To describe the clinical and molecular findings in ten unrelated African American patients with Stargardt disease.
Retrospective, observational case series.
We reviewed the clinical histories, examinations, and genotypes of 85 patients with molecular diagnoses of Stargardt disease. Three ABCA4 sequence variations identified exclusively in African Americans were evaluated in 300 African American controls and by in silico analysis.
ABCA4 sequence changes were identified in 85 patients from 80 families, of which 11 patients identified themselves as African American. Of these 11 patients, 10 unrelated patients shared 1 of 3 ABCA4 sequence variations: c.3602T>G (p.L1201R); c.3899G>A (p.R1300Q); or c.6320G>A (p.R2107H). The minor allele frequencies in the African American control population for each variation were 7.5%, 6.3%, and 2%, respectively. This is comparable to the allele frequency in African Americans in the Exome Variant Server. In contrast, the allele frequency of all three of these variations was less than or equal to 0.05% in European Americans. Although both c.3602T>G and c.3899G>A have been reported as likely disease-causing variations, one of our control patients was homozygous for each variant, suggesting that these are nonpathogenic. In contrast, the absence of c.6320G>A in the control population in the homozygous state, combined with the results of bioinformatics analysis, support its pathogenicity.
Three ABCA4 sequence variations were identified exclusively in 10 unrelated African American patients: p.L1201R and p.R1300Q likely represent nonpathogenic sequence variants, whereas the p.R2107H substitution appears to be pathogenic. Characterization of population-specific disease alleles may have important implications for the development of genetic screening algorithms.
Stargardt disease (OMIM# 248200) represents the most common autosomal recessive macular dystrophy, affecting approximately 1 in 1,000 to 10,000 individuals, depending on the population studied. The majority of patients present in the first to the third decade of life, although some forms of the disease may go undiagnosed until considerably later in life. Retinal findings vary widely and may include “beaten bronze” retinal pigment epithelium (RPE) granularity, a “bull’s eye” macular appearance, subretinal “pisciform” flecks, or pigmentary retinopathy. In 60% to 80% of patients, accumulation of lipofuscin in RPE cells obscures the normal early choroidal hyperfluorescence on fluorescein angiography, resulting in the so-called dark or silent choroid sign. Visual prognosis is correlated with best-corrected visual acuity (BCVA) at the age of onset of symptoms, presence of foveal sparing, and genotype.
In 1997, Allikmets and associates reported that Stargardt disease results from mutations in the ABCA4 (ATP-binding cassette transporter retina-specific) gene located on the short arm of chromosome 1 (1p21-p13). ABCA4 encodes an ATP-binding cassette transporter that acts as a flippase for retinylidene-phosphatidylethanolamine to be transported from the photoreceptor outer segment disk lumen to the cytoplasm, where it undergoes reduction to all- trans retinol. Dysfunction of the flippase leads to accumulation of RPE toxic retinoid derivatives and to subsequent loss of RPE and photoreceptors. Vast phenotypic heterogeneity has been reported in association with more than 700 distinct mutations in ABCA4 . ABCA4 mutations have been described not only as causative of a phenotype of classic Stargardt disease, but also of cone-rod dystrophies, retinitis pigmentosa, and age-related macular degeneration (AMD).
Molecular screening of patients with clinical features of Stargardt disease has been performed using single-strand conformation polymorphism, heteroduplex analysis, denaturing gradient gel electrophoresis, and microarray testing, with mutation detection rates of 40% to 65% as compared to a detection rate of approximately 80% or higher with next-generation sequencing and Sanger full-sequence analysis of all 50 exons of the ABCA4 gene. More recently, population studies have used microarray technology based on arrayed primer extension, mainly the ABCR400 chip that includes 400 of the disease-associated mutations. Detection rates using the ABCR400 chip depend upon the ethnic and clinical composition of the patient cohort analyzed. A population-specific microarray screening chip has been developed to evaluate for ABCA4 mutations specific to the Italian population. Following the discovery of seven mutations occurring in high frequency in the Afrikaner population in South Africa, a multiplex assay was developed for their initial screening. Such population-directed screening algorithms offer the potential for more efficient and cost-effective screening. Thus, efforts to determine mutations that occur with high frequency within specific patient populations are important endeavors.
In this study, we report three sequence variations unique to 10 unrelated African-American patients in a cohort of 85 patients with clinical and molecular diagnoses of Stargardt disease. Two of these sequence variations (c.3602T>G, c.3899G>A) were detected in a local African-American control population in the homozygous state and thus probably represent ethnic-specific variants. However, the third (c.6320G>A) is likely pathogenic and results in protein dysfunction.
Patients and Clinical Characteristics
This observational case series study was approved by the institutional review boards of the Cleveland Clinic and the Louis Stokes Cleveland Veterans Affairs Medical Center. Written informed consent was obtained from the patients and was in accordance with the regulations of the Health Insurance Portability and Accountability Act. We reviewed the medical records of 85 patients from 80 families with clinical and molecular diagnoses of Stargardt disease. One of the authors (E.I.T.) established the clinical diagnosis at the time of presentation, based on clinical examination and ancillary testing. The presence of at least one previously reported disease-associated sequence variation in the ABCA4 gene on subsequent genetic testing established a probable molecular diagnosis. From the medical record, we extracted the following: age at presentation, gender, race, BCVA, and retinal findings (graded as described by Fishman ). Fluorescein angiography, color testing, electroretinography, fundus autofluorescence, and optical coherence tomography were reviewed when they were available.
Blood was collected by venipuncture, and molecular DNA testing was performed at one of several research and commercial laboratories, depending on availability of testing at the time the patients were evaluated. Of the 85 patients in whom at least one disease-associated sequence variation was identified in the ABCA4 gene, 39 patients were evaluated using AsperChip version 1 (Asper Biotech, Tartu, Estonia),which tested 501 mutations/short nucleotide polymorphisms; 9 were evaluated using the ABCR400 microarray chip (Columbia University via the eyeGENE Research Project at the National Eye Institute, Bethesda, Maryland, USA); 10 were evaluated via targeted mutation analysis by the Carver Laboratory for Molecular Diagnosis at the University of Iowa Center for Macular Degeneration (Iowa City, Iowa, USA) (3 with 20 ABCA4 variants and 1 ELOVL4 variant; 1 with 68 ABCA4 variants and 1 ELOVL4 variant; 6 with 81 ABCA4 variants and 1 ELOVL4 variant); 3 with a combination of targeted mutation analysis and full sequencing (University of Michigan Ophthalmic Molecular Diagnostic Laboratory via the eyeGENE Research Project); and more recently, in 24 patients, by full sequencing of all 50 exons in the ABCA4 gene (GeneDx, Gaithersburg, Maryland, USA via the eyeGENE Research Project). With regard to the African American patients, 7 were screened using the AsperChip as above, 1 via the ABCR400 chip (Columbia University via eyeGENE), and 2 via full sequencing of all 50 exons in the ABCA4 gene (GeneDx via eyeGENE). The three sequence variations specific to the African American subset of patients were evaluated in 300 regional and race-matched controls by sequence analysis of select exons in our laboratory. For p.L1201R, exon 24 was amplified by polymerase chain reaction (PCR) and analyzed. The forward primer sequence was 5′ TAAATAAAGCGGGCGGTGACAGCA 3′ and the reverse primer sequence was 5′ TGACCTGCAGAAGTACCCAGTGTT.
For p.R1300Q, exon 27 was amplified by PCR and analyzed. The forward primer sequence was 5′ GGCATTAGAGATCCAGACCTTATAGGCA 3′ and the reverse primer sequence was 5′ TAAAGAGGGTGCTCCTTGCTGAGT 3. For p.R2107H, exons 46 and 47 were amplified together as one amplicon by PCR and analyzed. The forward primer sequence was 5′ CCTTCTGTCAGCTCATCCTCCACA 3′ and the reverse primer sequence was 5′ CCAAGTGTCAATGGAGAACACAGG 3′. For all exons, ChoiceTaq Master Mix (Denville Scientific, South Plainfield, New Jersey, USA) with 1.5 mM MgCl 2 was used with PCR cycling parameters at 62 ° C.
The allele frequency in our local African American control population was compared to that of both African and European Americans as compiled by the National Heart, Lung, and Blood Institute Exome Sequencing Project Exome Variant Server. The Exome Variant Server used the sequences of approximately 6,500 exomes and aggregated samples sequenced from a variety of studies of heart, lung and blood disorders. The Exome Sequence Project 6500 (ESP6500) version of the Exome Variant Server was utilized; it included samples from 2,203 African Americans and 4300 European Americans. Although the ophthalmologic statuses of the patients are unknown, the information can be applied to understand better the allele frequency of certain variants in the general population. The phenotypic information available for each study can be ascertained via the National Center for Biotechnology Information database of Genotypes and Phenotypes.
Three sequence homology-based programs were used to predict the functional impact of missense changes identified in this study: Polymorphism Phenotyping v 2 (PolyPhen-2, Harvard University, Cambridge, Massachusetts, USA) ( genetics.bwh.harvard.edu/pph2/ ) ; Sorting Intolerant from Tolerance (SIFT; sift.jcvi.org/ ); and PMut ( mmb2.pcb.ub.es:8080/PMut/ ). Polymorphism Phenotyping-2 analyzes the impact of an amino acid polymorphism on protein structure and predicts whether that amino acid change is likely to be deleterious to protein function by using straightforward physical and comparative considerations. For a given variant, Polymorphism Phenotyping-2 calculates the naive Bayes posterior probability that the mutation is damaging and reports estimates of false-positive rates (the chance that the mutation is classified as damaging when it is, in fact, nondamaging) and true-positive rates (the chance that the mutation is classified as damaging when it is, indeed, damaging). The polymorphism is then appraised qualitatively as being benign, possibly damaging, or probably damaging, based on the model’s false-positive rate thresholds. Results of the SIFT program are reported to be tolerant if the tolerance index is ≥0.05 or intolerant if the tolerance index is <0.05. PMut allows the accurate pathologic prediction of single amino acid mutations based on the use of neural networks. Following the input of a reference sequence and the amino acid substitution of interest, the algorithm provides an answer and a reliability index. An output value >0.5 is predicted to be a pathologic mutation, and a value <0.5 is neutral. The reliability value ranges from 0 to 9 and is considered good with a score of 6 or higher and highly reliable at the maximum score of 9.
Patient Population and Molecular and Clinical Characteristics
The average age of the 85 patients with molecular diagnoses of Stargardt disease was 29.5 ± 16.2 years; 45% were male and 55% were female. The racial identity for our population was as follows: 83% white, 13% African American, 1.1% Hispanic, 1.1% Asian, and 1.1% Arabic. A single mutant allele was detected in 43% of patients, and two or three mutant alleles were detected in 41% and 15% of patients, respectively.
Of 11 unrelated African American patients, 10 shared 1 of 3 alleles: c.3602T>G (n = 3); c.3899G>A (n = 3); and c.6320G>A, (n = 4) ( Table 1 ). The average age of self-reported onset in the 10 African American patients was 25 ± 13 years, although age at initial consultation with and diagnosis by one of the authors (E.I.T.) was 33.2 ± 12 years. On presentation, the logarithm of minimum angle of resolution (logMAR) BCVA was 0.588 ± 0.354 (≈20/80). The individual genotypic and phenotypic characteristics of the 10 patients are detailed in Table 2 and Table 3 . In the 11th African American patient, a single heterozygous mutation (c.4538A>G, p.Q1513R) was identified. The ability to draw genotype-phenotype correlations was limited by the small sample size and the presence of multiple sequence variations in each of several patients.
|Sequence Variation||Exon||Protein||Age of Onset (y)||Age of Presentation (y)||LogMAR BCVA (Snellen)|
|c.3602T>G||24||p.L1201R||33.3 ± 12.6||37.3 ± 12.5||0.583 ± 0.385 (≈20/75)|
|c.3899G>A||27||p.R1300Q||20.0 ± 12.8||37.7 ± 6.3||0.783 ± 0.189 (≈20/120)|
|c.6320G>A||46||p.R2017H||24.5 ± 14.2||26.8 ± 14.2||0.473 ± 0.401 (≈20/60)|
|Pt||cDNA (Protein Product)||% Popn (n = 75)||cDNA (Protein Product)||% Popn (n = 75)||cDNA (Protein Product)||% Popn (n = 75)|
|1||c.3602T>G (p.L1201R||0||c.3322C>T (p.R1108C)||4|
|3||c.3602T>G (p.L1201R)||0||c.4537delC (p.Q1513fsX1525)||0||c.5077G>A (p.V1693I)||0|
|5||c.3899G>A (p.R1300Q)||0||c.618C>T (p.S207S)||0||c.2546T>C (p.V849A)||0|
|6||c.3899G>A (p.R1300Q)||0||c. 3113C>T (p.A1038V)||15||c.1937+1G>C (N/A)||0|
|7||c.6320G>A (p.R2107H)||0||c.IVS38-10T>C (N/A)||10|
|8||c.6320G>A (p.R2107H)||0||c.174C>G (p.N58K)||0|
|9||c.6320G>A (p.R2107H)||0||c.6286G>A (p.E2096K)||0|
|Pt||Age of Onset (y)||Age of Consult (y)||LogMAR BCVA OD (Snellen)||LogMAR BCVA OS (Snellen)||Color Vision||Fundus Grade||Dark Choroid IVFA||ffERG Grade||OCT (Avg CMT μm)|
|1||20||25||1.00 (20/200)||1.00 (20/200)||ND||3||Present||ND||130.5|
|2||45||50||0.54 (20/70)||0.48 (20/60)||ND||1||Present||ND||100.0|
|3||35||37||0.30 (20/40)||0.18 (20/30)||Abnl||2||Present||ND||ND|
|4||34||34||1.00 (20/200)||1.00 (20/200)||Abnl||3||Present||III||206.2|
|5||9||45||0.70 (20/100)||0.70 (20/100)||Abnl||1||Present||III||ND|
|6||17||34||1.00 (20/200)||0.18 (20/30)||Full||3||Not Present||III||ND|
|7||42||45||0.51 (20/65)||0.60 (20/80)||Abnl||2||Present||ND||ND|
|8||30||31||0.30 (20/40)||0.18 (20/30)||Abnl||2||Present||ND||ND|
|9||11||16||0.10 (20/25)||0.10 (20/25)||ND||1||Not present||III||ND|
|10||15||15||1.00 (20/200)||1.00 (20/200)||ND||1||Present||ND||ND|
Of the 10 African American patients, 3 shared a c.3602T>G missense mutation located in exon 24 (p.L1201R). Despite varied fundus appearances and visual acuities across the group, all 3 patients exhibited a dark choroid sign on fluorescein angiography. Patient 1 was heterozygous for a second c.3322C>T missense mutation in exon 21 (p.R1108C), which was present in 4% of our patients and not unique to the African American subset ( Table 2 ). Several studies implicate this second variant as being pathogenic. We did not detect any additional sequence variations in patient 2. Patient 3 possessed two additional known disease-causing variations, c.4537delC (p.Q1513fsX1525) and c.5077G>A (p.V1693I), neither of which was detected in other members of our cohort. The c.4537delC reportedly causes a premature stop codon 12 amino acids downstream in the protein product. The c.5077G>A (p.V1693I) variation has been previously detected in Stargardt disease patients but not in controls.
Patients 4, 5 and 6 possessed a c.3899G>A sequence variation located in exon 27 (p.R1300Q). No other known disease-causing sequence variations were detected in patient 4. However, patient 5 possessed two additional sequence variations: c.618C>T (p.S206S), a synonymous sequence variation that has been found to cosegregate with disease in a family with Stargardt disease, and c.2546T>C (p.V849A). Patient 6 exhibited both a c.3113C>T mutation (p.A1038V), present in 15% of our cohort, and a c.1937+1G>C sequence variation that results in a splice site mutation in intron 13. The c.3113C>T mutation produces a biochemically altered protein product and has been detected in patients with Stargardt disease but not in control patients.
The third sequence variation, c.6320 G>A (p.R2107H), existed as a heterozygous sequence variation in patients 7, 8, 9, and 10. All patients except patient 10 were heterozygous for a second known disease-associated mutation. c.IVS38-10T>C was detected in patient 7 as well as in 10% of our cohort. This sequence variation has been found commonly in patients with ABCA4-associated disease and is proposed to be in linkage disequilibrium with a pathogenic mutation. Patient 8 had a c.174C>G sequence variation (p.N58K) that was predicted to be pathogenic in a study by Briggs and associates. The c.6286G>A (p.E2096K) variant identified in patient 9 is likely to be a pathogenic mutation and was located in nucleotide binding domain 2; it reduces adenosine triphosphate enzymatic (ATPase) activity in biochemical studies.
Sequence Variation Studies
The three sequence variations detected in the African American patients with Stargardt disease were studied in control African American patients. Our results were compared to the frequency of the minor allele in African Americans as compiled by the Exome Variant Server ( Table 4 ). Similar allele frequencies were found in our Cleveland-based control African American population when compared to the minor allele frequency determined by the Exome Variant Server. In contrast, the allele frequency of all sequence variations was much lower in patients of European ancestry ( Table 4 ). Both the c.3602T>G and c.3899G>A sequence variations were found in the homozygous state in two African American control individuals with no retinal disease. Although the ophthalmologic phenotypic data for patients included in the Exome Variant Server are not available, both sequence variations were reported in the homozygous state in 6 and 25 African American patients, respectively, per the Exome Variant Server, but in none of those of European ancestry. In contrast, the c.6320G>A was not found in the homozygous state in any of our phenotypically normal African American controls and was found in only one patient of African American ancestry per the Exome Variant Server. The ophthalmologic clinical status of this particular patient is unknown.