advertisement

---

Editors Selection IGR 7-2

Examination methods: Perimetric sensitivity

Comment by David Garway-Heath on:

13798 Visual field defects and retinal ganglion cell losses in patients with glaucoma, Harwerth RS; Quigley HA, Archives of Ophthalmology, 2006; 124: 853-859

Find related abstracts

---

The question of the relationship between structure and function is one that exercises the minds of glaucoma researchers and clinicians alike. Harwerth and colleagues have previously reported a model to describe the relationship between histological retinal ganglion cell (RGC) counts and perimetric sensitivity in a primate model of glaucoma. The model accounts for the effects of retinal eccentricity and may be used to predict RGC numbers from visual field thresholds. In a new report, Harwerth and Quigley (325) evaluate the same model in previously reported data of human visual fields and post-mortem RGC counts. The great strength of the study is that the model was applied unchanged and appears to predict well the general pattern of loss of function with loss of RGCs in human glaucoma.

Although the general pattern was well-predicted, the absolute predictions were less precise. The average difference (mean absolute deviation) between the predicted and histological RGC counts was &plm; 2.59 dB. This means that the average error was large - to predict either +80% or -45% of the histological count. The way in which the data are presented makes it difficult to establish whether or not there may be systematic deviations of the predictions at certain parts of the range (for instance, at high or low RGC density).

The prediction of RGC density from perimetric sensitivity in humans was remarkably good

However, the results need to be interpreted in the context of the many potential sources of variability that may result in imprecision in the predictions. Some of these are acknowledged by the authors. They include time between histology and last visual field (within one year for 11 eyes and within two years for a further five eyes, mean 1.2 years), threshold variability (which may be very large in damaged regions of the field), possible ganglion cell dysfunction (ie present, but poorly functioning RGCs), the lateral displacement of RGCs compared to their receptive field at the central test locations, possible inclusion of amacrine cells (which vary in density with eccentricity) in the RGC counts, the area of histological sampling for each location being considerably larger than that sampled in perimetry, and the considerable variation in the estimates of histological RGC density at each location.

On balance, given the numerous potential sources of error in studies of this sort, the prediction of RGC density from perimetric sensitivity in humans (using a model derived from experimental glaucoma) was remarkably good. Further work is needed to understand the model in the context of the psychophysical mechanisms underlying clinical perimetry and the changes that may occur in glaucoma.

---

• ← Previous
• Next →

---