Expanding the Genetic Landscape of Usher-Like Phenotypes

Abstract

Purpose: Usher syndrome (USH) is a rare disorder characterized by retinitis pigmentosa (RP) and sensorineural hearing loss. Several genes are responsible for the disease, but not all cases are explained by mutations in any of these, supporting the fact that there remain other unknown genes that have a role in the syndrome. We aimed to find the genetic cause of presumed USH patients lacking pathogenic mutations in the known USH genes.

Methods: Whole exome sequencing was performed on a priori USH-diagnosed subjects from nine unrelated families, which had shown negative results for an USH-targeted panel in a previous study.

Results: We identified possible pathogenic variants in six of the studied families. One patient harbored mutations in REEP6 and TECTA, each gene tentatively causative of one of the two main symptoms of the disease, mimicking the syndrome. In three patients, only the retinal degeneration causative mutations were detected (involving EYS, WDR19, and CNGB1 genes). Another family manifested a dementia-linked retinal dystrophy dependent on an allele dosage in the GRN gene. Last, another case presented a homozygous mutation in ASIC5, a gene not yet associated with USH.

Conclusions: Our findings demonstrate that pending cases should be clinically and genetically carefully assessed, since more patients than expected may be either related phenocopies or affected by a more complex disease encompassing additional symptoms rather than classical USH.

Usher syndrome (USH) is considered the principal hereditary disorder involving sensorineural bilateral hearing loss (HL) and retinitis pigmentosa (RP). It is an autosomal recessive disease, with a prevalence up to 6.2/100.000.¹ There are three clinical types of USH depending on the severity and progression of the symptoms. Type 1 (USH I) is the most severe category and it is characterized by a severe-profound congenital HL and onset of RP during childhood. Additionally, patients with this type also present with vestibular dysfunction. Type 2 (USH II) is the most common type and the patients display moderate-to-severe congenital hearing impairment and adult-onset RP. Type 3 (USH III) is defined by progressive postlingual HL, variable onset of RP, and sometimes also absence of vestibular function.¹ 

USH is a genetically heterogeneous disease, since there are currently 10 responsible genes, namely MYO7A, USH1C, CDH23, PCDH15, USH1G, CIB2, USH2A, ADGRV1, WHRN, and CLRN1.² In addition, another five genes also have been associated with the disease in some studies, yet it is more sensible still to consider them as causative of a non-USH form of deaf blindness, given that the symptoms of these cases do not fully match the usual clinical picture and some even manifest other comorbidities: ARSG, CEP250, CEP78, ESPN, HARS, and PDZD7.^3–8 

Consequently, the gold standard to date for the molecular diagnosis of USH patients is by means of targeted exome sequencing.^9–11 However, not all USH cases are explained by mutations in the coding regions of any of the USH genes, implying that there remain other unknown genes causing the disease or pathogenic variants in the noncoding regions of the USH genes. 

Furthermore, some cases could result from concurrence of the two main symptoms caused by mutations in separate genes; therefore, mimicking the syndrome,^12,13 or due to other syndromes that also associate a retinal dystrophy (RD) and HL, such as polyneuropathy, hearing loss, ataxia, retinitis pigmentosa and cataracts (PHARC) or Heimler syndrome.^11,14,15 Taking this prospect into account, the genes responsible for nonsyndromic RP (nsRP) and HL independently should be considered, as well as those associated with more complex diseases with those symptoms among its spectrum. 

We unveil the genetic landscape underlying cases with an USH-like phenotype yet negative for the USH genes, and unveil new insights of either new candidate genes or to resolve the genetic puzzle. For this purpose, we performed whole exome sequencing (WES) in probands from nine unrelated families that had been screened previously through our implemented targeted panel.¹¹ 

Methods

This study was approved by the Hospital La Fe and the FJD Ethics Committee, according to the tenets of the Declaration of Helsinki, and all research was performed in accordance with the relevant guidelines and regulations. Authorizations from all the patients and the participating relatives were obtained by signing an informed consent form. 

Subjects

A total of 10 Spanish patients with suspected Usher syndrome from nine unrelated families were selected for the WES analysis. These probands consisted of unsolved cases from a previous screening assay through a custom USH targeted panel.¹¹ Subjects from families FRP-480, FRP-581, FRP-613, FRP-616, and FRP-621 were recruited by the Hospital La Fe. Patients belonging to families FRP-70, FRP-426, FRP-520, and FRP-539 were referred from the Fundación Jiménez Díaz University Hospital. 

WES and Data Analytic Workflow

Genomic DNA (gDNA) from the probands was obtained and purified using standard procedures. The concentration of the resulting DNA samples was determined with Nanodrop and Qubit fluorometer (Thermo Fisher Scientific, Waltham, MA, USA). WES was performed in all USH probands, except for a large family where only one affected sibling was sequenced (FRP-480). Sequencing and data processing before the filtering was done at the Center for Applied Genomics of The Children's Hospital of Philadelphia. Exome enrichment of the samples was performed using Agilent Select Human All Exome V5 kit. 

The Dynamic Read Analysis for GENomics platform (DRAGEN, version 2.3.1; Illumina, San Diego, CA, USA) was used for the read mapping, duplicate reads marking, and haplotype-based variant calling in regards of the GRCh37 with decoy sequences (hs37d5) reference human genome. The ANNOVAR tool was used for the variant annotation.¹⁶ 

The resulting variants were filtered based on cutoff for the maximum minor allele frequency ≤ 1% in the International Genome Sample Resource (1000 Genomes Project, 1000g), gnomAD, and ExAC. Nonsense, frameshift, splice-site, and pathogenic-predicted missense and synonymous variants were prioritized. The pathogenicity of missense and synonymous variants was determined according to the prediction of several in silico tools depending on the type of mutation. Missense mutations were evaluated using the scores provided by SIFT, PolyPhen, and Mutation Taster. Synonymous variants were considered when located near the exon boundaries and with a predicted effect by the Human Splicing Finder 3.1. Additional deleteriousness analyses based on conservation parameters were performed for the missense variants, specially in regards to the assessment of novel amino acid changes. To this end, GERP++, phyloP, and SiPhy score values were taken into account,^17–19 and alignment variant-enclosing amino acid sequence across several species was performed. The pathogenicity interpretation of all the novel variants presented in this study was rated according to the American College of Medical Genetics and Genomics (ACMG). 

Additionally, a focalized survey was done, namely of variants already described as pathogenic in ClinVar or the Human Gene Mutation Database (HGMD), and of genes matching the phenotype. The latter assessment was performed using the genes associated with either nsRP, HL, or syndromes involving those symptoms, as listed in the Retinal Information Network (RetNet), the Deafness Variation Database, and OMIM. 

Furthermore, an analysis to explore possible copy number variations (CNVs) was conducted using the DECoN tool.²⁰ 

In general, variants concordant with the expected inheritance were more exhaustively scrutinized, but no genetic scenario was discarded. 

The validation of the candidate mutations and segregation analysis were performed through conventional Sanger sequencing. Primer sequences are available upon request. 

Mitochondrial Mutational Survey

The genetic causes producing sensorineural HL are very variable, and many cases presenting the symptom are due to mitochondrial mutations. The m.1555A>G and m.1494C>T in the MT-RNR1 gene are a frequent cause of aminoglycoside-induced and nonsyndromic HL.²¹ Another common mitochondrial variant associated to hearing impairment is m.3243A>G in MT-TL1, which usually leads to syndromic HL and to a lesser extent nonsyndromic HL.^22,23 In those cases where the HL was not explained by any of the variants found after the high-throughput sequencing screening, an additional targeted analysis through direct Sanger sequencing was performed looking for said deafness-associated mitochondrial variants, which were not contemplated in the WES design used in this study. 

Results

Six of the nine families investigated harbored either the causative variants in known disease-associated genes, or candidate mutations in new genes suggesting a possible role in the phenotype (see the pertinent pedigrees in Fig. 1). Basic clinical data were obtained before the variant analysis; however, the information was extended afterwards to check if the findings matched the related phenotype, as well as for those cases where the causative mutation was not detected firsthand. For an overview of the clinical evidences see Table 1. All causative mutations and the proposed putative pathogenic variants detected in this study are available in Table 2. Information about the conservation of the missense changes and ACMG classification of the novel variants is available at Supplementary Tables S1, S2, and S3, respectively. 

Figure 1

 

Pedigrees of the families with genetic findings. Left dark filling indicates affected by HL. Right dark filling indicates affected by RP. *Affected by dementia. Bracketed mutations are inferred from the confirmed genotype of the descendants.

Table 1

 

Clinical Phenotype of the Subjects Presenting With Symptoms

 

Table 2

 

Variants of Interest Found in This Study

 

Family FRP-70

The sporadic proband of this consanguineous family was referred as a typical USH I patient, as he presented with profound congenital HL and early onset RP. The homozygous nonsense variant c.58A>T (p.Lys20*) in ASIC5 (OMIM: 616693), a gene of unknown function, was identified in the patient and cosegregation among the family members was confirmed (Fig. 1). The variant was absent in all the databases (1000 Genomes Project, ExAC, gnomAD, dbSNP), albeit for a European (non-Finnish) carrier in heterozygosis among the gnomAD database excluded from the control dataset, accounting for an allele frequency of 0.000004. 

Family FRP-426

The proband from this family was referred to us as having USH II, since he presented with moderate-to-severe congenital HL and late-onset RP. This is a consanguineous family, since the parents of the patient are second cousins, and there were no other RP affected cases in the family. 

Analysis of the WES results revealed the novel homozygous pathogenic (as per AMCG standards) mutation c.598+1delG in the REEP6 gene (OMIM: 609346),²⁴ which is described as responsible for autosomal recessive RP (arRP) and cosegregated with this symptom in the family. 

In addition, we detected the heterozygous small deletion c.5383+5_5383+8delGTGA in TECTA (OMIM: 602574),^25–27 a gene causative of autosomal dominant and recessive deafness. After these results, new family assessment of HL demonstrated that the father, and the paternal aunt and grandfather also presented with deafness (Fig. 1). 

Family FRP-480

The referred patient from this family (II:9) presented with late-onset retinal degeneration with a very rapid progression and HL. In addition, she also suffered from an uncharacterized dementia starting in her seventies, which was referred to as severe Alzheimer's disease (AD). Analysis of the sequencing results exposed the splice-site mutation c.1414-1G>T in GRN (OMIM: 138945) in the homozygous state, a gene mainly associated with frontotemporal dementia (FTD) or neuronal ceroid lipofuscinosis (NCL) depending on the allele dosage.²⁸ Members within this same family presented with variable clinical manifestations (Table 1; Fig. 1). Both progenitors (I:1 and I:2) must have been obligate carriers of the mutation. The father died of lung cancer at 55 years of age, making impossible to know if he would have manifested the disease later in his life. The mother did not display any of the visual or neurologic symptoms, yet two of her brothers had manifested behavior changes compatible with FTD (not shown in the pedigree). Concerning the sibs, clinical history and the genetic screening were hampered by the fact that the majority were deceased, but some genotypes could be inferred by testing the alleles of the descendants. The only to date completely unaffected sister (II:11) also is a carrier of the mutation in progranulin. However, she is presently 74 years old, which is more or less the same age when most of her siblings started to manifest the dementia symptoms, and therefore, her health status remains unclear. 

Family FRP-520

Both affected siblings were included in the sequencing study, which revealed the homozygous mutation c.2939A>T (p.Asn980Ile) in the CNGB1 gene (OMIM: 600724), previously reported as causative of arRP.²⁹ No likely pathogenic variants explaining the HL were found. 

Family FRP-539

The only affected member of this family was diagnosed with RP at 28 years of age, but he stated he began noticing the beginning of the symptoms in childhood (Table 2). Sequencing analysis revealed that he was compound heterozygous for the two novel mutations c.1983-2A>T and c.2782A>T (p.Ile928Phe) in WDR19 (OMIM: 608151), a gene mostly related to multisyndromic RP.³⁰ In concordance with the evolutionary constraint scores obtained by the conservation-based programs (Supplementary Table S1), the interspecies alignment of the amino acid sequence comprising the novel missense variant revealed that the position is highly conserved (Supplementary Table S2). The segregation analysis could not be fulfilled, as none of the other family members was willing to collaborate in the study. As a result, only the splicing variant was ranked as pathogenic according to ACMG, but not so the c.2782A>T missense variant, which still must be classified as of unknown significance (Supplementary Table S3). 

Family FRP-621

The proband from this family began to notice the typical RP symptoms in his twenties, but he was not clinically diagnosed until he was 31 years old. In addition, he suffered from a mild sensorineural HL at high frequencies of adult onset. Variants c.3567delA (p.Gly1190Aspfs*39) and c.8779T>C (p.Cys2927Arg) in EYS (OMIM: 612424) were identified, which already are described as causative of arRP.^31,32 Segregation analysis was confirmed, except for the youngest sibling (III:6), who initially was referred to as unaffected (Fig. 1) and subsequently was determined to be a carrier of both mutations. 

No likely pathogenic variants explaining the HL were found. 

Discussion

USH Phenocopies

Family FRP-426

For this case, the distinct genetic origins of the RP and HL have been ascertained by the mutations identified in REEP and TECTA. Regarding the latter, the dominant causality of the splicing variant may prompt a distrust, as the general train of thought is that missense variants in the gene are the ones responsible for this type of inheritance pattern, whereas truncating mutations are characteristic of a recessive trait. It should be noted that a synonymous and another splice-site variant in the same exon-intron boundary have been reported to likewise result in the skipping of exon 16, in patients featuring a dominant HL phenotype.^25,33 The suppression of this exon, though, has been proved to maintain the reading frame.²⁷ Therefore, the aforementioned changes actually stand as noninactivating TECTA mutations, consistently with the initial mainstream theory. In addition, HL due to mutations in TECTA display a distinct audiometric profile that matches the audiogram of the patient (Fig. 2).³⁴ 

Figure 2

 

Audiograms from patient III:1 (FRP-426) at two different times. Results show a moderate-to-severe bilateral HL affecting all frequencies, that has remained stable in a 33-year time lapse. The audiograms present a “U” shape, which is a distinctive pattern associated with the TECTA gene. yo, years old; dB, decibels; Hz, Hertz.

These revelations supported the causative role of this variant for the deafness phenotype. 

Family FRP-621

Regarding the apparent genotype-phenotype discrepancy of the one sib carrying both pathogenic mutations, the reason might be that, due to his young age, either the RP has not yet debuted or that he has unperceived inceptive symptoms. Unfortunately, he was not willing to undergo ophthalmologic evaluations, thus making it impossible to determine his genuine phenotype. 

With respect to the genetically unsolved HL, other possibilities must be contemplated. Considering the mild condition that only comprises the high-frequencies, it could be a case of presbycusis, which can begin as early as the third decade of life and for which a great number of risk factors, not only genetic, are at stake.³⁵ Indeed, the subject suffered from loud occupational noise exposure; therefore, the deafness in this case also could be noise-induced. 

Partially Misdiagnosed Cases

Family FRP-480

The c.1414-1G>T variant in the progranulin gene was detected in the sequenced patient of this family. Mutations in this gene generally are reported to cause FTD when presented in heterozygosis, and adult onset NCL if both alleles are compromised.^28,36 The course of FTD includes cognitive and behavioral changes, whereas NCL is a more severe condition that usually involves additional neuropathologic features, such as motor deterioration or epilepsy. Apart from the severe neurologic implications, a common symptom of NCL is retinal degeneration,^36–38 which matches the retinopathy presented in our index case, who manifested a severe RP-like phenotype that quickly ended up compromising her central vision as well. 

The c.1414-1G>T mutation has been reported previously in the heterozygous state in a study where the subjects were first diagnosed with late-onset AD, though they were later reassessed by the authors as FTD cases due to the genetic findings.³⁹ The family presented in this study is reminiscent of those misdiagnoses, as it is not completely clear whether the symptomatic individuals suffer from either AD or FTD. Due to the molecular findings and symptoms of other relatives, we believe it to be FTD. Indeed, more cases have been reported to be inaccurately characterized with a posterior alternate diagnosis.⁴⁰ Therefore, the boundaries between these diseases should not be strict, at least when progranulin is involved. 

Taking into account the homozygosity of the variant, the proband manifested a rather mild phenotype when compared to other subjects with NCL carrying homozygous null mutations in the GRN, since the retinal and neurologic degeneration appeared in later stages. However, progranulin deficiency also has been repetitively associated with other related diseases, such as AD, amyotrophic lateral sclerosis, or corticobasal syndrome, depending on the allele dosage and type of mutation. Another study described a case of FTD with a mutation affecting the same splice-site domain, namely c.1414-2A>G.⁴¹ Remarkably, the investigators performed ex vivo splicing analysis of the variants and determined a pathogenic effect, yet the aberrant splicing process had only a partial effect. Thus, in vivo exon skipping due to homozygous mutations in that region could not be absolute in all cases; therefore, leading to a rather milder version of the phenotype. Not only has GRN a pleiotropic effect, but its age at onset is variable even within members of the same family,⁴² as occurs with the one presented herein. Other factors might be conditioning the appearance of the disease, and supporting this is the fact that some studies have nominated genetic modifiers.⁴³ In addition, the gene is estimated to have 90% penetrance by an advanced age, leaving a small amount of carriers possibly unaffected.⁴⁴ 

In brief, we believe that this particular mutation is responsible for standard FTD and, when both alleles are affected, for mild adult onset NCL. 

Nonetheless, this assumption should be considered carefully, since most of the clinical information comes from a retrospective interrogation of the relatives which could not be precise. Unfortunately, because this was an old family in which most of the members were deceased and some of the descendants were unwilling to participate, we could not perform a more extensive study that could have provided more insights. 

As for the hearing impairment affecting two of the sibs, it seems that the feature segregates independently from the other presented symptoms. We did not find any likely causative variant, but in view of the scarce clinical information, it could very well be due to presbycusis or other nongenetic factors. Even more, hearing deficits have been linked to patients with dementia,^45,46 so the progranulin deficiency also may be responsible for the HL in these cases. 

Family FRP-520

Due to the lack of genetic findings regarding the hearing impairment in this family, further clinical inquiries were made. We discovered that the index case (II:3) had been misdiagnosed as having USH, since the proband did not suffer from sensorineural HL, but rather from unilateral HL characterized as Meniere's syndrome, and the other affected brother did not manifest any hearing impairment at all. Therefore, this case finally was concluded as typical arRP due to the aforementioned mutations in CNGB1. 

Family FRP-539

The WDR19 gene has been reported as causative of a broad range of ciliopathies with RP, such as nephronophthisis and Senior-Løken syndrome, but in the last years it has been associated with nonsyndromic RP.^47,48 Again, no additional mutations explaining the HL were detected, neither in new nor in deafness-associated genes. A new interview with the patient revealed that this was another misdiagnosed case, since HL had been a self-diagnosed perception that was discarded after audiological examinations. The clinical reports of the patients and personal interrogation do not state any renal pathologies. For all these reasons, and cautious of the ACMG assessment of the missense variant, we considered this another putative solved case of nonsyndromic arRP caused by the compound heterozygous mutations in the WDR19 gene, which further supported the statements of previous studies of the gene being responsible for a wide phenotypic spectrum. In this way, there already was enough evidence to consider WDR19 not limited to nephropathy-associated RDs, but also responsible for isolated RP. 

Unsolved Cases

Family FRP-70

The study proband presented a stop gained mutation in the gene encoding for protein member 5 of the acid-sensing ion channel family. Not a single homozygous loss-of-function (LoF) variant in that gene was present in the public databases from control subjects, suggesting that truncation of the protein probably has a visible pathogenic impact in the phenotype. 

Acid-sensing ion channels (ASICs) are proteins that belong to the amiloride-sensitive Na+ channel and degenerin (NaC/DEG) family, comprising five paralogous genes (ASIC1, ASIC2, ASIC3, ASIC4, and ASIC5).⁴⁹ These genes are preserved among the chordate species and are expressed mainly in the central and peripheral nervous system,⁵⁰ being ASIC5 specific to a type of cells from the vestibulocerebellum and dorsal cochlear nucleus identified as unipolar brush cells, according to a gene reporter study in mice.⁵¹ Previous studies in rodents showed expression of the gene predominantly in brain, liver, and intestine.^52,53 Those findings neither support nor contradict the pathogenic role insinuated in this study, since the molecular mechanism is yet unknown and it could be involved in several pathways. In fact, other USH genes also are expressed in the intestine and brain,² as well as other ciliary tissues.⁵⁴ 

It also is noteworthy that other genes of this family of proteins, particularly ASIC2 and ASIC3, have been found to contribute to retinal and auditory function in animal models^55–58; thus, providing more hints of a possible equal role of ASIC5. 

Furthermore, ASICs also have been associated to mechanosensation and nociception,⁵⁰ in keeping with some other USH causative genes that have been proven to be involved in these processes.^59,60 

Apart from this truncating protein, a missense variant cosegregating in the family was detected and also should be contemplated (Table 2). The change c.2885G>T (p.Arg962Leu) in PCDH15, a gene responsible for USH I, was detected in the homozygous state in the proband. However, this is a variant of uncertain significance, mostly predicted to be benign, and found in homozygosity in a control subject from the population databases. Moreover, conservation analyses (Supplementary Tables S1, S2) suggest that this position is subject of considerable substitution rates. Therefore, we considered that this variant should be ruled out as the possible causative variant. 

Due to the lack of more supporting data and despite the fact that no further convincing variants were found in the patient, we were unable to strictly propose the ASIC5 gene as candidate for an USH-like phenotype at the present time. Nevertheless, it indeed could be responsible for the phenotype, and for that reason, we have highlighted this case to be of help to other present and future studies. We believe that it would be very interesting to test the functional role of this gene in an animal model if more cases harboring mutations in ASIC5 are found. 

Other Unexplained Cases

Three additional cases in this study could not be solved, namely FRP-581, FRP-613, and FRP-616. Probably, the genetic cause of these cases may be either due to variants in regions not contemplated in this study, such as introns or UTRs, or because the mutations are located in still uncharacterized genes. Though most of the genes are labeled to date, there presently is little to no information for many of them, which makes it difficult to designate them as responsible for a certain disease, despite the availability of other parameters, such as conservation, expression, variant segregation, and so forth. A sensible approach for another attempt to molecularly diagnose these patients could be the performance of whole genome sequencing. With this method, not only the (supposedly) noncoding regions are contemplated, but also the read-depth remains steadier across the covered genome given that no capture step is needed. This enables the survey of other types of large rearrangements, such as translocations and inversions, and statistically more robust CNVs tests. However, the massive output of variants requires that other members of the family are included in the study to help in the search by allowing segregation filtering, and samples of the relatives are not always available. 

Conclusions

USH is the major disorder accounting for combined hearing and visual impairment, yet it is not the only one. Other disorders also manifest with similar features, such as Wolfram, Alström, Norrie, Heimler, Refsum, or PHARC syndromes, although they usually present with additional symptoms. Therefore, a rigorous evaluation of each case is needed, clinically and genetically, to establish better phenotype-genotype correlations. Cases FRP-520 and FRP-539, both being inaccurately diagnosed, evidence this need of a robust clinical evaluation before a genetic screening. It not only is essential for selection of the best application, such as targeted or whole exome sequencing, but also for an efficient data curation based on the symptoms. The proficiency of the diagnosis, however, depends greatly on the health care system and resources of each country. Retinal dystrophies and hearing loss are presented in multiple forms on their own; hence, they become harder to characterize when syndromic and prone to be misdiagnosed, especially if patients do not undergo exhaustive examination by specialists. We believe that an interdisciplinary medical diagnosis of patients manifesting at least one rare feature is key to understanding the molecular role of new and already known disease-associated genes. The GRN (FRP-480) gene also stands as a fair example for this, since it is mainly associated with FTD and NCL, but this and other previous studies have supplied cases that suggest it to be responsible of an array of neurologic disorders or disease continuum. It seems clear that mutations in the granulin precursor are involved in neurodegenerative diseases, but further genotype-phenotype associations, or research on disease modifiers and gene penetrance, should be conducted. Altogether, the findings in GRN and WDR19 attest the need to reevaluate the phenotypic spectrum caused, depending on the type and state of the mutations. 

Despite RP having a low prevalence in the population, HL loss is not uncommon; therefore, the potential of uncovering an USH phenocopy must be kept in mind, as we showed with families FRP-426 and FRP-621. The latter lacks the molecular diagnosis for the HL, yet it remains unclear if the cause is genetic. Even so, hearing impairment is genetically extremely heterogeneous due to the variability of the causative elements and inheritance patterns known to date, making the diagnosis approach arduous. 

Nevertheless, it must be remarked that results suggest that a greater than expected number of cases may be due to a combination of mutated genes concealed as an USH appearance. The outcome of this study proved that WES is a powerful tool to genetically decipher unsolved or misdiagnosed USH cases, and the several genetic scenarios have brought new insights into the mutational spectrum of Usher-like phenotypes. 

Acknowledgments

The authors thank the patients involved in this study and the associations Retina CV, FARPE, and Asociación Retina Asturias. 

Supported by the Institute of Health Carlos III (ISCIII) and the European Development Regional Funds (Grants PI16/00539, PI16/00425), Fundación ONCE, by RAREGenomics, which is funded by Regional Government of Madrid (CAM; Grants B2017 and BMD37) and partially supported by FEDER (European Regional Development Fund), by a fellowship from the ISCIII and the European Social Fund (IFI14/00021; CFG), and by a senior postdoctoral contract from CIBERER (GGG). 

Disclosure: C. Fuster-García, None; G. García-García, None; T. Jaijo, None; F. Blanco-Kelly, None; L. Tian, None; H. Hakonarson, None; C. Ayuso, None; E. Aller, None; J.M. Millán, None