To investigate the relationships between clinical and full-field electroretinographic (ERG) findings and progressive loss of visual function in Stargardt disease.
Retrospective cohort study.
We performed a retrospective review of data from 198 patients with Stargardt disease. Measures of visual function over time, including visual acuity, quantified Goldmann visual fields, and full-field ERG data were recorded. Data were analyzed using SAS statistical software. Subgroup analyses were performed on 148 patients with ERG phenotypic data, 46 patients with longitudinal visual field data, and 92 patients with identified ABCA4 mutations (46 with 1 mutation, and 47 with 2 or more mutations).
Of 46 patients with longitudinal visual field data, 8 patients with faster central scotoma progression rates had significantly worse scotopic B-wave amplitudes at their initial assessment than 20 patients with stable scotomata ( P = .014) and were more likely to have atrophy beyond the arcades ( P = .047). Overall, 47.3% of patients exhibited abnormal ERG results, with rod–cone dysfunction in 14.2% of patients, cone–rod dysfunction in 17.6% of patients, and isolated cone dysfunction in 15.5% of patients. Abnormal values in certain ERG parameters were associated significantly with (maximum-stimulation A- and B-wave amplitudes) or tended toward (photopic and scotopic B-wave amplitudes) a higher mean rate of central scotoma progression compared with those patients with normal ERG values. Scotoma size and ERG parameters differed significantly between those with a single mutation versus those with multiple mutations.
Full-field ERG examination provides clinically relevant information regarding the severity of Stargardt disease, likelihood of central scotoma expansion, and visual acuity deterioration. Patients also may exhibit an isolated cone dystrophy on ERG examination.
Mutations in the adenosine 5′-triphosphate–binding cassette transporter gene ABCA4 are known to result in several phenotypic patterns of retinal degeneration. These include Stargardt disease, an autosomal recessive form of juvenile-onset macular dystrophy first described in 1905 by Karl Stargardt, who presented cases with foveal atrophy, paramacular yellow deposits, and fleck. Other phenotypic presentations of ABCA4 mutations include progressive cone–rod degenerations with rapidly expanding central scotomata, often called inverse retinitis pigmentosa , and rarely typical retinitis pigmentosa. The most common clinical manifestation of 2 or more ABCA4 mutations is Stargardt disease, which involves progressive loss of central vision, with near normal peripheral visual fields. Symptom onset of Stargardt disease typically occurs during the first and second decades of life, although there is a subgroup of patients who demonstrate symptoms later in life. Funduscopic findings include pisciform-shaped yellow deposits (or flecks) and macular involvement, which can manifest as either a beaten-bronze sheen, macular granularity, a bull’s-eye lesion, or retinal pigment epithelium depigmentation with atrophy. Our experience has demonstrated that the disease course in mutation-proven individuals can vary greatly from patient to patient, with variable age of onset as well as differential rates of symptomatic progression with respect to visual acuity and visual fields.
Full-field electroretinography (ERG) is one of the methods that has been used to describe the severity of Stargardt disease. The extent of full-field ERG abnormality seems to be related to disease severity and has been used to classify the type of photoreceptor dysfunction present in Stargardt disease (cone dysfunction vs rod–cone dysfunction vs cone–rod dysfunction). The presence of normal full-field ERG results often accompanies a clinical diagnosis of Stargardt disease. However, in a limited number of patients with more advanced disease, Fishman and associates reported a higher percentage of abnormal full-field ERG values. More recently, Simonelli and associates used full-field ERG values to group patients with ABCA4 -associated disease into mild and severe phenotypes. In contrast, Oh and associates were unable to correlate clinical appearance with full-field ERG characteristics in a large cohort of patients.
Given the variability seen in full-field electroretinographic patterns and Stargardt disease progression rates, we sought to examine the prognostic value of full-field ERG results and their association with visual field changes in our large cohort of Stargardt disease patients. We focused on defining groups based on the clinical progression over time of the central scotoma size.
The design of this study was retrospective, and approval from the University of Michigan Institutional Review Board was granted to perform a retrospective study. All patients signed informed consent forms for genetic testing as well as use of their clinical data for research purposes.
A retrospective review was performed of patients with a diagnosis of Stargardt disease who were seen at the Kellogg Eye Center, University of Michigan (UM; n = 131) and at the Jules Stein Eye Institute at the University of California, Los Angeles (UCLA; n = 67) by the same senior investigator (J.H.). The study was approved by the UM Institutional Review Board and the UCLA Human Subject Protection Committee. Patients qualified for the study if they had received a clinical diagnosis of Stargardt disease, had a pedigree consistent with autosomal recessive inheritance, and met one of the following inclusion criteria: (1) 2 causative mutations in ABCA4 ; (2) 1 causative mutation in ABCA4 and 1 of the following clinical findings: a dark choroid sign, documented pisciform flecks, macular atrophy, reduced central acuity, or the presence of central scotoma on Goldmann visual field testing; or (3) a clinical diagnosis based on the combined presence of fundus flecks and a dark choroid on fluorescein angiography, with mutational analysis pending. It should be noted that finding only 1 causative mutation in ABCA4 patients is not unusual. We did not include patients who had no mutations identified in ABCA4 after sequencing of all 50 exons.
All patients underwent comprehensive ocular examinations, Goldmann perimetry (every visit), standardized full-field ERG following International Society for Clinical Electrophysiology of Vision guidelines, fundus photography, and fluorescein angiography. ABCA4 mutational screenings were performed at UM, the University of Iowa Carver Lab, the University Medical Centre Nijmegen, or through the eyeGENE Research Project at the National Eye Institute.
Clinical data collected included: age of symptom onset, family history, best-corrected visual acuity, the area of central scotomata for standardized isopters (I4e, III4e, IV4e) on Goldmann visual field testing measured by digital planimetry (Placom 45C digital planimeter; Koizumi Sokki Mfg Co., Nagaoka Nagata, Japan), retinal distribution of flecks and atrophy on fundus photography, and the presence or absence of a dark choroid sign. Snellen visual acuities were transformed into logarithm of the minimal angle of resolution (logMAR) notation for the purposes of analysis.
The clinical phenotype was evaluated by examining all patient charts and color fundus photographs, which were graded by 2 retinal dystrophy specialists (J.H., K.T.J.). Fundus flecks and atrophy were scored according to their location on the fundus: presence or absence in the macula lutea in the posterior pole. These retinal features were used to stratify the patients into 3 distinct phenotypes (see Supplemental Figure ). Stage I involved flecks limited to the macula, whereas stage II included flecks limited to the posterior pole with foveal atrophy. Stage III patients had advanced disease and were defined by macular atrophy and diffuse flecks throughout the fundus. Stage III in this study is essentially an amalgamation of stages III and IV as described previously by Fishman.
Goldmann visual field data (ie, peripheral field, central scotoma, and physiologic blind spot areas) were obtained using digital planimetry. Data were recorded from standardized Goldmann visual field sheets (Haag-Streit, Bern, Switzerland) and entered by isopters I4e, III4e, and IV4e.
Patients also were separated into 4 distinct groups based on their full-field ERG results at presentation. Of note, given different protocols and normal ranges for full-field ERG values at UCLA and UM (there were 2 sets of normal reference values at UM depending on when the full-field ERG results were obtained), the normal means were derived for each center separately (eg, scotopic B-wave amplitude mean: UM 1 −274 μV, UM 2 −325.36μV, UCLA−375 μV; photopic B-wave amplitude mean: UM 1 −166 μV, UM 2 −125.36μV, UCLA−169 μV). Patient full-field ERG data were expressed as a percentage of the respective mean for each center to facilitate analysis of data compiled from the 2 centers. For the purposes of our study, abnormal was defined as 2 standard deviations less than the normal mean values for amplitude measures and 2 standard deviations more than the normal mean values for latency measures, which were derived separately from control patients seen at UM and UCLA. Patients with abnormal photopic and scotopic full-field ERG values whose scotopic deficiency was more severe were considered to have rod–cone dysfunction. Patients with abnormal photopic and scotopic full-field ERG values with a more severe photopic deficiency were considered to have cone–rod dysfunction. Patients with abnormal photopic full-field ERG values in the presence of normal scotopic full-field ERG values were considered to have cone dysfunction. Finally, patients with normal photopic and scotopic B-wave amplitudes were considered to have a normal full-field ERG phenotype.
Statistical analyses were performed using SAS software version 9.2 (SAS Institutes, Cary, North Carolina, USA). Statistical comparisons between groups were made using t tests or analyses of variance for continuous variables and chi-square or Fisher exact tests for categorical variables. Linear regression was used to estimate the slope of scotoma progression. The data for the right eye of each patient were included in statistical analyses, given that Stargardt disease is a dystrophy that affects eyes in a nearly symmetrical fashion (Tegins EO, et al. IOVS 2011;52:ARVO E-Abstract 5005).
The 198 patients with Stargardt disease we studied were on average 25.6 years of age (standard deviation [SD], 15.1 years) when symptoms first presented and were on average 34.5 years of age (SD, 16.2 years) when first seen at either UM or UCLA, and 59.1% (n = 117) were female. Follow-up was available on subjects for an average of 3.5 years (SD, 5.9 years) after their initial visit from a mean of 2.8 visits (SD, 4.1 visits). When first seen, 15.8% (n = 31) of subjects had stage I Stargardt disease, 54.4% (n = 108) had stage II Stargardt disease, and 29.9% (n = 59) had stage III Stargardt disease. Baseline and follow-up patient data are summarized briefly in Table 1 .
|Continuous Variables||Mean (SD)||Minimum, Maximum||Median|
|At first visit (n = 196)||34.5 (16.2)||7.2, 79.7||32.5|
|At symptom onset (n = 167)||25.7 (15.0)||5.0, 72.0||22.0|
|Follow-up (y, n = 198)||3.5 (5.9)||0.0, 29.7||0.5|
|No. of visits (n = 198)||2.8 (4.1)||1.0, 38.0||2.0|
|At first visit (n = 179)||0.71 (0.54)||0.00, 2.30||0.60|
|At last visit (n = 184)||0.81 (0.57)||−0.12, 2.30||1.00|
|I4e scotoma size (cm 2 )|
|At first visit (n = 160)||6.4 (12.0)||0.0, 76.8||1.6|
|At last visit (n = 160)||10.5 (16.7)||0.0, 86.5||3.5|
|Sex (n = 198)|
|Disease stage at presentation (n = 184)|
Visual Field Analysis: Differences in Scotoma Progression
We performed a subgroup analysis for those patients for whom we could describe rates of central scotoma progression for the I4e isopter using Goldmann visual fields (n = 46). All patients with at least 3 Goldmann visual field data points were included, and a slope of scotoma progression was calculated for each patient. Based on their rates of scotoma progression, we divided these patients into 3 distinct groups. Faster progressors were defined as those patients with a rate of scotoma size progression of more than 2 cm 2 /year (n = 8; mean number of scotoma measurements, 6.1; median number of scotoma measurements, 5; range number of scotoma measurements, 3 to 13; mean follow-up, 9.5 years; median follow-up, 6.4 years; follow-up range, 0.6 to 21.2 years). One outlying fast progressor exhibited a scotoma progression rate of 7.8 cm 2 /year over the course of 4.1 years after initial presentation at age 32 years. Nonprogressors exhibited a rate of scotoma size progression of less than 1 cm 2 /year (n = 27; mean number of scotoma measurements, 5.2; median number of scotoma measurements, 4.0; range number of scotoma measurements, 3 to 15; mean follow-up, 8.3 years, median follow-up, 6.9 years; follow-up range, 1.0 to 24.1 years). The rate of central scotoma progression using the I4e isopter was plotted against patient age at first reported scotoma to illustrate these 2 groups and the broad range of ages at first scotoma ( Figure 1 ). Comparing available full-field ERG data from these 2 groups, the faster-progressing patients (n = 8) had significantly worse scotopic B-wave amplitudes at initial presentation than the slowly progressing patients (n = 20; percent of mean, 53% and 83%, respectively; P = .014, t test). Of note, another group of patients progressed at an intermediate rate of 1 to 2 cm 2 /year (n = 11) and exhibited an intermediate mean for scotopic B-wave amplitude (percent mean, 66%). Interestingly, patients with faster clinical progression were more likely to have atrophic-appearing retinal pigment epithelium beyond the vascular arcades than slower progressors (2 of 8 vs 0 of 27, respectively; P = .047, Fisher exact test).
Age of onset in fast progressors averaged 35.7 years, whereas that of nonprogressors averaged 29.8 years ( P = .429, t test). There were minimal differences in clinical stage at presentation between progressors and nonprogressors ( P = .814, Fisher exact test). There was a statistically significant association between gender and scotoma progression. Specifically, all fast progressors were female (8/8; 100%), whereas 59% of slow progressors were female (16/27; 59%; P = 0.037, Fisher exact test). Average logMAR visual acuities in the fast progressors were somewhat worse than those in nonprogressors at the first visit (0.78 vs 0.54 logMAR units, respectively; P = .427, t test) and the last visit (1.19 vs 0.76 logMAR units, respectively; P = .139), although these differences were not statistically significant. There were statistically significant differences in scotoma size at the first visit ( P = .030, t test) and the last visit ( P = .004, t test), with the fast progressors exhibiting larger average central scotoma sizes at both time points.
Full-Field Electroretinography Phenotype Subgroup Analysis: Characterizing Electroretinographic Abnormalities
One hundred fifty-one patients had full-field ERG data, of which 148 had both scotopic and photopic B-wave amplitude data available for analysis. Of these patients, 47.3% (n = 70) had abnormalities of one or both waveforms. The remainder (52.7%; n = 78) had both rod and cone amplitudes within the normal range. We categorized the 70 patients with abnormal full-field ERG results into groups by the type of full-field ERG abnormality detected: 14.2% (21/148) of patients exhibited rod–cone dysfunction, 17.6% (26/148) had cone–rod dysfunction, and 15.5% (23/148) had isolated cone dysfunction with normal rod function. Full-field ERG phenotypic distribution and abnormalities in each full-field ERG parameter from all available data are shown in Figure 2 and Figure 3 , respectively. The percentage of patients showing abnormal full-field ERG values ranged from 19.6% (scotopic B-wave latency) to 63.4% (photopic A-wave latency).
Full-Field Electroretinography and Scotoma Size
Patients who exhibited abnormal photopic and scotopic B-wave amplitudes tended to have a higher average rate of central scotoma progression (1.88 cm 2 /year [n = 19; P = .066] and 2.05 cm 2 /year [n = 14; P = .064], respectively) compared with patients with respective normal amplitudes (0.78 cm 2 /year [n = 16] and 0.94 cm 2 /year [n = 22], respectively), although these differences were not statistically significant. Patients with abnormal maximum stimulation A-wave amplitudes exhibited a higher average rate of scotoma progression when compared with patients with corresponding normal ERG values (2.56 cm 2 /year [n = 11] and 0.87 cm 2 /year [n = 23], respectively; P = .008, t test). Similarly, patients with abnormal maximum stimulation B-wave amplitudes exhibited a higher average rate of scotoma progression compared with patients with corresponding normal values (2.08 cm 2 /year [n = 17] and 0.76 cm 2 /year [n = 17], respectively; P = .030, t test). See Figure 4 for a comparison of the average rate of scotoma progression between patients with abnormal and normal full-field ERG parameters.
Full-Field Electroretinography and Clinical Stage
When the various full-field ERG testing parameters were compared with the disease stage at presentation, patients with abnormal full-field ERG parameters were more likely to have more advanced stages of disease when compared with patients with normal full-field ERG values. The proportional differences were statistically significant for 7 particular full-field ERG parameters. These include photopic and scotopic B-wave amplitudes and latencies ( P < .0001 for both photopic amplitude and latency; P = .045 and P = .011 for scotopic amplitude and latency, respectively), as well as maximum stimulation A-wave amplitude ( P < .0001) and latency ( P = .049) and maximum stimulation B-wave latency ( P = .039). These data are shown in Figure 5 .
Differences Among Full-Field Electroretinography Phenotypes
On average, patients with abnormal full-field ERG values tended to exhibit worse logMAR visual acuity values than patients with normal full-field ERG values. This difference was statistically significant for the photopic B-wave amplitude and latency, scotopic B-wave amplitude, maximal stimulation A-wave amplitude and latency, and maximal stimulation B-wave amplitude ( P = .003, P = .014, P = .015, P = .001, P = .030, and P = .031, respectively; Table 2 ).
|Full-Field ERG Parameters in Patients with Stargardt Disease||Abnormal ERG Results a||Normal ERG Results a||P Value d|
|No.||Mean LogMAR b||SD c||No.||Mean LogMAR b||SD c|
|Photopic ERG B amplitude||57||0.88||0.64||86||0.58||0.46||0.003|
|Photopic ERG B latency||32||0.91||0.67||76||0.58||0.47||0.014|
|Photopic ERG A amplitude||49||0.79||0.61||86||0.62||0.50||0.090|
|Photopic ERG A latency||68||0.72||0.58||41||0.62||0.51||0.399|
|Scotopic ERG B amplitude||46||0.85||0.61||97||0.61||0.51||0.015|
|Scotopic ERG B latency||20||0.78||0.61||89||0.65||0.54||0.338|
|Maximum stimulation A amplitude||35||0.92||0.63||90||0.57||0.49||0.001|
|Maximum stimulation A latency||42||0.82||0.61||67||0.59||0.50||0.030|
|Maximum stimulation B amplitude||51||0.82||0.62||78||0.59||0.49||0.031|
|Maximum stimulation B latency||50||0.75||0.58||59||0.61||0.52||0.181|