Rates of change and longitudinal variability of using multiple OCT retinal nerve fibre layer ring sizes in glaucoma

• OCT segmentation errors were similar across three cp-RNFL ring sizes • Rates of change were similar across ring sizes for global and sectoral thickness • Longitudinal variability was significantly lower in the 4.7mm ring in three sectors To compare the proportion of segmentation errors, rates of change and longitudinal variability of optical coherence tomography (OCT) circumpapillary retinal nerve fibre layer (cp-RNFL) global and sectoral thickness measurements when using three different ring sizes in a cohort of healthy and glaucomatous eyes. Retrospective cohort study. 205 eyes of 105 patients with normal, suspect or manifest glaucoma. Subjects underwent cp-RNFL scans on four or more consecutive follow up visits (mean follow up 4.5 years). Scans were taken using the Spectralis® OCT (Heidelberg Engineering, Heidelberg, Germany). Global thickness and six sectoral thickness values in the three ring sizes (3.5, 4.1, 4.7mm) were extracted. Longitudinal variability was quantified by mean absolute regression residuals (MAR). Rates of change (RoC) and longitudinal variability of global and sectoral cp-RNFL thickness across three ring sizes. The rate of global cp-RNFL thinning in the 3.5mm ring was -0.40 µm/year (CI -0.58 to -0.22, p < 0.001), while the inferotemporal sector displayed the greatest rate of thinning (-0.97 µm/year, CI -1.26 to -0.68, p < 0.001). There was no significant difference in the RoC of thickness values in the 4.1mm and 4.7mm compared to the 3.5mm ring size in the global or any of the six sectors. Longitudinal variability was consistently lower in the 4.7mm compared to the smaller ring sizes, with a significant difference observed in the inferonasal, inferotemporal and superonasal sectors. While there was no significant difference in the global and sectoral RoC across ring sizes, the larger RNFL ring sizes demonstrated significantly lower longitudinal variability for inferonasal, inferotemporal and superonasal sectors compared to the smaller ring sizes. Using larger RNFL ring sizes may therefore provide more accurate estimates of the true rate of change overtime.