Influence of fovea position and retinal vessels in interindividual variability of healthy circumpapillary retinal nerve fiber layer measurements

Abstract

Purpose: To assess whether intersubject variability in healthy circumpapillary retinal nerve fiber layer (RNFL) measurements acquired with spectral domain optical coherence tomography (SD-OCT) may be reduced by considering the fovea position, either alone or together with a compensation based on the retinal vessel distribution (RVD).

Methods: A sample of 106 healthy volunteers underwent complete ophthalmological examination including SD-OCT (Cirrus® Carl Zeiss Meditec Inc.). For each subject, both SLO images centered in the optic disc (OD) and in the macula were acquired with SD-OCT. A proprietary software was developed in Matlab (Version R2012b) to manually assess OD contour, RVD at 3.4 mm diameter circle and fovea position. RVD is a function of retinal vessel thickness and its position relatively to the OD center. Both SLO images were manually registered and the angle between a line connecting fovea and OD centers and the horizontal axis passing through OD center was calculated (foveal angle). Compensation for interindividual variability in RNFL was based in two methods : 1^st Compensation: RNFL thickness is compensated according to RVD variation; 2^nd Rotation: we shifted RNFL measurements according to the foveal angle. The coefficient of variance (CoV) was calculated in 12 clock hour sectors for original, rotated and compensated RNFL (RNFLo, RNFLr and RNFLc, respectively), and both rotated and compensated RNFL (RNFLrc). CoV reductions were compared across the 12 sectors using paired t-tests.

Results: Compared to the mean CoV of RNFLo, the mean CoV of RNFLr, RNFLc and RNFLrc was changed by -0.47% (18.00% to 17.95%, p=0.905), -10.42% (18.00% to 16.08%, p<0.05) and -10.20% (18.00% to 16.16%, p<0.05), respectively. Compared to RNFLr, RNFLrc did significantly reduce the mean CoV (17.95% to 16.16%, p<0.05), while, compared to RNFLc, RNFLrc did not significantly change the mean CoV (16.08% to 16.16%, p=0.782).

Conclusions: Although reaching an improvement in some sectors, rotation of RNFL measurements according to the fovea angle, both original and compensated, do not present, on average, a reduction in intersubject variability of RNFL. However, RVD compensation significantly reduced the intersubject variability of RNFL. The results reinforce our work in assessing RVD as one of the most important anatomical factors responsible for intersubject variability.