Diagnostic utility of combined retinal ganglion cell count estimates in Japanese glaucoma patients

Mari Sakamoto, Sotaro Mori, Kaori Ueda, Azusa Akashi, Yukako Inoue, Takuji Kurimoto, Akiyasu Kanamori, Yuko Yamada, Makoto Nakamura

To assess the combined estimate of retinal ganglion cell (RGC) count developed by Medeiros et al. as a tool for diagnosis of glaucoma in Japanese patients. Cross-sectional study. Thirty-one eyes of 19 healthy controls and 106 eyes of 70 glaucoma patients underwent standard automated perimetry (SAP) and three types of spectral domain optical coherence tomography (SD-OCT) imaging using the Cirrus, RTVue, and 3D-OCT instruments. RGC counts derived from SAP and SD-OCT data were estimated using the Harwerth model (SAPrgc and OCTrgc, respectively), from which the combined RGC count estimates (CRGC) were calculated using the formula developed by Medeiros et al. Receiver operating characteristic curve (ROC) analyses were conducted for mean deviation (MD), retinal nerve fiber layer thickness (RNFLT), and CRGC. The mean OCTrgc derived from the Cirrus, RTVue, and 3D-OCT instruments were 1150, 1245, and 1316 (× 1000 cells), respectively, for the control group and 463, 519, and 516 (× 1000 cells), respectively, for the patient group. SAPrgc of the controls' group was 1526 and the patients' group, 731 (× 1000 cells), and were consistently greater than OCTrgc in both groups (a generalized estimating equation model, p < 0.001). Partial area under the curve (pAUC) of MD was 0.178, and that of RNFLT and CRGC for the three OCT instruments were 0.185, 0.18, 0.189 and 0.196, 0.196, 0.197, respectively. CRGC had larger pAUC than MD, whereas there was no or marginal difference in pAUC between CRGC and cpRNFLT, irrespective of OCT device used or glaucoma severity. CRGC proved well suited to discriminate glaucoma patients from controls. However, its clinical utility did not seem to overwhelm isolated structural measures in the tested Japanese patients.