Long-term natural history of ellipsoid zone width in USH2A-retinopathy

Abstract

Aims: To investigate the long-term natural history of ellipsoid zone (EZ) width in USH2A-retinopathy.

Methods: EZ width measurements from optical coherence tomography were retrospectively obtained from 110 eyes of 55 participants with molecularly confirmed biallelic USH2A-retinopathy. We used a hierarchical Bayesian method to construct and compare different mathematical models describing the long-term decline of EZ width.

Results: Compared with linear and quadratic models, exponential decline best represented the long-term loss of EZ width based on the deviance information criterion score. Log-transformed EZ width declined linearly over 30 years of inferred disease duration (median: 0.063 (IQR: 0.040-0.086) log (µm)/year). Compared with the raw EZ width decline rate, the log-transformed EZ width decline rate required 48% fewer patients to achieve an identically powered 1-year trial (38 vs 73 participants). Log EZ width decline rate was uncoupled from baseline EZ width (Spearman ρ=-0.18, p=0.06) and age (ρ=-0.10, p=0.31). Eyes with Usher syndrome exhibited earlier median onset ages of macular EZ width loss (18.8 (IQR: 13.1-24.7) vs 28.1 (IQR: 18.5-35.8) years, p<0.001) but comparable log EZ width decline rates (0.060 (IQR: 0.035-0.100) vs 0.065 (IQR: 0.050-0.079) log (µm)/year; p=0.42).

Conclusions: EZ width follows an exponential decline in USH2A-retinopathy. Compared with raw EZ width decline rate, log-transformed EZ width decline rate may be a superior endpoint for clinical trials. Syndromic eyes exhibit an earlier onset of macular EZ width loss but progress at comparable rates to non-syndromic eyes.

Keywords: Degeneration; Epidemiology; Imaging; Retina.

Figure 1. Representative set of spectral-domain optical coherence tomography (OCT) scans obtained over three longitudinal visits from a single study participant. We measured the width of the ellipsoid zone (EZ) band (marked by yellow arrows) at each visit. (A) OCT scan at baseline visit; EZ width=3900 µm. (B) OCT scan 3 years after baseline visit; EZ width=3337 µm. (C) OCT scan 4.75 years after baseline visit; EZ width=2912 µm.

Figure 2. Graphical representations of ellipsoid zone (EZ) width loss over time. USH2, n=68 eyes; NSRP, n=42 eyes. (A) EZ width plotted against time following enrolment (baseline visit). Due to limited follow-up durations (<5 years), the long-term progression is not discernible. (B) Same data as (A) with EZ width plotted against age and fitted to an exponential trendline (EZ=10 000×e^{−0.0406×age}). (C) Linearised version of (B) after applying natural log to EZ width. Log EZ width appears to decline linearly, although varies at the same age among different eyes, suggesting heterogeneity in the onset age of macular EZ width loss and potentially the rate of disease progression. (D) Log EZ width as a function of inferred duration of macular EZ width loss. Because the age did not sufficiently capture the disease duration, we used Bayesian entry time realignment to estimate the duration of macular EZ width loss for each eye and horizontally translated the datasets in (A) to convert the horizontal axis from time after enrolment to inferred duration of macular EZ width loss. Realignment of the data demonstrated that the log EZ width of individual eyes declined linearly at different rates over approximately 30 years. Eyes with similar log EZ width decline rates were in-line with one another, extending from a single origin at log (6000 µm). NSRP, non-syndromic retinitis pigmentosa; USH2, Usher syndrome type II.

Figure 3. Scatterplots showing the relationship of the ellipsoid zone (EZ) width decline rate with baseline age and EZ width. Solid red lines indicate linear trendlines. (A) EZ width decline rate was correlated with baseline age (Spearman’s correlation coefficient, ρ=−0.41, p<0.001). (B) The use of log-transformed EZ width decline rate reduced the ρ value between decline rate and baseline age to −0.10 (p=0.31). (C) EZ width decline rate was correlated with baseline EZ width (ρ=0.67, p<0.001). (D) Using log EZ width reduced the association between decline rate and baseline EZ width (ρ=−0.18, p=0.06).

Figure 4. Boxplots depicting estimated onset age of macular ellipsoid zone (EZ) width loss and log EZ width decline rates, stratified by presence of syndromic disease. USH2, n=68 eyes; NSRP, n=42 eyes. (A) Among all 110 eyes, the median and IQR of onset age of macular EZ width loss was 21.1 (14.8–29.0) years. Syndromic eyes exhibited an earlier onset age of macular EZ width loss (18.8 (IQR: 13.1–24.7) years) compared with non-syndromic eyes (28.1 (IQR: 18.5–35.8) years) (p<0.001). (B) Among all eyes, the median log EZ width decline rate was 0.063 (IQR: 0.040–0.086) log (μm)/year. The log EZ width decline rate was comparable between syndromic eyes (0.060 (IQR: 0.035–0.100) log (μm)/year) and non-syndromic eyes (0.065 (IQR: 0.050–0.079) log (μm)/year) (p=0.42). NSRP, non-syndromic retinitis pigmentosa; USH2, Usher syndrome type II.