The number of retinal ganglion cells (RGCs) in the human retina decreases with age, although estimates of the rate of decline vary widely.1 Behavioral sensitivity, as measured with standard automated perimetry (SAP), also decreases with age.1 It is generally thought that the loss of RGCs contributes to this loss in SAP sensitivity. While current technology does not allow for in-vivo counts of human RGCs, optical coherence tomography (OCT) provides a measure of the thickness of the retinal nerve fiber layer. Although RNFL thickness is dominated by RGC axons, it is also influenced by the presence of blood vessels, glial cells, and the software algorithm determining thickness.2 The stated purpose of this paper by Harwerth et al. (1332) is to test a model of age-related decreases in SAP visual sensitivity and OCT RNFL thickness. For a thorough understanding of this work, one needs to read earlier papers, especially Harwerth and Wheat1 and Harwerth et al.3 However, here are some major assumptions.
There is good agreement between the number of RGCs estimated from SAP sensitivity and the number of RGC axons estimated from OCT RNFL thicknessFirstly, based upon earlier work with monkeys, the authors assume that local human SAP sensitivity can be converted to the number of RGCs via a model.1,3 Secondly, they assume that the decrease in SAP sensitivity with age is due entirely to a loss in RGCs. That is, the decrease in SAP sensitivity with age does not depend on loss of post-retinal cells and/or cognitive factors. Thirdly, they assume, also based upon earlier work,1,4 that they can estimate the number of axons in a region of the optic disc based upon the thickness of the RNFL measurement. Fourthly, because RNFL thickness does not decrease as rapidly with age as does SAP sensitivity, they assume a model of RNFL thickness (Fig. 1 in ref. 1) that contains an age-related increase in the non-neuronal (glial) portion of the RNFL. Finally, as in their past work,1,4 they assume a map that relates optic disc sectors to SAP field regions. They conclude that there is good agreement between the number of RGCs estimated from SAP sensitivity and the number of RGC axons estimated from OCT RNFL thickness and, thus, that the data support their model of RNFL changes with age. One can quibble with the number of parameters involved, the ad-hoc nature of the map used, and their method of fitting the model. However, one must admire this attempt to relate both SAP sensitivity and OCT thickness to the same underlying variable, RGC number. Ultimately, the usefulness of this work will depend upon whether its assumptions re axon thickness, SAP to OCT mapping, as well as glial cell contribution to RNFL thickness with age, can be confirmed with independent (e.g., histological) evidence.