In this study, we employed a single genetic test, WES, for the genetic diagnosis of 168 unrelated Korean patients with different types of IRD. Using this single test, we molecularly diagnosed 86 (51.2%) patients. To the best of our knowledge, this is the first study to evaluate clinical diagnostic accuracy and causative genes in Korean patients with various IRD using a WES approach.
Targeted sequence capture is used to isolate and enrich specific genomic regions prior to massively parallel sequencing. Although the cost of WES is gradually decreasing, it remains expensive for clinical use. Herein, we used pre-capture pooling with targeted sequence capture to reduce the reagent cost and hands-on time. As pre-capture pooling using xGen Lockdown panel was not inferior to routine post-capture pooling using SureSelect Human All Exon v6, we successfully utilised the xGen panel for WES in Korean IRD patients. We identified 147 causative variants in 35 known IRD-related genes, including 51 (34.7%) novel variants. However, most of these novel variants were potentially causative which were annotated by computational tools. Functional validation would be required to derive a definite causality of these potentially causative variant.
Several studies have reported that the hereditary features and causative genes of IRD vary among ethnicities, even in geopolitically close Asian countries [5-7, 29, 71-73]. However, only a few reports have been published on the mutation spectrum of a large-scale Korean cohort with IRD. Our previous study using targeted exome sequencing (TES) of 53 RP-related genes in 62 Korean patients with non-syndromic RP revealed causal variants in 50.0% of the patients [7]. PRPF31 mutations (17.6%) were the most frequently found causative variants, followed by mutations in EYS, PDE6B, RHO, RP1, and RP2 (11.8% each). A recent study reported the TES results for 204 IRD-related genes in 86 Korean patients with IRD [74], as well as the molecular diagnoses rate for 44.2% of the patients. In RP, EYS mutations (22.2%) were the most frequent causative variants, followed by mutations in PED6B (16.7%), PED6A (11.1%), and USH2A (11.1%). In the present study, USH2A mutations (22.4%) were the most common causative variants in RP, followed by variants in EYS (10.3%), RP1 (6.9%), and ABCA4, PED6B, and RHO (5.2% each) mutations. The differences in the mutation spectra amongst the Korean studies appears to largely originate from selection of the sample population.
Moreover, the sample sizes of the two previous studies were much smaller than that of the present study making them more vulnerable to sampling error. In addition, the inclusion of many subjects from a single family can result in increased representation of a specific genetic mutation, thereby skewing the mutational spectrum. In the present study, we only included one proband from each family to maximise the representativeness of the cohort for the total population and to reduce sampling error. In addition, differing inheritance pattern distributions in the sampled populations may result in different distributions of causative mutations. Our previous study included a large proportion of subjects with AD inheritance, which resulted in a higher proportion of mutations in PRPF31, an AD inheritance gene [7]. In the present study, the distribution of the inheritance pattern for RP resembles that in a previous report on a large cohort of Korean RP patients [6].
The mutation spectrum of the patients in the present study resembles those established in recent large-scale studies in other Asian countries [21, 75]. For instance, the USH2A mutation was the most common in the present study, as in the Chinese study, and it was the second most common in the Japanese study. The second most common mutation was in the EYS gene, which was the most common in the Japanese study and the third most common in the Chinese study. These findings were similar to those of previous studies demonstrating ethnic differences between Asian and Caucasian populations, with a higher incidence of EYS mutations in arRP patients of an Asian background than Caucasian background [76]. The most common EYS mutation in our cohort was c.4957dupA; it was also common in Japanese RP patients, however, was rare, or not detected in European RP patients and Chinese RP patients [66, 77]. Meanwhile, the c.C8805A and c.C7394G EYS mutations, which were frequently observed in the Japanese RP cohort, were not detected in Korean studies [7, 74, 78]. This suggests that differences in the mutational spectra of IRD patients exist among East Asian countries, despite their geographical proximity, and emphasises the importance of obtaining reference data for individual nations or regions to determine the local genomic IRD landscape.
The prevalence of the RP1 mutation was 6.9% in arRP cases in this study, which was higher than those among Japanese and Chinese arRP patients (1.7%–2%) [29, 73]. Mutations in the RP1 gene cause both AR and AD forms of RP, accounting for 5.5% of adRP and less than 1% of arRP [76]. Different explanations for the dominant/recessive mutation effect of the RP1 gene have been proposed, however, the precise mechanism remains unclear. In addition, a recent study suggested that the phenotypic spectrum associated with RP1 mutations should be expanded to CORD and macular dystrophy [79, 80]. In our cohort, Case 103, who showed well-demarcated macular atrophy with normal electroretinography findings (see Additional file 3) carried compound heterozygous nonsense (c.C5797T) and frameshift (c.649delG) mutations, and received a molecular diagnosis of RP1-associated AR macular dystrophy.
WES has clear advantages over TES for the molecular diagnosis of IRD. The heterogeneity of the genotype and phenotype, as well as the unclear inheritance patterns of IRD, make it difficult to select target genes for TES. In addition, more than 270 causative genes have been identified for IRD to date, and new causative genes continue to be discovered. To keep pace with the literature, researchers must redesign panels to incorporate new genomic regions at additional expense. In contrast, WES provides the advantage of re-evaluating previously analysed datasets when a novel gene associated with IRD is reported. In a study comparing WES and three commercial gene panels, WES discovered causative gene mutations in 42% of cases, which were not included in at least one commercial panel [81]. In the present study, six genes (CYP4V2, NMNAT1, RP1L1, CACNA1F, BBS2, and REEP6) out of 35 causative genes detected in our cohort were not included in at least one of the TES studies in Korean IRD patients [7, 74].
Among patients with macular disease, patients with STGD had the highest detection rate for causative variants (90.0%), however, patients with other forms had a poorer detection rate (9.5%). This is likely due to the presence of pathognomonic findings for the accurate clinical diagnosis of STGD. Initially, 13 patients with clinical diagnosis of STGD were included in this study. Three patients who were redefined as having another disease based on their molecular diagnosis did not have dark choroid rings on fluorescein angiography. In contrast, the remaining ten patients had dark choroid signs and a (probable) molecular diagnosis of STGD. This pathognomonic finding, the dark choroid sign, can differentiate STGD from other similar conditions including non-hereditary diseases [82]. Meanwhile, all vitelliform macular dystrophy cases in our cohort were clinically diagnosed as adult-onset vitelliform macular dystrophy with clinical findings of submacular vitelliform material and normal electrooculogram. The mean age of the vitelliform macular dystrophy patients was 64.5 years. These clinical features are similar to those of exudative age-related macular degeneration, choroidal neovascularisation, or central serous chorioretinopathy, which do not belong to the category of IRD. There is a possibility that some of our vitelliform macular dystrophy cases may arise from non-genetic conditions. This finding suggests that accurate clinical diagnosis can increase the efficiency of molecular diagnosis.
The 17.9% (30/168) of patients who received a possible molecular diagnosis may carry a second pathogenic variant in the same gene, which occurs in no- or low-coverage genetic regions [73]. In the exome sequencing approach, some genetic regions have low or no coverage. For example, repetitive regions, GC-rich regions, and regions with high homology are difficult to enrich. In addition, small copy number variations and deep-intronic regions cannot be detected via exome sequencing [83, 84]. In these patients, further analysis, such as direct sequence analysis or whole genome sequencing, is required.
In this study, we were unable to identify any causative variants in 31.0% (52/168) of the cases. Several explanations may account for these. The first explanation is the limitation of WES. Approximately 85% of known causative mutations occur in exonic regions that encode proteins, indicating that WES is unable to discover the cause of the remaining ~15% of causative mutations. In addition, some genetic regions have low or no coverage in the exome sequencing approach, as described above. The second explanation is potentially inaccurate clinical diagnosis. It is possible that some of our cohort have non-hereditary retinal conditions that phenotypically resemble IRD. Therefore, further study via direct sequence analysis or whole-genome sequencing with detailed clinical diagnosis is required to achieve an optimal detection rate.