Genetic Spectrum of Negative Electroretinograms in a Predominantly Pediatric Cohort of 177 Patients

Abstract

Purpose

The purpose of this study was to elucidate the common and rare genetic causes of negative electroretinograms (nERGs) and their association with systemic disease and with myopia in a predominantly pediatric cohort.

Methods

Patients underwent electroretinogram (ERG) testing at the Hospital for Sick Children (Toronto) between 2007 and 2023. Negative ERG was defined as b/a amplitude of <1 to a dark-adapted 3.0 and/or 10.0 cd*s*m⁻² stimulus. Genetic testing results were reanalyzed using American College of Medical Genetics guidelines. Genes accounting for <2.5% of cases were defined as rare.

Results

Of 4347 ERGs performed, 293 (6.7%) cases had nERGs. Among these, 276 (94.1%) were classified as inherited; 193 had genetic testing. Of these, 177 (91.7%) had an established genetic diagnosis involving 41 genes. Major phenotypes included congenital stationary night blindness (CSNB, 55.4%), retinoschisis (18.6%), and photoreceptor dystrophies (17.5%). Both common (CACNA1F, RS1, TRPM1, NYX, and IDUA) and rare genetic associations were identified. Among patients with CSNB, postsynaptic ON-bipolar genes were associated with high myopia, increasing refractive error by −6.12 diopters (D). Collectively, rare causes affected the same number of cases as CACNA1F, the most frequently associated gene (27.7% each). Genetic etiology varied (23 genes; 31 cases) among photoreceptor dystrophies causing nERG. Systemic disease affected 31 cases. Five novel genetic associations (ABHD12, AP3B2, OAT, PCDH15, and PDE6A) were found.

Conclusions

This study expands the genetic spectrum underlying nERGs, confirming prior associations and identifying five novel genes associations. We provide evidence for the role of ON-bipolar pathway in myopia development. In photoreceptor dystrophies, nERGs rarely occur and likely represent a transient epiphenomenon independent of the underlying gene or mechanism.

The negative electroretinogram (nERG) is defined as the selective reduction of the scotopic bright flash b-wave amplitude below that of the photoreceptor-driven a-wave, irrespective of whether the a-wave amplitude is normal or reduced. Acquired causes of the nERG have been well-described, and include retinal vascular occlusions, vitamin A deficiency, toxic exposures (iron, methanol, quinine, and vigabatrin), melanoma-associated retinopathy, birdshot chorioretinitis, and others.^1–5 In contrast, the catalog of its genetic underpinnings continues to expand as genetic testing diagnostic yield and access improve. Understanding this complexity offers insight into the structure and function of retinal circuitry, its relationship to systemic disease and eye development, and is clinically relevant for increasing diagnostic precision and development of emerging gene-specific therapies.

Since Koh et al. published the first cohort of 128 patients with nERGs in 2000, follow-up studies (sample sizes ranging from 47–139) showed that the dominant categories of inherited retinal disorders causing nERG included congenital stationary night blindness (CSNB), X-linked retinoschisis (XLRS), and photoreceptor or retinal pigmental epithelium (RPE) dystrophies.^2–4 Recently, Katta et al. described the common genes known to cause nERG in 122 patients with CSNB, with CACNA1F, TRPM1, and NYX being the most frequent causes,⁶ which held true in a smaller cohort of pediatric patients by Huang et al.⁷ Apart from these reports, previous studies did not systematically examine the genetic etiology of nERGs. In addition, to our knowledge, the association of nERG with multisystem disease has not been investigated outside of focused case series or isolated case reports.

Here, we describe clinical and genetic associations of nERGs in a large predominantly pediatric cohort with a high representation of systemic disease. To improve the accuracy of variant classification, all genetic testing results were reinterpreted according to current ClinGen Sequence Variant Interpretation Working Group (ClinGen SVIWG) recommendations for using American College of Medical Genetics (ACMG) guidelines for variant interpretation.^8–13 We report both common and rare genetic associations with nERG, five novel associations (ABHD12, AP3B2, OAT, PCDH15, and PDE6A), and explore the role of nERG in systemic disease and association with myopia.

Methods

Subjects

This study was approved by the Hospital for Sick Children research ethics board (REB: 1000017804) and adheres to the Declaration of Helsinki. Charts of patients with nERG recordings at the Visual Electrophysiology Unit at the Hospital for Sick Children (HSC; Toronto, Canada) completed between January 2007 and January 2023 were reviewed. The nERGs were identified by a dark-adapted (DA) mixed rod-cone b:a wave ratio of less than 1 in at least one eye, as per International Society for Clinical Electrophysiology of Vision (ISCEV) standards at the time of testing.^14–17 Accordingly, cases with a b/a ratio <1 to (i) both the DA standard (2.29 cd·s·m⁻² or 3.0 cd·s·m⁻²) and bright flash (7.6 cd·s·m⁻² or 10.0 cd·s·m⁻²) stimuli, or (ii) bright flash stimulus only, or (iii) standard flash stimulus only (if bright flash was not available), were included. Data collected included: age and sex at the time of testing, best-corrected visual acuity, refraction, subjective complaints (such as nyctalopia, photophobia, dyschromatopsia, and photopsia), color vision testing (if available), fundus features, medical history including those from non-ophthalmic clinical specialties, ocular history, genetic testing results, and pedigree.

The cases were categorized as “inherited” or “acquired” by expert consensus of authors Elise Heon and Ajoy Vincent based on phenotype and history. Specifically, cases that were classified as inherited exhibited typical features consistent with an inherited retinal dystrophy for the suspected phenotype (e.g., bilaterality, symptoms present at a young age, and characteristic phenotype of a known retinal dystrophy), had a suggestive family history (e.g., other affected family members and consanguinity), and had no evidence of nutritional deficiency, no evidence of toxic exposure, no other known signs or symptoms of acquired disease (e.g., positive dysphotopsia), or other known acquired cause of nERG.

Subjects were categorized based on the clinical phenotype to facilitate analysis. CSNB included patients with normal fundus and imaging (fundoscopy, fundus autofluorescence [FAF] and optical coherence tomography [OCT]) and those with abnormal fundus diagnosed with Oguchi disease or fundus albipunctatus. The retinoschisis category included patients with retinal schisis on OCT, spoke-wheel appearance on fundus imaging, and a family history suggestive of X-linked inheritance. The photoreceptor/RPE dystrophy category included patients with imaging features consistent with degenerative photoreceptor or RPE disease on imaging, such as evidence of outer retinal and/or RPE atrophy on OCT, abnormal fundus hypo-autofluorescence, and bony spicules or pigment clumping on fundoscopy. Cases that could not be assigned into these categories were grouped as “others.” All patient charts were reviewed for the presence of systemic disease. Common (e.g., attention deficit hyperactivity disorder, hypothyroidism, eczema, and asthma) or age-related (e.g., hypertension and Parkinson's disease) systemic conditions were classified as unrelated to the underlying genetic diagnosis, unless they were previously associated with the disorder. Patients with disorders that resulted from metabolic dysfunction (e.g., mucopolysaccharidosis, neuronal ceroid lipofuscinosis, and others) confirmed either by metabolic assays or genetic diagnosis were classified as having metabolic disorders. Patients with features of neurologic or neurodevelopmental dysfunction without an established metabolic disturbance were classified as having neurodevelopmental disorders. Patients with immune deficiency or autoimmune disease were classified as having immune system disorders. Patients with hearing loss were further subclassified when appropriate to distinct types of Usher Syndrome based on clinical features and genetic cause. Overall, patients were classified based on clinical diagnosis into CSNB, photoreceptor/RPE dystrophy, retinoschisis, or “others.” Some patients within the first three categories and all patients within the “others” category had associated multisystem disease.

Electroretinography

Full-field electroretinograms (ffERG) were recorded according to the ISCEV standard at the time of recording. DA ERG a- and b-wave amplitudes and implicit times, and b:a ratios were extracted for analysis, and a-wave amplitudes were categorized as normal or reduced for all nERG cases. Light-adapted (LA) 3.0 or 4.1 a-wave morphology was categorized as normal, broadened flat trough (indicative of selective cone ON-bipolar dysfunction), and U-shaped broadened trough (indicative of cone ON- and OFF-bipolar cell dysfunction).

ERGs were predominantly performed on the Espion Diagnosys LLC (Lowell, MA, USA) system with the exception of two patients who were tested using the LKC Technologies (Gaithersburg, MD, USA) system. Recordings were performed using either the Burian-Allen Electrode (Hansen Ophthalmic Development Lab; Coralville, IA, USA), DTL Plus Electrode (Diagnosys LLC, Lowell, MA, USA), or Jet electrode (Fabrinal SA, La Chaux-de-Fonds, Switzerland). Gold-Cup electrodes (Grass Technologies, West Warwick, RI, USA) were placed at the forehead as a ground and at the orbital rim for reference (when DTL or Jet electrodes were used). ERGs were either performed with the patient awake, under sedation with chloral hydrate (Odon, Montreal, Quebec, Canada), or under general anesthesia. Binocular ERGs were performed on pharmacologically dilated pupils. DA testing preceded LA testing.^14–16 Data from the right eye was analyzed unless the only available eye was the left eye or the left eye recordings were judged as more reliable.

Normal amplitudes are based on internal controls from a cohort of non-sedated subjects in the clinic (n = 45), subjects under chloralhydrate sedation (n = 35), and subjects under general anesthesia (n = 28). The age range of non-sedated control participants was between 7 and 50 years, whereas those under sedation or general anesthesia were younger than 7 years. Each study patient's amplitude parameters were compared to the age- and equipment-matched control group at the time of the test. The same statistical methods were used to analyze all patients.

Imaging

Color fundus photographs were acquired using true color imaging using the Zeiss Clarus (Carl Zeiss AG., Oberkochen, Germany) or Topcon (Tokyo, Japan) camera systems or ultrawidefield pseudocolor scanning laser ophthalmoscopy (Optos, Dunfermline, UK) in awake and cooperative patients or the RetCam (Natus Sensory Inc., IL, USA) in patients under sedation or under anesthesia. When possible, shortwave autofluorescence images were acquired using the Zeiss Clarus or ultra widefield imaging systems (Optos) in patients not requiring sedation or anesthesia. Spectral domain OCT images were acquired using the Zeiss Cirrus (Carl Zeiss Meditec, Inc., Dublin, CA, USA) or Heidelberg Spectralis (Heidelberg Engineering GmbH, Heidelberg, Germany) systems in awake patients or the Bioptigen Envisu (Leica Microsystems, Wetzlar, Germany) in sedated patients or patients under anesthesia. Overall, imaging data were available for 179 of 193 patients with genetic testing.

Genetic Variant Curation and Re-Assessment

Given that the study period review spanned 17 years, during which new guidelines for genetic variant interpretation by the ACMG and several recommendations for their application by the ClinGen SVIWG were published,^8–11,13 we re-classified all detected variants according to current recommendations. Briefly, all variants were remapped to the most recent human genome GRCh38.p14 build and subsequently reannotated with SnpEff and SnpSift¹⁸ to extract associated ClinVar data (where available), in silico pathogenicity predictions (PP3 criterion), and population allele frequencies (PM2 criterion). We assigned PP4 evidence based on estimated diagnostic yield (probability of obtaining a genetic diagnosis) of the genetic test used, and reviewed patient pedigrees to assign evidence for variant segregation (PM3 for autosomal recessive variants in trans to a pathogenic variant, PM6 for presumed de novo variants, PS1 for confirmed de novo variants, and PP1 for variant co-segregation). Interpretation of pathogenicity for a given variant (e.g., “Pathogenic” or “Likely Pathogenic” interpretation) was then determined by adding up the strength of evidence as proposed in the 2015 ACMG guidelines (see Table 5 in Richards et al.). Interpretations were cross-referenced with ClinVar for variants with ClinVar records. Discrepancies with ClinVar annotations were resolved by consensus among authors Kirill Zaslavsky and Ajoy Vincent according to current guidelines and clinical context.

Data Analysis

A given gene was defined as a rare cause of nERG when it affected less than 2.5% of the entire cohort. Descriptive statistics (mean and standard deviation) are reported for baseline variables between groups. Fisher's exact test was used to assess for over-representation of genes within different groups. Linear regression was used to determine which factors influence refractive error in patients with CSNB. Alpha level of 0.05 was considered statistically significant. All analyses were completed in R (version 4.2.1). The code used for the following tasks is available at https://github.com/kzaslavsky/nERG_HSC.

Results

Between 2007 and 2023, 4,347 subjects underwent ERG testing as part of standard of care at HSC. Negative ERGs were recorded in 293 (6.7%) cases, of which 276 (94.1%) were classified as inherited. Genetic testing results were available for 193 individuals with 180 unique genetic variants reported in 67 genes. When re-analyzed using current ClinGen recommendations, classification of 81 (45%) variants remained consistent with prior Pathogenic (P) or Likely Pathogenic (LP) interpretations in ClinVar. (Supplementary Table S1). Next, we classified 74 (41.1%) as Pathogenic/Likely Pathogenic, of which 47 were novel to ClinVar (Supplementary Table S2), whereas 27 were listed in ClinVar as variants of unknown significance (VUS), Likely Benign (LB), or with conflicting interpretations of pathogenicity (Supplementary Table S3). The remaining 25 (13.9%) variants were re-classified as VUS or LB (Supplementary Table S4). These efforts altered interpretation in 18 cases, enabling genetic diagnosis for 177 patients involving 41 genes (155 variants) overall (Supplementary Fig. S1; Supplementary Table S5).

For these 177 patients, the mean age at presentation was 12.9 ± 10.9 years (median = 11 years), the majority were male patients (142 cases, 80.2%), mean best-corrected visual acuity (BCVA) was 0.5 ± 0.3 logMAR, and mean spherical equivalent for those aged 5 years and above was −2.5 ± 5.7 diopters (D). The most common clinical diagnosis was CSNB (98 cases, 55.4%), followed by retinoschisis (33 cases, 18.6%), photoreceptor/RPE dystrophy (31 cases, 17.5%), and mucopolysaccharidosis (MPS) retinopathy (10 cases, 5.6%; 9 of MPS type I and 1 case of MPS type III). Multisystem disease affected 31 patients (17.5%), with metabolic disorders affecting 18 (58.1%) of these subjects. The majority (137 cases, 77.8%) had normal DA 3.0 and 10.0 a-wave amplitudes, whereas 39 (22.2%) cases had reduced a-wave amplitudes (Table 1; Fig. 1A). The morphology of the LA single flash wave was analyzed in 176 cases; was normal in 82 (46.6%), broadened with a flat trough in 39 (22.1%), and showed U-shaped broadened trough in 55 (31.3%; Supplementary Fig. S2).

Table 1.

Characteristics of 177 Genotyped Patients With nERG Grouped by Phenotype

	CSNB (N = 98)	Retinoschisis (N = 33)	Photoreceptor/RPE Dystrophy (N = 31)	Others (N = 15)	Total (N = 177)
Age, y
Median (IQR)	9.0 (6.0 to 13.8)	14.0 (9 to 18)	15.0 (9.0 to 21.5)	9.0 (7.5 to 12.5)	11.0 (3.0 to 43.4)
Range	0.4 to 66.0	1.0 to 46.0	0.6 to 63.0	4.0 to 18.0	0.4 to 66.0
Sex
Male	83 (84.7%)	33 (100.0%)	19 (61.3%)	7 (46.7%)	142 (80.2%)
BCVA (logMAR)
Mean (SD)	0.5 (0.3)	0.6 (0.2)	0.6 (0.5)	0.6 (0.3)	0.5 (0.3)
Range	0.0 to 1.5	0.2 to 1.3	0.0 to 2.0	0.2 to 1.2	0.0 to 2.0
Refractive error, diopters
Mean (SD)	−5.0 (6.0)	1.5 (2.1)	−0.0 (3.6)	0.8 (3.0)	−2.5 (5.7)
Range	−19.0 to 7.2	−2.2 to 8.8	−11.4 to 4.5	−3.6 to 6.8	−19.0 to 8.8
A-wave type
Normal	85 (87.6%)	31 (93.9%)	9 (29.0%)	12 (80.0%)	137 (77.8%)
Reduced	12 (12.4%)	2 (6.1%)	22 (71.0%)	3 (20.0%)	39 (22.2%)
DA 0.01 b-wave amplitude, µV
Undetectable	45	22	22	7	96
Mean (SD)	61.9 (50.2)	97.9 (52.4)	53.0 (51.7)	57.0 (36.8)	65.4 (50.6)
Range	4.0 to 204.0	30.0 to 202.0	7.0 to 152.0	14.0 to 99.0	4.0 to 204.0
DA 3.0 a-wave amplitude, µV*
Mean (SD)	−186.1 (81.9)	−213.9 (81.1)	−78.2 (75.6)	−157.1 (76.1)	−170.3 (91.2)
Range	−504.0 to −40.0	−365.0 to −88.0	−373.0 to −16.0	−298.0 to −30.0	−504.0 to −16.0
DA 3.0 b-wave amplitude, µV*
Mean (SD)	105.5 (58.3)	157.0 (89.6)	63.4 (63.0)	124.7 (92.4)	109.6 (74.4)
Range	3.0 to 303.0	14.0 to 415.0	4.0 to 280.0	21.0 to 327.0	3.0 to 415.0
B:A ratio
Mean (SD)	0.6 (0.2)	0.7 (0.2)	0.7 (0.2)	0.7 (0.2)	0.6 (0.2)
Range	0.1 to 1.0	0.1 to 1.0	0.1 to 0.9	0.3 to 1.0	0.1 to 1.0
30 Hz flicker peak amplitude, µV
Mean (SD)	50.5 (36.2)	53.6 (26.1)	30.5 (21.0)	79.6 (62.1)	50.9 (37.5)
Range	4.0 to 143.0	14.0 to 140.0	3.0 to 80.0	5.0 to 216.0	3.0 to 216.0
30 Hz flicker implicit time, ms
Mean (SD)	31.6 (4.4)	32.8 (3.7)	36.0 (5.3)	30.3 (4.3)	32.4 (4.7)
Range	22.0 to 40.0	27.0 to 41.0	27.0 to 46.0	25.0 to 40.0	22.0 to 46.0

BCVA, best-corrected visual acuity; CSNB, congenital stationary night blindness; DA, dark-adapted; nERG, negative ERG; LA, light-adapted; logMAR, logarithm of the minimum angle of resolution; RPE, retinal pigment epithelium.

^*DA 3.0 includes cases with 3.0 or 2.29 cd * s/m2.

Figure 1.

Genetic etiology of nERG. (A) Alluvial graph of the major phenotypic categories and their relation to dark-adapted a-wave amplitude an underlying genotype. (B) Proportion of solved genetic etiologies for negative electroretinograms in the entire cohort and (C) for patients with congenital stationary night blindness.

The 5 genes most frequently associated with nERG were CACNA1F, RS1, TRPM1, NYX, and IDUA, accounting for 122 subjects (68.9%; Fig. 1B). Ninety-six (54.2%) patients had variants in 6 X-linked (XL) genes, 7 patients in 4 autosomal dominant (AD) genes, 72 patients in 30 autosomal recessive (AR) genes, and 2 patients in 1 gene (CRX) with both AD and AR patterns of inheritance. Rare genetic causes collectively and CACNA1F were the most frequent etiologies, accounting for more than a quarter of all genetically solved cases each (49, 27.7%; Table 2; Supplementary Table S6). Among rare genetic causes, 15 genes had well-established associations with nERG.^19–38 An additional 16 genes were previously reported in isolated cases or small case series.^39–53 Last, we report first observations of nERGs in five cases with variants in ABHD12, AP3B2, OAT, PDE6A, and PCDH15 (Supplementary Table S7).

Table 2.

Genetic Etiology of 177 Solved Cases Grouped by Phenotype

	CSNB (N = 98)			Photoreceptor/RPE Dystrophy (N = 31)
Genetic Etiology	Isolated (N = 93)	Multisystem Disease (N = 5)	Retinoschisis (N = 33)	Isolated (N = 20)	Multisystem Disease (N = 11)	Others (N = 15) Multisystem Disease (N = 15)	Total (N = 177)
CACNA1F	48 (51.6%)	1 (20.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	49 (27.7%)
TRPM1	18 (19.4%)	1 (20.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	19 (10.7%)
NYX	12 (12.9%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	12 (6.8%)
CABP4	6 (6.5%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	6 (3.4%)
RS1	0 (0.0%)	0 (0.0%)	33 (100.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	33 (18.6%)
IDUA	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	9 (60.0%)	9 (5.1%)
Rare	9 (9.7%)	3 (60.0%)	0 (0.0%)	20 (100.0%)	11 (100.0%)	6 (40.0%)	49 (27.7%)

Among the 98 patients with an established genetic etiology for CNSB, CACNA1F accounted for 50%, TRPM1 for 19.4%, NYX for 12.2%, and CABP4 for 6.1% (Fig. 1C). The remaining 12.2% of associations were rare, and included GNB3, GNB5, GPR179, RHO, GRK1, and RDH5. The latter two genes accounted for in three cases (1 for GRK1 and 2 for RDH5) with fundus findings supporting diagnoses of Oguchi disease and fundus albipunctatus, respectively. Eighty-three were male patients (84.7%), the average age was 10.8 ± 10 years, and the average refractive error was myopic at –5.0 ± 6.0 D of spherical equivalent (see Table 1). Five cases (5.1%) had multisystem disease, lower than expected by chance given the 17.5% frequency of multisystem disease in the overall cohort (odds ratio [OR] = 0.11, 95% confidence interval [CI] = 0.03 to 0.31, Fisher's Exact Test, P < 0.001). The underlying genetic diagnosis explained both retinal ON-bipolar signaling defect and systemic disease in two cases (POLG and GNB5) and could not explain multisystem disease in the other three (CACNA1F, GNB3, and TRPM1).

Patients in the CNSB group aged over 5 years were myopic with a bimodal distribution of refractive error (peaks at −6.61 D and −0.66 D; Fig. 2A). Specifically, genes involved in post-synaptic ON bipolar cell function (TRPM1, NYX, and GPR179) were associated with a significantly higher myopic refractive error (−6.12 D, 95% CI = −8.60 to −3.65, P < 0.001) relative to other genes after adjusting for age, sex, BCVA, and ERG parameters (Table 3). Spherical equivalent in cases secondary to variants in TRPM1 (−8.9 ± 1.6 D) and NYX (−10.7 ± 1.4 D) together was approximately 5.7 D more myopic than that secondary to CACNA1F (−3.9 ± 0.9 D) or other genes (−2.4 ± 1.4 D; Figs. 2B, 2C).

Figure 2.

Refractive error in patients older than 5 years of age with CSNB. (A) Mean refractive error (spherical equivalent in diopters) for patients with CSNB shows a bimodal distribution with peaks at −6.61 and −0.66. (B) Mean refractive error grouped by localization and (C) by associated gene; “others” include genes with less than three patients with recorded refractive errors (CABP4, GPR179, and GNB3). Each point is the mean refractive error from both eyes of one patient. Mean ± standard error of the mean is plotted.

Table 3.

Predictors of Refractive Error in Patients With CSNB Over Age 5 Years

	Refractive Error (Spherical Equivalent in Diopters)
Predictors	Estimates	CI	P Value
(Intercept)	−14.42	−22.15 to −6.69	<0.001
Post-synaptic gene	−6.12	−8.60 to −3.65	<0.001
Age	−0.05	−0.20 to –0.11	0.537
Sex, M	3.42	−0.74 to –7.58	0.105
BCVA, logMAR	0.31	−5.43 to –6.06	0.913
DA 3.0 a-wave amplitude, µV	−0.00	−0.02 to –0.01	0.713
DA 3.0 b:a ratio	−3.21	−9.08 to –2.66	0.277
LA 30 Hz flicker peak amplitude, µV	0.05	0.01 to –0.10	0.027
Observations	60
R2/R2 adjusted	0.478/0.408

P values meeting the statistical significance threshold (α = 0.05) are bolded.

All 33 patients with XLRS harbored pathogenic variants in RS1 and none had multisystem disease. The mean age at testing was 16.5 ± 11.1 years, visual acuity was 0.6 ± 0.2 logMAR, and refractive error was mildly hyperopic at +1.50 ± 2.1 D spherical equivalent (see Table 1). The ffERG demonstrated normal amplitude a-waves in 31 patients (93.9%), indicating largely preserved photoreceptor function. Among patients older than 5 years, visual acuity was associated with increasing age after adjusting for refraction and ERG parameters, with a loss of 3 lines (logMAR 0.3) every 3 decades (Table 4). In contrast to a previous report,⁵⁴ patient age was not predictive of b:a ratio (Supplementary Table S8).

Table 4.

Predictors of Visual Acuity in Patients With Retinoschisis Over Age 5 Years

	BCVA (logMAR)
Predictors	Estimates	CI	P Value
(Intercept)	0.56	0.29 to 0.84	<0.001
Age	0.01	0.00 to 0.02	0.007
Refractive Error (Spherical Equivalent in D)	0.03	−0.00 to 0.07	0.062
DA 3.0 a-wave amplitude (µV)	−0.00	−0.00 to 0.00	0.841
DA 3.0 b:a ratio	−0.19	−0.48 to 0.09	0.178
LA 30 Hz flicker peak amplitude, µV	−0.00	−0.01 to 0.00	0.061
Observations	26
R2	0.401

P values meeting the statistical significance threshold (α = 0.05) are bolded.

Among the 31 patients with photoreceptor/RPE dystrophies, all were associated with rare genetic causes. Eleven cases (35.4%) had multisystem disease, enriched relative to the overall cohort (OR = 3.43, 95% CI = 1.29 to 8.94], Fisher's Exact Test, P = 0.008). The ffERG showed cone-rod and rod-cone patterns of dysfunction in an equal number of cases (15 each). One case with a congenital glycosylation disorder associated with PMM2 had isolated rod bipolar cell dysfunction, representing an early stage of retinopathy in this condition. In at least four cases of rod-cone dystrophies, the photopic hill phenomenon in the dark was responsible for nERG.

Of the remaining 15 cases that did not meet criteria for CSNB, retinoschisis, or photoreceptor/RPE dystrophy, all had an associated multisystem disease, 10 (66.7%) of which were MPS. Overall, these patients were the youngest among the cohort, with the average age at testing of 10 ± 4 years, average refractive error of 0.8 ± 3.0 D, and preserved a-wave amplitudes in 12 cases (80%).

Thirty-one cases (17.5%) in this nERG cohort had multisystem disease. Eighteen (58.1%) had metabolic disorders. Ten of these were MPS, and nine had variants in IDUA (MPS type I), and one had a variant in HGSNAT (MPS type III). The remaining eight cases were neuronal ceroid lipofuscinoses (TPP1 and CLN3), as well as disorders of glycosylation (PMM2 and SRD5A3), peroxisomal disorders (PEX1), mitochondrial disorders (POLG), and ornithine aminotransferase deficiency (OAT). Six cases (19.4%) were diagnosed with neurodevelopmental disorders, with the involved gene explaining both the systemic disease and nERG in four (AP3B2, ERCC6L2, GNB5, and WDR73) and only the nERG in two (CACNA1F and GNB3). An additional three cases had immune system disorders, with the involved gene explaining both the systemic disease and the nERG in two cases (CD40LG and RNU4ATAC), and only the nERG in one (MERTK). Usher Syndrome was diagnosed in three cases (MYO7A, PCDH15, and USH2A). Last, anterior pathway glioma was diagnosed in a patient with CSNB secondary to biallelic TRPM1 variants, a deletion of exons 2 to 7 and c.2491T>C (p.[Phe870Leu]; NM_001252020.2; Table 5).

Table 5.

Multisystem Disorders in Cases With nERGs With Solved Genetic Etiology

Condition	Gene	Number of Cases
Metabolic disorder		18 (58.1%)
Mucopolysaccharidosis		10
Type I (Hurler)	IDUA	9
Type III (Sanfilippo)	HGSNAT	1
Neuronal ceroid lipofuscinosis		2
NCL type 2	TPP1	1
NCL type 3 (Batten)	CLN3	1
Glycosylation disorder		3
PMM2-CDG	PMM2	2
SRD5A3-CDG	SRD5A3	1
Others		3
Mitochondrial disorder (SANDO)	POLG	1
Peroxisomal disorder, Zellweger syndrome	PEX1	1
Gyrate atrophy	OAT	1
Neurodevelopmental disorder		6 (19.4%)
EOEE	AP3B2	1
Bone marrow failure syndrome 2	ERCC6L2	1
Lodder-Merla syndrome type 1	GNB5	1
Unspecified, seizures, ataxia, learning disability, ADHD, clinodactyly*	GNB3	1
Unspecified, dysmorphic features, developmental delay, spasms*	CACNA1F	1
GAMOS	WDR73	1
Immune system disorder		3 (9.7%)
Primary X-linked hyper IgM syndrome	CD40LG	1
Roifman syndrome	RNU4ATAC	1
SLE*	MERTK	1
Hearing loss		3 (9.7%)
Usher syndrome type 1A	MYO7A	1
Usher syndrome type 1F	PCDH15	1
Usher syndrome type 2A	USH2A	1
Anterior pathway glioma*	TRPM1	1 (3.2%)

ADHD, attention-deficit hyperactivity disorder; EOEE, early onset epileptic encephalopathy; GAMOS, Galloway-Mowat spectrum disorder; PMM2-CDG, phosphomannomutase deficiency, congenital disorder of glycosylation type 1a; SANDO, sensory ataxic neuropathy, dysarthria, ophthalmoparesis; SLE, systemic lupus erythematosus; SRD5A3-CDG, Steroid 5α-reductase congenital disorder of glycosylation type 3.

^*Unlikely to be explained by the involved gene.

Among the five cases with novel associations, four were classified as photoreceptor/RPE dystrophies (ABHD12, OAT, PDE6A, and PCDH15). For these, the average age at testing was 15.5 years (range = 7–24 years), and 3 were male patients. The most common complaint was nyctalopia (2 cases). Aside from OAT, which was associated with high myopia, the remaining cases had mild myopic or hyperopic refractive errors. One case, a 7-year-old male with biallelic variants in PCDH15, had sensorineural hearing loss, vestibular dysfunction and undetectable or reduced rod- and cone-isolated responses consistent with Usher Syndrome type 1F. Finally, the case secondary to biallelic variants in AP3B2 was a 5-year-old female patient with early infantile epileptic encephalopathy and neurodevelopmental delay. Examination revealed optic atrophy, foveal hypoplasia, but an otherwise normal fundus. The ffERG showed normal photoreceptor responses, isolating retinal dysfunction to the inner retina (Fig. 3; Table 6).

Figure 3.

ERG and multimodal imaging of patients with nERGs with variants in genes not previously associated with nERG. (A) A 23-year-old female patient with variants in ABHD12 with nERG and attenuated a-wave, diffuse outer retinal atrophy on OCT, parafoveal hyperAF, foveal hypoAF, and abnormal foveal reflex on color imaging (B) 5-year-old female patient with variants in AP3B2 with nERG and preserved a-wave, normal outer retinal structure, foveal hypoplasia, and inner retinal thinning on OCT, and normal appearing fundus (C) 13-year-old male patient with variants in OAT with nERG and attenuated a-wave, normal outer retinal structure and CME on OCT, scalloped hypoAF lesions starting in mid-periphery corresponding to areas of atrophy on color imaging (D) 7-year-old male patient with variants in PCDH15 with nERG and attenuated a-wave, preserved outer retinal structure in central macula with atrophy in the periphery on OCT, faint hyperAF macular ring, and vascular attenuation on color imaging (E) A 19-year-old male patient with variants in PDE6A with nERG and attenuated a-wave, preserved outer retinal structure in the sub-foveal region with atrophy in the outer macula on OCT, hypoAF paramacular ring with hypoAF starting within vascular arcades corresponding to atrophy with bone spicules on imaging.

Table 6.

Clinical Characteristics of Patients With nERGs With Variants in Genes Not Previously Associated With nERG

ID	Age, Sex	Gene	Variant(s)	Visual Complaints	BCVA, logMAR	Refraction, Diopters	Contrast Sensitivity, log CS	ERG*, R: rods, C: cones	Other Ocular Features	Systemic Features
N133	23, F	ABHD12	NM_015600.5:c.1063C>T (p.Arg355*);	Nyctalopia	OD: 1.0	OD: +1.00 + 1.00 × 90	OD: 0.5	R: ↓	OD: amblyopia, strabismus	None (no hearing loss, no ataxia)
			NM_015600.5:c.259C>A (p.Pro87Thr)	Photophobia	OS: 0.1	OS: +1.00 + 0.25 × 85	OS: 1.2	C: ↓
			Compound heterozygous
N109	5, F	AP3B2	NM_001278512.2:c.199C>T (p.Arg67*)	Unknown	Fix and follow	OD: −3.75 + 2.50 × 135	Not tested	R: WNL	Optic atrophy	Early infantile epileptic encephalopathy, cerebral palsy, hypotonia, microcephaly, developmental delay, oral motor dysfunction, choreoathetoid movements
			Homozygous			OS: −4.50 + 2.00 × 120		C: WNL	Foveal hypoplasia
N132	13, M	OAT	NM_000274.4:c.1058G>A (p.Gly353Asp);	Hemeralopia	OD: 0.3	OD: −10.25 + 1.75 × 90	OD: 1.4	R: ↓	CME	None
			NM_000274.4:c.897C>G		OS: 0.1	OS: −10.75 + 1.75 × 90	OS: 1.4	C: ↓
			(p.Tyr299*)
			Compound heterozygous
N130	7, M	PCDH15	NM_001142763.2:c.3373C>T (p.Arg1125*);	None	OD: 0.2	OD: −0.75 + 2.75 × 90	OD: 1.5	R: ND	None	Hearing loss, vestibular dysfunction
			Exon 2–26 deletion		OS: 0.2	OS: −1.75 + 2.75 × 90	OS: 1.5	C: ↓
			Compound heterozygous
N118	19, M	PDE6A	NM_000440.3:c.304C>A (p.Arg102Ser)	Nyctalopia	OD: 0.0	OD: +2.50 + 1.00 × 006	OD: 1.4	R: ND	None	Stroke, elevated Hb
			Homozygous		OS: 0.5	OS: +1.00 + 1.50 × 160	OS: 0.6	C: ↓

CS, contrast sensitivity; CME, cystoid macular edema; F, female; M, male; ND, nondetectable; logMAR, logarithm of the minimum angle or resolution.

^*R: rod-isolated ERG (DA 0.01 or 0.006), C: cone-isolated ERG (30 Hz flicker).

Discussion

Characterization of the clinical nature and genetic causes of nERGs can deepen mechanistic understanding of retinal signal processing, providing insights relevant for diagnosis and design of therapies. We studied the largest cohort of genotyped patients with nERGs to date, elucidating both common and rare genetic associations. We demonstrated that rare genetic causes collectively and CACNA1F-related CSNB comprise the most common etiologies (27.7% each). In line with prior non-genotyped nERG cohorts,^1–4 we find that CNSB, retinoschisis, and photoreceptor/RPE dystrophies are the most common phenotypes accounting for 55.3%, 18.6%, and 17.5% of cases, respectively. Genes important for post-synaptic ON-bipolar cell signaling are associated with significantly increased myopia. Further, we characterize genotypes associated with nERGs in patients with multi-system disease, which affected 17.5% of the cohort. Finally, we report novel associations of nERG with variants in five previously unassociated genes (ABHD12, AP3B2, OAT, PDE6A, and PCDH15).

Our dataset recapitulates known clinical and genetic features of CSNB, with CACNA1F as the most frequently associated gene followed by TRPM1, NYX, and CABP4. We confirm that patients with variants in TRPM1 and NYX have higher myopic refractive error than those with CACNA1F.^7,55 Given that some of these genes are important for post-synaptic ON-bipolar cell function, we investigated whether CSNB genes with known selective post-synaptic roles (TRPM1, NYX, and GPR179) that cause complete CSNB (ON-bipolar signaling defect) were associated with increased refractive error. We found that variants in these genes were associated with a significantly higher myopic refractive error compared with those predominantly expressed in photoreceptors. Together, these results suggest that normal function of the rod-ON bipolar pathway is important for preventing myopia, consistent with findings from experimental studies on model organisms.⁵⁶ In the absence of stimulation by rods, bipolar cells are sensitized by dopamine-dependent gamma-aminobutyric acid (GABA) release from amacrine cells.⁵⁷ Given the importance of dopamine signaling for regulation of axial length,⁵⁸ local modulation of these signaling pathways may represent a pharmacologic target for preventing progression of myopia.

All subjects with XLRS studied were representative of the typical retinoschisis phenotype. Visual acuity and slight hyperopic refractive error was similar to that reported in larger cohorts, in which 60% to 72.2% of patients had nERG.^54,59,60 Nearly 94% of our cohort had normal amplitude a-waves, slightly higher than the 72% to 80% observed previously.^54,60 In one study of 68 patients, Bowles et al. suggested that age and more severe RS1 variants⁵⁴ were correlated with the b:a ratio, but we were unable to detect this effect after adjusting for ERG parameters, although our cohort was smaller and 10 years younger on average.

As observed in previous studies, nERGs were frequently associated with photoreceptor/RPE dystrophies, albeit at a lower rate (17.5% of our cohort vs 34.5% in Sakti et al., 26% in Koh et al., 23% in Renner et al., and 9% in Kim et al.).^1–4 However, we did not find a unifying genetic etiology, as all underlying genes in our cohort were classified as rare causes. No gene affected more than four individuals (ABCA4) and most affected one to two cases. In addition, most of the genes do not have known roles in inner retinal function. Hence, it is possible that inner retinal remodeling at some stage of disease in photoreceptor/RPE dystrophies may lead to nERG.⁶¹ Alternatively, the impairment of the photoreceptor-bipolar cell synapse may precede the impairment of the phototransduction as photoreceptor degeneration progresses, causing a transient nERG.¹⁹ In at least 4 cases (2 novel), all in rod-cone dystrophies associated with RHO, ABHD12, NR2E3, and PCDH15, nERG was likely due to the photopic hill phenomenon in the dark produced by residual cone photoreceptor function in the setting of disproportionate rod loss. In the unaffected light-adapted retina, as light stimulus intensity increases, the a-wave amplitude continues to increase whereas the b-wave saturates and eventually decreases, such that very bright stimuli can create the appearance of selective b-wave reduction.⁴ Similarly, luminance-response series for the retina in the dark-adapted state also show progressive reduction of the b:a ratio.⁶² Therefore, in the retina with rod-cone pattern of photoreceptor degeneration, standard DA flashes for testing the mixed rod-cone response (e.g., 3 cd*s*m⁻²) may preferentially activate cones or function as “very bright” flashes and lead to observing nERG (defined as photopic hill phenomenon in the dark). This may be particularly relevant for PDE6A retinopathy characterized by rod dysfunction,⁶³ although we could not definitively confirm photopic hill phenomenon as the cause of nERG in our patient. Consistent with all these ideas, 71% of the photoreceptor/RPE dystrophies had diminished a-waves indicating a degenerating outer retina, suggesting that their cause conceptually differs from that underlying “traditional” nERGs with normal a-waves typically observed in conditions like CSNB.

Collectively, these considerations suggest that the nERG in most photoreceptor/RPE dystrophies is likely a transient phenomenon, not intrinsic to disease mechanism, independent of the underlying gene, and rare. Supporting this notion, we find that within the Hospital for Sick Children Ocular Genetics Program, only 1 of 29 patients (3.4%) with USH2A and 4 of 110 (3.6%) with ABCA4 retinopathy had nERG and are part of this cohort. Similarly, although USH2A and ABCA4 are among the common causes of IRDs, only a few cases with nERG have been published in literature.^19,46 For all genes in our cohort, an overview of protein localization, their role in synaptic signaling, and whether the inner or outer retina is the primary affected layer is provided in Supplementary Table S9.

Our work highlights the importance of nERG as a feature of multisystem disease. Metabolic disorders affected 58.1% of cases of nERG with associated systemic disease, suggesting that they are a major cause of nERG. Particularly, MPS type I/Hurler Syndrome due to variants in IDUA comprised ⅓ of cases with multisystem disease and 5% of the overall cohort. We reinforce prior evidence that nERG is associated with variants in WDR73,⁴⁹ which causes Galloway-Mowat Syndrome (GAMOS). We also describe a novel association of nERG with early onset epileptic encephalopathy resulting from a homozygous variant in AP3B2.⁶⁴ These results highlight the importance of nERG as a marker of metabolic and central nervous system dysfunction.

We describe three cases of nERG in distinct types of Usher Syndrome. We lend support to previous observations of nERG in cases with USH2A⁴⁶ and MYO7A⁴⁵ variants, and report nERG in a case with PCDH15 variants for the first time. Given that most patients with USH2A⁶⁵ and MYO7A variants⁶⁶ do not have recordable ERGs, the nERGs in these cases may be transient observations in earlier stages of photoreceptor degeneration.

Our group previously described a case series of patients with Roifman syndrome,⁵³ which is characterized in part by defective B-cell maturation. Here, we show one of these cases had nERG, which was not emphasized in the original study. In addition, we report nERG in X-linked hyper-IgM immunodeficiency syndrome (HIGM1) due to CD40LG exon 3 deletion. Schuster et al. reported nERG in a case of HIGM1 and showed that patient-derived antibodies bound to all retinal layers.⁴⁰ We also describe nERG in a case with systemic lupus erythematosus (SLE) and variants in MERTK. Although MERTK cannot be definitively linked to SLE in this case, emerging evidence highlights its importance in autoimmunity. MERTK is a susceptibility gene for end-stage renal disease in SLE,⁶⁷ is important for the macrophage anti-inflammatory response triggered by clearance of apoptotic cells,⁶⁸ is associated with autoimmune disease including uveitis,⁶⁹ and its absence causes a lupus-like autoimmune disease⁷⁰ and photoreceptor degeneration via RPE inflammation in mouse models.⁷¹ These findings add to the growing body of evidence that genetic susceptibility to immune dysfunction can cause inner retinal disease, as nERGs are also observed in autosomal dominant neovascular autoimmune retinopathy secondary to CAPN5⁷² and in birdshot chorioretinitis,⁷³ which only affects individuals with the HLA-A*29 allele.⁷⁴

Last, a patient with TRPM1-related CSNB also had optic pathway glioma (OPG). Given CSNB prevalence of 1 in 294,000⁷⁵ and OPG prevalence of 3 to 4 in 100,000,⁷⁶ there is a 1.3e–10% chance that both diagnoses were independent. Recently, heterozygous missense variants in TRPM1 were associated with risk of familial glioma,⁷⁷ and other members of the TRPM gene family have been implicated in tumorigenesis of various cancers, including gliomas.⁷⁸ This association, although intriguing, requires further validation.

Our work has limitations. The cohort of patients assessed at a quaternary pediatric hospital is not representative of patients with nERG in the general population. For this reason, it is enriched in cases with systemic disease requiring multidisciplinary care. In addition, only a single timepoint is described for patients in this dataset – as such, it is difficult to determine whether nERG is a constant or transient feature for a given case, especially relevant for photoreceptor/RPE dystrophies. Furthermore, as an investigation of genetic causes that starts with defining a cohort of patients with nERG, the study is subject to phenotypic ascertainment bias. This approach carries inherent risks of overestimating the strength of genetic associations, severity of associated disease, and may lead to the inference of false genotype-phenotype associations.⁷⁹ This is particularly relevant to nERGs in photoreceptor/RPE dystrophies, the genetic causes of which are highly heterogenous. Natural history studies of genetically ascertained individuals with these dystrophies can elucidate to what extent nERG is a relevant feature in the course of their disease.

In conclusion, this study represents the largest cohort of genotyped patients with nERGs to date, highlighting both common and rare genetic causes. Our findings underscore that rare genetic etiologies collectively constitute the most frequent cause of nERGs along with CACNA1F, expanding the spectrum of known genetic associations. Our work recapitulates prior studies on genetics of CSNB⁸⁰ and highlights the importance of post-synaptic inner retinal function in preventing myopia development. We describe genetic causes of photoreceptor/RPE dystrophies associated with nERG, highlighting their highly heterogenous nature. This study also reveals that systemic diseases, particularly metabolic disorders, are frequently present among patients with nERGs. Last, we report five novel genetic associations with nERGs. By elucidating the genetic architecture of nERGs, these findings deepen understanding of mechanisms of inner retinal dysfunction and support the utility of nERG as a biomarker in retinal and systemic disease.