Determinants of Agreement Between the Heidelberg Retina Tomograph (HRT) Topographic Change Analysis (TCA) and Standardized Assessment of Glaucomatous Progression in the Diagnostic Innovations in Glaucoma Study (DIGS)

Abstract

Purpose: : To estimate the agreement of HRT TCA with assessment of stereophotographs and standard automated perimetry (SAP) for detecting glaucomatous change and to identify factors associated with agreement between methods.

Methods: : HRTII series (≥ 4, mean follow-up 4 years), stereophotographs and SAP results from 246 eyes of 167 DIGS glaucoma patients, suspects and ocular hypertensives were included in the analysis. 36/246 eyes (14.6%) were classified as progressors by masked stereophotograph assessment or SAP Guided Progression Analysis (GPA, "likely progression"). 15 (6.1%) progressed by stereophotographs only, 18 (7.3%) progressed by GPA only and 3 (1.2%) progressed by both. TCA progression was determined for 3 previously described parameters (CSIZE^disc, number of progressed superpixels in the largest cluster; CAREA^disc, area of progressed superpixels in the largest cluster; and CVOL^disc, volume of progressed superpixels in the largest cluster; all 3 within the disc margin) at each of 3 previously determined specificity cut-offs (99%, 95%, 90%)¹. Agreement between progression by each TCA parameter, stereophotographs and GPA was assessed using the Kappa test. ANOVA with post-hoc analysis was applied to identify factors including image quality (standard deviation of the mean topography), disc size and disease severity (PSD and cup area) associated with agreement/non-agreement between methods.

Results: : Agreement was generally poor regardless of the TCA parameter and specificity cut-offs applied. For the parameters with the strongest agreement, CSIZE^disc and CAREA^disc, Kappa values were 0.14 (72.8%, agreement on 165 non-progressing eyes and 14 progressing eyes) at 95% cut-off and 0.15 (63.8%, agreement on 135 non-progressing eyes and 22 progressing eyes) at 99% cut-off. Most of the factors evaluated were not significantly associated with agreement/non-agreement between methods (all p>0.11). However, SAP PSD was greater in the progressors by stereophotograph only group compared to the progressors by TCA only group (5.8±4.6 and 2.7±2.3, respectively p=0.001 for CSIZE ^disc at 95% specificity and 5.3±4.6 and 2.6±2.3, respectively p=0.003 for CAREA ^disc at 99% specificity).

Conclusions: : Agreement between TCA, stereophotography and SAP GPA is poor. However, longer follow-up may result in better agreement among methods. Progressors by stereophotography tended to have more advanced disease at baseline than progressors by TCA only.1. Bowd C et al, IOVS Oct 3 2008 Epub.

Keywords: imaging/image analysis: clinical • optic disc • visual fields