X-linked congenital retinoschisis (XLRS) is an inherited retinal disorder that affects central vision and manifests in early childhood1. XLRS is the most common inherited retinal disorder; its highest reported prevalence is 14 per 10,000 individuals in Finland1. XLRS is characterized by foveal retinoschisis, which occurs in nearly 100% of patients, whereas peripheral schisis is present in 50% of XLRS patients1. Neuronal dysfunction, manifested as a reduction in the b-wave/a-wave ratio of dark-adapted electroretinograms (DA-ERGs), is also a characteristic of XLRS, although the detection rate for this sign has varied2,3. Clinical diagnosis is not easily determined in certain cases because of the wide range of phenotypes, which may include macular and retinal degeneration and secondary complications such as vitreous hemorrhage and retinal detachment; thus, genetic diagnosis is helpful.

XLRS is caused by mutations in the RS1 gene4, which encodes retinoschisin, a 24-kDa retina-specific protein secreted by photoreceptors and bipolar cells. Retinoschisin functions as a cell adhesion protein that maintains the synaptic structure of the retina5. To date, according to the Human Gene Mutation Database (HGMD; 2018.2 version, https://portal.biobase-international.com), 251 different mutations in this gene are known to cause XLRS. Because of variations in phenotypes among and within families, the genotype–phenotype relationship has not yet been definitively established 6,7.

We conducted a multicenter observational study at 12 institutions located throughout Japan; this study was approved by the ethics committee of each institution. Signed written informed consent was obtained from all participants or their parents.

Sixty-seven Japanese patients from 56 families with XLRS were studied (Table 1). All patients were male, and their average age was 19.1 years (range: 2 months to 57 years). XLRS was diagnosed based on retinal findings, including the presence of foveal schisis with or without peripheral schisis and a reduced b-wave/a-wave ratio on dark-adapted ERGs as well as family history1. The 56 patients included 14 patients with familial XLRS and 42 patients with sporadic XLRS. Medical records were reviewed for all patients who had been identified as carriers of mutations in the RS1 gene.

Table 1 Mutations in the RS1 gene and clinical features in patients with congenital retinoschisis

Genomic DNA was extracted from peripheral blood using DNA extraction kits or manual extraction with ethanol. Polymerase chain reaction (PCR) followed by Sanger sequencing was performed on 56 samples for six coding exons of the RS1 gene unless whole-exon deletions were detected via PCR. In brief, oligonucleotide primers for the flanking intron/untranslated region sequences were designed, and PCR was performed, followed by uni- or bidirectional sequencing depending on the quality of the PCR products. The primer sequences and annealing temperature for PCR for each exon are available on request. The other 11 samples were screened by whole-exome sequencing with at least 30× coverage for all exons. To identify sequence variations, reference sequences of RS1 (NM_000330.3) were used; variations were numbered based on the cDNA sequence, with +1 corresponding to the first nucleotide of the initiation codon (ATG).

Thirty-seven different mutations in the RS1 gene were identified in the 56 families, including 26 missense, 4 nonsense, 3 splicing, 1 deletion, 1 insertion, and 2 whole-exon deletion mutations (Table 1). Eleven point mutations were novel mutations, and 24 point mutations had previously been reported, based on the HGMD and one recent report (Table 1)8. A whole-exon deletion of exon 3 had been reported9, whereas a deletion of exons 2 and 3 has not been reported. In our study, DNA break points were not determined, and it is unknown whether the exon 3 deletion that we observed was identical to the known exon 3 deletion at the DNA level.

The frequency of the 11 novel point mutations was assessed using public domain databases. None of these variants were found in human genome variation databases for the Japanese population (the Human Genetic Variation Database (HGVD), http://www.hgvd.genome.med.kyoto-u.ac.jp/) or other population databases, such as the 1000 Genomes Project database (http://www.internationalgenome.org/1000-genomes-browers), the Exome Aggregation Consortium (ExAC) database (http://www.exac.broadinstitute.org), and the 6500-exome database of the NHLBI-ESP project (ESP6500, http://evs.gs.washington.edu/EVS/). The pathogenicity of the seven novel missense mutations was predicted in silico by nine programs and via folding energy assessments6,10,11,12,13,14,15,16,17. Overall, all variants were considered to be pathogenic (Table 2).

Table 2 Pathogenicity assessment of the novel missense mutations in the RS1 gene

Seven of the known mutations were detected in more than one family; in particular, p.Glu72Lys, p.Tyr89Cys, p.Arg182Cys, p.Arg102Trp, p.Arg197Cys, p.Arg200His, and p.Pro203Leu were observed in 6, 4, 4, 3, 3, 2, and 2 families, respectively. These mutations have previously been reported in the same population and in other populations4,7,9. Mutation hot spots were suggested instead of founder effects as an explanation of these mutations.

Overall, the clinical findings of this study were consistent with those of earlier reports, although detailed phenotype–genotype relationships remain undetermined1,3,6,7.

Of the 109 phakic eyes for which refractive error (in spherical equivalents) was measured, there were 60 (55.0%), 5 (4.6%), and 44 (40.4%) hypermetropic, emmetropic, and myopic eyes, respectively (Table 1). For the hypermetropic eyes, the refractive error ranged from 0.25 to 10.0 diopters (D), and the average error was + 2.7 D. For the myopic eyes, the refractive error ranged from −0.125 to −7.5 D, and the average error was −1.6 D. The average difference in refractive error between the two eyes was 1.0 D for 51 patients.

For 125 eyes, the decimal best-corrected visual acuity varied from counting fingers to 1.2, with a median of 0.3. For 131 eyes for which retinal status was determined, retinoschisis was present in the macula in 110 eyes (84.0%) and in the periphery in 88 eyes (61.8%).

DA-ERGs were recorded in 104 eyes using different stimulus intensities; intensities that tended to be higher than those recommended in the standard protocol from the International Society of Clinical Electrophysiology of Vision were used for certain patients18. Negative ERGs were more frequently observed in this study (84.6%, Table 1) than in earlier studies3, likely due to the use of higher-intensity light stimuli19.

The observed retinal complications included a need for pars plana vitrectomy (N = 16); macular or retinal degeneration, including Oguchi disease-like retinal surface abnormalities (N = 6); vitreous hemorrhage (N = 5); retinal folds (N = 4); and congenital glaucoma (N = 1).

We sought to establish a possible phenotype-genotype relationship for eyes with truncation mutations (i.e., nonsense, splicing, deletion, insertion, or exon deletion mutations) as opposed to missense mutations. The newly identified mutations do not appear to produce distinct clinical phenotypes compared with reported mutations. However, patients with novel missense mutations did present at an earlier age than those with reported missense mutations (data not shown).

Foveal schisis was more frequently found in eyes with truncation mutations than in those with missense mutations (100% versus 78%, P = 0.0035, Supplemental Table 1). It is possible that nearly normal foveal structure can only be seen in eyes with missense mutations20. Peripheral schisis was found in 50% and 67% of eyes with truncation and missense mutations, respectively (P = 0.107).

Compared with eyes with missense mutations, eyes with truncation mutations showed larger b-waves (P = 0.023) and higher b/a ratios (P = 0.019) on DA-ERG, whereas no significant difference was observed for the mean a-wave amplitude (Supplemental Table 2). Differences in patient age, visual acuity, refractive error, and light stimulus settings for DA-ERGs were not significant.

Vincent et al.20 reported that truncation mutations were associated with poor visual acuity and a higher probability of a b/a ratio < 1.0. Our data yielded contradictory results, with higher b-wave amplitude and a greater b/a ratio in eyes with truncation mutations than in eyes with missense mutations. One possible reason for this discrepancy is that the patients with truncation mutations presented at a younger age, which tends to be associated with better preservation of ERG findings21. Nonetheless, our study implies that it will be difficult to determine a phenotype–genotype relationship using ERGs.

This study has limitations. Because of the retrospective nature of this investigation, in which only mutation-proven cases were selected, the identification rate of the RS1 gene in XLRS has not been determined. A history of clinical findings, including vitreous hemorrhages, may have been missed in certain cases due to only reviewing medical records.

In summary, this study was the largest survey of patients with mutations in the RS1 gene in the Japanese population. The progress of gene therapy for XLRS has reached the clinical trial stage, and exact genetic determinations for each patient could lead to more efficient future treatments22.