Get access

Optimal discrimination of an Abyssinian cat recessive retinal degeneration: a short electroretinogram protocol is more efficient than a long one


Correspondence: Dr Vaegan, 2/12 Birriga Road, Bellevue Hill, NSW 2023, Australia. Email:


Aim: To determine the diagnostic efficiency of scores in a long protocol electroretinogram (ERG) for Abyssinian cat slow recessive rod/cone dystrophy.

Methods: Kittens (n = 22) were bred from homozygous, affected and heterozygous normals. Ophthalmoscopy was regularly performed and disease signs noted. Cats (age ≥ 8 months, 40 sessions, 1–3 repeats) were dark-adapted overnight, anaesthetized and simultaneous binocular ERG recorded using a long protocol. Conventional a- and b-amplitudes and peak implicit times were measured and b/a ratios calculated, initially only for ERG to the maximum photopic and scotopic stimulus. Principal components factor analysis was applied to various subsets of these scores plus age at testing.

Results: Six cats with ophthalmoscopic change were classed as affected. Three cats, one tested three times, were suspect. The rest were considered normal. The first analysis, of 80 eyes and 37 parameters, showed that the first factor was the only effective one. Using it, the groups overlapped 5%, scotopic amplitudes and b/a ratios loaded higher than peak times, the eyes were very similar, and age and photopic b/a ratios loaded poorly. The groups were discriminable with all the data. A second analysis, with eyes averaged and the 20 measures loading over 0.5 on the first factor, showed better group separation on factor 1 alone. An iterative search with varying data sets found that factor 1 was optimal, with eight ERG measures to the three brightest scotopic and one brightest photopic response. It produced a large absolute separation and classified the suspects consistently. With b/a ratios on these four ERG also included, 12 parameters gave better separation.

Conclusion: Twelve scores on four ERG separate affected from normal cats with a wide gap and consistently classify suspects. It may work for earlier ages. Additional data probably adds noise. This combination of optimal scores needs confirmation in new data.