advertisement

Editors Selection IGR 8-3

Structure and function measurements: mfVEP versus HRT II

Comment by Brad Fortune on:

20812 Topographic comparison of the visual function on multifocal visual evoked potentials with optic nerve structure on heidelberg retinal tomography, Punjabi OS; Stamper RL; Bostrom AG et al., Ophthalmology, 2008; 115: 440-446

Find related abstracts

In their recent study, Punjabi et al. (175) sought "to compare the visual field obtained using multifocal visual evoked potential (mfVEP) testing with the optic nerve parameters of Heidelberg retinal tomography (HRT II) in primary open-angle glaucoma (POAG) patients and normal controls and to determine which parameters correlate with visual function." To that end, the investigators collected mfVEP and HRT II data from 67 POAG patients and 14 controls. The average mean deviation (MD) derived from standard automated perimetry was ‐6.2 (± 0.8) in the POAG group, while the average pattern standard deviation (PSD) in that group was 0.9 (± 0.2). As an aside, this represents a rather unusual population of POAG patients for whom diffuse sensitivity loss was evidently more prevalent than localized loss, though the negative value reported for the control group PSD average suggests there might be errors in the data set. If error-free, the effect that this relatively high prevalence of diffuse loss might have had on the study outcomes remains unclear. The mfVEP data were collected using an AccuMap system. Thus there were 58 mfVEP responses, 29 each for the superior and inferior hemifields for each eye, which were then entered into mixed effects regression models to examine their relationship (i.e., potential correlation) to various parameters of optic nerve head (ONH) topography. The investigators report that mfVEP response amplitudes from each hemifield are correlated to various parameters of ONH topography for the corresponding hemi-sector of the ONH. This led the authors to conclude that their study "substantiates the theory that there is a relationship between optic nerve structural abnormality and failure of the neural pathways to transmit information to the cortex." Axiomatic as such a conclusion may be, it evades the striking truth of the results, which is that the correlations were all quite poor (despite being 'statistically significant'). Some of those relationships should make intuitive sense a priori (i.e., to fit the axiom well). For example, we accept the notion that cortical signals (like mfVEP responses) are driven by retinal ganglion cell inputs (via the LGn) and should therefore be reduced when those inputs are removed. Hence, in the most simplified scenario, mfVEP response amplitudes should be related to ONH parameters such as rim-to-disc area ratio, or inversely with cup-to-disc area ratio. Yet the data reported by Punjabi et al. reveal that correlation coefficients ranged from only 0.05 to 0.4. Moreover, the authors' lists of regression results reveal a profound difference between the strength of these relationships in one hemifield as compared with the other. That is, the relationships between the superior hemi-sector of the ONH and the inferior hemifield of the mfVEP were markedly worse than those linking the inferior ONH with the mfVEP superior hemifield.

Given these results, it would have been preferred (one might even argue required) that the authors discuss the possible reasons underlying the observed (poor) relationships and to speculate about the dramatic differences between inferior and superior hemifields, rather than to have discussed various aspects of 'good correlations'. To start, there are numerous reasons to anticipate that the simplistic scenario described above is inadequate, and thus, that relationships between mfVEP amplitude and ONH topography would be weaker than one-to-one. These include the complexities of architectural changes in the glaucomatous ONH (as compared with, for example, non-glaucomatous optic atrophy), the methodological derivation of mfVEP response 'amplitudes' as applied by the AccuMap system, the spatially incongruous samples being compared (central 50 degrees assessed by mfVEP versus whole retina being assessed by ONH topography), and more. Thus, it is not clear how the authors' ultimate conclusion that, "mfVEP and HRT showed correlation between visual function and optic nerve struture" really teaches us much more than we might have concluded a priori. In fact, a rigorous evaluation of the results might have taught us better why our a-priori assumptions had failed. In summary, the conclusions of this paper are not well supported by the reported results, the study design and its results are not particularly original, the paper appears to contain numerous errors, and the discussion appears to evade some admittedly difficult truths.

• ← Previous
• Next →