Abstract
Purpose :
To evaluate the significance of factors that may confound glaucoma diagnosis using structural OCT parameters such as retinal nerve fiber layer (RNFL) and ganglion cell complex (GCC) thickness measurements.
Methods :
This study analyzed the peripapillary RNFL thickness and macular GCC thickness of healthy participants in the Advanced Imaging for Glaucoma study. In this longitudinal study, Healthy eyes were scanned with spectral domain OCT every year. OCT Data were selected between 2009 and 2013. Confounding factors, including age, gender, race, axial length, and disc area, were evaluated. The disc area was obtained from OCT and corrected for magnification. A linear mixed model followed by ANOVA was used to get the relative variance (partial R2 ) explained by each factor, suggesting its contribution to the variance. In calculation of the contribution percentage of factors, variance due to repeated measurements was exclude from the total variance. To test the significance of magnification adjustment, we compared two models: 1) the full model with all confounding factors, and 2) the reduced model without axial length and disc area.
Results :
210 eyes from 106 Healthy participants were analyzed. The participants had an average of 3 visits. The average age was 61.3±9.7 years; 66% were female; 86.8% were white, 5.7% were black, and 7.5% were Asian. The average RNFL thickness was 99.2±8.2 µm; the average GCC thickness was 95.4±6.8 µm. In the reduced models, age contributed to most of the variance among confounding factors (Table 1). In the full model, axial length contributedto most of the variance. We also noticed that the contribution of age was reduced in the full model. Disc area and race had small borderline contributions to RNFL thickness. Subgroup analysis showed that Asians had significantly thicker RNFL thickness than white (p=0.032). The contribution of gender was negligible.
Conclusions :
Axial length variation affects the transverse image magnification and is by far the largest source of measurement bias in glaucoma diagnostic parameters measured by OCT. Age and Asian race are two significant demongraphic confounders, but of less importance than axial length variation. Accounting for axial length or other axial refractive error variables could greatly improve the diagnostic accuracy of OCT measurements.
This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.