Most previous studies examining the diagnostic ability of imaging devices in glaucoma have used eyes with visual field defects as cases and eyes with no suspicious findings of glaucoma as controls. In such circumstances, detecting glaucoma is relatively straightforward and it is not surprising that imaging devices perform well. Arguably, a more clinically relevant measure is how well an imaging device is able to differentiate eyes with early or preperimetric disease from those with suspicious looking discs, however few studies have addressed this question.
Rao et al. report the results of a study in which they examined the ability of spectral domain optical coherence tomography to differentiate eyes with preperimetric glaucoma from those with large physiological cupping. They examined optic nerve head, circumpapillary and macular thickness parameters.
The best performing parameters had area under the receiver operating characteristic curves (AUCs) of around 0.7, with sensitivities at a specificity of 95% of around 25%. AUCs for rim area, average circumpapillary RNFL thickness, and average macular ganglion cell complex thickness were 0.72, 0.75 and 0.72 respectively. As expected, the diagnostic ability of OCT was lower than previously reported for the detection of perimetric glaucoma. The AUCs were also lower than other diagnostic accuracy studies in preperimetric glaucoma likely because the other studies used controls with no suspicious findings of glaucoma.
Differentiating a suspicious optic disc from preperimetric glaucoma may be challenging without follow up examinations to determine if there are progressive changes
The study is important as it highlights the need to use appropriate cases and controls when evaluating the diagnostic ability of a test. The results should however be interpreted with some caution as the study included a relatively small sample size (only 34 glaucomatous eyes) and the diagnosis of preperimetric glaucoma was based on examination on a single visit. Differentiating a suspicious optic disc from preperimetric glaucoma may be challenging without follow up examinations to determine if there are progressive changes and so it is possible that misclassifications occurred. In fact, another recent study, which the authors discuss, has shown RNFL thickness to perform well even in differentiating glaucoma suspects from preperimetric glaucoma detected through progressive changes on optic disc stereophotographs (AUC = 0.89, sensitivity of 70% at 95% specificity). Further studies should provide further information.