January 2017
Volume 58, Issue 1
Open Access
Genetics  |   January 2017
Mutation Spectrum of the ABCA4 Gene in 335 Stargardt Disease Patients From a Multicenter German Cohort—Impact of Selected Deep Intronic Variants and Common SNPs
Author Affiliations & Notes
  • Heidi L. Schulz
    Institute of Human Genetics, University of Regensburg, Regensburg, Germany
  • Felix Grassmann
    Institute of Human Genetics, University of Regensburg, Regensburg, Germany
  • Ulrich Kellner
    Rare Retinal Disease Center, AugenZentrum Siegburg, MVZ ADTC Siegburg GmbH, Siegburg, Germany
    RetinaScience, Bonn, Germany
  • Georg Spital
    Department of Ophthalmology, St. Franziskus-Hospital, Münster, Germany
  • Klaus Rüther
    Augenarztpraxis, Berlin, Germany
  • Herbert Jägle
    Department of Ophthalmology, University Hospital Regensburg, Regensburg, Germany
  • Karsten Hufendiek
    University Eye Hospital Hannover Medical School, Hannover, Germany
  • Philipp Rating
    Department of Ophthalmology, University Hospital Essen, Essen, Germany
  • Cord Huchzermeyer
    Department of Ophthalmology, University Hospital Erlangen, Erlangen, Germany
  • Maria J. Baier
    Institute of Human Genetics, University of Regensburg, Regensburg, Germany
  • Bernhard H. F. Weber
    Institute of Human Genetics, University of Regensburg, Regensburg, Germany
  • Heidi Stöhr
    Institute of Human Genetics, University of Regensburg, Regensburg, Germany
  • Correspondence: Heidi Stöhr, Institute of Human Genetics, University of Regensburg, Franz-Josef-Strauß-Allee 11, D-93053 Regensburg, Germany; heidi.stoehr@klinik.uni-regensburg.de
Investigative Ophthalmology & Visual Science January 2017, Vol.58, 394-403. doi:10.1167/iovs.16-19936
  • Views
  • PDF
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to Subscribers Only
      Sign In or Create an Account ×
    • Get Citation

      Heidi L. Schulz, Felix Grassmann, Ulrich Kellner, Georg Spital, Klaus Rüther, Herbert Jägle, Karsten Hufendiek, Philipp Rating, Cord Huchzermeyer, Maria J. Baier, Bernhard H. F. Weber, Heidi Stöhr; Mutation Spectrum of the ABCA4 Gene in 335 Stargardt Disease Patients From a Multicenter German Cohort—Impact of Selected Deep Intronic Variants and Common SNPs. Invest. Ophthalmol. Vis. Sci. 2017;58(1):394-403. doi: 10.1167/iovs.16-19936.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Purpose: Stargardt disease (STGD1) is an autosomal recessive retinopathy, caused by mutations in the retina-specific ATP-binding cassette transporter (ABCA4) gene. To establish the mutational spectrum and to assess effects of selected deep intronic and common genetic variants on disease, we performed a comprehensive sequence analysis in a large cohort of German STGD1 patients.

Methods: DNA samples of 335 STGD1 patients were analyzed for ABCA4 mutations in its 50 coding exons and adjacent intronic sequences by resequencing array technology or next generation sequencing (NGS). Parts of intron 30 and 36 were screened by Sanger chain-terminating dideoxynucleotide sequencing. An in vitro splicing assay was used to test selected variants for their splicing behavior. By logistic regression analysis we assessed the association of common ABCA4 alleles while a multivariate logistic regression model calculated a genetic risk score (GRS).

Results: Our analysis identified 148 pathogenic or likely pathogenic mutations, of which 48 constitute so far unpublished ABCA4-associated disease alleles. Four rare deep intronic variants were found once in 472 alleles analyzed. In addition, we identified six risk-modulating common variants. Genetic risk score estimates suggest that defined common ABCA4 variants influence disease risk in carriers of a single pathogenic ABCA4 allele.

Conclusions: Our study adds to the mutational spectrum of the ABCA4 gene. Moreover, in our cohort, deep intronic variants in intron 30 and 36 likely play no or only a minor role in disease pathology. Of note, our findings demonstrate a possible modifying effect of common sequence variants on ABCA4-associated disease.

Stargardt disease (STGD1; Mendelian Inheritance in Man [MIM] 248200) is the leading cause of Mendelian macular dystrophy that often manifests in the first two decades, but may also become symptomatic only later in life. It is typically characterized by bilateral progressive loss of central vision, color vision defects, and photophobia. The fundus often presents with yellow-white autofluorescent flecks (fundus flavimaculatus) and chorioretinal atrophy in the macula, typically with an initial foveal sparing and variable extension into the periphery.1,2 
Autosomal recessive STGD1 is caused by mutations in the ABCA4 gene.3 Other retinal phenotypes, such as retinitis pigmentosa and cone–rod dystrophy, have also been linked to recessive mutations in this gene.4,5 ABCA4 encodes a member of the ATP-binding cassette (ABC) superfamily of transmembrane proteins specifically localized at the rims and incisures of outer segment discs of rod and cone photoreceptors.6 In the retina, ABCA4 transports N-retinylidene-phosphatidylethanolamine from the lumen to the cytoplasmic leaflet of photoreceptor disc membranes, a process that together with chemical isomerization facilitates the removal of potentially toxic retinoid compounds.7 Conversely, impaired ABCA4 function results in an accumulation of bisretinoids such as A2E in the RPE. This debris, also known as lipofuscin, is thought to disturb RPE metabolism and ultimately may cause cell death.8 Current genotype–phenotype models correlate the degree of residual ABCA4 activity with the severity of the retinal disease phenotype.5,9 In addition, toxic gain-of-function mutations were described in the ABCA4 gene and were proposed to contribute to disease severity.10 
To date, more than 900 different mutations including missense, splicing, truncating, and frameshift alterations have been reported in the ABCA4 gene and shown to be associated with retinal degeneration (HGMD database, build 2015.4) (Qiagen, Hilden, Germany). Of these, 72% were identified in STGD1 patients. Additionally, 1120 rare nonsynonymous variants (frequency ≤ 0.005) were found in coding regions or highly conserved exon-flanking intronic sequences of the ABCA4 gene by the Exome Aggregation Consortium (ExAC) (Cambridge, MA, USA; http://exac.broadinstitute.org; January 2016; in the public domain). These variants encompass 257 known and 863 potential novel disease-associated mutations. ABCA4 mutations are scattered across all 50 coding exons of the gene, although representative studies from different geographic regions reveal the existence of ethnic founder mutations.9,1117 
Despite detection rates of ABCA4 mutations greatly increasing over recent years by state-of-the-art analytical approaches such as next generation sequencing, homozygous or compound heterozygous mutations are regularly detected in no more than 70% of STGD1 patients, while a significant number of patients carry only a single ABCA4 mutation (17–30%).15,1721 It has been debated whether some of the carriers of monoallelic ABCA4 mutations could be combined into a subgroup of late-onset STGD1 or may present as age-related macular degeneration (AMD).22,23 Recently, several deep intronic variants in the ABCA4 gene were reported to segregate with disease and in some cases to affect correct splicing. It was suggested that these variants represent a novel and as yet rather unexplored type of pathogenic ABCA4 mutations.20,21,2426 
To further explore the genetic architecture of ABCA4 pathology in STGD1, we analyzed the ABCA4 gene in a large German cohort of 335 STGD1 patients referred to our molecular diagnostics unit at the Institute of Human Genetics, Regensburg, between 2008 and 2015. The molecular analysis included sequencing of the 50 coding exons/conserved splice sites of the ABCA4 gene as well as two regions in intron 30 and 36 reported earlier to harbor deep intronic disease-associated variants. Moreover, functional studies were performed to assess pathogenicity of selected ABCA4 variants. Finally, a multivariate logistic regression model was used to investigate the role of common ABCA4 variants with allele frequencies > 0.01. 
Materials and Methods
Patients
Sequence analysis included a total of 335 patients (191 female, 144 male) with definite or suspected clinical diagnosis of STGD1 who underwent routine genetic testing at the Institute of Human Genetics Regensburg. Referrals between 2008 and 2015 were included in the study. A positive family history of STGD1 was reported for 35 cases; 131 individuals had a negative family history, and no further information on the familial status of retinal degeneration was available for 169 patients. Information on ethnicity was mainly provided by the patients themselves or was estimated from their first or last name. A total of 283 patients were Caucasian (80%), 30 patients were from the Middle East, 1 patient was from Asia, and for 21 patients ethnicity could not unanimously be determined. A control group of 779 probands free of retinal disease was recruited from the Lower Franconian region in Germany27 and used to assess the role of common variants in STGD1 pathology. In accordance with the German Genetic Diagnostics Act, written informed consent was obtained from all subjects. This study adhered to the tenets of the Declaration of Helsinki. 
Molecular Analysis
For DNA extractions, EDTA peripheral blood samples were obtained from all patients. Mutational analyses of the coding exons and the flanking intronic sequences [−12 bp/+6 bp covering the key donor ((C|A) A G G T (A|G) A G T) and acceptor (Y10 N (C|T) A G (G|A)) splice signals at the exon–intron boundaries28] of the ABCA4 gene (NM_00350.2) were mainly performed (186 patients) by a custom-designed GeneChip CustomSeq Resequencing Array (RetChip) (Affymetrix, Santa Clara, CA, USA). Multiplex-PCR products were amplified by exon flanking oligonucleotide primer pairs (oligonucleotide sequences will be provided upon request). The data were analyzed with the Affymetrix GeneChip Analysis Software (GSEQ) and the SeqC module (JSI medical systems, Kippenheim, Germany) as previously described.29 For 149 patients, mutation testing was performed by next generation sequencing (NGS) on an ION Torrent semiconductor personal sequencing machine (Life Technologies, Darmstadt, Germany) upon multiplex-PCR amplification of the fragments with the STARGARDT MASTR kit (Multiplicom, Niel, Belgium). Regions with a depth of sequence coverage below 30× were reanalyzed by Sanger chain-terminating dideoxynucleotide sequencing on an ABI 3130xl instrument (Applied Biosystems, Carlsbad, CA, USA). Next generation sequencing data were analyzed within the CLC Genomics Workbench (CLC bio, Aarhus, Denmark). Potentially disease-relevant variants identified by RetChip or NGS were routinely confirmed by Sanger sequencing. To test for deep intronic sequence variants in 236 patients, primer pairs ABCA4-Exon30.1 fwd (5′-TCT GAA GTT CCC ACA TAG TAC TTC C-3′)/rev (5′-GAT TGT GTT TGA GTG GCT AGG G-3′) and ABCA4-Exon36.1 fwd (5′-ACT ACC ACC TTC CTG ACA ACC AA-3′)/rev (5′-CTA CAT CTC TCT CCA TAG GCT CAG A-3′) were used to amplify and sequence a 607- and 535-bp fragment within ABCA4 intervening sequences 30 and 36, respectively. 
Disease-related variants were submitted to the ClinVar database (Acc. Nos. SCV000281879–SCV000281965, http://www.ncbi.nlm.nih.gov/clinvar/; in the public domain). All variants are described according to current nomenclature guidelines (http://www.hgvs.org/mutnomen; in the public domain). 
Bioinformatic Analyses and Variant Classification
Variants were classified as recommended by the American College of Medical Genetics and Genomics (ACMG) standards and guidelines and the Association for Molecular Pathology (AMP) Clinical Practice Guidelines and Reports without discrimination between benign or likely benign and were based on population data, computational data, previous publications, and functional data. Variant frequency in non-Finnish Europeans was taken from the browser of the Exome Aggregation Consortium (ExAC). A variant was classified as benign if the minor allele frequency (MAF) was ≥0.005 except for variant c.2588G>C. Variants with MAF below 0.005 or with an unknown frequency were further tested for disease relevance by prediction algorithms of the programs MutationTaster (Mutation Taster, http://www.mutationtaster.org/; in the public domain) and PolyPhen-2 (PolyPhen2, http://genetics.bwh.harvard.edu/pph/; in the public domain). A sequence variant was classified as a variant of uncertain significance (vus) when none or only one of the two programs predicted a possible pathogenic effect. Exceptions to this rule included missense mutations well documented in the literature to be disease associated in STGD1 patients. Such variants were nonetheless classified as “likely pathogenic.” When both programs, MutationTaster and PolyPhen-2, predicted a pathogenic effect, the variants were classified as “likely pathogenic.” Stop and frameshift mutations as well as variants with experimental evidence of ABCA4 dysfunction were classified as pathogenic. In addition, rare variants were further analyzed for a potential effect on correct splicing by using the interface provided by the software package Alamut (Interactive Biosoftware, Rouen, France; Alamut Visual 2.7.1), which combines five algorithms (SpliceSiteFinder, MaxEntScan, NNSPLICE, GeneSplicer, and Human Splicing Finder). All tools provide prediction scores for the normal and the mutant allele. Variants were selected for further testing by in vitro minigene splicing assays when the score differences calculated by four prediction programs were >10%. 
Genetic Risk Score Calculation
Genotyping of controls was performed on a HumanExome Bead Chip (Illumina, San Diego, CA, USA) by the International AMD Genomics consortium.30 Association testing and genetic risk score (GRS) calculations were performed as described by Grassmann et al.31 In brief, logistic regression analysis was used to assess the association of common alleles (MAF > 0.01) present in the coding regions of the ABCA4 gene. Since variants rs1800717 and rs6666652 are highly correlated (pairwise r2 = 1, based on the EUR panel in the 1000 Genomes Project, Build 20130502, http://www.internationalgenome.org/; in the public domain), only rs6666652 was used in the analysis. Variants with a MAF ≤ 0.01 and rare deep intronic variants were not included in further calculations. Initially, we determined the odds ratios (OR), 95% confidence intervals (95% CI), and P values for each common variant. We adjusted the resulting P values according to the false discovery rate (FDR)32 and considered variants with an FDR < 0.05 to be statistically significant. A GRS was calculated from significantly associated variants for all controls and patients from a multivariate logistic regression model. This was done by multiplying the number of risk-modulating alleles (which can be either 0, 1, or 2) with the respective logarithmic odds ratio (log odds, computed from the multivariate model); then the sum of all weighted genotypes was calculated. Finally, a constant (−1.0503) was subtracted from the score to center the risk score distribution of the controls on zero. 
Minigene Assay
Functional consequences of putative splicing variants were analyzed with the pSPL3-based exon trapping system.33 Exons harboring spliceogenic variants and control exons were PCR amplified from genomic DNA using a 7:1 ratio of Taq:Pfu polymerase and oligonucleotide primer pairs flanked by restriction sites for EcoRI and BamHI (Supplementary Table S1). The amplified products were cloned into the pSPL3b vector. HEK293 cells were transfected in duplicate with 3 μg construct DNA using TransIT-LT1 Transfection Reagent (Mirus Bio, Madison, WI, USA) and harvested after 48 hours; total RNA was extracted using RNeasy Micro Kit (Qiagen). Reverse transcription–PCR was performed with the RevertAidH First Strand cDNA Synthesis Kit (Thermo Fisher Scientific, Waltham, MA, USA) and a pSPL3-specific SD6 (5′-TCT GAG TCA CCT GGA CAA CC-3′) and SA4 (5′-CAC CTG AGG AGT GAA TTG GTC G-3′) primer pair in combination with oligonucleotide primers for the housekeeping genes ACTB (forward 5′-GAC ATC CGC AAA GAC CTG TA-3′; reverse (5′-CAG GAG AGC AAT GAT CTT GA-3′) and GUSB (forward 5′-GAT CCA CCT CTG ATG TTC AC-3′; reverse (5′-TAT TCC CCA GCA CTC TCG TC-3′). Polymerase chain reaction products were separated by electrophoresis on a 2% agarose gel, excised, cloned, and sequenced. Fragment analysis was done by PCR amplification with fluorescein amidite (FAM)–labeled SD6 using Phusion Flash High-Fidelity PCR Master Mix (Thermo Fisher Scientific). Samples and a LIZ Size Standard (Applied Biosystems, Darmstadt, Germany) were run on an ABI 3130xl sequencer and analyzed with the Peak Scanner software (Applied Biosystems). 
Results
Identification of ABCA4 Mutations
Mutational analysis of the ABCA4 gene was performed in 335 STGD1 patients and identified 189 unique sequence variations, of which 171 were listed in the non-Finnish European ExAC cohort as a rare variant with MAF ≤ 0.005. Of the 171 rare ABCA4 variants, 148 were classified as pathogenic or likely pathogenic and 23 as vus (Supplementary Table S2). Twenty-eight percent (48/171) of the rare alleles constitute novel as yet unpublished variants (Table 1). Missense mutations represent 59.6% (102/171) followed by nonsense and frameshift mutations (15.8%; 27/171), splice site mutations (13.5%; 23/171) and synonymous variants (8.8%; 15/171). Of note, variant c.2588G>C is present with a MAF > 0.005 (0.0063; ExAC) but was classified as pathogenic based on its known deleterious effect on ABCA4 protein function.9,34,35 
Table 1
 
Novel Rare Sequence Variants in the ABCA4 Gene
Table 1
 
Novel Rare Sequence Variants in the ABCA4 Gene
The most common pathogenic or likely pathogenic mutations in our cohort were c.5882G>A (51 of 396 mutant alleles), c.[3113C>T(;)1622T>C] (36/396), c.2588G>C (28/396), and c.5461-10T>C (23/396), which is consistent with our earlier report of frequent mutations in 144 German STGD1 patients.12 Together, these sequence changes account for 34.8% (138/396) of disease-associated variants. Ninety-nine mutations were identified in only one patient (Supplementary Table S2). 
Impact of Selected ABCA4 Sequence Variants on RNA Splicing Efficiency
Twenty-eight sequence variants were considered to have a potential impact on splicing of the mature ABCA4 transcript based on results from the bioinformatic analysis (Supplementary Table S2). Experimental evidence for splicing defects was previously published for five of these (c.768G>T, c.2588G>C, c.4253+5G>A, c.5461-10T>C, c.5714+5G>A) (Supplementary Table S2). Here, functional pSPL3-based minigene assays were performed for another 22 variants (Table 2; Supplementary Fig. S1). The analysis of cDNAs derived from transfected HEK293 cells revealed aberrant pre-mRNA splicing for all of the 13 variants affecting canonical GT-AG splice sequences and for c.4773+3A>G located in the conserved donor splice site of ABCA4 exon 33. This includes three mutations changing the canonical GT to the weaker GC 5′ splice site, which is present in ∼1% of authentic introns.36 The absence of functional donor or acceptor splice sites led to exon skipping, intron retention, and use of cryptic splice sites (Table 2; Supplementary Fig. S1), likely leading to truncated protein products or nonsense-mediated mRNA decay (NMD). 
Table 2
 
Functional Analysis of Putative Splice Mutations in the ABCA4 Gene
Table 2
 
Functional Analysis of Putative Splice Mutations in the ABCA4 Gene
Of the eight exonic sequence changes included in the minigene assay, three stop mutations and four missense mutations were predicted to create cryptic canonical splice sites, whereas the synonymous mutation c.3813G>A could weaken the donor splice site of exon 25 (Table 2). Four exonic variants revealed both correct and aberrant splice products in various proportions. Synonymous mutation c.3813G>A and missense mutations c.4469G>A, c.4919G>A, and c.6647C>T led to incorrectly spliced transcripts that match the predictions from the bioinformatic analysis (Table 2; Supplementary Fig. S1). Hence, these four variants initially classified as likely pathogenic or vus were reclassified as being pathogenic (Supplementary Table S2). Sequence variants c.122G>A/p.(W41*), c.179C>T/p.(A60V), c.3808G>T/p.(E1270*), and c.5189G>A/p.(W1730*) revealed no effect on mRNA splicing in the in vitro assay. 
Disease-Associated ABCA4 Alleles in 335 STGD Patients
Two or more likely pathogenic or pathogenic changes in the ABCA4 gene were found in 151 of the 335 STGD1 patients. In another seven patients a deleterious mutation was detected in addition to a vus, eventually raising the overall detection rate of two expected mutations to 47.2% (Supplementary Table S3). This includes 14 cases with null mutations: c.[2692G>T(;)5461-10T>C(;)5606C>T], c.[5018+2T>C(;)6386+2C>G], c.[296dupA];[5018+2T>C], c.[3813G>A(;)3813G>A], c.[4234C>T(;)5714+5G>A], c.[4773+3A>G(;)5461-10T>C], c.[5461-10T>C];[5461-10T>C], c.[67-1G>C];[67-1G>C], c.[1A>G; 6089G>A];[5714+5G>A], c.[5714+5G>A(;)5714+5G>A], c.[5917delG];[5917delG], and c.[5714+5G>A];[5917delG]. The latter two combinations were found in two patients (ID 86 and ID 255), and an additional 58 cases were carrying a null and a missense mutation (Supplementary Table S3). Compound heterozygosity or homozygosity could be confirmed by segregation analysis in 27 patients. One pathogenic or likely pathogenic mutation was identified in 82 individuals, one vus in 13 cases resulting in a detection rate of 28.4% for carriers of a single ABCA4 mutation (Supplementary Table S3). Complex alleles verified by testing the parent(s) of the respective patients were c.[1A>G; 6089G>A] in patient 219, c.[1609C>T; 5881G>A] in patient 314, c.[1622T>C; 3113C>T] in patients 1, 13, 106, 124, 273, 282, and 311, c.[2588G>C; 3289A>T] in patient 75, and c.[2588G>C; 5153T>G] and c.[5318C>T; 6148G>C] together in patient 69. In a total of 89 patients (26.6%) no disease-associated variant was found in the ABCA4 gene. 
Frequency of Deep Intronic Variants in Putative Alternative Exons of the ABCA4 Gene
To investigate the frequency of deep intronic variants in the ABCA4 locus in our patient cohort, we sequenced genomic regions encompassing putative alternative exons in intron 30 and 36, namely, exon 30.1, exon 36.01, and exon 36.1. These regions were previously identified by RNA sequencing of retinal tissue (Supplementary Fig. S2).24,37,38 Ten different sequence variants were found in 236 patients, of which four (c.4539+2028C>T, c.4540-2036C>A, c.5196+1013A>G, c.5196+1056A>G) were defined as rare variants with a MAF ≤ 0.005 in the 1000 Genomes catalog, phase 3 (Table 3; Supplementary Fig. S2). These changes were each identified once in the 472 alleles analyzed. Among these, c.4539+2028C>T and c.5196+1056A>G have been described previously by Braun et al. (2013)24 as V5 and V3, respectively, and were later also found in another cohort of STGD1 patients.20 Deep intronic variants c.4540-2036C>A and c.5196+1013A>G are reported for the first time. The c.5196+1056A>G (V3) variant is predicted to strengthen a cryptic donor splice site. Reverse transcriptase–PCR analysis using keratinocyte mRNA of patients carrying the V3 variant revealed the inclusion of a 177-bp segment into the ABCA4 transcript (termed “alternative exon 36.01”), thus providing experimental evidence for a loss-of-function effect of this variation.24 Patient 230 in our cohort carried the V3 variant and a second variant (c.2588G>C) annotated as pathogenic. This suggests that the V3 mutation is indeed the second disease-associated allele in our patient. A novel sequence change, c.5196+1013A>G, was found in patient 79 in combination with c.1654G>A, a likely pathogenic missense mutation [p.(V552I)]. Similar to the V3 variant, c.5196+1013A>G is predicted to strengthen a cryptic donor splice site and thus may also result in deleterious effects upon splicing. Based on bioinformatics, variants c.4539+2028C>T (V5) and c.4540-2036C>A are not predicted to affect splicing. The c.4539+2028C>T (V5) sequence change was identified in patient 42 with no other probable mutant ABCA4 allele; variant c.4540-2036C>A was detected in patient 159 in addition to pathogenic mutation c.[1622T>C(;)3113C>T] and likely pathogenic mutation c.3261A>C. A disease association of these intronic variants in our patients is thus unclear. Six previously reported intronic sequence changes (c.4539+2001G>A, c.4539+2064C>T, c.5196+1136C>A, c.5196+1137G>A, c.5196+1159G>A, c.5196+1216C>A) were not found in our patient cohort (Table 3). 
Table 3
 
Deep Intronic Variants in the ABCA4 Gene
Table 3
 
Deep Intronic Variants in the ABCA4 Gene
The Contribution of Common ABCA4 Variants to STGD1
We noticed a striking difference in the frequencies of common ABCA4 variants (MAF ≥ 0.01) in the patient versus the control group (Supplementary Table S2; Fig. 1). This led us to investigate whether such variants may exert an effect on the risk for developing STGD1. Genotype information on 16 common ABCA4 variants was available for the 335 patients and 781 German controls from Lower Franconia. For each variant, risk association was computed applying a logistic regression model. We found statistically significant association for six variations (c.1268A>G, c.4203C>A, c.5603A>T, c.5682G>C, c.5843C>T, c.6249C>T) (FDR < 0.05, Supplementary Table S4). To estimate the independence of the associated variants, we fit a multivariate logistic regression model incorporating these variations. In the multivariate model, association of variant c.4203C>A (rs1801666) with disease risk was no longer statistically significant, probably due to linkage disequilibrium with the strongly associated variant c.5603A>T (rs1801466, r2 = 0.38). The remaining five risk-modulating variants were found to be independently associated with disease risk (Table 4). Repeating the analysis excluding the data from 52 samples of non-European or unknown ancestry, similar results were obtained, including the significant and independent association of the five risk-modulating variants (c.1268A>G, c.5603A>T, c.5682G>C, c.5843C>T, c.6249C>T) with similar effect sizes. 
Figure 1
 
Frequency of common ABCA4 variants in patients with Stargardt disease and European non-Finnish controls. Six variants that were found to be significantly enriched in either patients or controls are framed.
Figure 1
 
Frequency of common ABCA4 variants in patients with Stargardt disease and European non-Finnish controls. Six variants that were found to be significantly enriched in either patients or controls are framed.
Table 4
 
Association of Common ABCA4 Variants With Stargardt Disease
Table 4
 
Association of Common ABCA4 Variants With Stargardt Disease
Assuming that common pathogenic mutations emerged on common haplotypes tagged by associated variants, our analysis may measure their impact on disease risk rather than that of the common associated variant. We therefore evaluated the correlation between common mutations in our cohort (c.1622T>C, c.2588G>C, c.3113C>T, c.4234C>T, c.5461-10T>C, c.5714+5G>A, c.5882G>A, c.5917delG) and the five risk-modulating variants (c.1268A>G, c.5603A>T, c.5682G>C, c.5843C>T, c.6249C>T). Three frequent mutations (c.2588G>C, c.5461-10T>C, c.5882G>A) were found to be potentially correlated (R2 > 0.05) with common STGD1-associated variants. Subsequently, the association analysis was repeated by excluding individuals carrying any of three correlated pathogenic mutations c.5882G>A, c.2588G>C, and c.5461-10T>C from the analysis. Still, STGD1 association of common variants c.1268A>G and c.5603A>T remained highly significant (P values of 3.45E-05 and 4.82E-05, respectively). Similarly, removing patients carrying one of the eight most common STGD1 mutations (c.5882G>A, c.3113C>T, c.1622T>C, c.2588G>C, c.5461-10T>C, c.5714+5G>A, c.4234C>T, or c.5917delG) from the analysis resulted in a statistically significant association for the common variants c.5603A>T and c.1268A>G (P values of 0.000273 and 0.0109, respectively) (Supplementary Table S5). For variant c.6249C>T, the association was no longer significant after removal of patients carrying common mutations, although the variant still revealed a similar effect size. The latter finding suggests that loss of significance is likely due to lack of statistical power caused by the reduced number of individuals in the subsequent analyses (from n = 335 to n = 247 to n = 197 patients, Supplementary Table S5). In contrast, variant c.5682G>C showed a reduced effect size and also a lack of statistically significant association in both subsequent analyses. 
To further evaluate the role of these common variants in retinal pathology, a GRS was calculated for the STGD1 patients in our cohort and the 781 control individuals. This was computed as the sum of risk-increasing and protective alleles weighted by the relative effect size of each of the risk-modulating variants from the multivariate logistic regression model. An alpha estimate of −1.0503 was used to center the GRS on zero for the controls. Patients diagnosed with STGD1 have a significantly higher mean GRS (0.418; P value < 10−16, 95% CI: 0.3479–0.4887) compared to controls (0.0, 95% CI: −0.0466 to 0.0466) (Fig. 2). Importantly, patients with one, two, or more disease-associated ABCA4 mutations have a significantly higher GRS than those with none (P = 0.0011 and 0.0024; respectively) (Fig. 2), implying a risk-increasing effect of common ABCA4 alleles when combined with a single defective copy of the ABCA4 gene. 
Figure 2
 
Genetic risk score (GRS) for patients and controls stratified according to the number of ABCA4 mutations or disease status. An alpha estimate of −1.0503 was used to center the GRS on zero for the controls. Red bars represent the mean GRS and confidence intervals.
Figure 2
 
Genetic risk score (GRS) for patients and controls stratified according to the number of ABCA4 mutations or disease status. An alpha estimate of −1.0503 was used to center the GRS on zero for the controls. Red bars represent the mean GRS and confidence intervals.
Discussion
Almost 20 years after the identification of ABCA4 as the causative gene for autosomal recessive STGD1,3 the genetics of this common form of hereditary macular degeneration still have not fully been clarified. Our study describes the findings of mutational profiling of the ABCA4 gene in the largest cohort of STGD1 patients published so far. A total of 171 unique rare sequence variants, 48 of which constitute novel, so far unreported, mutations, were identified in the coding region, and an additional four rare deep intronic variants were detected in intron 30 and 36 of the ABCA4 gene. In 47% of patients, two causative ABCA4 mutations were found. This is at the lower end of the mutation detection rate in STGD1 patients, with other studies identifying biallelic mutations between 55% and 80%,17,19,21,39,40 but may be explained by a combination of factors. First, clinical heterogeneity and variable diagnostic criteria inherent to multiclinic retrospective cohort studies may have led to uncertainties in clinical diagnosis in some patients of our cohort. In addition, small deletions/insertions may have been missed due to limitations in the detection accuracy of this type of mutation by the resequencing array and ION Torrent semiconductor technology. Finally, large rearrangements within the ABCA4 locus that have been shown to occur in 0.3% to 2.2% of the analyzed chromosomes in previous reports9,26,41 would not have been detected by the mutational screening in this study. 
An important question relates to the high proportion of carriers of single ABCA4 mutations consistently observed in numerous studies regardless of the ethnic background of the patients.14,15,18,19,21,39,40,42 A single mutant ABCA4 allele was found in almost 30% of the 335 individuals in our patient group, a number representing the upper end of the reported ranges. Mutations in noncoding regions of the ABCA4 gene locus have been proposed as a common source for a second causative mutation. More specifically, several studies have described the occurrence of deep intronic variants in intron 30 and 36 in a significant proportion of patients (7.9–17.9%) with single ABCA4 mutations.20,21,2426 Indeed, there is some experimental evidence that a few of these variants activate/strengthen cryptic splice sites leading to the inclusion of alternative exons in the mature transcript.24 In our cohort, of 95 STGD patients with a single pathogenic/likely pathogenic ABCA4 mutation, only 2 were found to carry an additional rare intronic variant in intron 30 or 36 (2.1%). Two other novel rare intronic variants were detected in a patient with two causative ABCA4 mutations and a patient with no ABCA4 mutations, respectively. From these data, we conclude that causative ABCA4 mutations in intron 30 and 36 may contribute to STGD1 pathogenesis only in a minority of cases. Similarly, in a small German cohort of 46 patients with single heterozygous ABCA4 variants, no mutations were found in these intronic regions.25 
To further add to the discussion on the impact of deep intronic variants to STGD1 disease, the full sequence analysis of the entire ABCA4 locus including exonic as well as intronic sequences will be needed independent of the patient's mutation status. It was estimated that approximately 95% of multiexon transcripts undergo alternative splicing.43 While the key components and mechanisms involved in mRNA processing are reasonably well understood, high-confidence predictions on splicing effects of defined sequence changes remain inaccurate. In this context, it is of note that the extent and variability of alternative splicing in the human retina were found to be remarkable,37 although the functional consequences of the alternative transcripts remain obscure. Approximately 13% of novel mRNA junctions were expressed in the retina at levels similar to or higher than the reference transcript. Deep RNA sequencing of three healthy human retinas revealed at least 11 alternatively spliced exons and 71 alternative exon/intron junctions,37 implying the possibility of additional deep intronic regions as candidates to potentially harbor mutations affecting the correct splicing of the ABCA4 transcript. Considering that 7.6% of STGD1 patients of a Belgian cohort carried a founder deep intronic c.4539+2001G>A mutation,25 it is conceivable that a significant fraction of the German STGD1 patients carry as yet unknown mutations in the noncoding regions of the ABCA4 gene. 
To address the strikingly high frequency of carriers with only a single identified ABCA4 disease allele, we assessed the contribution of common ABCA4 variants on disease development. To this end, our findings reveal that variants c.1268A>G (OR 0.6682; CI 0.531–0.836), c.5843C>T (OR 0.1678; 0.063–0.377), and c.6249C>T (OR 0.6483; CI 0.407–1.001) exert a protective effect, while variants c.5603A>T (OR 1.9652; CI 1.335–2.909) and c.5682G>C (OR 1.3973; CI 1.082–1.800) confer an increased risk for developing STGD1. These results are in agreement with and extend previous studies that identified c.5603A>T as a risk-increasing allele44 and c.1268A>G as a protective variant.40 Our data now facilitate a GRS estimate that coalesces the total effect of protective and risk-increasing alleles similarly to the model developed to assess the genetic risk for AMD.31 Interestingly, the GRS in the group of patients with only a single disease-associated ABCA4 mutation is significantly higher than in patients with no ABCA4 mutation, suggesting that common variants are involved in conferring risk for STGD1. 
Simultaneous loss of significant association with STGD1 and effect size was observed for common variant c.5682G>C when the three most frequent pathogenic mutations were excluded from the analysis. This could be due to linkage disequilibrium between certain pathogenic mutations and c.5682G>C, implying that the common ABCA4 allele either has no effect on the pathogenicity of the rare variant or adds to the damaging effect of the mutation. Furthermore, c.5682G>C may act as a risk-increasing allele only in trans when combined with specific pathogenic mutations. The assumption that the presence of a common genetic variant with an adverse effect on ABCA4 function actually triggers the deleterious consequences of the pathogenic mutation could explain why some of the common mutations have a frequency of >1% in certain populations but do probably not cause STGD1 in the respective carriers. For example, the allele frequency of pathogenic mutation c.5882G>A/p.(G1961E) is 1.4% in the South Asian population (ExAC) and even higher in Somalia,45 although a higher prevalence of STGD1 in these countries has not been reported. Alternatively, a protective ABCA4 allele may ameliorate a pathogenic mutation in trans, although only in combination with a less severe ABCA4 allele. In this context, it should be noted that protective variant c.6249C>T is five times more frequent in the African than in the European or Asian populations. 
The GRS model may also explain another discrepant aspect in our patient cohort. Fourteen STGD1 patients in our study each harbor two null mutations in the ABCA4 gene. Compound heterozygosity/homozygosity could be confirmed in seven of these patients. In accordance with the hypothesis by Maugeri et al.,9 the presence of the severe mutations in our patients would strongly suggest a cone–rod dystrophy or an atypical retinitis pigmentosa phenotype, thus reflecting the greatly reduced or even absent functional activity of ABCA4. A review of the clinical data, however, verified that all 14 patients presented with typical symptoms of STGD1 disease with age of onset ranging between 4 and 60 years. More precisely, eight patients presented with initial symptoms at the age of 11 or younger; two patients were under 20; and the remaining four patients were 30 to 60 years old when they first noticed vision problems. In another report, one-third of the nontruncating ABCA4 alleles were found to cause more severe disease than premature truncations.10 Thus, specific common ABCA4 variants may alleviate deleterious mutations leading to asymptomatic, mild, or less severe disease and vice versa. Taken together, future studies aiming at investigation of the influence of common variations and rare mutations on disease risk are necessary, ideally in a prospective cohort setting. 
In summary, despite certain shortcomings in terms of possible uncertainties in clinical diagnosis and the paucity of detecting large copy number variations (CNVs) or mutations in noncoding regions, the present study further broadens our knowledge about the genetic basis of STGD1 disease. First, we expanded the mutational spectrum of STGD1 by identifying an additional 48 novel pathogenic/likely pathogenic gene variants in ABCA4. Our findings also add to the clarification of the role of deep intronic variants in putative alternative exons within intron 30 and 36, which seem to represent only a minor proportion of disease-associated variants. Finally, an in-depth analysis of common variants at the ABCA4 gene locus revealed risk-altering properties for several common variants, particularly in those patients who carry only a single disease-associated ABCA4 allele. Furthermore, GRS data can be used to ascertain the risk of a patient with a single ABCA4 mutation and may further refine our understanding of ABCA4-associated retinal pathology. 
Acknowledgments
The authors thank the International AMD Genomics Consortium (IAMDGC, http://eaglep.case.edu/iamdgc_web/) for providing the genotypes in the ABCA4 gene for our control samples. They also thank Jennifer Loewen-Horsch and Romy Walker for their help with the in vitro splicing assays and intronic sequencing. 
Supported in part by grants from the ProRetina Foundation Germany and the Federal Ministry of Education and Research (BMBF) (Ref. IDs 01GM0851 and 01GM1108B). The samples were genotyped as part of the IAMDGC exome-chip project supported by CIDR contract HHSN268201200008I and funded by EY022310 (to Jonathan L. Haines) and 1X01HG006934-01 (to Gonçalo R. Abecasis). 
Disclosure: H.L. Schulz, None; F. Grassmann, None; U. Kellner, None; G. Spital, None; K. Rüther, None; H. Jägle, None; K. Hufendiek, None; P. Rating, None; C. Huchzermeyer, None; M.J. Baier, None; B.H.F. Weber, None; H. Stöhr, None 
References
Stargardt KB. Über familiäre, progressive Degeneration in der Makulagegend des Auges. Albrecht von Graefes Archiv für Ophthalmologie. 1909; 71: 534– 550.
Fishman GA, Farbman JS, Alexander KR. Delayed rod dark adaptation in patients with Stargardt's disease. Ophthalmology. 1991; 98: 957– 962.
Allikmets R, Shroyer NF, Singh N, et al. Mutation of the Stargardt disease gene (ABCR) in age-related macular degeneration. Science. 1997; 277: 1805– 1807.
Cremers FP, van de Pol DJ, van Driel M, et al. Autosomal recessive retinitis pigmentosa and cone-rod dystrophy caused by splice site mutations in the Stargardt's disease gene ABCR. Hum Mol Genet. 1998; 7: 355– 762.
Klevering BJ, Deutman AF, Maugeri A, Cremers FPM, Hoyng CB. The spectrum of retinal phenotypes caused by mutations in the ABCA4 gene. Graefes Arch Klin Exp Ophthalmol. 2005; 243: 90– 100.
Papermaster DS, Schneider BG, Zorn MA, Kraehenbuhl JP. Immunocytochemical localization of a large intrinsic membrane protein to the incisures and margins of frog rod outer segment disks. J Cell Biol. 1978: 78: 415– 425.
Quazi F, Molday RS. ATP-binding cassette transporter ABCA4 and chemical isomerization protect photoreceptor cells from the toxic accumulation of excess 11-cis-retinal. Proc Natl Acad Sci U S A. 2014; 111: 5024– 5029.
Molday RS. ATP-binding cassette transporter ABCA4: molecular properties and role in vision and macular degeneration. J Bioenerg Biomembr. 2007; 39: 507– 517.
Maugeri A, van Driel MA, van de Pol DJ, et al. The 2588G>C mutation in the ABCR gene is a mild frequent founder mutation in the Western European population and allows the classification of ABCR mutations in patients with Stargardt disease. Am J Hum Genet. 1999; 64: 1024– 1035.
Cideciyan AV, Swider M, Aleman TS, et al. ABCA4 disease progression and a proposed strategy for gene therapy. Hum Mol Genet. 2009; 18: 931– 941.
Lewis RA, Shroyer NF, Singh N, et al. Genotype / phenotype analysis of a photoreceptor-specific ATP-binding cassette transporter gene, ABCR, in Stargardt disease. Am J Hum Genet. 1999; 64: 422– 434.
Rivera A, White K, Stöhr H, et al. A comprehensive survey of sequence variation in the ABCA4 (ABCR) gene in Stargardt disease and age-related macular degeneration. Am J Hum Genet. 2000; 4: 800– 813.
Briggs CE, Rucinski D, Rosenfeld PJ, Hirose T, Berson EL, Dryja TP. Mutations in ABCR (ABCA4) in patients with Stargardt macular degeneration or cone-rod degeneration. Invest Ophthalmol Vis Sci. 2001; 42: 2229– 2236.
Chacón-Camacho OF, Granillo-Alvarez M, Ayala-Ramírez R, Zenteno JC. ABCA4 mutational spectrum in Mexican patients with Stargardt disease: identification of 12 novel mutations and evidence of a founder effect for the common p.A1773V mutation. Exp Eye Res. 2013; 109: 77– 82.
Riveiro-Alvarez R, Lopez-Martinez M-A, Zernant J, et al. Outcome of ABCA4 disease-associated alleles in autosomal recessive retinal dystrophies: retrospective analysis in 420 Spanish families. Ophthalmology. 2013; 120: 2332– 2337.
Zernant J, Collison FT, Lee W, et al. Genetic and clinical analysis of ABCA4-associated disease in African American patients. Hum Mutat. 2014; 35: 1187– 1194.
Jiang F, Pan Z, Xu K, et al. Screening of ABCA4 gene in a Chinese cohort with Stargardt disease or cone-rod dystrophy with a report on 85 novel mutations. Invest Ophthalmol Vis Sci. 2016; 57: 145– 152.
Bocquet B, Lacroux A, Surget M-O, et al. Relative frequencies of inherited retinal dystrophies and optic neuropathies in Southern France: assessment of 21-year data management. Ophthalmic Epidemiol. 2013; 20: 13– 25.
Zernant J, Schubert C, Im KM, et al. Analysis of the ABCA4 gene by next-generation sequencing. Invest Ophthalmol Vis Sci. 2011; 52: 8479– 8487.
Zernant J, Xie YA, Ayuso C, et al. Analysis of the ABCA4 genomic locus in Stargardt disease. Hum Mol Genet. 2014; 23: 6797– 6806.
Zaneveld J, Siddiqui S, Li H, et al. Comprehensive analysis of patients with Stargardt macular dystrophy reveals new genotype-phenotype correlations and unexpected diagnostic revisions. Genet Med. 2015; 17: 262– 270.
Fritsche LG, Fleckenstein M, Fiebig BS, et al. A subgroup of age-related macular degeneration is associated with mono-allelic sequence variants in the ABCA4 gene. Invest Ophthalmol Vis Sci. 2012; 53: 2112– 2118.
Westeneng-van Haaften SC, Boon CJF, Cremers FPM, Hoefsloot LH, den Hollander AI, Hoyng CB. Clinical and genetic characteristics of late-onset Stargardt's disease. Ophthalmology. 2012; 119: 1199– 1210.
Braun TA, Mullins RF, Wagner AH, et al. Non-exomic and synonymous variants in ABCA4 are an important cause of Stargardt disease. Hum Mol Genet. 2013; 22: 5136– 5145.
Bauwens M, De Zaeytijd J, Weisschuh N, et al. An augmented ABCA4 screen targeting noncoding regions reveals a deep intronic founder variant in Belgian Stargardt patients. Hum Mutat. 2015; 36: 39– 42.
Bax NM, Sangermano R, Roosing S, et al. Heterozygous deep-intronic variants and deletions in ABCA4 in persons with retinal dystrophies and one exonic ABCA4 variant. Hum Mutat. 2015; 36: 43– 47.
Rivera A, Fisher SA, Fritsche LG, et al. Hypothetical LOC387715 is a second major susceptibility gene for age-related macular degeneration, contributing independently of complement factor H to disease risk. Hum Mol Genet. 2005; 14: 3227– 3236.
Bhasi A, Pandey RV, Utharasamy SP, Senapathy P. EuSplice: a unified resource for the analysis of splice signals and alternative splicing in eukaryotic genes. Bioinformatics. 2007; 23: 1815– 1823.
Schroeder C, Stutzmann F, Weber BH, Riess O, Bonin M. High-throughput resequencing in the diagnosis of BRCA1/2 mutations using oligonucleotide resequencing microarrays. Breast Cancer Res Treat. 2010; 122: 287– 297.
Fritsche LG, Igl W, Bailey JN, et al. A large genome-wide association study of age-related macular degeneration highlights contributions of rare and common variants. Nat Genet. 2016; 48: 134– 143.
Grassmann F, Fritsche LG, Keilhauer CN, Heid IM, Weber BH. Modelling the genetic risk in age-related macular degeneration. PLoS One. 2012; 7: e37979.
Yekutieli D, Benjamini Y. The control of the false discovery rate in multiple testing under dependency. Ann Stat. 2001; 29: 1165– 1188.
Church DM, Stotler CJ, Rutter JL, Murrell JR, Trofatter JA, Buckler AJ. Isolation of genes from complex sources of mammalian genomic DNA using exon amplification. Nat Genet. 1994; 6: 98– 105.
Sun H, Smallwood PM, Nathans J. Biochemical defects in ABCR protein variants associated with human retinopathies. Nat Genet. 2000; 26: 242– 246.
Suárez T, Biswas SB, Biswas EE. Biochemical defects in retina-specific human ATP binding cassette transporter nucleotide binding domain 1 mutants associated with macular degeneration. J Biol Chem. 2002; 277: 21759– 21767.
Thanaraj TA, Clark F. Human GC-AG alternative intron isoforms with weak donor sites show enhanced consensus at acceptor exon positions. Nucleic Acids Res. 2001; 29: 2581– 2593.
Farkas MH, Grant GR, White J, Sousa ME, Consugar MB, Pierce E. Transcriptome analyses of the human retina identify unprecedented transcript diversity and 3.5 Mb of novel transcribed sequence via significant alternative splicing and novel genes. BMC Genomics. 2013; 14: 486.
Brandl C, Grassmann F, Riolfi J, Weber BH. Tapping stem cells to target AMD: challenges and prospects. J Clin Med. 2015; 4: 282– 303.
Miraldi Utz V, Coussa RG, Marino MJ, et al. Predictors of visual acuity and genotype-phenotype correlates in a cohort of patients with Stargardt disease . Br J Ophthalmol. 2014; 98: 513– 518.
Ścieżyńska A, Oziębło D, Ambroziak AM, et al. Next-generation sequencing of ABCA4: high frequency of complex alleles and novel mutations in patients with retinal dystrophies from Central Europe. Exp Eye Res. 2016; 145: 93– 99.
Yatsenko AN, Shroyer NF, Lewis RA, Lupski JR. An ABCA4 genomic deletion in patients with Stargardt disease. Hum Mutat. 2003; 21: 636– 644.
Xin W, Xiao X, Li S, Jia X, Guo X, Zhang Q. Identification of genetic defects in 33 probands with Stargardt disease by WES-based bioinformatics gene panel analysis. PLoS One. 2015; 10: e0132635.
Pan Q, Shai O, Lee LJ, Frey BJ, Blencowe BJ. Deep surveying of alternative splicing complexity in the human transcriptome by high-throughput sequencing. Nat Genet. 2008; 40: 1413– 1415.
Aguirre-Lamban J, González-Aguilera JJ, Riveiro-Alvarez R, et al. Further associations between mutations and polymorphisms in the ABCA4 gene: clinical implication of allelic variants and their role as protector/risk factors. Invest Ophthalmol Vis Sci. 2011; 52: 6206– 6212.
Guymer RH, Héon E, Lotery AJ, et al. Variation of codons 1961 and 2177 of the Stargardt disease gene is not associated with age-related macular degeneration. Arch Ophthalmol. 2001; 119: 745– 751.
Figure 1
 
Frequency of common ABCA4 variants in patients with Stargardt disease and European non-Finnish controls. Six variants that were found to be significantly enriched in either patients or controls are framed.
Figure 1
 
Frequency of common ABCA4 variants in patients with Stargardt disease and European non-Finnish controls. Six variants that were found to be significantly enriched in either patients or controls are framed.
Figure 2
 
Genetic risk score (GRS) for patients and controls stratified according to the number of ABCA4 mutations or disease status. An alpha estimate of −1.0503 was used to center the GRS on zero for the controls. Red bars represent the mean GRS and confidence intervals.
Figure 2
 
Genetic risk score (GRS) for patients and controls stratified according to the number of ABCA4 mutations or disease status. An alpha estimate of −1.0503 was used to center the GRS on zero for the controls. Red bars represent the mean GRS and confidence intervals.
Table 1
 
Novel Rare Sequence Variants in the ABCA4 Gene
Table 1
 
Novel Rare Sequence Variants in the ABCA4 Gene
Table 2
 
Functional Analysis of Putative Splice Mutations in the ABCA4 Gene
Table 2
 
Functional Analysis of Putative Splice Mutations in the ABCA4 Gene
Table 3
 
Deep Intronic Variants in the ABCA4 Gene
Table 3
 
Deep Intronic Variants in the ABCA4 Gene
Table 4
 
Association of Common ABCA4 Variants With Stargardt Disease
Table 4
 
Association of Common ABCA4 Variants With Stargardt Disease
Supplement 1
Supplement 2
Supplement 3
Supplement 4
Supplement 5
Supplement 6
Supplement 7
×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×