We read with interest the article by Nouri-Mahdavi and associates evaluating the ganglion cell/inner plexiform layer (GC/IPL) measurements in early glaucoma. They found no significant improvement in performance of GC/IPL parameters with both regional and global GC/IPL performing equally well for glaucoma detection. This is very encouraging and informative where other algorithms may be used for improving glaucoma detection. Such information regarding clinical application of imaging in glaucoma diagnosis is indeed useful for any clinician. Yet there are a few concerns that we would like to communicate.

The authors have stated that they included eyes with reproducible field defects as glaucoma regardless of intraocular pressure or disc appearance. This does not apply well to clinical diagnosis of glaucoma, where field defects should correspond to optic nerve head changes of glaucoma. Though neurologic or retinal lesions were excluded by the authors, glaucomatous changes of the disc is primal for defining glaucoma since field defects could be simulated by other causes in this study, such as high myopia (cutoff for refractive errors was <8 diopters in this study).

The authors have reported that a lower GC/IPL predicted a worse mean deviation in control subjects. The implication for such a finding in a normal eye (keeping in mind the inclusion criteria cited above) is not easily understandable. A careful look at the plots (Figure 1) shows that many glaucoma eyes had average retinal nerve fiber layer thickness (RNFLT) and GC/IPL measurements similar to controls. This again raises the question of whether this could have been because of visual fields being the inclusion criteria for glaucoma rather than disc changes. If the diagnosis of glaucoma could have included structural changes rather than functional damage, this discrepancy could have possibly not arisen.

It is interesting that the authors found that adding the 2 best GC/IPL to RNFL parameters improved the performance of spectral-domain optical coherence tomography (pseudo-R ² improving from 65% to 75%); a quick re-look into Table 2 shows that glaucoma eyes had a wide confidence interval, implying that these may not be a true glaucoma cohort.

Although imaging techniques are changing trends of glaucoma diagnosis, it is important to remember that clinical diagnosis of glaucoma still has to be with respect to disc changes, while visual fields or imaging only complement our findings rather than act as a substitute. The results of this study showing that the average RNFL or inferior RNFLT actually performs equally well with GC/IPL measurements, a feature of newer versions, only reiterates the fact that newer imaging techniques can only complement our findings and relying entirely on them would actually be fallacious for good glaucoma practice.