Electronegative electroretinography in inherited retinal disease: a systematic review of genotype–phenotype correlations Authors: Ibrahim Taha, Khalil Huraibat, Liana Al-Labadi Published: 2026 ABSTRACT AbstractBackgroundThe electronegative electroretinogram (ERG)—in which the dark-adapted bright-flash b-wave amplitude falls below the a-wave (b: a ratio < 1.0)—localizes dysfunction to the inner retina or photoreceptor–bipolar synapse. In inherited retinal disease (IRD), this waveform has been associated with a restricted set of genetic etiologies, but the evidence has not been systematically synthesized.ObjectiveTo systematically identify and appraise the evidence on genetically confirmed IRD associated with electronegative ERG, characterizing reported gene-specific electrophysiologic patterns, structural correlates, and clinical features.MethodsA systematic search of PubMed/MEDLINE, Scopus, Web of Science, and the Cochrane Library was conducted from inception through March 2026, following PRISMA 2020 guidelines. Studies reporting patients with genetically confirmed IRD and electronegative full-field ERG were included. For this review, electronegative ERG was defined quantitatively as a dark-adapted bright-flash b: a amplitude ratio below 1.0 where numerical data were provided; studies using only qualitative descriptors without numerical ratios were included but flagged in sensitivity analysis, as qualitative labelling introduces subjectivity and reduces cross-study comparability. Quality was assessed using Joanna Briggs Institute checklists. Evidence was synthesized narratively, stratified by gene group, because clinical and methodological heterogeneity precluded meta-analysis.ResultsEighty-seven studies (approximately 1,250 patients; 23 countries; 1986–2025) met inclusion criteria. The literature was dominated by case reports and small case series (83%), with substantial heterogeneity in ERG protocols and inconsistent ISCEV compliance (documented in 61%). The most frequently reported associations were with genes causing congenital stationary night blindness (NYX, CACNA1F, TRPM1, GRM6, GPR179, CABP4, LRIT3), X-linked retinoschisis (RS1), and cone dystrophy with supernormal rod responses (KCNV2). Electronegative ERG patterns were commonly described across the CSNB and XLRS literature, with the greatest consistency for complete CSNB and KCNV2. Structural imaging correlates showed gene-group specificity, particularly foveal schisis in RS1-associated disease. Evidence for rarer gene associations was limited to isolated case reports. Risk of bias was high across most included studies. These limitations directly inform the confidence ratings assigned to each gene–phenotype association throughout the review.ConclusionsElectronegative ERG appears to be a potentially useful phenotypic marker that may help narrow the genetic differential in IRD, particularly for CSNB, XLRS, and KCNV2-associated disease. This utility is most applicable in clinical settings where comprehensive genetic sequencing is not immediately accessible; ERG findings should be interpreted in conjunction with structural imaging and clinical context rather than in isolation. However, the evidence base is constrained by small sample sizes, heterogeneous methodology, likely publication bias, and limited longitudinal data. Importantly, conclusions regarding rarer gene associations should be treated as preliminary, given that they rest on isolated case reports with high risk of bias. Prospective studies with standardized ERG protocols and reporting are needed to validate the diagnostic patterns identified in this review.Keywords: Electroretinography, Electronegative ERG, Inherited retinal disease, Congenital stationary night blindness, X-linked retinoschisis, KCNV2, Genotype–phenotype correlation, ON-bipolar cell, ISCEV, Systematic review Introduction The full-field electroretinogram (ERG), recorded according to the standards of the International Society for Clinical Electrophysiology of Vision (ISCEV), provides a non-invasive measure of mass retinal function that can distinguish between photoreceptor-driven and post-receptoral responses [1, 2]. Among the clinically most informative ERG abnormalities is the electronegative waveform—a configuration in which the b-wave amplitude of the dark-adapted bright-flash response fails to exceed the a-wave, yielding a b: a ratio below 1.0. Because the a-wave is generated predominantly by photoreceptor hyperpolarization and the b-wave primarily by ON-bipolar cell depolarization via the mGluR6–TRPM1 signaling cascade, an electronegative ERG suggests dysfunction at or distal to the photoreceptor–bipolar synapse while indicating relative preservation of phototransduction [3–5]. In inherited retinal disease (IRD), the electronegative ERG has been associated with a relatively restricted group of genetic disorders. These include the congenital stationary night blindness (CSNB) subtypes, in which mutations in genes encoding components of the ON-bipolar signaling complex (NYX, TRPM1, GRM6, GPR179, LRIT3) or the presynaptic ribbon synapse (CACNA1F, CABP4) disrupt signal transmission from photoreceptors to bipolar cells [6–18]. X-linked retinoschisis (XLRS), caused by RS1 mutations, produces electronegative ERG through structural disruption of inner retinal lamination [19, 20]. Cone dystrophy with supernormal rod responses (CDSRR), caused by KCNV2 mutations, generates a distinctive intensity-dependent electronegative pattern reflecting altered photoreceptor membrane dynamics [21–23]. Additional, less well-characterized associations have been reported, including electronegative ERG in the context of retinitis pigmentosa [3, 24], though these are supported by very limited genetically confirmed data. Several important prior works have addressed components of this landscape. Zeitz et al. provided a comprehensive review of CSNB genotype–phenotype correlations [6]. Robson et al. reviewed genetic and acquired causes of electronegative ERG [3]. Bijveld et al. genotyped and phenotyped 101 Dutch CSNB patients [25]. O’Gorman et al. directly compared b: a ratios across genetically confirmed electronegative ERG conditions [26]. However, none of these works applied systematic search methodology, standardized quality assessment, or evidence-stratified confidence grading across the full spectrum of IRD associated with electronegative ERG. Although individual studies and disease-specific narrative reviews have described these associations, no prior systematic review has attempted to synthesize the evidence across the full electronegative ERG–IRD spectrum with standardized quality assessment. The published literature is fragmented: evidence is distributed across case reports, small case series, and heterogeneous observational studies, often with incomplete genotypic data, variable ERG protocols, and inconsistent adherence to ISCEV standards. This fragmentation makes it difficult for clinicians to assess the reliability and specificity of electronegative ERG as a diagnostic tool for prioritizing genetic testing. The present review addresses this gap by providing the first PRISMA-aligned systematic synthesis of genetically confirmed IRD associated with electronegative full-field ERG, integrating quality assessment to explicitly differentiate high-confidence from low-confidence gene–phenotype associations. Specifically, this review aims to: (1) determine which genes are most frequently and consistently associated with electronegative ERG in the published literature; (2) describe the reported gene-specific electrophysiologic patterns; (3) summarize the structural imaging and clinical correlates; (4) assess the quality and consistency of the evidence base; and (5) propose an evidence-informed, rather than expert-opinion-derived, framework for interpreting electronegative ERG in the context of suspected IRD. By grounding this framework in explicit quality assessment rather than expert opinion alone, the review aims to provide clinicians—particularly those in settings without immediate access to comprehensive genetic testing—with a transparent, calibrated tool for genetic test prioritisation. Methods Protocol and reporting This systematic review was conducted and reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 statement [27]. The review was not prospectively registered with PROSPERO; this is acknowledged as a methodological limitation: the absence of a priori registration increases the theoretical risk of post hoc modification of eligibility criteria or analytic decisions. To mitigate this risk, all eligibility criteria, quality assessment instruments, and synthesis approach were defined and documented before data extraction commenced, and no changes were made to these decisions after extraction began. The rationale for narrative rather than quantitative synthesis was determined a priori based on anticipated heterogeneity in ERG protocols and outcome reporting. The review was conducted in full accordance with the PRISMA 2020 reporting items. Eligibility criteria Population Patients of any age with inherited retinal disease confirmed by molecular genetic testing (Sanger sequencing, targeted gene panel, whole-exome sequencing, whole-genome sequencing, or segregation-supported linkage analysis). Electrophysiology Full-field ERG demonstrating an electronegative waveform on at least one dark-adapted bright-flash recording condition (typically DA 3.0 or DA 10.0 cd·s·m⁻²), defined as a b: a amplitude ratio below 1.0. Studies describing the waveform qualitatively as ‘electronegative’ or ‘negative’ without quantitative ratio data were included but flagged for sensitivity analysis. This distinction is critical: qualitative labels depend on individual clinical judgement and are not directly comparable across centres or ERG equipment platforms, whereas a b: a ratio below 1.0 provides an objective, reproducible threshold. The impact of including qualitatively defined studies is discussed in the limitations. Genetic confirmation At least one pathogenic or likely pathogenic variant identified in a gene plausibly associated with the electrophysiologic phenotype. Studies relying solely on clinical diagnosis without molecular confirmation were excluded. Study designs Prospective and retrospective cohort studies, cross-sectional studies, case–control studies, case series (≥ 2 patients), and individual case reports. Excluded: narrative reviews without original patient data, editorials, conference abstracts without sufficient extractable data, and animal or in vitro studies. Language No language restrictions were applied. Non-English articles were evaluated when an English abstract or translated data permitted extraction. Information sources and search strategy Four electronic databases were searched from inception through March 2026: PubMed/MEDLINE, Scopus, Web of Science Core Collection, and the Cochrane Library. The search strategy combined three concept blocks using Boolean operators: (1) electrophysiology and electronegative ERG terms; (2) inherited retinal disease terms and specific gene symbols; and (3) genetics and genotype–phenotype terms. The complete search strings used for each database are available from the corresponding author upon reasonable request. Additional sources included: hand-searching of reference lists of all included studies and relevant narrative reviews [3, 6, 28]; screening of conference proceedings from ARVO, ISCEV, and AAO (2021–2026); grey literature from ClinicalTrials.gov and the WHO ICTRP; and the first 200 results of a Google Scholar search using simplified free-text terms. Study selection Titles and abstracts were screened independently by the author and a trained research assistant using Rayyan systematic review software. Full texts of potentially eligible records were assessed against the predefined inclusion and exclusion criteria. Disagreements were resolved by discussion and consensus review; where consensus could not be reached, a third-party opinion was sought from a colleague with systematic review experience. Inter-rater agreement for full-text screening was quantified using Cohen’s kappa. The selection process is documented in a PRISMA 2020 flow diagram. Data extraction Data were extracted by the author using a standardized, pilot-tested extraction form. The extraction form was piloted on five randomly selected studies before full extraction commenced, informing minor refinements to variable definitions. Owing to resource constraints inherent in a single-author review, full dual-independent extraction was not feasible; however, a second reviewer independently verified a random 25% sample for accuracy and completeness, with discrepancies triggering re-review of the full extraction for that study. This approach, while less robust than dual-independent extraction, represents a structured verification process appropriate for the review’s scope. This is acknowledged as a methodological limitation: single-reviewer extraction is susceptible to consistent directional errors that dual-reviewer processes are designed to detect. In the verified 25% sample, discrepancies were identified in 4 of 22 studies (18%), all resolved by consensus discussion; the most common discrepancies involved ACMG pathogenicity classification and ISCEV compliance status. These rates are reported transparently to allow readers to calibrate their confidence in the extracted dataset. Variables extracted included: study identification (author, year, country, journal), study design, sample size, patient demographics (age, sex), genetic data (gene, specific variant, zygosity, ACMG pathogenicity classification, inheritance pattern, genetic confirmation method), ERG protocol details (ISCEV compliance stated [yes/no/not reported], stimulus intensities, dark-adaptation duration, equipment, definition of electronegative used), ERG metrics (DA 0.01, DA 3.0/10.0 amplitudes and implicit times, b:a ratio where reported, oscillatory potentials, LA 3.0, 30-Hz flicker), clinical data (best-corrected visual acuity, refraction, nyctalopia, color vision), structural imaging (SD-OCT, FAF, fundus photography), and disease progression data where available. Management of overlapping cohorts Where multiple publications appeared to report overlapping patient cohorts from the same institution, the most comprehensive or most recent report was included and earlier or partial publications were excluded. Potential overlap was identified by cross-checking author names, institutions, recruitment periods, and patient demographic data across studies. When uncertainty remained, both publications were included but flagged to avoid double-counting in summary statistics. Risk of bias and quality assessment Quality was assessed using the Joanna Briggs Institute (JBI) critical appraisal checklists: the 8-item checklist for case reports and the 10-item checklist for case series [29]. For any comparative observational studies, the Newcastle–Ottawa Scale was applied [30]. These instruments were selected based on the anticipated study-design landscape; ROBINS-I was not used because it is designed for non-randomized intervention studies, a design category largely absent from this descriptive phenotypic literature. QUADAS-2 was available for application to any study formally evaluating the diagnostic accuracy of electronegative ERG against a genetic reference standard, but no such studies were identified. Quality assessment was performed by the author and verified on a 25% random sample by a second reviewer. Results are reported domain by domain in Table 2 and summarized narratively. Risk-of-bias findings were not treated as background information but were directly integrated into the confidence ratings assigned to each gene–phenotype association in Sect. 3.4. Specifically, associations supported primarily by studies with high risk of bias—for example, those lacking quantitative b: a ratios, ISCEV compliance documentation, or ACMG variant classification—were explicitly assigned lower confidence, described gene by gene. Readers are therefore advised not to interpret all reported associations equally: a ‘commonly described’ finding supported by well-reported studies carries substantially different evidentiary weight than the same descriptor applied to findings resting on case reports with pervasive methodological deficiencies. Importantly, risk-of-bias findings were integrated into the interpretation of gene-specific evidence: associations supported primarily by low-quality studies were identified explicitly as lower-confidence findings. Table 2Risk-of-bias assessment summary by study designStudy Design n Quality ToolMet ThresholdMost Common DeficienciesCase reports34JBI case report checklist (8 items)12/34 (35%)Missing ERG stimulus parameters or ISCEV compliance; no quantitative b: a ratio; incomplete ACMG variant classification; no follow-up dataCase series38JBI case series checklist (10 items)24/38 (63%)Consecutive/defined ascertainment not reported (37%); incomplete ERG protocol details; no control comparisonRetrospective cohort studies12Newcastle–Ottawa Scale9/12 (75%)Referral/ascertainment bias; retrospective design; variable follow-up durationCross-sectional studies3Newcastle–Ottawa Scale2/3 (67%)Small sample size; single-centre recruitment; no longitudinal dataALL STUDIES – cross-cutting deficiencies87——ISCEV compliance not documented: 34/87 (39%); b: a ratio not reported quantitatively: 58/87 (67%); ACMG pathogenicity classification incomplete: 50/87 (58%); no progression/follow-up data: 63/87 (72%)ACMG = American College of Medical Genetics and Genomics; ISCEV = International Society for Clinical Electrophysiology of Vision; JBI = Joanna Briggs Institute; NOS = Newcastle–Ottawa Scale. ‘Met threshold’ for case reports: ≥6/8 JBI criteria; for case series: ≥7/10 JBI criteria; for cohort/cross-sectional: NOS score ≥ 6. The overall high risk of bias across the literature reflects the predominance of case reports and small series with heterogeneous reporting standards rather than necessarily indicating poor clinical quality of individual studies Risk-of-bias assessment summary by study design ACMG = American College of Medical Genetics and Genomics; ISCEV = International Society for Clinical Electrophysiology of Vision; JBI = Joanna Briggs Institute; NOS = Newcastle–Ottawa Scale. ‘Met threshold’ for case reports: ≥6/8 JBI criteria; for case series: ≥7/10 JBI criteria; for cohort/cross-sectional: NOS score ≥ 6. The overall high risk of bias across the literature reflects the predominance of case reports and small series with heterogeneous reporting standards rather than necessarily indicating poor clinical quality of individual studies Data synthesis Formal meta-analysis was judged inappropriate for the primary and secondary outcomes of this review. The rationale for this decision was assessed a priori and confirmed during extraction: the included studies exhibited substantial clinical heterogeneity (different age ranges, disease stages, referral populations), methodological heterogeneity (different ERG stimulus protocols, equipment platforms, dark-adaptation durations, definitions of ‘electronegative’), and reporting heterogeneity (quantitative b: a ratios reported in a minority of studies; most studies reporting ERG findings descriptively). Pooling quantitative ERG data under these conditions would produce misleading summary estimates. Instead, a structured qualitative synthesis was performed, organized along three axes: (1) a gene-grouped evidence summary, presenting the number of studies, approximate patient counts, consistency of electrophysiologic findings, and confidence level for each gene or gene group; (2) a phenotype-domain synthesis, addressing structural imaging, visual function, and disease course by gene group; and (3) a diagnostic-pattern analysis, mapping reported ERG subtypes to candidate gene groups as an evidence-informed framework rather than a validated diagnostic algorithm. Within each axis, findings were graded qualitatively as ‘consistently reported,’ ‘commonly described,’ ‘suggestive but based on limited data,’ or ‘isolated reports only.’ To ensure transparency and consistency, these descriptors were operationally defined before synthesis commenced and applied uniformly across gene groups: ‘consistently reported’ required ≥ 3 independent studies with largely overlapping characteristics, AND the finding present in ≥ 75% of clearly reported cases, AND at least one cohort study (≥ 10 patients) contributing to the evidence; ‘commonly described’ required ≥ 3 studies with some variation, or fewer studies with high internal consistency in at least one larger series; ‘suggestive but limited’ applied to 2 studies or ≤ 10 patients total; and ‘isolated reports only’ to a single study or fewer than 5 patients without independent corroboration. Findings meeting the numerical criteria for a higher grade but supported solely by studies with high risk of bias were downgraded by one level unless corroborated by at least one lower-risk study. Assessment of evidence certainty Overall certainty of each gene–phenotype association was assessed narratively using principles adapted from GRADE, evaluating risk of bias, consistency, directness, and precision [31]. Formal GRADE ratings were not assigned, given the non-comparative, predominantly descriptive nature of the evidence. Sources of imprecision and potential bias—including small sample sizes, heterogeneous ERG methodology, probable publication bias (favoring unusual or severe phenotypes), and historical variation in genetic testing approaches—were documented for each gene group and integrated into the discussion of findings. Availability of additional materials To support transparency and reproducibility, the complete electronic search strings for all four databases, the fully completed PRISMA 2020 checklist, the full study-level data extraction table listing all 87 included studies with all extracted variables, and the full per-study risk-of-bias assessment table with domain-level JBI and Newcastle–Ottawa Scale ratings are available from the corresponding author upon reasonable request. Table 1 and Table 2 in the main manuscript provide summarised presentations of study characteristics and risk-of-bias findings, respectively. Table 1Summary of study characteristicsCharacteristicValueNotesTotal included studies87 Case reports34 (39%)JBI case report checklist applied Case series38 (44%)JBI case series checklist applied Retrospective cohort studies12 (14%)Newcastle–Ottawa Scale applied Cross-sectional studies3 (3%)Newcastle–Ottawa Scale applied Prospective studies / RCTs0None identifiedPublication years1986–2025Countries represented23Approximate total patients~ 1,250Estimates; some cohort overlap possibleMedian study sample size6IQR 1–18ISCEV compliance documented53 (61%)34 studies (39%) not stated or unclearQuantitative b: a ratio reported29 (33%)Remainder described ERG qualitativelyMost represented gene: CACNA1F28 studiesX-linked incomplete CSNBNYX24 studiesX-linked complete CSNBRS122 studiesX-linked retinoschisisTRPM119 studiesAR complete CSNBKCNV216 studiesCone dystrophy with supernormal rod responsesGRM69 studiesAR complete CSNBGPR1795 studiesAR complete CSNBCABP44 studiesAR incomplete CSNBLRIT33 studiesAR complete CSNBOther genes (RHO, NRL, CRX)7 studiesIsolated / secondary associationsAR = autosomal recessive; CSNB = congenital stationary night blindness; IQR = interquartile range; ISCEV = International Society for Clinical Electrophysiology of Vision; RCT = randomised controlled trial. Patient counts are approximate owing to possible cohort overlap between publications. Gene-level study counts sum to more than 87 because many studies reported more than one gene Summary of study characteristics AR = autosomal recessive; CSNB = congenital stationary night blindness; IQR = interquartile range; ISCEV = International Society for Clinical Electrophysiology of Vision; RCT = randomised controlled trial. Patient counts are approximate owing to possible cohort overlap between publications. Gene-level study counts sum to more than 87 because many studies reported more than one gene Results Study selection The systematic search identified 4,217 records, of which 2,893 were unique after deduplication. Title and abstract screening excluded 2,614 records. Two hundred seventy-nine full-text articles were assessed, and 87 met all inclusion criteria. The most frequent reasons for full-text exclusion were: absence of molecular genetic confirmation (n = 94), inability to determine electronegative ERG status from reported data (n = 58), duplicate or overlapping cohorts (n = 22), and ineligible study design (n = 18). Inter-rater agreement was substantial (κ = 0.84). Study- and patient-level totals should be interpreted as estimates, because some studies reported multiple genes and some patient cohorts may overlap between publications despite efforts to identify duplicates. Study characteristics Included studies were published between 1986 and 2025, originated from 23 countries, and comprised 34 case reports (39%), 38 case series (44%), 12 retrospective cohort studies (14%), and 3 cross-sectional studies (3%). No prospective cohort studies or randomized trials were identified. The aggregate sample encompassed approximately 1,250 genetically confirmed patients. Median study sample size was 6 patients (interquartile range 1–18). ISCEV-compliant ERG protocols were explicitly documented in 53 studies (61%); in 34 studies (39%), ISCEV compliance was not stated or could not be determined. Quantitative b: a ratios were reported in only 29 studies (33%); the remainder described the electronegative waveform qualitatively. Quantitative b: a ratios—essential for consistent application of the electronegative criterion—were thus available in only a minority of studies. The remaining 58 studies (67%) relied on qualitative characterisation of the waveform. This major reporting gap substantially limits cross-study comparability and is one of the most important methodological deficiencies of the current evidence base. The most frequently represented genes were CACNA1F (28 studies), NYX (24 studies), RS1 (22 studies), TRPM1 (19 studies), KCNV2 (16 studies), GRM6 (9 studies), GPR179 (5 studies), CABP4 (4 studies), and LRIT3 (3 studies). Because many studies reported more than one gene, the sum of gene-specific study counts exceeds the total number of included studies. A small number of studies (n = 7) reported electronegative ERG in association with RHO, NRL, CRX, or other genes. Key characteristics of included studies are summarised in Table 1. Risk of bias Quality assessment revealed pervasive methodological limitations (Table 2). Among the 34 case reports assessed with the JBI checklist, 22 (65%) met fewer than 6 of 8 quality criteria. Common deficiencies included: failure to specify ERG stimulus parameters or ISCEV compliance (39% of all studies), absence of quantitative b: a ratios (67%), incomplete reporting of variant pathogenicity classification according to ACMG guidelines (58%), and lack of follow-up or progression data (72%). Among the 38 case series, 14 (37%) did not report consecutive or clearly defined patient ascertainment. The 12 retrospective cohorts were generally better reported but were subject to referral bias and retrospective ascertainment. Publication bias was judged probable across the literature, with a likely overrepresentation of diagnostically striking, genetically novel, or severe phenotypes. These quality limitations were integrated into the gene-specific evidence synthesis: the confidence assigned to each gene–phenotype association reflects not only the number of supporting studies and patients but also the methodological quality of the underlying evidence. Readers are therefore cautioned that the volume of literature for a given gene does not, by itself, indicate strong or reliable evidence. Publication bias was judged likely, with probable overrepresentation of severe, unusual, or genetically novel phenotypes—a pattern that may cause the case-report literature to overstate both the frequency and the distinctiveness of electronegative ERG in certain conditions. A summary of risk-of-bias findings by study design is presented in Table 2; the full per-study risk-of-bias ratings are available from the corresponding author upon reasonable request. Gene-grouped evidence synthesis Complete congenital stationary night blindness (NYX, TRPM1, GRM6, GPR179, LRIT3) Complete CSNB was supported by approximately 48 studies encompassing over 400 patients across the five implicated genes. The electronegative dark-adapted bright-flash ERG—with absent DA 0.01 rod-isolated response, profoundly reduced b-wave, and moderately abnormal photopic responses with broadened troughs—was consistently reported across studies as a characteristic electrophysiologic feature of cCSNB, representing the best-supported association in this review [6, 13–18, 25, 32]. NYX mutations (X-linked cCSNB) were the most frequently reported cause, supported by 24 studies. The ERG phenotype was described as highly consistent across publications: absent rod-isolated response, profoundly electronegative DA 3.0/10.0 with a well-formed a-wave, absent or vestigial b-wave, and severely attenuated oscillatory potentials [13, 25]. TRPM1 mutations (autosomal recessive cCSNB) were reported in 19 studies and produced an ERG phenotype described as indistinguishable from NYX-associated cCSNB on standard full-field recording [14, 33, 34]. GRM6 (9 studies), GPR179 (5 studies), and LRIT3 (3 studies) were reported with similar ERG patterns, though published cohorts were substantially smaller and phenotypic characterization correspondingly less detailed [12, 15–18]. SD-OCT was reported as normal or near-normal in most cCSNB patients, consistent with a functional rather than structural disorder [35]. Visual acuity was generally stable over available follow-up periods, with the largest cohort (122 patients, Grigg et al.) demonstrating mean BCVA of 0.46 logMAR with no clinically significant change over 8.3 years [35]. High myopia was commonly reported but not universally quantified. Confidence assessment: The association between cCSNB genes and profoundly electronegative ERG is well supported by a moderate volume of generally consistent evidence. The indistinguishability of NYX and TRPM1 ERG phenotypes on routine testing is consistently reported across studies, though formal comparative studies are lacking. Incomplete congenital stationary night blindness (CACNA1F, CABP4) Incomplete CSNB was represented by approximately 30 studies and over 350 patients. CACNA1F mutations (X-linked iCSNB) constituted the majority of the evidence (28 studies). The ERG pattern commonly described was: DA 0.01 present but reduced (not absent), DA 3.0/10.0 electronegative with residual b-wave, disproportionately reduced photopic and 30-Hz flicker responses, and reduced rather than absent oscillatory potentials [6, 25, 32]. This pattern was generally distinguishable from cCSNB in studies that reported both subtypes, though one recent study found that the mean b: a ratio differed significantly between genetically confirmed iCSNB (0.60) and the clinical iCSNB cohort (0.80), suggesting that phenotypic diagnosis alone may be unreliable [26]. CABP4 mutations (4 studies, approximately 15 patients) were reported to produce an iCSNB-like phenotype with an emphasis on cone-related dysfunction, though the very small number of published cases limits confidence in detailed phenotypic characterization [25, 32]. Confidence assessment: The association between CACNA1F and electronegative ERG is well supported. The specific iCSNB ERG pattern—distinguishable from cCSNB—is commonly described, though quantitative b: a data are available in a minority of studies. The CABP4 association is suggestive but based on limited data. X-linked retinoschisis (RS1) RS1-associated electronegative ERG was reported in 22 studies encompassing approximately 250 patients. The electronegative waveform in XLRS was commonly described, though the mechanism differs from CSNB: retinoschisin deficiency disrupts retinal laminar integrity, physically impeding signal propagation rather than disrupting molecular synaptic transduction [19, 20]. The DA 0.01 response was typically present, the DA bright-flash b-wave was reduced but generally less severely than in cCSNB, and photopic responses occasionally showed electronegative morphology as well [3, 26]. Notably, one study specifically comparing b: a ratios across genetically confirmed electronegative ERG conditions found that a substantial proportion of XLRS patients with missense mutations retained b: a ratios above 1.0 [26], indicating that the electronegative ERG is not invariably present in XLRS. The co-occurrence of electronegative ERG and foveal schisis on SD-OCT was frequently described and provides a strong combined phenotypic clue suggestive of XLRS. Longitudinal data, available in a limited number of studies, suggest relatively stable visual acuity with slowly progressive outer retinal atrophy, and null RS1 genotypes associated with worse outcomes [36, 37]. Two phase I/II gene therapy trials have been conducted; neither met primary efficacy endpoints at reported timepoints [38, 39]. Confidence assessment: The association between RS1 and electronegative ERG is well supported for classic presentations. The observation that electronegative ERG is not present in all XLRS patients is important and tempers the strength of ERG as a sole screening tool. Cone dystrophy with supernormal rod responses (KCNV2) KCNV2-associated CDSRR was reported in 16 studies encompassing approximately 180 patients. The ERG pattern described is arguably the most distinctive in the review: at standard DA 3.0 intensity, the response is electronegative or shows a markedly reduced b: a ratio; at higher stimulus intensities (DA 10.0 and above), the rod-driven b-wave was consistently reported to amplify to supernormal levels [21–23, 40]. The KCNV2 Study Group Report 1 (117 patients, 75 variants) represents the largest cohort and confirmed that all patients were symptomatic before age 12 [41]. Unlike CSNB, KCNV2-associated disease was consistently described as progressive, with SD-OCT showing evolving outer retinal thinning and FAF showing perifoveal hyperautofluorescent rings [40, 42]. Confidence assessment: The KCNV2–supernormal rod association is well supported and arguably the most consistently reported ERG–genotype correlation in this review. However, the evidence relies on case series rather than prospective studies, and normative intensity-response data from healthy controls were not uniformly available for comparison. Other reported associations A small number of studies (n = 7, fewer than 30 patients in total) reported electronegative ERG in other genetic contexts. Cideciyan and Jacobson described electronegative waveforms in patients with retinitis pigmentosa, interpreted as secondary to inner retinal remodeling, though genotype confirmation was incomplete in that early series [24]. Littink et al. reported electronegative ERG features in patients with autosomal recessive NRL mutations and enhanced S-cone syndrome [43], and electronegative ERG was observed in an NRL-positive patient within a broader inherited retinal disease cohort [44]. Isolated reports have also described electronegative ERG in association with CRX-related cone–rod dystrophy, though the primary evidence is sparse and derived mainly from single case-level observations noted in a recent review [8]. In each case, the electronegative morphology likely reflects secondary inner retinal compromise rather than primary synaptopathy. Given the very limited evidence, these associations should be regarded as isolated observations rather than established gene–phenotype correlations and are assigned low confidence. Electrophysiologic pattern taxonomy Five broad ERG pattern archetypes emerged from the literature, each mapping to a proposed mechanistic category: (1) profoundly electronegative DA with absent rod-isolated response and broadened photopic troughs (consistent with cCSNB); (2) electronegative DA with residual b-wave and disproportionately reduced photopic/flicker responses (consistent with iCSNB); (3) electronegative DA with preserved rod-isolated response and coexistent structural schisis on OCT (consistent with XLRS); (4) intensity-dependent electronegative-to-supernormal DA with cone dysfunction (consistent with KCNV2); and (5) electronegative DA in the context of advanced photoreceptor degeneration (secondary inner retinal dysfunction). These patterns are proposed as a descriptive taxonomy derived from the reviewed literature rather than as validated diagnostic criteria; their sensitivity and specificity for individual genes have not been formally evaluated in prospective studies. A summary of gene-specific electrophysiologic findings and confidence ratings is provided in Table 3. Table 3Gene-specific electrophysiologic findings: evidence summaryGene(s)StudiesPatientsKey ERG Pattern DescribedConsistencyMain LimitationsConfidence NYX 24≈ 200Absent DA 0.01; profoundly electroneg. DA 3.0; broadened photopic trough; absent OPsHighly consistent across studiesMostly case series; limited quantitative b: a data High* TRPM1 19≈ 120Indistinguishable from NYX on standard ERGConsistent; confirmed across multiple cohortsCannot distinguish from NYX without genetics; mostly retrospective Mod–High GRM6 / GPR179 / LRIT39 / 5 / 3≈ 70Canonical cCSNB pattern reportedLimited but consistent where reportedVery small cohorts; phenotypic detail sparse Moderate CACNA1F 28≈ 340Electroneg. DA with residual b-wave; disproportionate photopic/flicker lossCommonly described; some quantitative confirmationPhenotypic overlap with other causes High CABP4 4≈ 15iCSNB-like; primarily cone-related problemsSuggestive but limitedVery few cases; incomplete characterization Low RS1 22≈ 250Electroneg. DA; preserved DA 0.01; photopic electroneg. variable; OCT schisisCommonly described; not invariableNot all XLRS patients have electroneg. ERG; mostly retrospective Mod–High KCNV2 16≈ 180Electroneg. at standard intensity → supernormal at high intensity; cone dysfunctionHighly consistent; most distinctive patternRequires high-intensity stimulation not always performed High RHO / NRL / CRX7< 30Secondary electroneg. in context of progressive photoreceptor degenerationIsolated reports onlyMechanism likely secondary; very sparse evidence Low Confidence was assigned based on: (a) number of independent studies and patients, (b) consistency of ERG findings across publications, (c) methodological quality of supporting evidence, and (d) availability of genetically confirmed cohort data (as opposed to case reports only). High = consistent findings across multiple independent genetically confirmed cohorts; High* = high confidence for the phenotype-family pattern (e.g., cCSNB), but gene-level discrimination from other genes within the same family is not possible on standard ERG alone; Moderate–High = consistent findings across multiple studies but predominantly retrospective, or not invariably present; Moderate = generally consistent but limited by very small cohorts; Low = isolated case reports only. Patient counts are approximate estimates owing to possible cohort overlap between publications. DA = dark-adapted; electroneg. = electronegative; OPs = oscillatory potentials Gene-specific electrophysiologic findings: evidence summary Confidence was assigned based on: (a) number of independent studies and patients, (b) consistency of ERG findings across publications, (c) methodological quality of supporting evidence, and (d) availability of genetically confirmed cohort data (as opposed to case reports only). High = consistent findings across multiple independent genetically confirmed cohorts; High* = high confidence for the phenotype-family pattern (e.g., cCSNB), but gene-level discrimination from other genes within the same family is not possible on standard ERG alone; Moderate–High = consistent findings across multiple studies but predominantly retrospective, or not invariably present; Moderate = generally consistent but limited by very small cohorts; Low = isolated case reports only. Patient counts are approximate estimates owing to possible cohort overlap between publications. DA = dark-adapted; electroneg. = electronegative; OPs = oscillatory potentials Structural imaging correlations Structural imaging findings showed meaningful gene-group specificity in the reviewed literature. In cCSNB, SD-OCT was described as normal in most reports, consistent with a functional disorder sparing retinal architecture [35]. In XLRS, foveal schisis (inner nuclear layer and/or outer plexiform layer splitting) was the most consistently reported imaging finding across studies and provides the strongest combined diagnostic clue when present alongside electronegative ERG [19, 26, 37]. In KCNV2-associated disease, progressive outer retinal thinning, ellipsoid zone disruption, and perifoveal hyperautofluorescent rings on FAF were commonly described, reflecting the progressive nature of this condition [40–42]. These imaging patterns were reported with reasonable consistency but should be interpreted with the caveat that most data derive from cross-sectional observations in small series. Structural imaging correlations by gene group are summarized in Table 4. Table 4Structural imaging correlations by gene groupGene GroupSD-OCT FindingsFAF PatternConsistencyDiagnostic SignificancecCSNB genesNormal or near-normal; intact EZ; normal CMTUnremarkable in most reportsCommonly describedNormal OCT in profoundly electroneg. ERG strongly suggests CSNBCACNA1F / CABP4Normal or subtle foveal hypoplasia reported in some seriesUnremarkableVariable; limited dataFoveal hypoplasia, if present, may favor iCSNBRS1Foveal schisis (INL/OPL); spoke-wheel pattern; peripheral schisis in ~ 50%Hypo-AF at schisis areasHighly consistentFoveal schisis + electroneg. ERG = strong combined clue for XLRSKCNV2Progressive outer retinal thinning; EZ disruption; macular atrophyPerifoveal hyper-AF ring; central hypo-AFCommonly describedProgressive changes distinguish KCNV2 from stationary CSNBAF = autofluorescence; CMT = central macular thickness; EZ = ellipsoid zone; INL = inner nuclear layer; OPL = outer plexiform layerNote: This framework is derived from patterns reported in the reviewed literature and is proposed as a clinical heuristic to guide genetic testing prioritization. It has not been prospectively validated and should not be used as a standalone diagnostic instrument. Sensitivity and specificity for individual gene identification are unknown Structural imaging correlations by gene group AF = autofluorescence; CMT = central macular thickness; EZ = ellipsoid zone; INL = inner nuclear layer; OPL = outer plexiform layer Note: This framework is derived from patterns reported in the reviewed literature and is proposed as a clinical heuristic to guide genetic testing prioritization. It has not been prospectively validated and should not be used as a standalone diagnostic instrument. Sensitivity and specificity for individual gene identification are unknown Discussion Principal findings This systematic review synthesizes the available evidence on electronegative ERG across the spectrum of genetically confirmed inherited retinal disease. Eighty-seven studies, predominantly case reports and small case series, were included, encompassing approximately 1,250 patients across 23 countries. The review identifies electronegative ERG as a commonly described electrophysiologic feature associated with a defined subset of IRD genes, with the strongest and most consistent evidence supporting associations with the cCSNB genes (NYX, TRPM1, GRM6, GPR179, LRIT3), the iCSNB genes (CACNA1F, CABP4), RS1, and KCNV2. Gene-specific ERG patterns were frequently described in the literature, though formal quantitative validation of these patterns is lacking. The strongest and most consistently supported associations are: (1) cCSNB genes (NYX, TRPM1), where profoundly electronegative ERG with absent rod-isolated response was consistently reported; (2) KCNV2, producing the most diagnostically distinctive ERG pattern—intensity-dependent electronegative-to-supernormal rod response; and (3) RS1 (XLRS), where the combination of electronegative ERG and foveal schisis provides a strong combined diagnostic clue, with the caveat that electronegative ERG is not invariably present. What can be concluded with relatively high confidence Several findings emerge with reasonable consistency across independent studies. First, profoundly electronegative ERG with absent rod-isolated response is consistently described as a feature of cCSNB across multiple genes, with NYX and TRPM1 being the most frequently reported. Second, the KCNV2-associated supernormal rod response at high stimulus intensity is arguably the most distinctive and consistently reported gene–ERG correlation in this review. Third, the combination of electronegative ERG and foveal schisis on SD-OCT has been commonly reported in RS1-associated XLRS and, where both findings coexist, provides a strong clinical pointer toward this diagnosis. Fourth, normal SD-OCT in the setting of profoundly electronegative ERG is commonly described in CSNB and helps distinguish it from progressive retinal dystrophies. Collectively, these findings support a practical clinical interpretation: when electronegative ERG is identified in the context of suspected IRD, the differential can be meaningfully narrowed from the full catalogue of IRD genes to the subset associated with ON-bipolar cell dysfunction, retinal laminar disruption, or KCNV2-related channelopathy. This narrowing can guide targeted gene panel selection or prioritise specific genes for earlier sequencing—a meaningful advantage in centres where whole-exome or whole-genome sequencing is not the first-line approach. What remains uncertain Important uncertainties persist. The sensitivity of electronegative ERG for individual gene identification has not been formally evaluated; it is likely lower than the published case-report literature suggests, given the probable overrepresentation of classic presentations in published cohorts. The specificity of the proposed ERG pattern subtypes (cCSNB vs. iCSNB vs. XLRS vs. KCNV2) for distinguishing between specific genes requires prospective validation. The electronegative ERG is not invariably present in XLRS [26], and the boundary between electronegative and electropositive ERG (b: a ratio near 1.0) has not been rigorously studied. Associations with rarer genes (CABP4, LRIT3, RHO, NRL, CRX) remain at the level of isolated case reports and cannot be considered established. Clinical implications Despite the limitations of the evidence base, the practical clinical value of recognizing electronegative ERG in the context of suspected IRD is supported by the consistency of associations across independent studies and diverse populations. When a patient presents with nyctalopia, reduced acuity, and high myopia, and ERG reveals an electronegative bright-flash waveform, the differential diagnosis can be meaningfully narrowed from hundreds of potential IRD genes to a focused set. Three illustrative clinical scenarios are worth considering. First, a young male with congenital nyctalopia, high myopia, profoundly electronegative ERG with absent rod-isolated response, and normal SD-OCT has a clinical profile highly consistent with complete CSNB; targeted sequencing of NYX (X-linked) or TRPM1, GRM6, GPR179, or LRIT3 (autosomal recessive) is a rational first-line approach. Second, a young male with reduced acuity, foveal schisis on SD-OCT, and electronegative ERG should prompt RS1 sequencing; absence of electronegative ERG does not exclude XLRS in missense-mutation cases. Third, a child with early photophobia, cone dysfunction, and an electronegative-to-supernormal rod pattern on high-intensity stimulation has a profile highly specific for KCNV2; this pattern requires stimulation beyond standard ISCEV protocols, and ERG laboratories should supplement standard recordings accordingly. This guidance is most valuable when sequencing is not immediately available; in settings where rapid comprehensive sequencing is accessible, genetic testing may appropriately precede ERG, and the framework here is complementary to, not competitive with, genetic testing. The diagnostic framework proposed in Fig. 1—structured around the presence or absence of rod-isolated response, relative photopic dysfunction, structural schisis, and supernormal rod responses—represents an evidence-informed heuristic for guiding genetic testing prioritization. It is explicitly not presented as a validated diagnostic algorithm; prospective studies are needed to establish its sensitivity and specificity. Proposed evidence-informed diagnostic framework Mechanistic interpretation The mechanistic framework (Fig. 2) organizes electronegative ERG disorders along five functional nodes of the photoreceptor–bipolar signaling unit. This taxonomy—photoreceptor membrane (KCNV2), presynaptic ribbon (CACNA1F, CABP4), postsynaptic ON-bipolar complex (GRM6, TRPM1, NYX, GPR179, LRIT3), retinal laminar architecture (RS1), and secondary inner retinal remodeling (RHO and others)—is derived from the known molecular biology of these proteins rather than directly from the clinical studies included in this review. It provides a conceptual framework for understanding why different genetic defects may produce different ERG subtypes, but its application to individual patient diagnosis remains inferential and has not been validated in a clinical accuracy study. Mechanistic map of electronegative ERG by functional node Relationship to prior reviews Several important prior works have addressed components of this landscape. Zeitz et al. provided a comprehensive review of CSNB genotype–phenotype correlations [6]. Robson et al. reviewed genetic and acquired causes of electronegative ERG with physiologic insights [3]. Bijveld et al. genotyped and phenotyped 101 Dutch CSNB patients [25]. Grigg et al. reported the largest CSNB cohort with longitudinal data [35]. Robson et al. and Wissinger et al. defined the KCNV2 electrophysiologic phenotype [22, 40]. O’Gorman et al. provided the first direct b: a ratio comparison across genetically confirmed electronegative ERG conditions [26]. The present review extends these contributions by applying systematic search methodology, standardized quality assessment, and evidence-stratified confidence ratings across the entire electronegative ERG–IRD spectrum. Strengths This review has several methodological strengths. To our knowledge, no prior PRISMA-aligned systematic review has specifically synthesised electronegative full-field ERG across genetically confirmed IRD with integrated quality assessment and explicit confidence grading—this constitutes the primary novel contribution of the work. The gene-grouped evidence synthesis with explicit confidence grading allows clinicians to differentiate between robust and tentative associations. The pre-specified operational definitions for qualitative evidence terms (Sect. 2.8) ensure that descriptors such as ‘consistently reported’ reflect transparent, reproducible criteria rather than subjective impressions. The structured qualitative synthesis approach is a deliberate and appropriate choice given the heterogeneity of the underlying evidence—it avoids the well-documented hazard of spurious precision from meta-analyses of incompatible data. Limitations Several limitations must be acknowledged. Each is discussed in terms of its potential impact on the conclusions of this review. First, the review was not prospectively registered, introducing a risk of post hoc methodological decisions; the absence of a time-stamped public record means post hoc modification cannot be ruled out by external verification, reducing reproducibility and increasing theoretical risk of reporting bias in the synthesis. Second, the evidence base is dominated by case reports and small case series (83% of included studies), which inherently limit generalizability and are subject to publication bias; described ERG patterns likely represent the more severe or atypical end of the phenotypic spectrum. Publication bias is judged probable, with likely overrepresentation of severe, unusual, genetically novel, or diagnostically striking phenotypes; this has two distinct consequences: (a) frequencies of electronegative ERG for individual genes may be overestimated; and (b) reported ERG patterns may reflect classic rather than typical presentations. Third, ISCEV compliance was documented in only 61% of studies, and quantitative b: a ratios were reported in only 33%, introducing substantial measurement heterogeneity; in the majority of included studies, the electronegative designation was applied qualitatively, likely introducing inter-study inconsistency in which patients were considered electronegative. Fourth, the era of genetic testing varied widely; older studies using single-gene sequencing may have missed patients with variants in untested genes or undetected deep-intronic mutations, while newer studies using comprehensive panels may capture milder phenotypes not previously reported. Fifth, the single-author design, while supplemented by independent verification on subsamples, is less robust than dual-reviewer extraction; discrepancies were found in 18% of the verified sample, and systematic errors in the unverified 75% cannot be excluded. Sixth, thediagnostic framework proposed in Fig. 2 has not been prospectively validated and should be interpreted as a clinical hypothesis requiring formal evaluation. Research priorities Four priorities emerge from this synthesis. First, the field requires standardized ERG reporting in IRD: quantitative amplitudes, b:a ratios, stimulus parameters, and ISCEV compliance should be reported in tabular format as a minimum dataset, analogous to STROBE extensions for observational research. Second, prospective genotype-stratified cohort studies with longitudinal follow-up are needed to establish natural history data and refine phenotypic ranges. Third, the diagnostic framework proposed here requires formal prospective validation: a study enrolling consecutive patients with newly identified electronegative ERG, applying the framework prospectively, and comparing its gene-prioritization accuracy against unbiased comprehensive genetic testing would establish its clinical utility. Fourth, as gene therapy trials progress for XLRS and potentially KCNV2, standardized ERG endpoints including the b: a ratio should be incorporated as functional biomarkers. Conclusion Electronegative ERG appears to be a potentially clinically valuable phenotype-guided marker in inherited retinal disease, with the strongest evidence supporting its potential diagnostic utility for the CSNB spectrum, X-linked retinoschisis, and KCNV2-associated cone dystrophy with supernormal rod responses. Gene-group-specific ERG patterns have been commonly described in the literature and, when integrated with structural imaging and clinical context, may help prioritize genetic testing—particularly in settings where comprehensive sequencing is not readily available. Electronegative ERG is best understood not as a definitive diagnostic test but as a phenotypic filter: when identified in the appropriate clinical context—nyctalopia, reduced acuity, high myopia, or early-onset cone dysfunction—it can narrow the differential from hundreds of potential IRD genes to a focused, testable subset. This guidance requires calibration to study quality: high-confidence associations (NYX/TRPM1 cCSNB; KCNV2; RS1/XLRS in classic presentations) should be weighted more heavily than low-confidence associations (CABP4; RHO/NRL/CRX) that rest on isolated case reports with high risk of bias. However, the current evidence is constrained by reliance on case-level reporting, heterogeneous ERG methodology, incomplete adherence to ISCEV standards, limited quantitative data, and probable publication bias. The diagnostic framework proposed in this review represents an evidence-informed clinical heuristic that requires prospective validation before it can be endorsed as a formal diagnostic tool. Standardized electrophysiologic reporting—including quantitative amplitudes, b:a ratios, and explicit ISCEV compliance status—remains the most impactful immediate step toward strengthening the evidence base. Standardized reporting of electrophysiologic data in genetically confirmed IRD remains the most important step toward strengthening the evidence base for phenotype-guided genetic diagnosis. Acknowledgements The author thanks the research assistant who contributed to independent screening and data verification. No other individuals or organizations contributed to this work beyond what is described in the Author Contributions section. Author contributions Ibrahim Taha: conceptualization, methodology, literature search, screening, data extraction, quality assessment, data synthesis, writing — original draft, writing — review and editing, visualization, and final approval of the manuscript. Khalil Huraibat & Liana Al-Labadi: writing — review and editing, and visualization. Funding This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. Data availability All data analysed in this review were extracted from published studies. The complete data extraction dataset and supporting materials–including the electronic search strings for all four databases, the completed PRISMA 2020 checklist, the full study-level extraction table covering all 87 included studies, and the full per-study risk-of-bias ratings–are available from the corresponding author upon reasonable request. Code availability Not applicable. Declarations Ethics approval Not applicable. This study is a systematic review of previously published data and did not involve human participants, human tissue, or animals. Nevertheless, in accordance with the principles of the Declaration of Helsinki and applicable institutional standards for research ethics, this review was conducted with full respect for the ethical principles governing research involving human data, including the requirements for data confidentiality and appropriate use of published patient information. Consent for publication Not applicable. Consent to participate Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement All data analysed in this review were extracted from published studies. The complete data extraction dataset and supporting materials–including the electronic search strings for all four databases, the completed PRISMA 2020 checklist, the full study-level extraction table covering all 87 included studies, and the full per-study risk-of-bias ratings–are available from the corresponding author upon reasonable request. Not applicable.