David Geoffrey Birch

This report summarises the Clinical Endpoints Special Session held during the 62nd annual symposium of the International Society for Clinical Electrophysiology of Vision (ISCEV), convened at Tivoli Vredenburg, Utrecht, The Netherlands. The session brought together clinicians, regulators, industry representatives, and a patient voice to consider the state of clinical endpoints in trials for inherited retinal disorders (IRDs). Discussions covered the adequacy of current endpoints, challenges of disease heterogeneity, regulatory expectations, operational feasibility, and the perspectives of patients. The session highlighted the need to anchor emerging endpoints to clinically meaningful outcomes and to minimise assessment burdens.

Small nuclear RNAs (snRNAs) combine with specific proteins to generate small nuclear ribonucleoproteins (snRNPs), the building blocks of the spliceosome. U4 snRNA forms a duplex with U6 and, together with U5, contributes to the tri-snRNP spliceosomal complex. Variants in RNU4-2, which encodes U4, have recently been implicated in neurodevelopmental disorders. Here we show that heterozygous inherited and de novo variants in RNU4-2 and in four RNU6 paralogs (RNU6-1, RNU6-2, RNU6-8 and RNU6-9), which encode U6, recur in individuals with nonsyndromic retinitis pigmentosa (RP), a genetic disorder causing progressive blindness. These variants cluster within the three-way junction of the U4/U6 duplex, a site that interacts with tri-snRNP splicing factors also known to cause RP (PRPF3, PRPF8, PRPF31), and seem to affect snRNP biogenesis. Based on our cohort, deleterious variants in RNU4-2 and RNU6 paralogs may explain up to ~1.4% of otherwise undiagnosed RP cases. This study highlights the contribution of noncoding RNA genes to Mendelian disease and reveals pleiotropy in RNU4-2, where distinct variants underlie neurodevelopmental disorder and retinal degeneration.

The U4 small nuclear RNA (snRNA) forms a duplex with the U6 snRNA and, together with U5 and ∼30 proteins, is part of the U4/U6.U5 tri-snRNP complex, located at the core of the major spliceosome. Recently, recurrent de novo variants in the U4 RNA, transcribed from the RNU4-2 gene, and in at least two other RNU genes were discovered to cause neurodevelopmental disorder. We detected inherited and de novo heterozygous variants in RNU4-2 (n.18_19insA and n.56T>C) and in four out of the five RNU6 paralogues (n.55_56insG and n.56_57insG) in 135 individuals from 62 families with non-syndromic retinitis pigmentosa (RP), a rare form of hereditary blindness. We show that these variants are recurrent among RP families and invariably cluster in close proximity within the three-way junction (between stem-I, the 5’ stem-loop and stem-II) of the U4/U6 duplex, affecting its natural conformation. Interestingly, this region binds to numerous splicing factors of the tri-snRNP complex including PRPF3, PRPF8 and PRPF31, previously associated with RP as well. The U4 and U6 variants identified seem to affect snRNP biogenesis, namely the U4/U6 di-snRNP, which is an assembly intermediate of the tri-snRNP. Based on the number of positive cases observed, deleterious variants in RNU4-2 and in RNU6 paralogues could be a significant cause of isolated or dominant RP, accounting for up to 1.2% of all undiagnosed RP cases. This study highlights the role of non-coding genes in rare Mendelian disorders and uncovers pleiotropy in RNU4-2 , where different variants underlie neurodevelopmental disorder and RP.

Phosphoribosyl pyrophosphate synthetase 1 (PRPS1) is an X-linked gene critical for nucleotide metabolism. Pathogenic PRPS1 variants cause three overlapping phenotypes: Arts syndrome (severe neurological disease), Charcot-Marie-Tooth type 5 [CMTX5] (peripheral neuropathy), and non-syndromic sensorineural hearing loss (SNHL). Each may be associated with retinal dystrophy. Multicenter phenotypic studies are limited. A multicenter retrospective clinical case series of 15 patients from 12 pedigrees with PRPS1-associated retinal degeneration is presented. Of 15 patients, 11 (73.3%) were female. Mean age of ocular disease onset was 8.5 years (range, 0.5-35 years). Many were diagnosed with Leber congenital amaurosis prior to genetic testing (n = 5). Five patients had clinical diagnoses of CMTX5 and Arts syndrome, two had isolated ocular disease, and one was asymptomatic. Mean initial VA (LogMAR) was 0.74, 0.74, 0.83, and 0.85 for isolated ocular disease, CMTX5, Arts, and SNHL, respectively. Ten patients were hyperopic and eight had asymmetric VA. Macular atrophy (n = 13), optic atrophy (n = 13), bone spicules (n = 10), and parafoveal outer retinal atrophy (n = 12) were common findings. Electroretinogram showed delayed and attenuated photopic and scotopic responses (n = 10). Median follow-up of 2.9 years (range, 1.5-11.6 years) in six patients showed retinal disease progression in two patients. PRPS1-associated retinal degeneration predominantly manifests as a bilateral asymmetric cone and rod dystrophy, commonly associated with hyperopia and optic atrophy.

The manual segmentation of retinal layers from OCT scan images is time-consuming and costly. The deep learning approach has potential for the automatic delineation of retinal layers to significantly reduce the burden of human graders. In this study, we compared deep learning model (DLM) segmentation with manual correction (DLM-MC) to conventional manual grading (MG) for the measurements of the photoreceptor ellipsoid zone (EZ) area and outer segment (OS) volume in retinitis pigmentosa (RP) to assess whether DLM-MC can be a new gold standard for retinal layer segmentation and for the measurement of retinal layer metrics. Ninety-six high-speed 9 mm 31-line volume scans obtained from 48 patients with RPGR-associated XLRP were selected based on the following criteria: the presence of an EZ band within the scan limit and a detectable EZ in at least three B-scans in a volume scan. All the B-scan images in each volume scan were manually segmented for the EZ and proximal retinal pigment epithelium (pRPE) by two experienced human graders to serve as the ground truth for comparison. The test volume scans were also segmented by a DLM and then manually corrected for EZ and pRPE by the same two graders to obtain DLM-MC segmentation. The EZ area and OS volume were determined by interpolating the discrete two-dimensional B-scan EZ-pRPE layer over the scan area. Dice similarity, Bland–Altman analysis, correlation, and linear regression analyses were conducted to assess the agreement between DLM-MC and MG for the EZ area and OS volume measurements. For the EZ area, the overall mean dice score (SD) between DLM-MC and MG was 0.8524 (0.0821), which was comparable to 0.8417 (0.1111) between two MGs. For the EZ area > 1 mm², the average dice score increased to 0.8799 (0.0614). When comparing DLM-MC to MG, the Bland–Altman plots revealed a mean difference (SE) of 0.0132 (0.0953) mm² and a coefficient of repeatability (CoR) of 1.8303 mm² for the EZ area and a mean difference (SE) of 0.0080 (0.0020) mm³ and a CoR of 0.0381 mm³ for the OS volume. The correlation coefficients (95% CI) were 0.9928 (0.9892–0.9952) and 0.9938 (0.9906–0.9958) for the EZ area and OS volume, respectively. The linear regression slopes (95% CI) were 0.9598 (0.9399–0.9797) and 1.0104 (0.9909–1.0298), respectively. The results from this study suggest that the manual correction of deep learning model segmentation can generate EZ area and OS volume measurements in excellent agreement with those of conventional manual grading in RP. Because DLM-MC is more efficient for retinal layer segmentation from OCT scan images, it has the potential to reduce the burden of human graders in obtaining quantitative measurements of biomarkers for assessing disease progression and treatment outcomes in RP.