Longitudinal Rates of Structural Change Measured with Confocal Scanning Laser Ophthalmoscopy in Glaucoma

Abstract

Purpose : To determine the longitudinal rates of structural change measured with confocal scanning laser ophthalmoscopy (CSLO) in glaucomatous eyes.

Methods : Suspected or confirmed glaucoma patients who had 6 or more acceptable quality (global pixel SD ≤ 50 um) CSLO (Heidelberg Retina Tomograph II [HRT]; Heidelberg Engineering, Germany) examinations were included. Global and regional rim area was regressed over time with linear and exponential models. The root mean square errors (RMSEs) of the two models were compared to determine best fits. For predictions, models were fit with all data excluding the either the final or the last 3 observations, and observed vs. predicted values were compared. The rates of change for global and sectoral rim areas were calculated. Correlations between rate of rim area change and Glaucoma Rate Index (GRI), MD rate, and VFI rate were calculated.

Results : 822 eyes of 486 patients were included. The median (±SD) follow-up time and number of CSLO images were 14.0 years (± 3.3) and 7.1 (± 1.4), respectively. The linear regression fit the model better (linear RMSE 0.0580 (± 0.0538) vs exponential RMSE 0.0588 (±0.0563); p < 0.001). However, the exponential model provided better predictions (linear 0.132 (± 0.16) vs exponential 0.128 (±0.15); p = 0.006); this was true for both shorter- and longer-term predictions. Global rim area changed by −0.0068 (±0.019) mm²/year with linear regression. The rate of rim area change was fastest for the inferotemporal and superotemporal sectors. The rate of global rim area change was weakly correlated with GRI (r =0.109, p=0.0017), and the MD rate (r=0.133, p < 0.001), but not with VFI rate (r=0.027, p=0.452)

Conclusions : While a linear regression model fits better for fitting CSLO parameters over time, the exponential model provides better predictions. Rates of rim area change were greatest in the inferotemporal and superotemporal sectors, as expected. Rim area change was significantly correlated with GRI and MD rate, but not with VFI rate. Modeling structural parameters to better estimate rates of change and predict future deterioration help to inform appropriate treatment decisions. Monitoring structural changes with CSLO is useful to measure and interpret glaucomatous deterioration.

This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.