . This applies to patient 9369, in whom an ERG was recorded at age 12 (all ERG responses had been nondetectable since the age of 21), patient 9378 at age 33 (all ERG responses at age 46 were nondetectable), and patient 10125 at age 8 (in 1998, at age 28, the ERG responses were no longer detectable). Recent ERG findings were not available for patients 9650 and 9371. Their ERGs were recorded in 1989 and 1985, respectively. The remaining ERG data were derived from ERG recordings performed in the past 4 years. Of patient 9371 only the ERG data in the left eye were available.  Two patients warrant a more detailed description, due to the unusual course of their retinal dystrophies. Patient 9378, at the age of 12, had blurred vision with fundoscopic evidence of irregular chorioretinal atrophy in the posterior pole. At that time, there were no peripheral abnormalities on ophthalmoscopy, and there was no history of night blindness. The ERG demonstrated an equal reduction of both cone- and rod-mediated responses. Later in life, however, fundoscopic changes developed that were characteristic of RP, and the patient reported a decrease in night vision. With fluorescein angiography partly confluent patches of chorioretinal atrophy were visible (Fig. 1C) .  The clinical picture of patient 10125 differed from that of the other patients, despite the mutation in the ABCA4 gene. Initially, disease in this patient was diagnosed as STGD because of the bull’s eye maculopathy, the granular pigment alterations in the macular area, and the pisciform flecks surrounding the posterior pole. At age 8 his visual acuity had decreased to 20/200 in both eyes. When he was referred to our clinic in 1998 at the age of 28, peripheral degeneration in the form of narrow retinal vessels and deposition of peripheral bone spicula had developed, in addition to the earlier described disease of the central retina. A fluorescein angiogram showed typical findings: a central small hypofluorescent spot enclosed by an ellipsoid—a markedly hyperfluorescent area that in turn was surrounded by hyperfluorescent dots against a dark background, most likely caused by obscuration of choroidal background fluorescence (Fig. 1D) . Early ERG recordings were not available, and the ERG tracings recorded at age 28 represent the final stage of the degenerative process, with absence of both cone and rod responses. This retinal dystrophy seemed to have evolved from STGD into more widespread retinal degeneration, resulting in loss of function of both rods and cones.  Discussion Progressive CRD is a clinically heterogeneous retinal disorder, but typical findings include reduced visual acuity, impairment of the central visual field, color vision deficits, and fundoscopic evidence of maculopathy, with no or few midperipheral retinal pigment deposits. 3 4 7 8 There is some dispute about typical ERG findings in CRD. Some state that the diagnosis of CRD must be based on the reduction or absence of cone responses in the presence of quantitatively less reduction in rod responses, whereas others state that an equal impairment of both photoreceptor systems, if accompanied by the characteristic features, suffices to justify the diagnosis of CRD. 3 7 8 22 Several propositions have been made in the past to classify cone–rod disorders. Some classification systems have focused on individual case reports and were based on nosologic aspects; others have made a distinction according to the various patterns of inheritance. 3 6 23 24 In recent studies, Szlyk et al. 7 and Yagasaki et al. 8 made use of full-field ERGs, dark adaptometry, and modified perimetric techniques to identify functionally distinct subtypes of CRD. Finally, over the past few years, a molecular genetic classification of CRD has emerged.  At the moment, four genes and three loci have been implicated in autosomal dominant CRD, whereas one X-linked locus has been described. 25 26 27 28 29 30 31 32 Thus far, two loci and one gene (ABCA4) have been associated with autosomal recessive CRD. 12 33 34 The genetic heterogeneity seen in CRD is matched by the range of the clinical findings attributed by various investigators to this type of retinal dystrophy.  Whatever the classification system used, some patients display retinal disorders that cannot be classified satisfactorily. Often, these retinal degenerations involve overlapping features. Krill et al. 5 reported that 9 of 45 patients with cone degenerations showed typical features associated with fundus flavimaculatus. Heckenlively 2 described 76 patients with cone–rod patterns on the ERG in whom retinal disease otherwise met the standard definition of RP (progressive peripheral visual field loss with ring scotoma). Alternatively, as seen in patient 10125 in this study, patients with STGD have been described who had progressive peripheral retinal degeneration with severe abnormalities in the ERG and electro-oculogram (EOG) later in life—a condition that has been described by Fishman 4 as secondary progressive cone–rod dysfunction.  The association of CRD and a dark choroid has also been described previously. 35 36 The atypical pattern of retinal degeneration with confluent patches of chorioretinal atrophy in patient 9378 resembles that in another previously described unrelated patient with CRD-like disease caused by mutations in ABCA4. 37 In the molecular genetic study by Maugeri et al., 14 in which 11 of the 12 patients with autosomal recessive CRD described in this study were analyzed, ABCA4 mutations were found in 13 of 20 unrelated patients, strongly suggesting that ABCA4 mutations are the major cause of this disorder. If this is true, the genetic heterogeneity in autosomal recessive CRD, compared with, for example, classic RP, is surprisingly low. Because autosomal recessive inheritance is believed to be the most frequent mode of inheritance of monogenic chorioretinal disorders, it is very possible that a large fraction of the patients with CRD who have been clinically studied previously carry ABCA4 mutations. In that case, the explanation for the high variability of the clinical findings in autosomal recessive CRD would not be genetic heterogeneity but rather the genotype–phenotype model for ABCA4. According to this model, there is an inverse relationship between the presumed residual ABCA4 function as an N-retinylidene-PE flippase and the severity of the disorder. 12 37 38 As a consequence, a continuum of phenotypes is to be expected, ranging from STGD to CRD to RP. Although this is probably a simplified representation of reality and needs corroboration by detailed biochemical studies of individual mutations, as described previously, this model explains why mutations in the ABCA4 gene could give rise to phenotypes that do not satisfy the standard classification of retinal dystrophies. 39   Two patients in this study may reflect borderline CRD phenotypes. Patient 9553 carries a combination of a mild (2588G>C) and severe ABCA4 mutation, which, according to the genotype–phenotype model described earlier, should be associated with STGD. We have previously discussed that most likely, one of the pathologic mutations has not yet been identified in this patient. 14 However, the age of onset in this patient (25 years) is relatively high, and although other features such as visual acuity, perimetry, and ERG findings are typical of CRD, this may indicate a relatively mild subtype. Another more convincing example of blending of ABCA4-associated phenotypes is patient 10125. The molecular findings in this patient have not yet been described elsewhere. He carries a severe splice site mutation (IVS30+1G→T) in combination with a nucleotide change leading to a stop codon at Gln1029. A patient with RP who was homozygous for the IVS30+1G→T mutation has been described, 12 whereas the Q1029X mutation has not been described. Both mutations can be considered to be null alleles. According to the proposed ABCA4 model, the clinical phenotype in patient 10125 should be RP. Instead, this patient exhibits a typical retinal dystrophy, which gradually progresses from STGD to a more widespread degeneration of photoreceptors in a cone–rod pattern later in life. At present, both rod and cone ERG responses are not detectable, indicative of a final stage similar to that in many patients with RP. Functional studies are necessary to clarify whether these specific ABCA4 mutations are responsible for the particular progression of the retinal degeneration in this patient, or whether other as yet unknown modifying factors play a role.  In this study we have described 12 unrelated patients with retinal dystrophy resembling CRD caused by mutations in the ABCA4 gene. In a previous study we described the ophthalmic features in five siblings with CRD-like retinal dystrophy who were carrying ABCA4 mutations. 37 From the clinical data of these patients and previous molecular studies in patients with autosomal recessive CRD, two important conclusions can be drawn. 12 14 First, the genetic basis of autosomal recessive CRD is less heterogeneous than was thought, based on the variability in clinical features, because mutations in the ABCA4 gene seems to be the major pathologic cause. Second, given the wide clinical spectrum of CRD-like phenotypes associated with ABCA4 mutations, detailed clinical subclassifications are difficult and may not be very useful.   Supported by the British Retinitis Pigmentosa Society, the Rotterdamse Vereniging Blindenbelangen, the Algemene Nederlandse Vereniging ter Voorkoming van Blindheid, the Stichting Blindenhulp, the Stichting de Drie Lichten, the Gelderse Blindenvereniging and the Landelijke Stichting voor Blinden en Slechtzienden and the Stichting voor Ooglijders.  Submitted for publication June 15, 2001; revised December 21, 2001; accepted January 2, 2002.  Commercial relationships policy: N.  The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.  Corresponding author: B. Jeroen Klevering, Department of Ophthalmology, University Medical Centre Nijmegen, PO Box 9101, 6500 HB, Nijmegen, The Netherlands; b.klevering@ohk.azn.nl. Table 1. View Table Patients with Cone–Rod Degeneration and ABCA4 MutationsTable 1. Patients with Cone–Rod Degeneration and ABCA4 Mutations Patient Sex Current Age (ys) ABCA4 Mutations* Visual Acuity Age of Onset (ys) Fundoscopy Color Vision Perimetry ERG Cone (μV), † ERG Rod (μV), † OD OS OD OS OD OS 9250 M 30 1622T→C; 3113C→T 194G→A CF CF 6 Pigment clumping in the macula NP Large central scotoma over 40° ND, ‡ ND, ‡ 9303 M 21 1622T→C; 3113C→T 20/400 20/400 7 Granular pigmentary changes in the macula Diffusely disturbed Central scotoma of 20° Severely decreased, § Severely decreased, § 9369 F 40 6601-6602deIAG LP LP 8 Irregular hypopigmentation, mainly in the posterior pole. In later stages: attenuated vessels and bone spicula temporal to the macula Red-green defect Central scotoma varying from 10–30° 65 (65%) 80 (80%), ∥ 140 (90%) 160 (nl), ∥ 9370 M 15 1622T→C; 3113C→T 20/200 20/200 7 Granular aspect of the macula NP Concentric central scotoma of 8° 10 (13%) 9 (13%), ¶ 29 (29%) 29 (23%), ¶ 9371 M 38 1622T→C; 3113C→T 1622T→C;3113C→T 20/400 20/400 10 Bull’s eye maculopathy Red-green defect Concentric central scotoma of 20° NP 29 (16%), ‡ NP 54 (30%), ‡ 9378 F 50 768G→T CF CF 12 Bull’s eye maculopathy, narrow vessels in periphery with mild granular changes of the pigment epithelium and confluent patches of chorioretinal atrophy Severely disturbed Large, absolute, paracentral scotomas, relative scotoma centrally 20 (20%) 30 (30%), ∥ 70 (47%) 90 (60%), ∥ 9553 F 45 2588G→C IVS35del-2→+2del4 20/400 20/400 25 Bull’s eye maculopathy. Peripheral diffuse motting of RPE Severely disturbed Large central scotoma over 40° 14 (14%) 19 (19%), ¶ 41 (41%) 24 (24%), ¶ 9633 M 22 1622T→C; 3113C→T 4469G→A 20/400 20/200 12 Atrophy of retinal pigment epithelium in posterior pole. Early stages of bull’s eye maculopathy Red-green defect Central scotoma of 20° 12 (16%) 12 (16%), ¶ 61 (62%) 39 (39%), ¶ 9650 F 20 3364G→A 20/400 20/400 5 Central granular aspect Red-green defect Large central scotoma of 30° and relative constriction of III-4 70 (39%) 106 (59%), ‡ 272 (nl) 115 (76%), ‡ 10125 M 30 IVS30+1G→T 3085C→T 20/200 20/200 8 Central hypopigmentation with dark surrounding, resembling bull’s eye. Later in life: peripheral changes characteristic of RP Severe red-green defect; mild blue-yellow defect Central scotoma of 10–15° with mild peripheral restriction 75 (75%) 80 (80%), ∥ 130 (87%) 140 (93%), ∥ 11872 M 30 634C→T 20/200 20/200 10 Bull’s eye pattern Severely disturbed; blue-yellow more than red-green Central scotoma of 25° 23 (23%) 35 (35%), ∥ 110 (73%) 95 (63%), ∥ 13163 M 15 1622T→C;3113C→T IVS36+1G→A 20/400 20/200 6 Granular aspect of retinal pigment epithelium in macula. Slightly pale optic disc Severely disturbed Central scotoma of 10–15,° no peripheral involvement ND, ¶ ND, ¶  CF, count fingers; LP, light perception; ND, not detectable; NP, not performed. *  Allele 1, first line; allele 2, second line. †  Between parentheses: percentage of the ERG value compared to the lower limit of the normality; normal ERG values are indicated nl. ‡  Minimal values for ERG recordings: 150 μV for the photopic ERG, 180 μV for the scotopic ERG. §  ERG performed with skin electrodes. ∥  Minimal values for ERG recordings: 100 μV for the photopic ERG, 150 μV for the scotopic ERG. ¶  Minimal values for ERG recordings: 99 μV for the photopic ERG, 75 μV for the scotopic ERG. Figure 1. View OriginalDownload Slide (A–D) Fundus photographs and fluorescein angiograms in eyes of patients with (atypical) CRD. (A) Patient 11872 with bull’s eye maculopathy. (B) Patient 9369, demonstrating CRD in the later stages with attenuation of the retinal arterioles and irregular pigmentation temporal to the macula. (C) Fluorescein angiograms in patient 9378 showing confluent patches of chorioretinal atrophy. (D) Patient 10125 with central hypofluorescence enclosed by an ellipsoid hyperfluorescent area. In the surrounding area, hyperfluorescent flecks are visible, and the choroidal background fluorescence seems blocked, as seen in STGD.

Subjects All subjects provided written informed consent for this research study, which was approved by the University of Iowa Human Subjects Committee. The study included 176 patients with SD, 457 patients with RP, 60 patients with CRD and 272 normal control subjects. Seventeen members of this cohort (13 with SD, 1 with CRD and 3 with RP) had disease-causing mutations identified on one or both alleles in previous studies that employed single-strand conformational polymorphism analysis as the primary screening method (3). All patients and control subjects were ascertained in the outpatient ophthalmology clinic at the University of Iowa. All patients received a complete eye examination including measurement of Snellen visual acuity, slit lamp biomicroscopy of the anterior segment and fundus, and binocular indirect ophthalmoscopy. Most patients had fundus photography and Goldmann perimetry performed as well.Molecular characterization of the ABCA4 gene A multi-platform screening approach was used to genotype the ABCA4 gene in all 693 research subjects and 272 controls. Thirty-two of the most common disease-causing ABCA4 variations were selected for the initial screen. The entire research cohort was assayed for these 32 variations using a combination of SNPlex, SSCP analysis, TaqMan and automated DNA sequencing (3). The SNPlex and TaqMan assays were performed as previously described (22,23). Fourteen of the variations were compatible with Applied Biosystems SNPlex allele-specific assay platform. Nine variations were screened by SSCP analysis, three were screened using an Applied Biosystems' TaqMan assay and six were screened by automated DNA sequencing. All CRD and RP patients who had one plausible disease-causing allele identified in the first tier of screening were assayed by automated sequencing of the entire coding region of the ABCA4 gene in an effort to identify their second disease-causing allele. In addition, Stargardt patients who had one of the three most common alleles (Gly863Ala, Gly1961Glu or IVS38-10) were also sequenced through the entire coding region of the ABCA4 gene.Visual acuity The best-corrected visual acuity was recorded in each patient's medical record as a Snellen fraction (normal = 20/20). Before statistical analysis, these values were converted to the logarithm of the minimal angle of resolution (logMAR) by calculating the base 10 logarithm of the Snellen fraction [e.g. normal = log(20/20) = 0]. For patients with multiple hospital visits, the acuity measurements taken closest to the patient's 27th birthday were used for the statistical analysis. Values from the left eye and right eye were averaged.Visual field volume scores Goldmann visual fields were scanned with a Sharp scanner and analyzed with ImageJ software (available at http://rsbweb.nih.gov/ij/) as follows: transparent layers were added to each field, and the isopters of the visual fields were manually traced onto these layers. Each isopter was assigned a z-axis value according to relative luminous energy of the stimulus (I2e = 100, I3e = 31.7, I4e = 10, III4e = 0.49, V4e = 0.024, no detection = 0). The volume scores were calculated by multiplying the area of each isopter by its associated z-axis value and then summing the values for all isopters. For patients with multiple hospital visits, the visual field measurements taken closest to the patient's 27th birthday were used for the statistical analysis. Values from the left eye and right eye were averaged.Statistical analyses The frequencies of disease-causing alleles observed among the three groups of retinal disease patients were compared with the frequency in controls using Fisher's exact test. The phenotype of each patient was assumed to result from the additive effect of two alleles and one or more additional factors that were cumulatively represented by a single residual value for each subject. The values for the quantitative contribution of each allele were estimated using a multiple linear regression analysis (24). Specifically, to dissect the effect of each allele, we decomposed the mean phenotype of a person whose genotype consists of allele i and allele j into ai + aj, where ai is the effect of the i-th allele. To do this, we created a vector Yva of length 51 containing the average logMAR visual acuities for each subject, and a 51x16 matrix, X. Each cell of this matrix, Xij, indicated the number of occurrences of the j-th allele for each subject i. This system of 51 equations was solved with multiple linear regression to give estimates of a, using the statistics program R (available at http://www.R-project.org). To verify the allelic data, each row of X summed to 2, and each column summed to the corresponding number of occurrences of each allele for this cohort. These calculations were performed in the same manner for Yvol, the vector of containing each subject's visual field quantitative phenotype. The estimates of a, the coefficients for each allele, for both Yva and Yvol are given in Table 1.

Eighty-four unrelated STGD1 patients were found to carry two or more disease-causing ABCA4 mutations, and cosegregation analyses were performed in 54 (64.3%). (See Supplementary Table S3 for a summary of clinical phenotype of STGD1 patients). The mean disease onset age of the patients was 13.1 years (range, 2–44 years), and half of these patients experienced their symptoms of visual impairment in their first decade. We classified the patients into three groups by their disease onset age. For the patients in the first group, whose disease onset ages were between 1 and 10 years, the fraction (45.3%, 19/42) of patients carrying compound heterozygous or homozygous deleterious mutations (nonsense, frameshift insertion or deletion, or splicing mutations) or complex alleles was much higher than those observed for the patients in the group 2 (24.1%, 7/29) and group 3 (15.4%, 2/13), whose disease onset ages were from 11 to 20 years or older than 20 years, respectively. In contrast, the percentage of patients carrying compound heterozygous or homozygous missense mutations was higher in group 3 than in group 1 or group 2 (Table 2). The most common mutation (p.Y808X) was detected in 12 patients, and all were heterozygous compound with missense (6 patients), splicing (1 patient), and insertion or deletion (5 patients) mutations. The two most frequent missense mutations (p.F2188S and p.N965S) were identified as heterozygous. In 84 patients, 9 patients carried complex mutations, and 4 of those 9 patients carried a common complex allele p.E328V/p.E1036K. All these patients had early onset age and relatively severe visual acuity defects. One patient (010222) also carried three heterozygous mutations (p.E328, p.E1036K, and p.R1843W); however, he did not harbor the common complex allele p.E328V/p.E1036K, as only p.E1036K was detected in his son in the subsequent cosegregation analysis. Compared to the four patients carrying the common complex alleles (p.E328V/p.E1036K), patient 010221 had a late onset age (44 years old).  Table 2 View Table Correlations Between Onset Age of STGD Patients and Their Carrying Mutations

9369a AR 6601–6602delAG Frameshift

Supplementary File pnas.1909378117.sd01.xlsx (5.4MB, xlsx)

Supplementary tables (1.1MB, pdf)

024 m Caucasian STGD ABCA4 NM_000350.2 c.[6601_6602delAG];[=], c.[4253+43G>A];[=] p.Arg2201fs* p.Ile1377Hisfs*3 Not found 0.004694 AR Pathogenic Pathogenic 2 [36] [19]

Supplementary TableS10

Supplementary TableS10