Clinical and Molecular Features of CNGA1-Associated Retinitis Pigmentosa in an East Asian Cohort: Insights for Gene Therapy Trials

Abstract

Purpose

To characterize the natural history of CNGA1-associated retinitis pigmentosa (RP) in an East Asian cohort and to identify potential endpoints for clinical trials.

Design

A multicenter, retrospective cohort study.

Participants

Patients diagnosed with CNGA1-associated RP.

Methods

Medical records were reviewed for age at onset, symptoms, best-corrected visual acuity, ultra-widefield retinal imaging, spectral-domain OCT, full-field electroretinography, full-field stimulus threshold (FST), multiluminance mobility testing (MLMT), and static perimetry.

Main Outcome Measures

Molecular genetic findings, visual acuity decline rate, FST measurements, MLMT score, retinal sensitivity, and image findings.

Results

The cohort included 57 patients (mean age 42 ± 16 years). Eighteen disease-causing CNGA1 variants were identified, 11 of which were novel. The c.253delC variant was the predominant mutation in both the Chinese and Korean cohorts, with an allele frequency of 69%. A fundus autofluorescence (FAF) staging system was applied to stratify disease severity. Despite high prevalence of epiretinal membranes (57.9% of eyes) and inner-retinal cysts (26.3% of eyes), OCT imaging demonstrated long-term preservation of the foveal ellipsoid zone until stage 4, indicating a broad therapeutic window. Best-corrected visual acuity decline was minimal and varied across stages, whereas cataracts in the fourth to fifth decades significantly impaired visual acuity, with 70% presenting as anterior subcapsular cataract. Multiluminance mobility testing showed a ceiling effect among potential trial candidates (stages 1-3). The decline rate of mean deviation in static perimetry was 0.17 dB per year. The estimated number of progression loci (≥7 dB decline) across the central field was 5.83 over 1 year, supporting the ability to meet US Food and Drug Administration thresholds for clinical meaningful change. Full-field stimulus threshold showed a cone-mediated pattern across different stages but remained valuable for detecting rod functional restoration.

Conclusions

CNGA1-associated RP in East Asians is characterized by a rod–cone dystrophy phenotype with a broad therapeutic window. The FAF-based staging system provides a reliable framework for severity assessment and reveals stage-specific progression patterns, with no significant genotype–phenotype association observed. The high prevalence of the c.253delC variant suggests that gene editing is a promising therapeutic strategy alongside gene augmentation. Static perimetry appears to be the most promising primary endpoint for future clinical trials.

Financial Disclosure(s)

Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Retinitis pigmentosa (RP) represents a genetically and clinically heterogeneous group of inherited retinal dystrophies (IRDs).¹ It accounts for more than 50% of IRDs and is the leading cause of visual disability in the working-age population.² To date, 143 genes associated with RP have been identified (https://retnet.org/). Retinitis pigmentosa has long been managed primarily through supportive care. With the success of Luxturna and rapid advancements in gene replacement therapy,³ precise treatments for RP have now become feasible. Accordingly, detailed characterization of gene-specific natural history has become essential for defining therapeutic windows, optimizing clinical trial design, and identifying sensitive outcome measures.

The CNGA1 gene (OMIM 123825) encodes the alpha subunit of the cyclic nucleotide-gated channel, which is specifically expressed in rod photoreceptors.^4,5 Cyclic nucleotide-gated channels play a critical role in the phototransduction cascade by mediating cyclic guanosine monophosphate-dependent ion influx under dark conditions. Dysfunction of CNGA1 impairs the assembly and trafficking of cyclic nucleotide-gated channels, thereby disrupting calcium and cyclic guanosine monophosphate homeostasis and ultimately leading to rod photoreceptor apoptosis.^6,7 CNGA1-associated RP is characterized by a relatively slow rate of progression and prolonged preservation of central vision.⁸ This extended disease course positions a considerable proportion of affected individuals as suitable candidates for gene therapy. Indeed, the first-in-human gene therapy trial for CNGA1-associated RP (NCT06291935) has recently been initiated.

While the therapeutic prospects are promising, current studies remain limited, with small sample sizes and predominantly cross-sectional designs that do not adequately characterize the disease progression essential for trial readiness. The global prevalence of CNGA1 mutations is approximately 1% in autosomal recessive RP.8, 9, 10, 11, 12, 13, 14 However, this gene represents a more prominent contributor to autosomal recessive RP in East Asian populations, with reported prevalence reaching 5.1% in Japan and 7.6% in China.^8,15 The absence of large-scale, longitudinal data from these relatively high-prevalence populations leaves a critical knowledge gap.

To address this unmet need, we conducted a retrospective natural history study in an East Asian cohort harboring CNGA1 variants. In addition to characterizing the genotype and phenotype spectrum, we established a clinical staging system based on multimodal imaging. This study provides important data to support prognostic counseling and offers a reference framework for patient stratification and the efficacy assessment in emerging gene therapy trials.

Methods

Study Population

Patients were recruited from the genetic database for hereditary eye diseases at Shanghai General Hospital (Shanghai, China) and the Department of Ophthalmology at Seoul National University Bundang Hospital (Seoul, Korea). Patients’ clinical information and ophthalmic examination data were collected from February 2013 to December 2025 and were analyzed from June 2025 to December 2025. The inclusion criteria were as follow: (1) a clinical diagnosis of RP; (2) a confirmed molecular diagnosis with 2 likely disease-causing CNGA1 variants; and (3) the availability of data from at least 1 ophthalmic examination. For patients with longitudinal data, follow-up duration was defined as the interval between the first and most recent clinical examinations available in the database.

This retrospective cohort study (ClinicalTrials.gov identifier: NCT03691168) adhered to the tenets of the Declaration of Helsinki and was approved by the institutional review boards of the participating centers. Written informed consent was obtained from all participants or their legal guardians.

Genetic Analysis

Genetic diagnosis was performed using whole-exome sequencing for the Chinese subcohort. For the Korean subcohort, targeted gene panel sequencing was employed using a custom retinal panel (Celemics, Probe ID: 2007PD004-V1), covering the coding sequences of 294 IRD-related genes, the untranslated regions of MIR204, and 82 intronic hotspot regions associated with IRDs. In 1 patient (CN13) with an initially negative whole-exome sequencing result, whole-genome sequencing was subsequently performed to identify causative variants. Familial segregation analysis was conducted where possible to confirm biallelism (homozygosity or compound heterozygosity). Variants were annotated against Reference Sequence NM_001379270.1 and classified according to American College of Medical Genetics and Genomics guidelines.¹⁶ The pathogenicity of missense variants was assessed using in silico prediction algorithms (PROVEAN, PolyPhen-2, SIFT, and AlphaMissense).17, 18, 19, 20 Variant allele frequencies were cross-referenced against population databases including gnomAD, jMorp, ChinaMAP, and WBBC.21, 22, 23 A systematic curation of previously reported CNGA1 variants was also performed (see Supplementary Methods, available at www.ophthalmologyscience.org).

Clinical Assessment

Ophthalmic Examination

Data extracted included age at onset, symptomatology, and medical history. Standard examinations included refraction (spherical equivalent), slit-lamp biomicroscopy, and best-corrected visual acuity (BCVA), which was converted into logarithm of the minimum angle of resolution (logMAR) values when necessary. Visual impairment was stratified according to World Health Organization criteria: mild or no impairment (≥0.3 decimal), moderate (<0.3–≥0.1), low vision (<0.1–≥0.05), and blindness (<0.05). Best-corrected visual acuity values for counting fingers, hand motion, and light perception were converted to 2.0, 2.3, and 2.7 logMAR units, respectively.²⁴ Asymmetry in BCVA between the 2 eyes was defined as a difference of 0.3 logMAR or more.²⁵ Myopia was defined as a spherical equivalent of <–0.50 diopters (D), high myopia as <–6.00 D, and hyperopia as >+0.50 D.

Multimodal Imaging

Multimodal imaging included ultra-widefield fundus photography (Optos), 488-nm wavelength ultra-widefield fundus autofluorescence (FAF; Optos), and spectral-domain OCT (Carl Zeiss Meditec AG). Ellipsoid zone (EZ) width and central foveal thickness were quantified on horizontal foveal B-scans.

To stratify disease severity, a ultra-widefield FAF-based staging system was developed. Three autofluorescence features were defined as follows: a hyperautofluorescent ring (Feature A, arrows in Fig 1), extramacular hypoautofluorescence (Feature B, asterisk in Fig 1), and the extension of the hyperautofluorescent ring beyond the vascular arcade (Feature C, circle in Fig 1). Eyes were categorized into 4 stages. Stage 1 was defined by the presence of all features (A, B, and C), indicating a large preserved central area with the ring outside the vascular arcade. Stage 2 was defined by the presence of Features A and B, where the ring constricted within the arcades while the hypoautofluorescence has not yet invaded the macula. Stage 3 was defined by the presence of Features A, representing a small residual ring with extensive surrounding atrophy. Finally, stage 4 was defined as generalized hypoautofluorescence without a hyperautofluorescent ring. Two graders (Z.L. and J.S.S.) who were masked to the genetic information evaluated the ultra-widefield FAF images. Disagreements were adjudicated by a third reviewer (T.L.).

Figure 1.

Multimodal imaging and UWF-FAF–based staging of CNGA1-associated retinitis pigmentosa. Representative multimodal retinal imaging of a healthy control eye and eyes with CNGA1-associated retinitis pigmentosa stratified into 4 stages based on 488-nm UWF-FAF patterns. Arrow: Feature A with a hyperautofluorescent ring; asterisk: Feature B with extramacular hypoautofluorescence; circle: Feature C with the hyperautofluorescent ring extending the vascular arcade; arrowhead: the boundary of the ellipsoid zone. UWF-FAF = ultra-widefield fundus autofluorescence.

Functional Assessment

Visual fields were assessed using standard automated perimetry (Humphrey Field Analyzer, 30-2 SITA Standard, stimulus size III). Progressing loci were defined as retinal sensitivity decline of ≥7 dB. Functional vision was assessed using the multiluminance mobility test (MLMT).²⁶ Full-field electroretinography (ERG) was performed according to the International Society for Clinical Electrophysiology of Vision standards.²⁷ Full-field stimulus threshold (FST) testing (Espion E3 system; Diagnosys LLC) was performed using white, blue, and red stimuli after 50 minutes of dark adaptation.28, 29, 30

Statistical Analysis

R and RStudio (version 4.5.0) was used for statistical analyses. To compare clinical characteristics across 4 stages, analyses were performed within a linear mixed-effects modeling framework. Stage was modeled as a fixed effect, and repeated measurements were handled using a nested random-intercept structure (1 |patient/eye) to account for within-subject and intereye correlation. Pairwise stage contrasts were obtained from model-based estimated marginal means with Bonferroni adjustment for multiple comparisons. Age-related longitudinal trends in BCVA, EZ width, visual field index (VFI), and mean deviation were analyzed using linear mixed-effects modeling. Age was mean-centered to improve model stability and interpretability. Lens status (clear lens, cataract, or pseudophakia) was included as a fixed-effect covariate for BCVA. Potential nonlinearity was tested and retained based on model comparison. Stage-stratified models were further fit to characterize within-stage age trends, with candidate functional forms (linear, quadratic, and spline) compared and the best model selected per stage. Sensitivity analyses were performed to assess outliers and influential observations. Model assumptions were evaluated using residual diagnostics, including histograms and normal Q-Q plots. To assess the impact of lens opacity on BCVA, the eyes were categorized into a "cataract" group and a "noncataract" group, and evaluated within the linear mixed-effects modeling framework. Spearman correlation analysis was performed to assess the associations among clinical parameters. Genotype–phenotype correlations were analyzed using Fisher exact test. For descriptive reporting, continuous data were expressed as either means with standard deviations or medians with interquartile ranges. Statistical significance was defined as two-sided P values < 0.05 for primary model estimates, and as Bonferroni-adjusted P < 0.05 for multiple pairwise comparisons.

Results

Fifty-seven patients from 57 unrelated East Asian families (40 Chinese and 17 Korean) were recruited, with 30 (53%) being male. The mean age at baseline was 42 ± 16 years (range, 2–71). Thirty-seven patients (65%) underwent follow-up as part of the study. The median follow-up duration was 3.1 years (interquartile range: 1.9–5.0). Detailed clinical and molecular findings are summarized in Table 1.

Table 1.

Demographic, Genetic, and Clinical Findings of Patients with CNGA1 Variants^∗

Patient	Sex	Age, Y	CNGA1 Variant 1		CNGA1 Variant 2		Segregation	Visual Acuity, LogMAR		FAF Stage		Other Ophthalmic Features
			Nucleotide Change	Amino Acid Change	Nucleotide Change	Amino Acid Change		Right Eye	Left Eye	Right Eye	Left Eye
KR16	M	<18	c.1666C>T	p.Arg556∗	c.179delG	p.Gly60Valfs∗29	No	0.1	0.2	Stage 1	Stage 1	None
CN01	F	<18	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	Yes	0.0	0.0	Stage 1	Stage 1	Myopia
KR13	M	<18	c.253delC	p.Leu85Phefs∗4	c.1313G > A†	p.Trp438∗	No	–0.1	–0.1	Stage 1	Stage 1	None
CN02	F	20s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	0.0	0.1	Stage 1	Stage 1	None
KR11	F	20s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	0.2	0.1	Stage 1	Stage 1	None
CN03	F	30s	c.253delC	p.Leu85Phefs∗4	c.1480delC†	p.Gln494Lysfs∗41	Yes	0.0	0.1	Stage 1	Stage 1	None
KR05	M	30s	c.253delC	p.Leu85Phefs∗4	c.179delG	p.Gly60Valfs∗29	No	0.2	0.1	Stage 1	Stage 1	None
CN04	F	30s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	0.0	0.0	Stage 1	Stage 1	High myopia
KR10	M	30s	c.253delC	p.Leu85Phefs∗4	c.179delG	p.Gly60Valfs∗29	No	–0.1	–0.1	Stage 1	Stage 1	None
CN05	F	30s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	0.0	0.0	Stage 1	Stage 1	Hyperopia
KR17	F	30s	c.179delG	p.Gly60Valfs∗29	c.179delG	p.Gly60Valfs∗29	No	–0.2	–0.1	Stage 1	Stage 1	None
KR14	M	40s	c.253delC	p.Leu85Phefs∗4	c.179delG	p.Gly60Valfs∗29	No	0.0	0.0	Stage 1	Stage 1	None
CN07	M	40s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	0.2	0.2	Stage 1	Stage 1	Myopia; IOL in BE
KR03	F	50s	c.253delC	p.Leu85Phefs∗4	c.1915C > T†	p.Arg639∗	No	0.3	0.6	Stage 1	Stage 1	None
KR12	F	50s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	1.0	0.0	Stage 3	Stage 1	None
CN40	M	60s	c.253delC	p.Leu85Phefs∗4	c.659_661dup†	p.Leu220_Glu221insVal	No	0.0	0.2	Stage 1	Stage 1	Myopia; IOL in BE
CN08	M	<18	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	Yes	0.2	0.2	Stage 2	Stage 2	Myopia
CN09	M	20s	c.253delC	p.Leu85Phefs∗4	c.1528 C > T	p.Arg510Ter	Yes	0.1	0.2	Stage 2	Stage 2	None
CN10	M	20s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	0.4	0.2	Stage 2	Stage 2	High myopia; cataract in RE; IOL in LE
CN38	F	20s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	0.2	0.3	Stage 2	Stage 2	None
KR15	F	30s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	0.8	0.7	Stage 2	Stage 2	Keratoconus in RE, PKP in LE; cataract in BE
CN12	F	30s	c.253delC	p.Leu85Phefs∗4	c.628G > A†	p.Asp210Asn	Yes	0.2	0.2	Stage 2	Stage 2	High myopia
CN13	F	30s	c.253delC	p.Leu85Phefs∗4	txStart-intron3del(X1)†	NA	No	0.2	0.4	Stage 2	Stage 2	Myopia; glaucoma in LE; IOL in BE
CN06	M	40s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	0.0	0.0	Stage 2	Stage 2	Myopia; IOL in BE
KR09	F	40s	c.253delC	p.Leu85Phefs∗4	c.179delG	p.Gly60Valfs∗29	No	0.2	0.2	Stage 2	Stage 2	LASEK in BE
CN14	F	40s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	0.5	0.5	Stage 2	Stage 2	Cataract in BE
CN15	F	50s	c.253delC	p.Leu85Phefs∗4	c.827G > A	p.Arg276His	Yes	0.3	0.3	Stage 2	Stage 2	Myopia; cataract in BE
CN21	F	50s	c.1537G > A†	p.Gly275Arg	c.1537G > A†	p.Gly275Arg	No	0.8	0.3	Stage 3	Stage 3	Myopia; IOL in BE
CN16	M	50s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	0.2	0.2	Stage 2	Stage 2	IOL in BE
CN17	M	40s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	0.2	0.2	Stage 3	Stage 3	High myopia
KR04	F	40s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	0.3	0.3	Stage 3	Stage 3	None
CN18	M	40s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	0.3	0.3	Stage 3	Stage 3	IOL in BE
CN19	M	40s	c.253delC	p.Leu85Phefs∗4	c.1258C > T	p.Arg420Ter	Yes	0.7	0.2	Stage 4	Stage 3	Myopia
CN39	M	50s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	0.4	1.3	Stage 3	Stage 3	Hyperopia; cataract in BE
CN20	M	50s	c.253delC	p.Leu85Phefs∗4	c.396_397delGA†	p.Glu132fs	No	0.2	0.3	Stage 3	Stage 3	None
KR07	M	50s	c.179delG	p.Gly60Valfs∗29	c.1915C > T†	p.Arg639∗	No	0.5	0.5	Stage 3	Stage 3	None
KR08	F	50s	c.179delG	p.Gly60Valfs∗29	c.827G > A	p.Arg276His	No	0.5	0.5	Stage 3	Stage 3	IOL in BE
CN22	F	50s	c.253delC	p.Leu85Phefs∗4	c.396_397delGA†	p.Glu132fs	No	0.5	0.7	Stage 3	Stage 3	Hyperopia; IOL in BE
CN23	M	70s	c.253delC	p.Leu85Phefs∗4	c.776T > C†	p.Leu259Ser	No	0.3	0.2	Stage 3	Stage 3	Myopia; IOL in BE
CN24	M	40s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	1.3	1.3	Stage 4	Stage 4	Myopia; cataract in BE
CN25	M	40s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	HM	1.0	Stage 4	Stage 4	Cataract in RE; IOL in LE
CN26	F	40s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	1.3	1.3	Stage 4	Stage 4	Cataract in BE
CN27	M	40s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	HM	0.7	Stage 4	Stage 4	None
CN28	M	40s	c.253delC	p.Leu85Phefs∗4	c.1666C > T	p.Arg556Ter	Yes	HM	CF	Stage 4	Stage 4	Myopia; strabismus in BE; IOL in BE
KR02	M	50s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	0.3	CF	Stage 4	Stage 4	IOL in BE
CN29	F	50s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	Yes	1.0	0.8	Stage 4	Stage 4	IOL in BE
CN30	F	50s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	CF	CF	Stage 4	Stage 4	IOL in BE
CN31	F	50s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	HM	HM	Stage 4	Stage 4	Cataract in BE
CN32	F	50s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	HM	CF	Stage 4	Stage 4	Cataract in RE; IOL in LE
CN33	M	50s	c.253delC	p.Leu85Phefs∗4	c.1028G > T†	p.Arg343Ile	No	CF	1.0	Stage 4	Stage 4	Cataract in BE
CN34	F	50s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	0.8	1.0	Stage 4	Stage 4	Myopia; cataract in BE
CN35	M	60s	c.253delC	p.Leu85Phefs∗4	c.518T > G†	p.Met173Arg	No	HM	HM	Stage 4	Stage 4	Myopia; IOL in BE
CN36	F	60s	c.253delC	p.Leu85Phefs∗4	c.253delC	p.Leu85Phefs∗4	No	1.3	1.0	Stage 4	Stage 4	IOL in BE
KR01	M	60s	c.253delC	p.Leu85Phefs∗4	c.179delG	p.Gly60Valfs∗29	No	0.5	HM	Stage 4	Stage 4	IOL in BE
CN37	M	70s	c.851C > G†	p.Thr284Arg	c.851C > G†	p.Thr284Arg	No	LP	HM	Stage 4	Stage 4	IOL in BE
KR06	M	20s	c.179delG	p.Gly60Valfs∗29	c.179delG	p.Gly60Valfs∗29	No	0.0	0.0	NA	NA	None
CN11	M	30s	c.253delC	p.Leu85Phefs∗4	c.1525G > A	p.Gly509Arg	Yes	0.3	0.2	NA	NA	Myopia; cataract in BE

BE = both eyes; CF = counting fingers; CN = Chinese subcohort; F = female; FAF = fundus autofluorescence; HM = hand movements; KR = Korean subcohort; IOL = intraocular lens; LASEK = laser epithelial keratomileusis; LE = left eye; LP = light perception; M = male; PKP = penetrating keratoplasty; RE = right eye.

^∗Patients are sorted by age and severity grade of fundus autofluorescence.

^†Previously unreported variant.

Molecular Genetics

Eighteen CNGA1 variants were identified, including 11 (61%) novel variants (Fig 2 and Table S1, available at www.ophthalmologyscience.org). The variants included missense (39%), nonsense (28%), frameshift (22%), in-frame duplication (6%), and multi-exon deletion (6%) variants. The c.253delC (p.Leu85Phefs∗4) mutation was identified in 50 patients (88%), accounting for 69% of the disease-causing alleles identified in this cohort. It was the most common variant in both Chinese and Korean subcohorts, consistent with other East Asian population databases (Table S2, available at www.ophthalmologyscience.org). c.179delG (p.Gly60Valfs∗29) was the second most common variant in the Korean subcohort (representing 35% of alleles in the subcohort), while being absent in the Chinese subcohort.

Figure 2.

Distribution of identified variants on the CNGA1 protein structure. Schematic representation of the CNGA1 protein structure with functional domains highlighted in color. All variants identified in the present study are mapped to their respective amino acid positions. Novel variants are indicated in bold. For splice-site and frameshift variants, arrows denote the position of the first affected amino acid. The most common variant in the cohort, c.253delC (p.Leu85Phefs∗4), is highlighted in red. The large deletion spanning from the transcription start site to the third intron (txStart-intron3del) is not depicted due to its genomic extent. CNBD = cyclic nucleotide-binding domain; IDR = intrinsically disordered region.

Disease Onset and Visual Acuity

Nyctalopia was the initial symptom in all patients (100%), with a mean age at onset of 4.4 ± 2.7 years. Peripheral visual field loss was reported in majority of patients (96%). Although 2 patients (ages 5 and 11) did not report field defects, perimetry confirmed a reduced VFI in 1 patient. Visual acuity decline was noted later, with a mean age at onset of 38.4 ± 13.5 years (n = 27). The majority of patients (23/27, 85%) experienced gradual loss of central vision after age 30.

According to World Health Organization criteria for visual acuity, 64.0% of eyes had mild or no visual impairment. Blindness (14.0% of eyes) was observed exclusively in patients older than 40 years. The mean BCVA at baseline was 0.61 ± 0.75 logMAR (range: –0.10 to 2.70). Mean BCVA stratified by stage was 0.05 ± 0.14 (stage 1), 0.23 ± 0.17 (stage 2), 0.43 ± 0.30 (stage 3), and 1.61 ± 0.73 logMAR (stage 4). Mixed-effects modeling revealed significant BCVA difference across stages after adjusting for lens status. Bonferroni-adjusted pairwise comparisons showed no significant differences among stages 1–3 (all P = 1.000), whereas stage 4 had significantly worse BCVA than each of the other stages (all P ≤ 0.0002) (Fig 3A; Table S3, available at www.ophthalmologyscience.org). Longitudinal analysis showed that BCVA remained stable in stages 1 and 2 (P > 0.05), while stage 4 showed significant worsening over time (0.046 logMAR/year; P = 0.022) (Fig S1, available at www.ophthalmologyscience.org). In contrast, stage 3 progression followed a nonlinear quadratic trend (P < 0.001), which should be interpreted cautiously, particularly in light of possible survival bias. Noncataract patients had significantly better BCVA than cataract patients throughout stage 1-3, whereas the lens opacity showed no significant difference in BCVA in Stage 4 (Fig S2, available at www.ophthalmologyscience.org). Interocular symmetry was high across the cohort (R² = 0.66, P < 0.001). Clinically significant asymmetry was observed in 11 patients (19%), all of whom were aged over 40 years.

Figure 3.

Structural and functional assessments of CNGA1-associated retinitis pigmentosa. (A) Boxplot showing best-corrected visual acuity (BCVA) across stages. Horizontal black lines indicate medians and colored boxed indicate interquartile ranges (IQRs). (B) Boxplot showing ellipsoid zone (EZ) width across stages. Horizontal black lines indicate medians and colored boxed indicate IQRs. ∗∗P < 0.01, ∗∗∗P < 0.001, ∗∗∗∗P < 0.0001. (C) Scatter plot of full-field stimulus thresholds (FST) white stimulus thresholds versus the threshold difference (blue - red). The blue–red difference determined whether FST responses were rod-mediated (difference of < –20 dB), cone-mediated (difference > –10 dB), or mixed-mediated (difference between –20 and –10 dB). (D) Bubble plot of multiluminance mobility test (MLMT) performance relative to age and stage. Asterisks (∗) denote patients unable to complete the test; for visualization, their completion times were set to 125 s. (E) Heatmap of the mean change in retinal sensitivity across the 30-2 visual field. The color intensity at each of the 76 test loci represents the mean sensitivity change. The black dashed rectangle delineates the region of interest (ROI), identified as the area with the most significant sensitivity loss. (F) Comparison of annual retinal sensitivity decline between the central field and the ROI. The bar chart displays the annual retinal sensitivity decline per patient within the full 30-2 field (blue) versus the ROI alone (red). (G) Heatmap of the frequency (%) of ≥7 dB retinal sensitivity loss across the 30-2 visual field. The color intensity at each of the 76 test loci represents the frequency of a sensitivity decline ≥7 dB. (H) Comparison of progressing loci between the central field and the ROI. The bar chart displays the expected number of progressing loci (defined as ≥7 dB decline) per patient within the full 30-2 field (blue) versus the ROI alone (red). All data are standardized to a right-eye format (left-eye data were mirrored) in (E) and (G).

Ophthalmic and Multimodal Imaging Findings

Refractive data were available for 28 patients (56 eyes). The mean spherical equivalent was –2.40 ± 2.66 D (range, –8.88 to +2.25), and the mean cylinder was –1.30 ± 1.33 D. Myopia was the most common refractive error (19/28, 67.9%), with 14.3% (4/28) classified as high myopia. Hyperopia was identified in only 10.7% (3/28) of patients. Cataract (including pseudophakia) was highly prevalent, affecting 31 of 57 patients (54%). The mean age at initial cataract diagnosis was 47.3 ± 12.0 years. Surgical extraction had been performed in 21 patients at a mean age of 53.6 ± 11.0 years. Cataract morphology data were available for 10 patients. Anterior subcapsular cataracts were identified in 7 patients while posterior subcapsular cataracts were identified in 5 patients. Other ocular complications were rare, including single cases of keratoconus, glaucoma, and strabismus.

Fundus photography (40 patients available) revealed generalized vascular attenuation in all eyes. Waxy pallor of the optic disc was present in 82.5% (33/40) of patients, whereas peripheral bone-spicule pigmentation was present in 90% (36/40). Macular phenotypes were heterogeneous: 20% (8/40) presented with a normal-appearing macula, 32.5% (13/40) exhibited a blunted or opaque macular reflex, and 47.5% (19/40) showed frank signs of atrophy and pigmentary changes (Fig 4).

Figure 4.

Representative ultra-widefield fundus images of of CNGA1-associated retinitis pigmentosa and healthy controls. (A–B) Fundus images from a 24-year-old healthy control. (C–D) A 36-year-old patient (CN04). Note the vascular attenuation, despite the absence of pigmentary deposits and macular sparing. (E–F) A 38-year-old patient (CN13). Scattered bone-spicule pigmentation in the periphery and a blunted macular reflex were evident. (G–H) A 49-year-old patient (CN18). Extensive, confluent pigment deposition, and macular atrophy were present.

Fundus autofluorescence data were available for 55 patients. Although findings were highly symmetrical in both eyes, 2 patients (KR12 and CN19) showed discordant FAF stages between eyes. Of note, patient CN39 presented with an atypical phenotype characterized by an isolated hypoautofluorescence area in the nasal macula bilaterally. Although a large hyperautofluorescent ring was preserved, the presence of macular atrophy resulted in classification of both eyes as stage 3 (Fig S3, available at www.ophthalmologyscience.org). The FAF staging system demonstrated strong correlations with all structural and functional parameters except age (|r| > 0.8, P < 0.001) (Fig S4, available at www.ophthalmologyscience.org).

OCT imaging was available for all 57 patients (114 eyes). Epiretinal membranes were identified in 57.9% of eyes (66/114), affecting 63.2% of patients, with bilateral involvement observed in 83.3% of cases. Morphologically, 51.5% of these epiretinal membranes involved the fovea. The prevalence of epiretinal membranes increased with advancing disease stage (P < 0.001). Minor cystoid spaces were observed in 14.0% of eyes (16/114), while clinically significant cystoid macular edema (CME) involving the fovea was present in 12.3% (14/114). The majority (93.8%) of minor cystoid space cases were localized to the inner nuclear layer (INL). In contrast, CME was associated with progressive disease (stages 2-3) and frequently extended beyond the INL to the outer nuclear layer or the ganglion cell layer (42.9%).

The mean central foveal thickness at baseline was 199.9 ± 77.5 μm (range: 51–392) in eyes without CME. A reduction of 13.2 μm (95% confidence interval: –22.8 to –3.7; P = 0.004) was revealed over a mean follow-up of 2.03 years. Regarding EZ width, a significant negative correlation was observed with increasing disease stage (Fig S5 and Table S4, available at www.ophthalmologyscience.org), with significant Bonferroni-adjusted pairwise differences for all stage comparisons except stage 2 versus stage 3 (Fig 3B and Table S3, available at www.ophthalmologyscience.org). In stage-stratified mixed-effects models, stages 1 through 3 were best described by linear trends, with estimated declines of 88.6, 52.8, and 60.2 μm/year, respectively (all P ≤ 0.001), indicating a stage-dependent age-related decline pattern with the most rapid in stage 1 (Fig S6, available at www.ophthalmologyscience.org). In contrast, stage 4 exhibited a nonlinear age effect best fit by a quadratic model (P = 0.0336), with a continuing but decelerated decrease in EZ width across the observed age range.

Full-Field Electroretinography and Full-Field Stimulus Testing

Full-field ERG recordings were available for 39 patients. Consistent with a diagnosis of RP, all patients exhibited rod–cone dysfunction with high interocular symmetry (Fig S7, available at www.ophthalmologyscience.org). Residual responses were detected in only 9 of 39 patients (23%), and these patients were significantly younger than those with nonrecordable responses (mean age: 35.5 ± 13.7 vs. 48.7 ± 13.1 years; P = 0.015). Scotopic responses in these patients were severely reduced or nonrecordable, whereas photopic responses were relatively preserved, although attenuated. Recordable ERG responses were primarily observed in stage 1 (7/10 patients, 70%), whereas only 2 patients in later stages (stage 2: patient CN06; stage 3: patient KR08) exhibited recordable responses.

Full-field stimulus threshold testing was performed in 19 patients. According to the previous studies,30, 31, 32 a blue–red FST threshold difference of ≤10 dB indicates a cone-mediated FST threshold pattern. In our cohort, 18 patients (95%) exhibited a cone-mediated pattern. These patients exhibited a near-zero difference (0.59 ± 2.41 dB) between the blue and red thresholds and a white threshold higher than –30 dB (Fig 3C). As detailed in Table S4, available at www.ophthalmologyscience.org, FST response exhibited a progressive decline correlating with advancing stages.

Multiluminance Mobility Testing

Multiluminance mobility test was conducted in 19 patients. Except for one stage 3 patient, all patients in stages 1 to 3 achieved the maximum score of 6. Furthermore, all patients in stages 1 and 2 completed the course within 40 s. In contrast, no patients in stage 4 achieved a score of 6. All patients with scores below 6 were older than 40 years (Fig 3D).

Standard Automated Perimetry

Standard automated perimetry (Goldmann size III stimulus) was performed in 29 patients. At baseline, the mean VFI was 21.71% ± 20.14% (range, 0.00%–88.00%), the mean deviation was –27.55 ± 6.15 dB (range, –32.87 to –6.56), and the pattern standard deviation was 6.35 ± 3.19 dB (range, 1.89-13.06). The annual rate of decline was 0.64% for VFI (95% confidence interval, 0.32% to 0.97%; P < 0.001) and 0.17 dB for mean deviation (95% confidence interval, 0.06 to 0.28 dB; P = 0.007).

Sensitivity analysis was conducted in patients with serial follow-up (n = 6; mean duration 1.05 ± 0.19 years). An inferior-temporal region of interest (ROI) was identified (Fig 3E). The ROI exhibited an annual decline of 2.35 ± 1.48 dB, when compared with a central field decline of 0.89 ± 1.29 dB (P < 0.001) (Fig 3F). The distribution frequency of progressing loci revealed a mean of 5.83 progressing loci per patient across the entire 30-2 field over 1 year (Fig 3G and H). The defined ROI contributed 4.00 of these loci.

Genotype–Phenotype Correlations

No significant genotype–phenotype correlations were observed. Disease severity did not differ significantly between patients with 2 null mutations and those with at least 1 non-null mutation (P = 0.42), nor between homozygous and compound heterozygous carriers of c.253delC (P = 0.24) (Table 2).

Table 2.

Genotype–Phenotype Correlation Analysis in Patients with CNGA1 Variants

Phenotype Classification	Genotype Grouping∗ (N = 53)		Further Stratification (N = 47)
	2 Null Mutations, n (%)	≥1 Non-Null Mutations, n (%)	Homozygous c.253delC, n (%)	Heterozygous c.253delC, n (%)
Earlier stage (stage 1-2)	22 (47.8%)	2 (28.6%)	12 (42.9%)	12 (63.2%)
Later stage (stage 3-4)	24 (52.2%)	5 (71.4%)	16 (57.1%)	7 (36.8%)
P value†	0.42		0.24

^∗There were 55 patients with fundus autofluorescence stage classification and 2 patients (CN19, KR12) were excluded due to interocular asymmetry.

^†Calculated using Fisher exact test.

Discussion

In this retrospective cohort study, we characterize the genotypic and phenotypic landscape of CNGA1-associated RP in a large East Asian cohort. Based on the FAF staging, we evaluated the structural and functional changes at different stages and identified the outcome rate of CNGA1-associated RP. Despite a decline exceeding the conventional US Food and Drug Administration therapeutic efficacy threshold of 15 ETDRS letters (0.3 logMAR) per year in 1 eye (aged 67 years), the overall stability of visual acuity limited its utility as a sensitive endpoint for short-term gene therapy trials. Interocular asymmetry in measures for future clinical trials.

Analysis of visual acuity showed slow progression BCVA was noted in 19% of patients; only 18% of these cases were attributable to fundus asymmetry. These findings indicate that CNGA1-associated RP manifested symmetrically in the majority of the cohort and that the use of the contralateral eye as an internal control for future interventional trials might be possible.

OCT imaging confirmed that the outer nuclear layer and the EZ were preserved until advanced disease stages (stage 4), despite early onset of rod dysfunction. The prolonged structural preservation suggests a broad therapeutic window for CNGA1-associated RP. Furthermore, we found that EZ width exhibited stage-dependent, age-related decline patterns. This phenomenon may result from a shift in cell types at the transition zone, which corresponds to the boundary of the hyperautofluorescent ring.³³ As the hyperautofluorescent ring contracts within the vascular arcades (stages 2-3), the transition zone involves a greater proportion of cones. Cones are spared from the primary molecular defect due to the absence of CNGA1 expression, which may contribute to the slower decline in EZ width.

Regarding visual fields, we identified an inferior-temporal ROI exhibiting a rapid sensitivity loss of 2.35 dB/year. Current US Food and Drug Administration guidelines for therapeutic efficacy require a sensitivity decline of ≥7 dB at ≥5 prespecified loci.³⁴ Our longitudinal analysis revealed a mean of 5.83 progressing loci per patient over a 1-year period of which 4.00 were located within the defined ROI. This pattern of rapid focal degeneration suggests that ROI-based retinal sensitivity may serve as a sensitive outcome measure.

Functionally, the predominance of cone-mediated thresholds on FST indicated a floor effect in rod sensitivity, rendering FST insensitive to monitoring degenerative progression at early disease stages. Unlike previous studies which uniformly reported nonrecordable ERG responses,8, 9, 10, 11 the larger sample size of our cohort allowed identification of responses in a subset of younger patients. However, these residual responses were significantly below the lower limit of normal. The restricted dynamic range made it unlikely for ERG to effectively reflect the disease progression. Moreover, while ERG represents the entire functioning retina, FST represents the most sensitive regions of the functioning retina.²⁸ Therefore, FST may be still useful in future studies for detecting functional restoration of rods, which is consistent with its successful application in prior gene therapy trials.^3,35 The MLMT demonstrated a ceiling effect in our cohort. Potential trial candidates (stage 1-3) retain sufficient cone-mediated central vision to navigate effectively under the tested lighting conditions. This finding suggests that MLMT lacks sufficient score range to detect functional changes in this population. Given the slow disease progression and the limitations of these functional assessments in monitoring early-stage CNGA1-associated RP, more sensitive parameters for detecting macula dysfunction are warranted. Specifically, multifocal ERG and microperimetry could provide earlier detection of subtle macular impairment.^36,37 Future studies incorporating these modalities may be instrumental in identifying sensitive biomarkers, thereby enhancing clinical trial readiness.

Among ocular complications, although the overall prevalence of cataracts in our cohort aligns with the approximately 50% reported in general RP populations,³⁸ the morphological subtypes were atypical. The proportion of anterior subcapsular cataracts is high (70% in our cohort), which contrasts with the posterior subcapsular opacities (50% in our cohort) commonly reported in RP.^38,39 Although CNGA1 is traditionally considered as a rod-specific gene, further analysis of the iSyTE database revealed Cnga1 expression in the lens epithelium of adult mice (Table S5, available at www.ophthalmologyscience.org). Epithelial–mesenchymal transition has been postulated to promote epithelium fibrosis in anterior subcapsular cataract formation,⁴⁰ and CNGA1 variants may be involved in this process by affecting ion channel function. This finding revealed a previously unrecognized aspect of CNGA1-associated pathology, distinguishing it from the classic RP phenotype. In addition, the minor cystoid spaces and CME in CNGA1-associated RP were localized to the INL, differing from the pattern observed in general RP, where edema typically involves the outer nuclear layer.^41,42 Since the therapeutic efficacy of carbonic anhydrase inhibitors is primarily attributed to enhanced fluid transport across retinal pigment epithelium,^43,44 fluid predominantly accumulating within the INL may be less accessible to this clearance mechanism in CNGA1-associated RP.

The c.253delC variant is the most common variant in this East Asian cohort. This genetic homogeneity implies that mutation-specific genome editing or prime editing strategies targeting this variant may be feasible. In contrast, the c.947C > T variant is the predominant mutation in other ethnic populations, including European [non-Finnish], African, and Middle Eastern patients (Table S6, available at www.ophthalmologyscience.org). No significant genotype–phenotype correlation was observed in our cohort. This is likely due to the strong pathogenicity of the founder mutation. The c.253delC variant causes a frameshift and a premature termination codon at position 89 (p.Leu85Phefs∗4) within the 686-amino acid CNGA1 protein. Due to the proximity of the truncation to the N-terminus, nonsense-mediated decay was expected to degrade the mRNA transcript.⁴⁵ Even if translation were to occur, the resulting truncated peptide would lack essential functional domains. Consequently, this variant may induce the clinical phenotype by reducing channel expression below a functional threshold. The residual activity provided by the milder allele in compound heterozygotes seems insufficient to exceed the functional threshold, resulting in a phenotype comparable to that of individuals with biallelic null variants.

Despite its multicenter design and inclusion of 57 patients, this study is limited by its retrospective and uncontrolled nature. Nonetheless, the relatively large cohort partially mitigates these constraints, providing a valuable dataset for an otherwise rare condition. Caution is warranted when generalizing the current study findings to populations with different CNGA1 mutational spectra.

In conclusion, this study represents the first multicenter retrospective natural history study in patients with CNGA1-associated RP. We characterized stage-dependent structural and functional changes and evaluated the application of candidate outcome measures for clinical trials. A broad therapeutic window was identified prior to stage 4, and the high prevalence of the c.253delC variant highlights the potential for mutation-specific gene editing approaches in East Asian populations. Among functional endpoints, BCVA and MLMT lacked the sensitivity required for clinical trials, whereas FST exhibited a floor effect but may remain a sensitive indicator for detecting rod functional restoration. Static perimetry, especially ROI-based retinal sensitivity, demonstrated the greatest potential as an efficacy endpoint. Collectively, these findings support clinical trial readiness by informing patients selection and outcome measure optimization in future gene therapy studies.

Manuscript no. XOPS-D-26-00028.