Abstract
Purpose: :
Ganglion cell losses are correlated with visual field (VF) defects in glaucoma. An indirect measurement of the in vivo ganglion cell count is the ganglion cell layer thickness (GCLT) as measured by spectral domain optical coherence tomography. This study analyzed the accuracy of RTVue (OptoVue Inc, Fremont, CA) GCLT analysis for the presence of a VF defect on automated perimetry.
Methods: :
GCLT data between July 15 and October 15, 2008 were retrospectively analyzed. Glaucoma suspects and glaucoma patients with early to moderate glaucomatous VF defects were included to the analysis if they had a reliable VF within 6 months. Poor quality GCLT images and eyes with other eye diseases which may affect VF or GCLT analysis were excluded. A glaucomatous VF defect was described as the presence of a cluster of 3 or more points depressed below the 5% level, at least 1 of which is depressed below the 1% level. GCLT abnormality was described in 3 different ways and sensitivity and specificities were calculated for each description: 1) presence of significantly thin focal areas in GCLT, 2) abnormal average GCLT in each hemisphere (p<5%), and 3) severely abnormal average GCLT in each hemifield (p<1%).
Results: :
There were 266 eyes of 146 patients that met the criteria for inclusion in the study. The presence of GCLT defects had 78.9% sensitivity and 61.9% specificity to detect the VF defects. Superior GCLT defects had 71.7% sensitivity and 70.7% specificity to detect inferior VF defects while inferior GCLT defects had 73.3% sensitivity and 65.5% specificity to detect superior VF defects. When description 2 is used, abnormal superior GCLT can predict the presence of inferior VF defects with 34.9% sensitivity and 89.3% specificity. Abnormal inferior GCLT can predict the presence of superior VF defects with 43.6% sensitivity and 89% specificity. When description 3 is used, severely abnormal superior GCLT can predict the presence of inferior VF defects with 27.4% sensitivity and 95.7% specificity. Severely abnormal inferior GCLT can predict the presence of superior VF defects with 34.7% sensitivity and 95.2% specificity.
Conclusions: :
Focal areas of abnormal GCLT correlate with the presence of defects in the corresponding VF hemifield with moderate sensitivity and specificity. Although classifications of average thicknesses as normal or abnormal for each hemifield were found to be specific, their low sensitivity limits their use to predict the presence of visual field defects.
Keywords: visual fields • ganglion cells • imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound)