Purpose
To compare grading results between short-wavelength reduced-illuminance and conventional autofluorescence imaging in Stargardt macular dystrophy.
Design
Reliability study.
Methods
setting : Moorfields Eye Hospital, London (United Kingdom). patients : Eighteen patients (18 eyes) with Stargardt macular dystrophy. observation procedures : A series of 3 fundus autofluorescence images using 3 different acquisition parameters on a custom-patched device were obtained: (1) 25% laser power and total sensitivity 87; (2) 25% laser power and freely adjusted sensitivity; and (3) 100% laser power and freely adjusted total sensitivity (conventional). The total area of 2 hypoautofluorescent lesion types (definitely decreased autofluorescence and poorly demarcated questionably decreased autofluorescence) was measured. main outcome measures : Agreement in grading between the 3 imaging methods was assessed by kappa coefficients (κ) and intraclass correlation coefficients.
Results
The mean ± standard deviation area for images acquired with 25% laser power and freely adjusted total sensitivity was 2.04 ± 1.87 mm 2 for definitely decreased autofluorescence (n = 15) and 1.86 ± 2.14 mm 2 for poorly demarcated questionably decreased autofluorescence (n = 12). The intraclass correlation coefficient (95% confidence interval) was 0.964 (0.929, 0.999) for definitely decreased autofluorescence and 0.268 (0.000, 0.730) for poorly demarcated questionably decreased autofluorescence.
Conclusions
Short-wavelength reduced-illuminance and conventional fundus autofluorescence imaging showed good concordance in assessing areas of definitely decreased autofluorescence. However, there was significantly higher variability between imaging modalities for assessing areas of poorly demarcated questionably decreased autofluorescence.
Stargardt macular dystrophy is the most common form of juvenile macular degeneration. It is an autosomal recessively inherited disorder caused by disease-causing variants in the ABCA4 gene, encoding a photoreceptor-specific ATP-binding cassette transporter involved in active transport of all- trans -retinal across the disc membranes within photoreceptor outer segments. Failure of this process leads to the accumulation of N-retinylidene-N-retinyethanolamine (A2E), one of the major components of lipofuscin in the retinal pigment epithelium. High concentrations of N-retinylidene-N-retinyethanolamine and lipofuscin are believed to be cytotoxic, leading to dysfunction and cell death of the retinal pigment epithelium and photoreceptors. Clinically, one of the early hallmarks of Stargardt macular dystrophy is retinal “flecks,” which represent areas of lipofuscin accumulation. Flecks can resorb over time, and with disease progression, macular atrophy occurs with deterioration of retinal function. These characteristic features of Stargardt macular dystrophy can be easily and noninvasively imaged by confocal scanning laser fundus autofluorescence using signals originating from fluorophores (such as lipofuscin) within the retina and the retinal pigment epithelium after excitation by short-wavelength light. The accumulation of lipofuscin leads to areas of increased autofluorescence, whereas areas of atrophy are associated with decreased autofluorescence. Since its early descriptions, fundus autofluorescence has been widely explored, and recent technological advances, especially the introduction of confocal laser ophthalmoscopy and frame averaging techniques, have led to a significantly improved signal-to-noise ratio and enhanced quality of fundus autofluorescence images.
For these reasons, and because the U.S. Food and Drug Administration is considering fundus autofluorescence as a possible surrogate endpoint for clinical trials, fundus autofluorescence has been chosen as the primary outcome measure in the “Natural History of the Progression of Stargardt Disease: Retrospective and Prospective Studies” (ProgStar; ClinicalTrials.gov Identifier NCT01977846 ). These multicenter studies aim to characterize the natural course of Stargardt macular dystrophy and to validate possible outcome measures for emerging clinical trials including gene therapy, stem cell therapy, and pharmacotherapy.
However, the appropriate fundus autofluorescence imaging protocols remain controversial, in terms of both limiting potential toxicity from short-wavelength light and ensuring optimum image quality and thereby measurement sensitivity to monitor disease progression. One mechanism of the aforementioned cytotoxicity of N-retinylidene-N-retinyethanolamine is its mediation through blue light–induced damage to retinal pigment epithelial cells by photooxidative damage. The high-intensity and short-wavelength excitation light used in conventional fundus autofluorescence imaging could, at least in principle, increase the rate of lipofuscin accumulation and/or its toxicity. Cideciyan and associates were the first to describe the concept of using short-wavelength reduced-illuminance autofluorescence imaging in ABCA4 -associated retinopathy to reduce potential toxicity ; however, the question remains whether image integrity is compromised. Although short-wavelength reduced-illuminance autofluorescence imaging results have been reported to be both qualitatively and quantitatively comparable to those in the literature on conventional fundus autofluorescence imaging, one described practical shortcoming was noisier and darker images apparent on the acquisition screen, leading to more difficult imaging.
In the context of the ProgStar study, we acquired images where the power of the imaging laser beam was reduced to 25% of its conventional setting. For this purpose, a special software tool was developed and provided by Heidelberg Engineering (Heidelberg, Germany) to all participating sites in the ProgStar study.
The purpose of this study is therefore to compare the grading results of images obtained using the short-wavelength reduced-illuminance autofluorescence imaging method with those obtained with conventional fundus autofluorescence imaging in the same patient cohort and determine the correlation and areas of disagreement. This has implications for both routine clinical care and clinical trial endpoint design.
Methods
This reliability study was approved by the local Ethics Committee of Moorfields Eye Hospital, NHS Foundation Trust; adhered to the provisions of the Declaration of Helsinki; and complied with the Health Insurance Portability and Accountability Act. Ethics committee approval for the ProgStar study was granted by the Western Institutional Review Board, the local Ethics Committee of Moorfields Eye Hospital, NHS Foundation Trust, the institutional review board of Johns Hopkins University School of Medicine, and the Human Research Protection Office of the U.S. Army Medical Research & Materiel Command, prior to enrollment of the first patient, respectively. These studies have been registered at www.clinicaltrials.gov (Identifier NCT01977846 ). Written informed consent was obtained by all participants and (if applicable) their guardians prior to enrollment.
Subjects
Inclusion criteria were as follows:
- •
Age of at least 6 years
- •
Molecularly confirmed Stargardt macular dystrophy with at least 2 likely disease-causing mutations in ABCA4 ; if only 1 mutation was present, the patient had to have the typical phenotype for Stargardt macular dystrophy (ie, flecks at the level of the retinal pigment epithelium)
- •
Atrophic lesion of at least 300 μm in diameter; all lesions together must total less than 5 standard disc diameters (= 12.00 mm 2 )
- •
Clear ocular media and adequate pupillary dilation to permit good-quality fundus autofluorescence imaging
- •
Participation in the prospective ProgStar study.
Exclusion criteria were as follows:
- •
Other ocular disease, such as choroidal neovascularization, diabetic retinopathy, and degenerative retinal dystrophies other than Stargardt macular dystrophy
- •
Intraocular surgery 90 days prior to the imaging visit
- •
Current or previous participation in an interventional study for Stargardt macular dystrophy, such as gene or stem cell therapy.
Study Design and Image Acquisition
Pupils were dilated with 1% tropicamide and 2.5% phenylephrine hydrochloride and fundus autofluorescence images were acquired using a Spectralis FA + OCT device (Heidelberg Engineering, Heidelberg, Germany). The device was equipped with a custom-developed software tool provided by Heidelberg Engineering to reduce laser intensity during acquisition of fundus autofluorescence images. Using this software tool, the laser power can be reduced to 25%, 50%, or 75% of the preset 100% laser power. Three images with different acquisition parameters were obtained in the study eye of each patient by 1 of 4 experienced photographers certified for the ProgStar study. All images were acquired with a 30-degree field of view in the high-speed mode centered on the anatomic fovea, not normalized, and an automatic real-time (ART) averaging of ≥15 frames. Differences in the parameter settings were in the laser power and total sensitivity: (1) the first image was obtained with a laser power of 25% and a fixed total sensitivity of 87 over an imaging duration of ∼5 seconds; (2) the second image was obtained with 25% laser power; however, total sensitivity was not fixed, but adjusted by the photographer to optimize image illumination (“freely adjusted”); (3) the last image was obtained with laser power 100% (conventional) and total sensitivity that was adjusted for an optimal image exposure (“freely adjusted”) over an imaging duration of ∼30 seconds. The 25% setting corresponds to a retinal illuminance of 2.5 × 10 5 scot-trolands and the 100% setting to 10 × 10 5 scot-trolands. These 2 settings would be predicted to result in rhodopsin bleaches of 12% and 95%, respectively, in normal eyes. Time interval between each capture of images was at least 5 minutes.
Image Grading and Analysis
Deidentified images were sent to the Doheny Imaging Reading Center, David Geffen School of Medicine at the University of California Los Angeles, Los Angeles, California, and graded using a semi-automated software tool (Heidelberg Engineering RegionFinder). The quality of the images submitted for grading was assessed by evaluating focus and clarity (“blurriness” or “fuzziness”).
Quantitative Parameters
Three different categories of areas of decreased autofluorescence were graded. The reference points for a scale of decreased autofluorescence were the optic nerve head and blood vessels as “100% level of darkness” and the peripheral retinal background fundus autofluorescence as “0% level of darkness.” The term “definitely decreased autofluorescence” was defined for areas in which the level of darkness was close to 100% (at least 90%) in reference to the optic nerve head/blood vessels; regions with levels between 50% and 90% darkness were defined as “questionably decreased autofluorescence”; in these lesions, the sharpness of the corresponding lesion border defined a lesion either as “well-demarcated questionably decreased autofluorescence” or “poorly demarcated questionably decreased autofluorescence.” In cases with multiple types of lesion, the areas of all subtypes were summed, respectively. Images were independently graded by 2 reading center–certified graders, with at least 1 grader being a senior-level grader. Discordant initial assessments underwent adjudication by a reading center investigator if consensus could not be reached.
Qualitative Parameters
Qualitative parameters included the following: (1) focus and clarity of images; (2) presence of increased autofluorescence at the lesion edge; (3) background uniformity (a homogeneous background was defined as an even distribution of background autofluorescence; a heterogeneous background signal was defined as widespread small foci of increased or reduced autofluorescence, as previously described ); and (4) presence of flecks.
Statistical Analysis
The primary comparisons were areas of decreased autofluorescence (definitely decreased autofluorescence, well-demarcated questionably decreased autofluorescence, and poorly demarcated questionably decreased autofluorescence) in the 3 different image acquisition settings. Kappa coefficients (κ) were used for the assessment of intergrader agreement in qualitative grading (eg, the presence/absence of each type of decreased autofluorescence, background uniformity, etc) in the respective image, and kappa coefficients >0.61 were considered to be indicative of good agreement. Intraclass correlation coefficients (ICC) with 95% confidence intervals (CI) were used for the calculation of quantitative assessments (comparison between different image acquisition parameters). Differences in area sizes were compared using the paired t test. Statistical analyses were performed in SAS Statistical Analysis Software Version 9.4 (SAS Institute, Cary, North Carolina, USA) and R version 2.15.1 (The R Foundation for Statistical Computing, Vienna, Austria).
Results
Eighteen eyes of 18 patients were enrolled in this reliability study at Moorfields Eye Hospital, London (United Kingdom). If both eyes of a patient were eligible and enrolled into the ProgStar study, 1 eye was randomly chosen for the purpose of this study. Mean age (± standard deviation) was 38.3 (± 14.2) years ( Supplemental Table 1 ; Supplemental Material available at AJO.com ). Figure 1 provides illustrative examples of images obtained with all 3 different parameter settings and respective grading results of areas of definitely decreased autofluorescence and poorly demarcated questionably decreased autofluorescence.
Presence/Absence of Categories of Decreased Autofluorescence
Based on the grading of images acquired with 25% laser power and 87 total sensitivity (short-wavelength reduced-illuminance autofluorescence imaging), 15 eyes had definitely decreased autofluorescence lesions and 12 eyes had poorly demarcated questionably decreased autofluorescence lesions. Only 1 eye of 1 patient showed a lesion of well-demarcated questionably decreased autofluorescence ( Table 1 ). Based on the grading of images acquired with 25% laser power and freely adjusted total sensitivity (short-wavelength reduced-illuminance autofluorescence imaging), 15 eyes had definitely decreased autofluorescence lesions, 12 eyes had poorly demarcated questionably decreased autofluorescence lesions, and no eyes had lesion(s) of well-demarcated questionably decreased autofluorescence ( Tables 1 and 2 ). In contrast, with conventional fundus autofluorescence, 14 eyes had definitely decreased autofluorescence lesions and 12 eyes had poorly demarcated questionably decreased autofluorescence lesions. One eye had lesion(s) of well-demarcated questionably decreased autofluorescence ( Tables 1 and 2 ).
| Acquisition Parameters | Number of Eyes With Definitely Decreased Autofluorescence |
|---|---|
| Laser power 25%, 87 total sensitivity | |
| Absent | 3 |
| Present | 15 |
| Laser power 25%, freely adjusted total sensitivity | |
| Absent | 3 |
| Present | 15 |
| Laser power 100%, freely adjusted total sensitivity | |
| Absent | 4 |
| Present | 14 |
| Number of Eyes With Well-Demarcated Questionably Decreased Autofluorescence | |
| Laser power 25%, 87 total sensitivity | |
| Absent | 17 |
| Present | 1 |
| Laser power 25%, freely adjusted total sensitivity | |
| Absent | 17 |
| Present | 1 |
| Laser power 100%, freely adjusted total sensitivity | |
| Absent | 17 |
| Present | 1 |
| Number of Eyes With Poorly-Demarcated Questionably Decreased Autofluorescence | |
| Laser power 25%, 87 total sensitivity | |
| Absent | 6 |
| Present | 12 |
| Laser power 25%, freely adjusted total sensitivity | |
| Absent | 6 |
| Present | 12 |
| Laser power 100%, freely adjusted total sensitivity | |
| Absent | 6 |
| Present | 12 |
| Acquisition Parameters | Presence/Absence of Areas of Definitely Decreased Autofluorescence | |||
|---|---|---|---|---|
| Laser power 25%, freely adjusted total sensitivity | Laser power 100%, freely adjusted total sensitivity | |||
| Absent | Present | Absent | Present | |
| Laser power 25%, 87 total sensitivity | ||||
| Absent | 2 | 1 | 3 | 0 |
| Present | 1 | 14 | 1 | 14 |
| Kappa (95% confidence limits) = 0.60 (0.10–1.00) | Kappa (95% confidence limits) = 0.82 (0.49–1.00) | |||
| Absent | Present | |||
| Laser power 25%, freely adjusted total sensitivity | ——————- | |||
| Absent | 3 | 0 | ||
| Present | 1 | 14 | ||
| Kappa (95% confidence limits) = 0.82 (0.49–1.00) | ||||
| Acquisition parameters | Presence/Absence of Areas of Well-Demarcated Questionably Decreased Autofluorescence | |||
|---|---|---|---|---|
| Laser power 25%, freely adjusted total sensitivity | Laser power 100%, freely adjusted total sensitivity | |||
| Absent | Present | Absent | Present | |
| Laser power 25%, 87 total sensitivity | ||||
| Absent | 17 | 0 | 17 | 0 |
| Present | 1 | 0 | 0 | 1 |
| Kappa n/a a | Kappa = 1.0 | |||
| Absent | Present | |||
| Laser power 25%, freely adjusted total sensitivity | ——————- | |||
| Absent | 16 | 1 | ||
| Present | 0 | 0 | ||
| Kappa n/a a | ||||
| Acquisition parameters | Presence/Absence of Areas of Poorly Demarcated Questionably Decreased Autofluorescence | |||
|---|---|---|---|---|
| Laser power 25%, freely adjusted total sensitivity | Laser power 100%, freely adjusted total sensitivity | |||
| Absent | Present | Absent | Present | |
| Laser power 25%, 87 total sensitivity | ||||
| Absent | 5 | 1 | 5 | 1 |
| Present | 1 | 11 | 1 | 11 |
| Kappa (95% confidence limits) = 0.75 (0.42–1.00) | Kappa (95% confidence limits) = 0.75 (0.42–1.00) | |||
| Absent | Present | |||
| Laser power 25%, freely adjusted total sensitivity | ——————- | |||
| Absent | 6 | 0 | ||
| Present | 0 | 12 | ||
| Kappa = 1.0 | ||||
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
