Relationship of Optic Nerve Structure and Function to Peripapillary Vessel Density Measurements of Optical Coherence Tomography Angiography in Glaucoma.

PURPOSE

To evaluate the sectoral and global structure-structure (vessel density-retinal nerve fiber layer thickness) and structure-function (vessel density-visual sensitivity loss) relationships of peripapillary vessel density measurements on optical coherence tomography angiography in primary open-angle glaucoma and to determine if fractional polynomial (FP) models characterize the relationships better than linear models.

MATERIALS AND METHODS

In a cross-sectional study, structure-structure and structure-function relationships of peripapillary vessel densities were determined in 227 eyes of 143 subjects (63 control and 164 primary open-angle glaucoma eyes) who had undergone standard automated perimetry and optical coherence tomography testing within 6 months of each other, using linear and FP models. FP model evaluates the relationship between the dependent and the best-fitting fractional powers of the independent variable. Strength of relationship was reported as coefficient of determination (R).

RESULTS

R values for structure-structure associations using linear models (0.53 for superotemporal sector, 0.61 for inferotemporal, and 0.53 for average measurements) were significantly less (P<0.05) than that determined using FP models (0.57, 0.65, and 0.55, respectively). R values for structure-function associations using linear models (0.35 for superotemporal vessel density-inferotemporal visual sensitivity loss, 0.49 for inferotemporal vessel density-superotemporal visual sensitivity loss, and 0.39 for average vessel density-average visual sensitivity loss) were significantly less than that determined using FP models (0.43, 0.58, and 0.47, respectively).

CONCLUSIONS

The inferotemporal peripapillary vessel density showed the strongest association with the corresponding retinal nerve fiber layer thickness and visual sensitivity loss in the global and sectoral regions studied. The FP models were significantly better than linear models in describing these relationships.