Abstract
Purpose:
We have previously described a method of estimating retinal ganglion cell (RGC) counts in glaucoma using a combination of standard automated perimetry (24-2) and retinal nerve fiber layer (RNFL) thickness assessment by optical coherence tomography. However, the SAP 24-2 test evaluates only a few points in the macular region, which contains the largest proportion of RGCs. Therefore, we hypothesized that improved estimates of RGC counts could be obtained by incorporating 10-2 data along with 24-2 and OCT measurements.
Methods:
Observational cross-sectional study including 259 glaucomatous and 132 control eyes, who were followed as part of the Diagnostic Innovations in Glaucoma (DIGS) study. Eyes were classified as glaucomatous if they had evidence of progressive glaucomatous optic neuropathy on stereophotos or repeatable field defects. Controls were recruited from the general population. All eyes were tested with 24-2 and 10-2 SAP and spectral domain OCT within 6 months. Estimates of RGC counts were obtained according to a previously described algorithm (Medeiros et al. Arch Ophthalmol. 2012 Sep;130(9):1107-16.). Receiver operating characteristic (ROC) curves were used to evaluate diagnostic accuracy.
Results:
Mean estimated RGC counts were 575,726 ± 218,407 and 1,031,715 ± 184,458 in glaucomatous and healthy eyes, respectively (P<0.001). Estimated RGC counts performed significantly better than SDOCT for discriminating glaucomatous from healthy eyes with ROC curve areas of 0.934 versus 0.876, respectively (P <0.001). However, no improvement was seen when estimates of RGC counts were obtained incorporating both 24-2 and 10-2 data versus estimates obtained using only 24-2 data (ROC curve areas of 0.934 vs 0.933, respectively, P=0.559).
Conclusions:
Estimates of RGC counts obtained from combining structure and function had excellent accuracy for discriminating glaucomatous from healthy subjects. However, no improvement in diagnostic accuracy was seen when 10-2 data was added to the original model.
Keywords: 550 imaging/image analysis: clinical •
758 visual fields •
552 imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound)