Statistics from Altmetric.com
The primate visual system is designed to provide maximum performance under a variety of conditions. Much of the enhanced sensitivity of the rod system comes from post-receptor neuronal organisation that pools impulses from thousands of receptors. In the retinal periphery multiple rods converge onto a single ganglion cell resulting in poor acuity but high sensitivity. At the fovea a one to one relation between cone and ganglion cell produces fine discrimination, colour perception and, by fusing the images of both eyes, binocular vision.
The “receptive field” for an individual ganglion cell is determined by the spatial arrangement of the connecting receptor fibres. These are designed to enhance perception of the stimulus and is achieved by an antagonistic arrangement of a concentric centre surround response where the surround is opposite in polarity to the centre.1
Ganglion cells have been subdivided according to their temporal response properties. Primate ganglion cells have been divided intotonic and phasictypes.2 Phasic cells respond transiently to changes in stimulus, while tonic ones provide a more sustained response. Phasic ganglion cells have both rod and cone input, make up about 10% of the total, and have large overlapping receptive fields. They subserve motion and contrast function and project via large diameter fibres to the magnocellular layers of the lateral geniculate body.3
Tonic ganglion cells by contrast are small, with small receptive fields and slower conducting axons. They comprise 80% of all ganglion cells with maximum representation for the foveal cones. They subserve colour and discriminatory function and project to the parvocellular layers of the lateral geniculate body.3 The axon size would appear to differ according to the cone wavelength perceived, with short wavelength cones having the largest diameter fibres.4These cells connecting to the magnocellular and parvocellular layers of the lateral geniculate body are also known as parasol cells (feeding to the magno system, and therefore “M” cells) and midget cells (feeding to the parvo system, “P” cells).
The idea of separate visual function based on different receptors and different post-receptor organisation has led to the concept of “parallel visual processing”. This concept has proved useful for identifying different aspects of visual loss in ocular disease.
The search for specific tests which made use of this psychophysical separation was greatly stimulated by the discovery that glaucoma was associated with selective ganglion cell death. Histological examination of eye with “early glaucoma” revealed that substantial numbers of ganglion cells were lost before defects appeared with white on white perimetry.5 In the experimental monkey model large ganglion cells were found to be more vulnerable, although the magnitude of this selective loss varied.6 Human necropsy material showed relatively greater loss of magnocellular cells in glaucoma.7 These pieces of evidence pointing to selective cell death have led to the development of tests of large fibre ganglion cells including contrast sensitivity,8 motion detection,9 and, perhaps, frequency doubling.10
Recently, the theory that selective cell death first affects large fibre ganglion cells has been questioned. Sample and associates reported on the results of testing glaucoma patients with short wavelength perimetry and motion automated perimetry, and found that both tests successfully diagnosed glaucomatous damage.11They suggested that in glaucoma the damage that occurs may affect either the magnocellular or parvocellular system, or that there may be individual differences between eyes that decide which system is damaged first. The numeric difference between the two types of ganglion cells could mean a disproportionate effect on visual function if equal numbers of M and P cells were lost in disease. Willis and Anderson noted that the age related rate of decline was faster in the glaucomatous eye than the normal eye.12 Morgan has suggested that the appearance of differential cell death could be an artefact due to postmortem shrinkage producing the impression of large cell loss.13
In the March issue of the BJO, Morgan and coworkers contributed further to this debate.14 They induced ocular hypertension in six primate eyes and then examined the retinal ganglion cell populations after retrograde labelling with horseradish peroxidase. They did not find a significant reduction in the proportion of parasol to midget cells. They did find an overall reduction in mean size for both the surviving M and P cells suggesting cell compromise before cell death. They pointed out that the use of a tracer to study the ganglion cell changes induced by ocular hypertension was not subject to the potential for artefact inherent in the methodology of earlier studies. They noted that another similar primate study15 also found a reduction in cell size before cell loss (although the exact significance of this finding was open to interpretation).
Where does this debate leave the clinician who is looking to detect early glaucomatous visual loss? Selective cell death may still occur in early glaucoma, although alternative explanations are possible. However, this doubt as to mechanism does not remove the validity of selective psychophysical testing. Whether the tests target the magno or the parvo system they are still likely to provide earlier diagnosis than white on white perimetry and as such should continue to be introduced into clinical practice.
Morgan and co-workers are to be thanked for reminding us that in biological systems all is rarely black and white, and that even certainties in medicine are worth re-examination. Their work has caused us to rethink the process by which ganglion cells are damaged in hypertensive eyes. In addition, by identifying premortem changes in the ganglion cell size that suggested compromise, they may have also provided a marker for worthwhile studies on neuroprotection in the future.
Note added at proof stage: Kerrigan-Baumrind et al(Invest Ophthalmol Vis Sci2000;41:741–8) looked at the ganglion cell loss for the entire retina in 17 eyes of 13 people with well documented glaucoma. They looked for retinal ganglion cell loss associated with visual field defects. They also found evidence to corroborate their previous findings that ganglion cells with larger axons preferentially die in glaucoma.