Patient Eligibility and Exclusion Criteria

All inclusion and exclusion criteria are detailed in ProgStar Report No. 1. The most relevant inclusion criteria for this analysis include a minimum age of ≥6 years and the presence of two or more pathogenic mutations in the ATP binding cassette subfamily A member 4 ( ABCA4 ) gene or one pathogenic mutation with a typical STGD1 phenotype. The primary end point of ProgStar is the progression of atrophic lesions as measured on fundus autofluorescence (FAF) imaging, and therefore, a well-demarcated area of atrophy on FAF imaging with a minimum lesion diameter of 300 µm was required. At the same time, the summed area of FAF atrophy had to be ≤12 mm ² (equivalent to no more than five disc areas) in the primary study eye. A minimum BCVA of 20 Early Treatment Diabetic Retinopathy Study (ETDRS) letters (20/400 Snellen equivalent) was also required for inclusion.

Data Collection and Management

Clinical information, including visual acuity, demographics, and dilated fundus examination findings, were entered into Research Electronic Data Capture (REDCap; Institute of Translational Health Sciences) by the clinical sites and Doheny Imaging Reading Center (DIRC), the central reading center. The presence or absence of flecks outside the arcades was determined by clinical examination. The data coordinating center (Dana Center for Preventive Ophthalmology, Wilmer Eye Institute, Johns Hopkins University School of Medicine, and Johns Hopkins Bloomberg School of Public Health) then verified completeness and consistency of the collected data.

All microperimetry testing was performed with the Nidek MP-1 (Nidek Technologies Srl), and all machines were certified by the DIRC. The DIRC also checked the submitted examinations for quality and completeness and notified the sites if data were missing or had poor quality. Two certified graders graded all examinations, and if discordant, a DIRC investigator made a final assessment.

Genetic Categorization

All patients underwent genetic screening between 2000 and 2014. As part of the ProgStar study, detected variants were confirmed in 2016. The predicted severity of dysfunction of the ABCA4 gene product was used to categorize all patients into three groups. The classification was presented in detail previously, and was updated as new experimental/clinical evidence became available. In brief, patients with two or more severe/null variants were included in group A, those with one severe/null variant and at least one missense or in-frame insertion/deletion in group B, and those with two or more missense or in-frame insertion/deletion variants in group C.

Microperimetry

All patients underwent microperimetry testing using the Nidek MP-1, as described previously, ^, with 6-month follow-up visits for a total of 24 months. All testing was performed monocularly with the contralateral eye patched. Static fixation data were recorded during a fixation task that preceded the microperimetric sensitivity test. During this task, the patients were asked to fixate as steadily as possible on the center of a single red cross (2° in extension and 1° in thickness) on a white background. The testing lasted approximately 30 seconds after the target was located. During the test, real-time fundus images were acquired at a frequency of 25 Hz, and the locations of fixation events ( x, y coordinates) were registered for each image. A training examination was available to novice patients to minimize the effects of learning.

Description of the Preferred Retinal Locus

The cloud of recorded fixation events was then further analyzed, and the distance between their barycenter and the foveal center was measured. This distance in degrees defines the preferred retinal locus (PRL) eccentricity. The Nidek NAVIS software (1.7.0 or higher) can automatically center a concentric grid on this barycenter of all fixation events, and two graders manually determined the distance to the fovea by reading the grid. Optical coherence tomography imaging was used as an auxiliary technology for difficult cases with foveal atrophy, whereas the point of maximal inner retinal layer convergence of the adjoining B scans immediately superior and inferior to the approximate foveal center were viewed to determine the foveal location as closely as possible.

We previously performed a subgroup analysis of 45 randomly selected eyes. The PRL distance from the fovea for dynamic testing (based on the full macular sensitivity test) and for the separate fixation examination were not significantly different (ρ _PRL = 0.87; P < .001; n = 45). Because more data from the dynamic test were available and the results did not appear different from the static test, we used PRL eccentricity measurements from dynamic testing for all PRL analyses.

Description of the 1-SD Bivariate Contour Ellipse Area

The 1-SD Bivariate Contour Ellipse Area (1 SD-BCEA) is the smallest elliptical area that encompasses 1 SD (68.3%) of all recorded fixation events ( Figure 1 ). It is commonly calculated with the following formula:

Grading of the 1-SD bivariate contour ellipse area (1 SD-BCEA). The 1 SD-BCEA is the smallest possible ellipse that encompasses 1 SD of all fixation events (blue dots) in two perpendicular axes. A. The dark blue innermost ellipse represents the 1 SD-BCEA. The eccentricity of the preferred retinal locus (PRL) was determined in degrees from the anatomical fovea by overlying a grid centered on the barycenter of all fixation events (not shown). In A compared with B, fixation is more unstable, with a larger 1 SD-BCEA of 2.57 log deg (vs 0.59 log deg in B) and PRL eccentricity of 5° (vs 0° in B). C. The spaghetti plot shows the change of all study eyes over the five study visits. Each eye was assigned a color in the plot depending on its baseline 1 SD-BCEA. Those eyes with a baseline logBCEA in the 33rd percentile or below are indicated in red, those between the 33rd and 67th percentile in green, and those in the 67th percentile or above in blue. The black lines connect the means of the logBCEAs in the three groups between the visits over time. D. The plot visualizes the corresponding information for the eccentricity of the preferred retinal locus in degrees from the anatomical fovea, also described as fixation location (FL).

We calculated a global 1 SD-BCEA using the raw data of all recorded fixation events and their x and y coordinates. For this analysis and as in most of the available literature on fixation stability, we used data from the static fixation test to calculate the 1 SD-BCEA after export of raw data from the Nidek MP-1 in the “.mfd” format. All calculations were performed in Excel software (Microsoft).

Statistical Analyses

The two main outcome variables in this analysis are the eccentricity of the PRL from the anatomical fovea in degrees and the logarithm of the 1 SD-BCEA in log deg . All logarithmic calculations are based on the 1 SD-BCEA. The logBCEA was chosen because the distribution of the 1 SD-BCEA is strongly positively skewed, whereas its natural logarithm has a relatively normal distribution. Linear mixed-effects models were used for statistical analysis to estimate the yearly change for each outcome.