Modelling normal age-related changes in individual retinal layers using location-specific OCT analysis

Abstract

Current descriptions of retinal thickness across normal age cohorts are mostly limited to global analyses, thus overlooking spatial variation across the retina and limiting spatial analyses of retinal and optic nerve disease. This retrospective cross-sectional study uses location-specific cluster analysis of 8 × 8 macular average grid-wise thicknesses to quantify topographical patterns and rates of normal, age-related changes in all individual retinal layers of 253 eyes of 253 participants across various age cohorts (n = 23–69 eyes per decade). Most retinal layers had concentric spatial cluster patterns except the retinal nerve fibre layer (RNFL) which displayed a nasal, asymmetric radial pattern. Age-related thickness decline mostly occurred after the late 4th decade, described by quadratic regression models. The ganglion cell layer (GCL), inner plexiform layer (IPL), inner nuclear layer (INL), and outer nuclear layer + Henle’s fibre layer (ONL+HFL) were significantly associated with age (p < 0.0001 to < 0.05), demonstrating similar rates of thickness decline (mean pooled slope = − 0.07 µm/year), while the IS/OS had lesser mean pooled thickness slopes for all clusters (− 0.04 µm/year). The RNFL, OPL, and RPE exhibited no significant age-related thickness change, and the RNFL were significantly associated with sex. Analysis using spatial clusters compared to the ETDRS sectors revealed more extensive spatial definition and less variability in the former method. These spatially defined, clustered normative data and age-correction functions provide an accessible method of retinal thickness analysis with more spatial detail and less variability than the ETDRS sectors, potentially aiding the diagnosis and monitoring of retinal and optic nerve disease.

Location-Specific Thickness Patterns in Intermediate Age-Related Macular Degeneration Reveals Anatomical Differences in Multiple Retinal Layers

“…OCT macular cube scans, covering an area of 8600 µm × 7167 µm or 30° × 25° across 61 B-scans, were acquired with Spectralis SD-OCT (Heidelberg Engineering, Heidelberg, Germany) as previously described. 37 , 53 , 54 In the presence of multiple scans per participant, the earliest scan meeting inclusion criteria without significant artefacts and with signal strength above 15dB were selected. Scans were corrected for ocular tilt and automatic segmentation applied to each individual retinal layer using the HRA/Spectralis Viewing Module 6.9.5.0 (Heidelberg Engineering, Heidelberg, Germany).…”

“…according to previous studies. 37 , 53 , 54 One of two randomized blocks of all participants had their scans reviewed independently by M.T. and V.K.…”

“… 55 – 57 Manual correction of segmentation for iAMD eyes in this study were performed to the same standard as manual correction of segmentation for normal eyes, which were completed in previous studies. 37 , 53 , 54 …”

“…Recently, we developed a method of location-specific OCT analysis using cluster analysis that provides greater spatial detail and less variability than the ETDRS sectors. 37 Spatial clustering involves assigning location-specific data into groups of statistical likeness. Use of this method has enabled the formation of OCT-derived topographical thickness maps for each individual retinal layer.…”

“…Use of this method has enabled the formation of OCT-derived topographical thickness maps for each individual retinal layer. 37 These maps are reflective of retinal neuronal distributions described via histological studies more so than the standard ETDRS sectors, which assume neuronal distribution to be concentric and symmetric around the fovea. 37 Application of the GCL topographical thickness map to iAMD eyes using cluster analysis has subsequently led to observation of unique location-specific patterns of thinned GCL toward the central macula and thickened GCL toward the peripheral macula when compared to normal eyes, both statistically significant and large in magnitude of effect.…”