Purchase this article with an account.
Hemma Resch, Ivania Pereira, Stephan Holzer, Barbara Kiss, Georg Fischer, Clemens Vass; Multiple linear regression analysis between circumpapillary retinal nerve fiber layer thickness profiles measured with scanning laser polarimetry and retinal blood vessel thickness, optic disc parameters and age. Invest. Ophthalmol. Vis. Sci. 2013;54(15):4807.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Variations of the normal quantity and distribution of retinal nerve fiber layer (cRNFL) thickness may compromise accurate glaucoma diagnosis. Following the idea that vessels mark regions of thicker cRNFL, the purpose of the present work is to correlate cRNFL with circumpapillary retinal vessel thickness (cRVT) together with optic disc (OD) size and shape as well as age.
cRNFL thickness values of 106 healthy subjects were measured using scanning laser polarimetry (SLP), (GDx-ECC ® Carl Zeiss Meditec Inc.) and OD HR-OCT scans (Cirrus® Carl Zeiss Meditec Inc.) were obtained. A proprietary software was developed in MATLAB® (Version R2009b, The Mathworks Inc.) for manual identification of the OD contour and assessment of the cRVT at OD border, using the SLO image from Cirrus HR-OCT centered in the OD. The angle of the intersection between a horizontal line passing through the OD center and a line between OD center and the center point of each vessel measured was calculated. cRVT data were convoluted with a Gaussian window, generating a 64-sector profile dependent on the 64 sectors for GDxECC. Intersubject correlations were calculated using SPSS Software Package (version 17.0). In a stepwise multiple linear regression analysis we used cRVT, OD area (ODA), OD vertical to horizontal diameter ratio (VHR) and age as independent and cRNFL thickness as dependent variables for each of the 64 sectors. The significance level was p<0.05.
52 out of 64 sectors had a significant intersubject correlation in multiple linear regression analysis. cRVT, ODA, VHR and age were included either alone or together with other variables in 30/64, 16/64, 14/64, 17/64 sectors with a maximum regression coefficient (Rmax) of 0.51, 0.31, 0.27 and 0.35. Both cRVT and VHR, ODA and age, cRVT and age were included in 11/64, 1/64 and 12/64 with Rmax of 0.55, 0.33 and 0.44.
In multiple linear regression we found significant impact of one or more of the investigated independent variables on the cRNFL in 80% of the sectors. The single most important variable was the cRVT, significantly contributing to the cRNFL in 47% of the sectors. The cRVT profile together with age and other OD parameters, as ODA and VHR might explain up to 30% of the interindividual variation of the cRNFL thickness profiles.
This PDF is available to Subscribers Only