advertisement

Editors Selection IGR 7-3

Comments

Comment by William Swanson on:

48060 Retinal Ganglion Cell Layer Thickness and Local Visual Field Sensitivity in Glaucoma, Raza AS; Cho J; De Moraes CGV et al., Archives of Ophthalmology, 2011; 129: 1529-1536

Find related abstracts

The study by Reza and colleagues is well-conceived, carefully performed and appropriately analyzed and discussed. Identified structure/ function associations were compared with an existing model that has explicit assumptions and the identified associations were found to deviate somewhat from the model.

Given that several studies have demonstrated that measurement of the macular 'RGC complex' (retinal nerve fiber layer [RNFL] + retinal ganglion cell [RGC] + inner plexiform layer [IPL]) thickness has at least as good sensitivity to identify early glaucoma as peripapillary RNFL thickness, the macula is a logical target for structure/function studies. Furthermore, the advantage of the RGC+IPL thickness as the structural measure is that the location of the structural measurement is spatially closer to the measurement of function, compared with the RNFL thickness. Thus, one may expect stronger associations. In this study, the authors associate RGC+IPL thickness with visual function, as measured by standard automated perimetry (SAP) with Goldmann size III targets and 10-2 VF pattern on a point-by-point basis. The RGC+IPL layer were automatically segmented in spectral- domain OCT images and then manually adjusted where segmentation errors were suspected. The lateral displacement of RGCs from the foveal centre was accounted for using previously-published histological data.

Data were analyzed by eccentricity; pooling all data would assume that the structure/function associations are the same for all eccentricities. This approach is wise, because previous work has suggested that the eccentricity of measurement affects the structure/function association.^1,2 The likely reason for this eccentricity effect is 'probability summation' where RGC density is high.

The plots in Figure 6 are worthy of further comment. The general distribution data points appears to concord reasonably well with authors' 'simple linear model'. However, as noted by the authors, most of the data points from the patients lie to the right of the model line, so that retinal sensitivity is higher than would be expected if there is a simple linear decline in retinal sensitivity (in linear units) from mean normal (as RGCs are lost during the glaucomatous process). This observation may be explained by spatial summation effects, as suggested by the authors. In the central retina (within about 15º of the fovea), there is not a simple linear relationship between the area illuminated by the Goldmann size III SAP target (and, therefore, the number of RGCs stimulated) and the retinal sensitivity: in the central retina, retinal sensitivity changes by less than a unit for each unit change in area (RGC number).^1,3 This is seen in the plots. This hypothesis (that spatial summation accounts for the deviation of the observed relationship from the model) may be tested directly. If smaller perimetric stimuli that are within the area of complete (linear) summation are employed for a similar study, then closer concordance with the 'simple linear model' would be expected. Such an approach may also reduced the scatter in the data and allow retinal function to be predicted from RGC+IPL thickness.

Potential biases in the study design: selecting subjects in a way that avoids bias for structure/function studies is difficult. If selection is on the basis of normal structure for one group and abnormal structure in the other, then this may artificially polarize the structure measurements and distort the true structure/function association (in other words, it prevents the selection of healthy subjects with anatomy that looks abnormal, but in reality may be associated with normal function, and vice versa in patients). In this study, healthy subjects had 'a normal optic disc appearance' and patients had 'glaucomatous optic neuropathy'. The requirement for patients to have visual field loss (in the context of a glaucomatous neuropathy) may further polarize the data and make the structure/function associations artificially strong. The effect of this potential bias is likely to be small, because disc appearance may not be strongly correlated with RGC+IPL thickness. However, the potential bias may be mitigated by selecting subjects on the basis of risk for being normal or abnormal (for instance, based on intraocular pressure level) and then examining the structure/function association across the range of values for each that happen to be present.

• ← Previous
• Next →