A natural history study of autosomal dominant GUCY2D-associated cone–rod dystrophy

Cone (CD) and Cone Rod dystrophies (CRD) are a group of inherited retinal dystrophies (IRD) that are characterised by the degeneration of the photopic (cone) system. The phenotypic naming of the condition is based on the patient symptoms and visual electrophysiology findings. Patients with Cone dystrophy usually present in adolescence or early adulthood with decreased visual acuity, photophobia, and dyschromatopsia [1, 2]. In many cases patients will progress, developing nyctalopia and loss in peripheral vision at low light levels due to rod degeneration, thereby conferring a subsequent diagnosis of cone–rod dystrophy (CRD) [1]. Several genes associated with CRD have been described that display autosomal recessive (AR), autosomal dominant (AD) and X-linked patterns of inheritance, as listed on RetNet, the University of Texas School of Public Health Retinal Information Network [3]. Many of the genes identified have been associated with multiple phenotypes, or disease in multiple inheritance patterns, with a subgroup associated with multiple phenotypes in multiple inheritance patterns.

One member of this subgroup is the GUCY2D gene, which encodes the photoreceptor guanylate cyclase (GC-E). GC-E is a membrane-bound protein that restores photoreceptor cGMP levels following the light-induced signal cascade, and is indirectly regulated by Ca²⁺ levels. More than 140 disease-causing GUCY2D mutations have been identified, of which 88% are associated with AR Leber congenital amaurosis (LCA) and 9% with AD CD/CRD [4]. Recently a further phenotype described as congenital night blindness has been associated with biallelic GUCY2D variants [5]. The mutations associated with AD CD/CRD are largely confined to exon 13 of GUCY2D, which encodes the GC-E dimerisation domain, with a hotspot at Arg838. In vitro [6] and in vivo [7] studies have shown that Arg838 mutations generate a functional cyclase with reduced sensitivity to Ca²⁺-mediated activation. In contrast, the mutations associated with LCA are found scattered throughout the gene [4], and generate GC-E with impaired catalytic ability [8] or misfolded protein that is subsequently degraded [9].

Recently, results from a Phase I/II human clinical trial of subretinal gene therapy for AR GUCY2D-LCA were published [10], demonstrating early indication of safety and efficacy of the treatment. Since variants in AD GUCY2D lead to a presumed dominant-negative disease mechanism, gene editing strategies have been tested in mouse and macaque as a therapeutic approach [11]. To prepare for potential therapeutic interventions for AD GUCY2D-CRD, it is imperative that outcome measures are identified that can reliably monitor disease progression and patient visual function. Two recently published studies described cohorts from Europe and the United Kingdom with GUCY2D-CRD, highlighting disruption of the ellipsoid zone (EZ) and visual acuity (VA) as two important endpoints for future therapeutic trials [12, 13]. Our study describes an Australian cohort of AD GUCY2D-CRD patients, highlighting key functional and structural biomarkers and their change with disease duration, providing further insights into disease progression.

All available medical records were reviewed and clinical data including patient age, gender and best corrected visual acuity (BCVA) were recorded from all visits. Visual acuity was measured using the Early Treatment Diabetic Retinopathy Study Chart [14], and letter acuity was recorded using metric scale, as per standard clinical recordings. To assist with data analysis, all BCVA values are expressed as logarithm of minimum angle of resolution (logMAR) notation.

Imaging was also collected, including ultra-widefield fundus autofluorescence (UW-FAF) and spectral domain optical coherence tomography (SD-OCT). UW-FAF imaging was obtained using Optos California (Optos plc, Dunfermline, UK) encompassing up to 200° (532 nm excitation) (Fig. 1). Dimensions of hyperautofluorescent areas were measured using the caliper tool embedded into the UW-FAF software. SD-OCT scans of the macula were captured using the Spectralis SD-OCT (Heidelberg Engineering, Heidelberg, Baden-Württemberg, Germany) or Cirrus SD-OCT (Carl Zeiss Meditec, Dublin, California, United States). Poor quality images were rejected at the time of examination, as per standard protocols. Manual review and correction of segmentation for all macular volume scans was performed to ensure quality control. Average macular thickness (AMT) was calculated from the nine areas depicted in the Early Treatment Diabetic Retinopathy Study (ETDRS) [14] map generated by the SD-OCT software. Integrity of the ellipsoid zone (EZ) was assessed via high definition cross-sectional scans through the macula, and the caliper tool embedded within the SD-OCT software was used to measure the length of disruption of the EZ (Fig. 1). Cirrus SD-OCT values were corrected using conversion tables and made comparable to those collected with Spectralis OCT in order to minimize the thickness differences existing between the two instruments [15].

Results from International Society for Clinical Electrophysiology standard full-field electroretinogram (ffERG) and pattern electroretinogram (pERG) results were collected [16]. The ffERG parameters analyzed include light-adapted 30 Hz flicker amplitude (µV) and peak time (ms), and fused flicker amplitude (µV). The fused flicker is a non-ISCEV standard waveform that enables assessment of residual cone function, as described by Berson. The flicker fusion ERG is the amplitude of the fundamental of the ISCEV 30 Hz flicker ERG, produced by narrow band filtering after averaging (i.e. not Fourier techniques post hoc.) [17, 18]. The pERG parameters analyzed include P50 amplitude (µV) and N95 amplitude (µV) to a 15° field.

Statistical analyses were performed using SAS version 9.4. Continuous data were reported as mean ± standard deviation (SD), range and quartiles as appropriate. Categorical data were expressed in proportions. To account for correlation between repeated measurements from the same individual, structure–function relationships were analyzed using generalised estimating equations (GEE) assuming a normal distribution of the outcome and exchangeable correlation structure, from which we report the model co-efficient β.

16 patients (32 eyes) with CRD and GUCY2D mutations were identified (Table 1). No patients with confirmed GUCY2D mutations were excluded. There were six (37.5%) males and ten females (62.5%) from six families, with a mean age of 30 years at first presentation to the clinics (range = 5–70 years). The length of follow-up ranged from 0 to 17 years (mean ± standard deviation [SD], 7 ± 6.18 years).

All patients in this cohort had disease-causing variants identified in GUCY2D. The variants identified were exclusively at amino acid position 838 (Fig. 2), which has previously been identified as a hotspot for mutation causing CRD. Ten patients from three families were heterozygous for p.Arg838His, and three patients from two families were heterozygous for p. Arg838Cys. Both of these variants have been previously associated with AD CRD [19‐25]. One patient (P6) had an additional mutation in ABCA4 p.Gly1961Glu.

Cone function was also demonstrated to decrease with increasing age, with the amplitude of ffERG light-adapted (LA) 30 Hz flicker and fused flicker decreasing over time (β = − 0.8 µV per year, p = 0.005; β = − 0.2 µV per year, p = 0.018, respectively; Supplementary 2B and 2D). However, no evidence of age-related delay was observed for the ffERG LA 30 Hz peak time (β = 0.09 µV per year, p = 0.10; Supplementary 2C), and no association between age and the amplitude of pERG P50 and N95 parameters (β = 0.0001 µV per year, p = 0.98; β = 0.01 µV per year, p = 0.25; Supplementary 2E and 2F). Representative ISCEV standard ffERG, wide field fundus autofluorescence and OCT scans are presented in Fig. 4 which illustrates the variability in both the photopic ffERG responses and the multimodal imagining.

This retrospective cohort study describes the natural history of AD-CRD associated with mutations in the GUCY2D gene in 16 patients from an Australian cohort. Given the recently promising results of subretinal gene therapy in autosomal recessive GUCY2D-LCA [10], and the progress towards gene editing approaches to treat AD GUCY2D-CRD [11], reliable outcome measures of the slow disease progression of AD GUCY2D-CRD will be useful in designing future therapeutic interventions.

At first presentation with a mean age of 30 years, the visual acuity of 75% of the patients in this cohort was below the threshold for holding a driver’s licence in Australia [27], indicating the significant impact this disease can have on quality of life even in early adulthood. Over the period of observation, this vision decreased by 0.019 logMAR per year (p =  < 0.0001), which is similar to the recently reported European/United Kingdom cohort (0.022 logMAR [13] and 0.017 logMAR [12] per year). Despite the average length of observation of seven years, we only observed three patients that crossed the threshold defining legal blindness which made it difficult to accurately calculate an age at which patients with AD GUCY2D-CRD are expected to cross this threshold. This is a key threshold from both a functional perspective but also from the perspective of accessing services and assistance from public-funded resources [26]. The two recent studies in a similar cohort from Europe/United Kingdom determined that by close to 40 years of age, approximately 50% of patients with AD GUCY2D-CRD will be blind or have severe vision loss [12, 13], which is in keeping with the trends observed in this study. In this study we have mapped individual patient visual acuity with the course of their disease with many patients having multiple timepoints recorded. In addition, we have identified that a visual acuity level between LogMAR 0.5–0.6 is maybe another marker of impending accelerated visual decline adding another variable to determine the therapeutic window. A larger sample and longer follow up beginning from a younger age will assist in refining the therapeutic window.

Cone dysfunction as measured by ffERG was shown to worsen with age, and was correlated with decreased visual acuity, confirming the role of cone degeneration in the progression of this disease. Decreasing visual acuity and ffERG parameters were both associated with structural changes in the retina, including lengthening disruption of the EZ and reduced AMT, demonstrating a strong structure–function relationship in the progression of AD GUCY2D-CRD. The LA30Hz and FFAmp both showed similar relationships. This suggests that in GUCY2D retinopathy the FFAMP does not add additional information to that obtained by the ISCEV standard LA30Hz. These correlations also suggest that visual acuity is reliable as a key functional outcome marker of disease progression, which is in agreement with the recent proposal by Neubauer et al. [12] in which they propose visual acuity as the most relevant endpoint.

Previous studies have demonstrated changes in fundus autofluorescence associated with progression of disease [13, 28, 29], however ours is the first study to our knowledge that quantifies these changes over time in patients with GUCY2D-CRD. The structural changes detected as fundus hyperautofluorescence increased by 0.05 mm² per year (p = 0.027), however no evidence of correlation was found between FAF and visual acuity, which is in keeping with previous findings in CRD populations [30]. There was a correlation between area of fundus hyperautofluorescence and pERG P50 representing retinal and macula dysfunction, however no association was identified with ffERG parameters, which is in contrast to previous studies of CRD populations [31]. This may suggest that the structural changes detected by FAF in GUCY2D-CRD follow a different course to that of the measured functional outcomes. In addition, although fundus hyperautofluorescence was correlated with ellipsoid zone disruption and macula thinning, these relationships were confounded by age. Indeed, 5 of the 16 patients demonstrated hyperautofluorescence without any disruption to the EZ, all of whom were below the mean age of 30 years on first visit. When the data was graphically depicted by age quartiles, those in the first and second quartile showed small areas of stable but present hyperautofluoresence with normal macula thickness and no or small changes in the EZ. Following this stable area, the slope of the curve changes to show a more linear relationship between increasing fundus hyperautofluorescence and deterioration of the ellipsoid zone and is dominated by those in the 3rd and 4th age quartiles. These findings suggest that UW-FAF may be useful in detecting early structural changes to the retina that precede other structural and functional deficits, and therefore may be a useful early structural biomarker to assist in developing a timeline of optimal therapeutic treatment. However, it must be noted that manual measurements used in this study are prone to error given the lack of distinct margins surrounding the area of hyperautofluorescence. A recent study using automated measurements of FAF suggests that quantifying FAF changes is more useful than OCT parameters in analysing visual field changes in patients with CRD [32]. Future studies would benefit from utilizing such automated method of detecting area and intensity of hyper/hypoautofluorescence to validate this as a useful structural biomarker.

It is noteworthy that significant variance in phenotype was demonstrated within families, and even between siblings. For example, of the brothers P2 and P3, at age 27 years P2 had a BCVA of 6/9 and no measurable EZ disruption, while P3 had a BCVA of 6/18 and EZ gap of 1.1 mm. (Fig. 4) Variable expressivity and incomplete penetrance has been demonstrated in previous studies [13, 22], and is likely multifactorial, however a clear mechanism determining phenotypic variability has yet to be established.

Limitations of this study include inherent variability in data availability and intervals between visits due to the retrospective nature of the study, as well as limited sample size. Future studies would benefit from a prospective method in which visits and data collection can be standardised across the sample. In addition, ideally participants would be studied from a young age, with length of observation over multiple decades to capture the time at which vision crosses important thresholds that impact the daily living of these patients, including ability to hold a driver’s licence and legal blindness. This study spanned on average seven years of observation, which provides one of the longest observation periods in this population, however the data most often did not cross these important thresholds. In any case, this study presents one of the few longitudinal studies with a large, multigenerational cohort of patients with GUCY2D-CRD, and the only study to quantify FAF changes over time in this population.

In this study we have described the natural history of autosomal dominant GUCY2D-CRD in an Australian cohort using longitudinal data. Structural changes measured using OCT correlate with BCVA and ffERG parameters, highlighting their possible utility as reliable biomarkers for future therapeutic clinical trials. We have also highlighted the different rates of change in BCVA, EZ and FAF with disease duration which will assist in determining windows for therapeutic intervention.

This figure presents ffERG ISCEV standard responses, widefield fundus autofluorescence and macular OCT raster scans in a subset of the patient cohort compared to a normal subject. The cases chosen highlight the differences within families P1–P3, and with various ages P4, P6 and P15.

P1 illustrates normal scotopic (DA 0.01, DA 3.0 & DA 12.0) and reduced photopic (LA30Hz and LA 3.0) responses with a well-defined bulls eye pattern on FAF and loss of EZ line and outer retinal structures on OCT. P2 & P3 are the sons of P1. P2 has preserved scotopic responses and moderately reduced photopic responses. The FAF shows a blotchy central hyper autofluorescent pattern and OCT demonstrating blurring of the EZ line but no loss. P3 has normal scotopic and moderately reduced photopic ffERG responses. The FAF has an early bulls eye appearance with a bright hyperautofluorescent central region surrounded by ring of moderate hyperautofluorescence. The OCT has loss of the EZ line in parafoveal region corresponding to the boundary of the bulls eye macular hyperautofluorescence. P4 shows the normal scotopic and moderate reduction in photopic ffERG. FAF is diffuse blotch involving central fovea. The OCT shows early disruption of the EZ line. P6 similarly has normal scotopic ffERG and moderately reduced photopic ffERG. The FAF shows a compact bulls eye pattern of hyperautofluorescence. The OCT shows a relatively intact central subfoveal island and gaps in the EZ line in the parafoveal region. P15 is the oldest patient in the cohort the ffERG shows the normal scotopic but significantly reduced photopic ffERG. A well defined bulls eye pattern of fundus autofluorescence is show with rings of hyper and hypo autofluorescence. The OCT shows loss of the EZ line and outer retinal structures at the fovea.

ffERG full field electroretinogram, ISCEV International society for clinical electrophysiology of vision, DA dark adapted, LA light adapted, FAF fundus autofluorescence, OCT optical coherence tomography, EZ ellipsoid zone,