Clinical and Genetic Evaluation of a Cohort of Pediatric Patients with Severe Inherited Retinal Dystrophies

We performed a clinical and genetic characterization of a pediatric cohort of patients with inherited retinal dystrophy (IRD) to identify the most suitable cases for gene therapy. The cohort comprised 43 patients, aged between 2 and 18 years, with severe isolated IRD at the time of presentation. The ophthalmological characterization also included assessment of the photoreceptor layer integrity in the macular region (ellipsoid zone (EZ) band). In parallel, we carried out a targeted, next-generation sequencing (NGS)-based analysis using a panel that covers over 150 genes with either an established or a candidate role in IRD pathogenesis. Based on the ophthalmological assessment, the cohort was composed of 24 Leber congenital amaurosis, 14 early onset retinitis pigmentosa, and 5 achromatopsia patients. We identified causative mutations in 58.1% of the cases. We also found novel genotype-phenotype correlations in patients harboring mutations in the CEP290 and CNGB3 genes. The EZ band was detectable in 40% of the analyzed cases, also in patients with genotypes usually associated with severe clinical manifestations. This study provides the first detailed clinical-genetic assessment of severe IRDs with infantile onset and lays the foundation of a standardized protocol for the selection of patients that are more likely to benefit from gene replacement therapeutic approaches.


Introduction
Inherited retinal dystrophies (IRD) are a clinically and genetically heterogeneous group of disorders that affect the retina, and mainly photoreceptor cells. They represent the most frequent cause of blindness of genetic origin in the Western population [1]. To date, the clinical classification of IRDs relies on a variety of features, i.e., (a) primary target cell (cones vs. rods); (b) primarily targeted retinal region (macula vs. periphery); (c) age of onset and severity of visual dysfunction; and (d) absence or presence of extra-retinal clinical involvement (isolated vs. syndromic forms). As a result, IRDs are traditionally divided into retinitis pigmentosa (RP), achromatopsia (ACHM), and cone or cone-rod dystrophies. Among the syndromic forms, the Usher and Bardet-Biedl syndromes are the most frequent ones.
IRDs are characterized by a notable extent of genetic heterogeneity with approximately 200 responsible genes identified to date (http://www.sph.uth.tmc.edu/RetNet/). Such heterogeneity renders molecular diagnosis challenging. Next-generation sequencing (NGS) approaches, both targeted and whole exome-based [2], offer an effective solution to the problems related to the molecular characterization of highly genetically heterogeneous disorders, including IRDs. NGS-based approaches have been successfully applied to the study of the genetic basis of patients with specific forms of photoreceptor degenerations, such as RP, cone dystrophy, and Leber congenital amaurosis (LCA) [3]. Nevertheless, it is currently possible to determine the molecular defect underlying IRDs in about 50-70% of patients [4][5][6], which suggests the existence of additional genes responsible for these conditions. A significant number of IRD cases are characterized by infantile/juvenile onset, i.e., prior to 18 years of age. This is the case of LCA, early onset retinitis pigmentosa (EORP), and ACHM. However, to the best of our knowledge, there are no reports that describe clinical phenotypes related to genotypes in a cohort of IRD patients with infantile/juvenile onset. The absence of similar reports could be justified by the many difficulties hindering a precise classification of the specific disease type in pediatric patients. Fundamental tests for a proper clinical diagnosis, such as electroretinogram (ERG), visual field (VF), or optical coherence tomography (OCT), cannot be easily carried out in these patients. Furthermore, clinical features are often shaded in young patients.
Currently, there is no effective cure for IRDs. The experimental treatment that provided initial evidence of success and currently holds higher promises is gene therapy [7][8][9]. Gene therapy is particularly relevant for those forms in which the retina retains a significant degree of morphological and functional preservation [7][8][9]. Therefore, a detailed phenotypic and genotypic characterization of pediatric IRD patients is essential because the latter represent an important reservoir of cases with high potential of successful outcomes in gene therapy-based trials. To obtain information on both genetic and clinical features of early-onset IRD cases, we analyzed a cohort of Italian IRD patients of pediatric age. We performed NGS-based genetic testing and an extensive clinical evaluation including analysis of the macular thickness (MT) and the ellipsoid zone (EZ) band [10] by OCT. We believe that the results of our study will pave the way towards the definition of a standardized pipeline to select IRD patients with the highest potential of successful outcomes in gene therapy-based approaches.

Ophthalmological Analysis
A total of 43 Italian patients aged between 2 and 18 years, with isolated non-syndromic retinal degeneration at the time of presentation, were recruited for the study. The medical records of the patients were taken at the Referral Center for Inherited Retinopathies of the Eye Clinic of the Università degli Studi della Campania Luigi Vanvitelli from June 2013 to March 2015. Inclusion criteria were: disease onset ≤10 years of age (age of symptoms' onset), age ≤ 18 years, best corrected visual acuity (BCVA) ≤20/70, standard electroretinogram (ERG) abnormalities and macular thickness (MT) ≥100 µm in cases where ERG and OCT were performable.
Prior to enrollment in the study, all patients underwent full ophthalmological examination, which included: BCVA measured using LEA symbols in patients aged between 3-5 years or Snellen chart in patients >5 years old, Farnsworth D-15 color test, slit lamp anterior segment examination, fundus examination, Goldmann visual field examination, ERG and OCT. ERG was recorded according to the International Guidelines of the International Society of Clinical Electrophysiology of Vision (ISCEV) [11]. OCT was performed with the spectral domain OCT (SD-OCT) (Cirrus HD-OCT, Carl Zeiss, Dublin, CA, USA) by an experienced operator. The acquisition protocol comprised both a five-line raster scan and a macular cube scan pattern (512 × 128 pixels) in which a 6 × 6 mm region of the retina was scanned within a scan time of 2.4 s.

Clinical Diagnosis
Clinical diagnosis was performed according to the following criteria. Diagnostic criteria of LCA, according to Chacon-Camacho and Zenteno [12], included: (a) Functional signs: nystagmus already present at 6 weeks of life, photophobia, night blindness, oculodigital signs, sluggish or nearly absent pupillary responses, severe visual loss at birth; (b) Fundus appearance ranging from normal or mild retinal involvement (like "salt and pepper" dystrophy) to maculopathy or macular coloboma, bone-spicule pigment migration, marbleized fundus; (c) ERG: severely subnormal or non-detectable scotopic and photopic responses; (d) Age at diagnosis: at birth or shortly after birth [12].
Diagnostic criteria of EORP, according to Hamel [13], included: (a) Functional signs: night blindness, photophobia with visual acuity preserved in early and mid-stages; (b) Fundus appearance ranging from "salt and pepper" dystrophy to pigmentary deposits resembling bone spicules, initially in peripheral retina, attenuation of the retinal vessels, waxy pallor of the optic disc, and various degrees of retinal atrophy; (c) VF: patchy losses of peripheral vision evolving to ring shape scotoma, and eventually tunnel vision; (d) ERG: dramatic decrease in a-and b-wave amplitudes; ERG is usually unrecordable in scotopic conditions and photopic responses (30-Hz flickers, bright light) are markedly hypovolted; (e) Age: EORP is diagnosed when symptoms of mid-stage RP are already present at the age of two years [13].
Diagnostic criteria of ACHM, according Kohl et al [14], included: (a) Functional signs: nystagmus, photophobia, reduced or complete loss of color discrimination, reduced visual acuity, eccentric fixation; (b) Fundus appearance: ranging from normal to absent foveal reflex, pigment mottling, narrowing of the retinal vessels, frank atrophy of the retinal pigment epithelium (RPE) in the fovea; (c) VF: small central scotoma; (d) ERG: absent or markedly reduced photopic responses, normal or mildly abnormal scotopic responses; (e) OCT: variable degree of foveal hypoplasia, disruption and/or loss of inner/outer photoreceptor segment junction (IS/OS), now described as EZ band, and an attenuation of RPE layer [14].

Selection of RETplex Genes and Enrichment Procedures
In the RETplex targeted sequencing panel, we included all genes responsible for isolated forms of IRDs that were listed in the RETnet website (http://www.sph.uth.tmc.edu/RetNet/; accessed on 31 March 2014) as well as some genes responsible for both isolated and syndromic forms of IRD. The panel comprised all the coding exons of the above genes as well as the genomic regions covering some previously reported deep intronic mutations [15,16]. Finally, we also included the coding exons of genes and the precursor sequences of microRNAs that, based on our previous results [17,18], can exert a candidate pathogenic role in IRDs. The complete list of genes and sequence elements (n = 159) represented in the RETplex panel is reported in Table S1. Capture oligonucleotide probes covering the selected target regions were designed using the HaloPlex TM Target Enrichment System (Agilent Technologies Inc., Santa Clara, CA, USA) according to previously reported protocols [19].

Targeted NGS Analysis
Genomic DNA was extracted from peripheral blood using standard procedures. All procedures were approved by the Ethics Boards of the Università degli Studi della Campania Luigi Vanvitelli (Project No. 0006282/2015 approved on 17 March 2015) and adhered to the tenets of the Declaration of Helsinki. All samples were acquired after written informed consent was obtained from the patient or, in the case of children, their legal guardians. DNA quality was assessed and RETplex sequencing libraries were prepared as previously described [20]. Libraries were sequenced using the HiSeq1000 system (Illumina Inc., San Diego, CA, USA). The generated sequences were analyzed using an in-house developed pipeline [19]. Briefly, the generated paired sequencing reads were aligned to the reference genome (UCSC, hg19 build) using the Burrows-Wheeler Alignment (BWA) tool [21] and sorted with SAMtools [22] and Picard (http://picard.sourceforge.net). Genome Analysis Toolkit (GATK) [23] with parameters adapted to the Haloplex-generated sequences was then used to identify insertions-deletions (indel) and single nucleotide variants (SNV). The called SNV and indel variants were annotated using ANNOVAR [24] with: the relative position in genes using the RefSeq87 gene model, amino acid change, presence in dbSNP v137, frequency in the EXAC database (http://exac.broadinstitute. org), 1000 genomes project [25], presence in the Human Gene Mutation Database (HGMD) [26], Clinvar database [27], multiple cross-species conservation [28] and prediction scores of damage on protein activity [29][30][31][32]. The annotated variants were also checked for their presence in an internal variation database, which stores all the variations found in sequencing projects carried out in our Institute. The alignments at candidate positions were visually inspected using the Integrative Genomics Viewer (IGV). All the candidate variants identified were validated by Sanger sequencing.

Patient Selection
We selected for our study a cohort of 43 Italian IRD patients (representative of 41 different families) according to the inclusion criteria mentioned above. Based on a first-level ophthalmological assessment (see Materials and Methods), this cohort was composed of 24 patients affected by LCA (55.8%), 14 by EORP (32.6%), and 5 by ACHM (11.6%) ( Figure 1a). All patients were sporadic with the exception of two pairs of sibs.

RETplex Analysis
We set up a targeted NGS-based procedure (RETplex) to identify the genetic basis of disease in IRD patients. In particular, we included in the platform 159 genes, of which 137 with an already demonstrated pathogenic role and 20 genes with a candidate pathogenic role in IRDs (see Table S1). To generate the sequencing libraries, we used the HaloPlex TM Target Enrichment System (Agilent Technologies Inc.) that we have successfully used for other targeted sequencing efforts [19]. We first tested the efficacy of the procedure on a small training subset of three genomic DNAs from IRD patients with already known molecular defects. RETplex allowed us to successfully detect all the previously known mutations in the three patients analyzed.
We then applied the RETplex procedure to the entire cohort of 43 selected patients. In particular, we carried out the analysis on the 41 probands. In the two familial cases, the candidate pathogenic variants were then validated by Sanger sequencing in the affected sib not analyzed by RETplex. This analysis led to the complete identification of the presumably pathogenic variants in 25/43 cases (58.1%) ( Table 1). Interestingly, in three additional cases we found single heterozygous mutations in the CEP290 gene (Table S2). Since the latter variants were all displaying a loss-of-function effect and two of them were previously described as causative [33], we hypothesize that these three patients may harbor a second mutation in CEP290 that is not detectable using this approach, e.g., a deep intronic mutation or a large copy number variation.

ACHM Patients
Finally, the five ACHM cases had a mean age of 12.6 years (range between 10 and 16 years). The onset was around the first year of age. In all patients, color vision was tested by Farnsworth D-15: one patient (20%; pt. 40) had total color vision loss, two (40%; pt. 8,9) had deutanopia, and the other two (40%; pt. 24, 39) protanopia. Visual acuity (VA) was 20/200 in all patients. Fundus appearance was normal in two patients (40%; pt. 24,40) and pigment mottling was observed in three patients (60%; pt. 8,9,39) (Figure 2g). ERG examination showed that the photopic response was absent or markedly diminished while the scotopic response was normal or mildly abnormal. MT was between 128 and 294 μm. OCT imaging revealed a wide spectrum of photoreceptor integrity, ranging from a continuous ellipsoid zone band at the fovea in two patients (40%; pt. 24, 40) to outer retinal atrophy in three patients (60%; pt. 8, 9, 39) (Figure 2h,i). All of the ACHM patients received a molecular

ACHM Patients
Finally, the five ACHM cases had a mean age of 12.6 years (range between 10 and 16 years). The onset was around the first year of age. In all patients, color vision was tested by Farnsworth D-15: one patient (20%; pt. 40) had total color vision loss, two (40%; pt. 8,9) had deutanopia, and the other two (40%; pt. 24, 39) protanopia. Visual acuity (VA) was 20/200 in all patients. Fundus appearance was normal in two patients (40%; pt. 24,40) and pigment mottling was observed in three patients (60%; pt. 8,9,39) (Figure 2g). ERG examination showed that the photopic response was absent or markedly diminished while the scotopic response was normal or mildly abnormal. MT was between 128 and 294 µm. OCT imaging revealed a wide spectrum of photoreceptor integrity, ranging from a continuous ellipsoid zone band at the fovea in two patients (40%; pt. 24,40) to outer retinal atrophy in three patients (60%; pt. 8,9,39) (Figure 2h,i). All of the ACHM patients received a molecular diagnosis. ACHM patients showed CNGB3, CNGA3, and PDE6C mutations. The clinical and molecular diagnoses were always concordant. Table 4 summarizes the clinical findings in the cohort of ACHM patients. One patient refused to perform ERG examination (pt. 8). Patient 8, bearing mutations in the CNGB3 gene, showed some overlapping clinical features between LCA and ACHM. The disease onset in this patient was at 1 year of age with photophobia and nystagmus. He had deuteranopia at the Farnsworth D-15 test and his BCVA was 20/200. Pigment mottling was observed at fundus examination. However, MT was quite preserved (258 µm in the RE and 294 µm in the LE) and OCT revealed a disruption of the ellipsoid zone band.

Discussion
To the best of our knowledge, this is the first study that reports an integrated clinical/genetic evaluation in a pediatric IRD cohort and provides an estimation of the frequency of specific gene defects, as well as an analysis of the ellipsoid zone (EZ band) by OCT. A case series of pediatric patients with degenerative retinal disease causing severe visual impairment due to LCA, EORP, and ACHM was recruited. In analyzing these patients, we took into account their different ages (ranging from 2 to 18) that can be relevant for progressive diseases such as IRDs. In our cohort, LCA was the most frequent diagnosis (55.8%) causing severe loss of vision. Following LCA were EORP (32.5%) and ACHM (11.7%). The RETplex success rate was 58.1% (25/43), in line with previous reports [6].
In our LCA patients, the most frequently mutated gene was CEP290, when taking into consideration the three cases in which a single heterozygous mutation was found. LCA affects 1 in approximately 30,000 to 80,000 people in the general population, and accounts for more than 5% of all cases of severe visual damage [54]. Mutations in CEP290 are reported as the most common (20%) cause of LCA [12,55], in agreement with our results.
In EORP, the most frequently mutated gene observed was IQCB1 (15.4%), which is responsible for syndromic forms of IRDs. In ACHM the most frequently mutated genes were CNGA3 and CNGB3 (40%). The ACHM gene distribution in our cohort is in agreement with previous reports [12], while for EORP, our results cannot be compared with other studies because, to the best of our knowledge, this report represents the first example of a detailed analysis of this type of patients.
In this study, we found two notable cases of possible discordancy between clinical and molecular diagnosis. The first one is represented by patient 31 (clinical diagnosis of EORP) who has two putative pathogenic variants (p.R2306*; p.K663*) in the CEP290 gene, mutations which have been mostly linked to LCA [55]. Previously, CEP290 mutations in patients with RP had rarely been reported, except for the compound heterozygous mutations (c.4705-1G>T and c.3559delC) in an autosomal recessive RP patient [3] and in another patient with (c.4040G>A and c.3104-2delA) mutation [56]. Patient 31 indeed displays clinical signs highly suggestive of EORP, with preserved visual acuity, absence of oculodigital sign, and abnormal but preserved EZ band [13], which further strengthen the involvement of the CEP290 gene also in severe forms of RP. The second case of clinical/molecular discordance involves patient 10, who has two frameshift variants in the CNGB3 gene. So far, CNGB3 mutations have been reported only in ACHM. Yet, patient 10 has a phenotype compatible with a diagnosis of LCA, in particular considering his ERG, under noise level for both scotopic and photopic response at 1 year of age [12]. Therefore, the present study, for the first time to our knowledge, indicates that mutations in CNGB3 may also be associated with LCA and not only with ACHM. In light of this finding, it would be advisable to include ACHM genes in LCA-targeted sequencing panels.
The partial discordance between the genotype and the corresponding clinical features found for CEP290 and for CNGB3 suggests that additional novel genotype/phenotype correlations can be found for other genes as well. For several patients, the identified gene defect led to a re-evaluation of the patients' phenotype and the identification of additional abnormalities either not present or too mild to be noted at first evaluation. In fact, patients with mutations in the IQCB1, SPATA7, CEP290, and CLN3 genes should be re-evaluated also for the systemic involvement correlated to the mutation which is not always detectable at the first observation. For example, the nephronophthisis linked to IQCB1 mutations could be clinically unmasked belatedly [34,57]. In patient 21, bearing IQCB1 mutations, a subsequent clinical evaluation led to a diagnosis of polycystic kidney. The latter finding strengthens the need for a careful re-evaluation of the clinical diagnosis following molecular analysis even in cases with apparently isolated forms of IRDs.
We were able to evaluate the macular region by OCT in a notable number of patients, 30 out of 43 (69.8%), despite their young age [58]. In analyzing the macular region, we considered the presence of a relevant MT and of the EZ bands as two important criteria for successful gene therapy [9]. It was possible to evaluate the ellipsoid EZ band in twelve of our LCA patients; in eight of them it was absent, while in four it was irregular. The EORP patients with an irregular ellipsoid EZ band were only three, two with no detected mutations and one with mutations in CEP290; in ten EORP patients the EZ band was absent while it was not analyzed in one patient. In our cohort, the patients with ACHM presented the most preserved EZ band (three disruptions and two irregular bands) even if they are the oldest individuals analyzed. It is tempting to speculate that MT and EZ bands are linked to specific genotypes rather than to clinical conditions. Therefore, the evaluation of MT and EZ band is extremely useful, both therapeutically and clinically, to identify a time-frame for effective gene therapy. Our findings suggest that patients with mutations in CEP290, CNGB3, CNGA3, and PDE6C genes present fairly well-preserved MT and EZ bands and may benefit more from gene therapy. It must be underlined that when the EZ band is disrupted in the foveolar area and only preserved outside, its reliability as a predictor of the possible improvement in visual acuity may be compromised. This is particularly relevant in the case of ACHM patients. Nevertheless, we believe that our study provides useful information to identify early enough IRD patients that are more likely to reap the benefits of gene therapy. For many conditions that exhibit quick degeneration in combination with functional defects, early gene therapy may be useful to both prevent retinal degeneration and restore visual function. To date, successful proof-of-principle gene therapy data on pre-clinical models are available for most of the genes found to be mutated in the analyzed patients, except for IQCB1, PCYT1A, TULP1, IMPG1, and PDE6C [55,[59][60][61]. Overall, our results clearly indicate that the clinical evaluation of pediatric IRD patients in the absence of a genetic characterization may have limited value in terms of prognostic assessment and of putative therapeutic options.

Conclusions
This study provides the first detailed clinical and genetic assessment of severe IRDs with infantile onset. We illustrate the distribution of the mutated genes in an Italian IRD pediatric cohort and the importance of novel sequencing technologies to unravel the etiology of IRDs. Genetic results are crucial for a clinical re-evaluation of patients, particularly for children that are often difficult to evaluate clinically because they are unwilling or unable to undergo instrumental testing such as ERG.
Supplementary Materials: The following are available online at www.mdpi.com/2073-4425/8/10/280/s1. Table S1: List of genes and sequences represented in RETplex; Table S2: LCA patients with a single heterozygous variant in the CEP290 gene; Table S3: In silico characterization of the putative pathogenic missense sequence variants newly-identified in this report.