Optical coherence tomography derived macular volume loss over 5 years in Stargardt disease

Abstract

Purpose : To examine the variability in the rate of macular volume loss (MVL) in Stargardt disease derived from spectral domain-optical coherence tomography (SD-OCT).

Methods : Subjects with clinical Stargardt disease and biphasic ABCA4 mutations were eligible. Assessments included ETDRS visual acuity, fundus examination, widefield fundus autofluorescence (FAF, Optos) and SD-OCT (30°×25°, 61 slices with 119µm separation, Heidelberg Spectralis OCT) imaging. FAF pattern was divided into types A and B, defined by lesions confined to within or extended beyond the central 50°, respectively. Manual centration of the ETDRS grid and correction of segmentation errors in the inner limiting and Bruch’s membranes were performed. Patients with choroidal neovascular membrane, severe epiretinal membrane and decentred scans leading to image truncation within the ETDRS grid boundary were excluded. Annualised MVL was calculated by linear regression after confirming the rate of decline was constant. Inter-ocular (OD-OS) and inter-individual variabilities (OD only) in baseline total macular volume (TMV) and annualised MVL were explored.

Results : Forty eyes of 20 patients (mean age 39 years, mean symptom duration 13 years) were recruited and followed for a mean (SD) of 5.0 (1.0) years (range: 3.5-6.6 years). Five patients had late-onset disease (defined as symptom onset >40yrs) and 8 had a Type A FAF pattern. Mean (SD, range) number of OCT scanning sessions was 7.3 (2.1, 4-12) during follow up. Inter-ocular differences and agreements (bias, SD, 95% CI) in TMV (mm³) and MVL (mm³/year) were +0.03, 0.31, -0.12 to +0.18 and +0.02, 0.10, -0.03 to +0.06, respectively. Baseline TMV (OD) was negatively correlated with duration of symptoms (Pearson correlation, r = -0.76, p < 0.001) and associated with FAF pattern (7.55 vs 6.01 mm³, A vs B, p = 0.01). In contrast, there was no significant relationship between MVL (OD) and duration of symptoms or FAF patterns. A trend for lower rates of MVL in those with longer symptom duration (r = 0.27, p = 0.25) and FAF pattern A (mean (SD): -0.09 (0.07) vs -0.16 (0.12) mm³/year, p = 0.16, respectively) were noted.

Conclusions : MVL in Stargardt disease requires extensive and time-consuming manual correction of segmentation errors. Duration of symptoms and FAF patterns are associated with baseline TMV but not with MVL. These estimates in MVL can be used for sample size calculation in clinical trials.

This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.