Author: Danial Roshandel (Australia)
Co-authors: Rachael C. Heath Jeffery, Jason Charng, Danuta M. Sampson, Samuel McLenachan, David A. Mackey, Fred K. Chen
Spectral-domain optical coherence tomography (SD-OCT) is the standard method for evaluating retinal structure in various pathologies, including retinitis pigmentosa (RP). The span and area of the ellipsoid zone (EZ) decline at a rate of approximately 4–10% and up to 13% per year, respectively, in patients with RP. Current SD-OCT devices lack the resolution to visualize individual photoreceptors. Adaptive optics (AO) imaging is an evolving technology enabling in vivo photoreceptor imaging in the clinical setting. Multiple AO retinal cameras have been used to measure cone density (CD) at different retinal eccentricities in healthy individuals and a range of retinal pathologies, including RP. However, longitudinal changes in CD in patients with RP have not been reported previously. We investigated the feasibility and utility of AO flood-illumination ophthalmoscope (FIO) in analysing the cone mosaic and monitoring parafoveal CD in RP.
This study was the single-centre prospective observational cohort; a part of the Western Australian Retinal Degeneration study. All examinations were performed between September 2018 and December 2020 at the Lions Eye Institute, Perth, Western Australia, Australia.
Patients with RP and healthy controls were evaluated for eligibility for AO imaging. Exclusion criteria included visual acuity (VA) worse than 20/40, poor fixation, history of ocular disease or surgery, significant cataract or other media opacity, nystagmus, cystoid macular oedema, epiretinal membrane and history of using systemic medications with known photoreceptor toxicity. All patients and controls underwent a complete ophthalmic examination with right eye SD-OCT and AO-FIO imaging using a commercial AO retinal camera (rtx1, Imagine Eyes, Orsay, France). Nasal and temporal boundaries of the EZ were marked on the foveal-centred SD-OCT scan and used to measure residual EZ span. Commercial analysis software (AODetect version 3.0, Imagine Eyes, Orsay, France) was used for cone segmentation and measurement of CD at 2° temporal from the foveal centre using regions of interest selected from four partially overlapping tiles. Coefficient of repeatability (CoR) was calculated using partially overlapping images taken at the same session. For each case, the highest CD value from one of the four tiles was used for statistical analysis. All patients were followed for 6 months and measurements were performed on the same location using the same protocol. CD and CoR were reported in angular (cells/deg2) and metric (cells/mm2) units.
Eight patients and 10 age-matched healthy controls were enrolled. There was no significant difference between the control and patient groups in baseline age (p=0.72), spherical equivalent (p=0.99) and axial length (p=0.60). CoR of CD using the same session overlapping tiles was 530 cells/deg2 (6307 cells/mm2) in the control group and 659 cells/deg2 (8364 cells/mm2) in the patient group. Mean baseline CD in the control and patient groups was 2094 cells/deg2 (24732 cells/mm2) and 1750 cells/deg2 (21689 cells/mm2), respectively (p=0.09 and 0.28 for the angular and metric values, respectively). Mean CD declined by 198 cells/deg2 or -11.3% (2356 cells/mm2 or -10.9%) at 6 months follow up in the patient group (p=0.01 for both angular and metric values). EZ span was beyond the SD-OCT imaging field (30°) in two patients. Baseline and follow up residual EZ span of the remaining 6 patients was 3189 µm and 3065 µm, respectively, which was not statistically significant (-3.9%, p=0.08). CD declined by -8.4% and -25.9% in the two patients with greater than 30° EZ span. Mean baseline and follow-up logMAR best-corrected VA in the patient group were 0.08 (0.12) and 0.03 (0.13), respectively (p=0.08).
There was 15% difference in baseline parafoveal CD between the patient and control groups, irrespective of EZ span. Although not statistically significant, this finding suggests further investigation is warranted in exploring the potential use of AO imaging in early stage RP. Significant decline in CD at 6 months despite a stable EZ span suggests that the parafoveal CD may be more sensitive than EZ span for detecting RP progression. CoR using partially overlapping tiles was greater than previous reports, suggesting potential errors in using overlapping tiles and performing CD analysis on montage images. Our results support further studies are required to explore the reliability and short-term test-retest variability of CD measurements in larger patient samples with longer follow up durations.