Abstract
Purpose:
To find and to validate a multivariable predictive model to detect glaucoma using a combination of Retinal Nerve Fiber Layer (RNFL), Optic Disc, and macular Retinal Ganglion cell (RGC) parameters measured with Cirrus OCT.
Methods:
Five hundred eyes from 500 patients and 187 eyes from 187 patients were respectively included in the study and validation groups. Glaucoma patients were classified into five groups according to the visual field damage (Glaucoma Stage System). The sensitivity/specificity of all parameters was analysed. Receiver operating characteristic curves (ROC) and areas under the ROC (AUC) were compared. A predictive multivariate model from quantitative and qualitative parameters was performed using a combination of the best parameters. This model was compared for the same parameters in all glaucoma groups, including early and advanced glaucoma.
Results:
The best AUCs parameters from Cirrus OCT were: Inferior RNFL, Average RNFL, Vertical Cup/disc Ratio, Minimum RGC and Inferior RGC. The comparison among them did not demonstrate that the RGC parameters were better than RNFL, neither in early nor in advanced glaucoma. The highest AUC was found in the predictive model (0.937; 95% Confidence Interval: 0.911-0.957) which was statistically superior to the other isolated parameters considered. Also, this predictive model had better AUC than the other parameters, both in initial and advanced glaucoma. The validation group displayed similar results than the study group.
Conclusions:
The RGC analyses do not improve the ability of RNFL or Optic Disc analysis to detect glaucoma. The predictive formula using a combination or parameters from Cirrus OCT improve the detection of glaucoma.