Skip to main content
snıff
Canine Mendelian disease record

Collie Eye Anomaly (CEA)

Collie Eye Anomaly (CEA). Autosomal recessive. Observed in 44 of 266 breeds tested in the Sniff Atlas, with measured carrier frequencies drawn from 242,663 dogs (Donner 2023). Per-dog phenotype outcome depends on penetrance, modifiers, and environment; the carrier frequencies below describe variant prevalence, not disease incidence.

OMIA identifier
OMIA:000218-9615
Autosomal recessive
Source dataset
Sniff Atlas v1.0.1 / DOI
About this disease

From OMIA's curated record

Documented in OMIA (Online Mendelian Inheritance in Animals). This describes the disease as recorded in the published literature, not a prediction for any individual dog. As of 2026-06-03.

Summary

Collie eye anomaly (CEA) is a complex developmental defect of the posterior segment of the eye characterized by choroid hypoplasia. Malformations are present at birth and may include inadequate development of the choroid (choroidal hypoplasia), defects of the choroid, sclera, and/or optic nerve (coloboma/staphyloma), and complete retinal detachment (with or without hemorrhage). Mildly affected animals will have no detectable vision deficit. (Genetics Committee of the American College of Veterinary Opthalmologists, 2021).

Clinical features

CEA presents as varying levels of visual impairment, from no visual impairment to complete blindness. As summarised by Fredholm et al. (2016), "Collie eye anomaly (CEA) is a congenital, inherited ocular disorder which is widespread in herding breeds. Clinically, the two major lesions associated with CEA are choroidal hypoplasia (CH) and coloboma, and both lesions are diagnosed based on ophthalmological examination". Choroidal hypoplasia is noted as a defect in the ocular fundus temporal to the optic nerve, which can be detected during ophthalmoscopic exam. Some affected dogs have tortuous retinal vessels and multiple retinal folds. More severely affected dogs can develop retinal detachments leading to blindness. These dogs can also develop subretinal and preretinal neovascularization and intraocular hemorrhage (Lowe et al., 2003).

Molecular genetics

By a neat bit of detective work that makes use of between-breed variation in linkage disequilibrium in tracking down mutations that are common to several breeds, Parker et al. (2007) discovered that a likely causal variant for Collie Eye Anomaly in four breeds (Collie, Border Collie, Australian Shepherd, and Shetland Sheepdog) is a deletion of 7.8kb in the NHEJ1 gene: 37:g.28,697,542–28,705,340del. A warning that the causality of this mutant may not be as straightforward as initially thought, was provided by Fredholm et al. (2016) who reported that " the deletion in NHEJ1 is not predictive for CH [choroidal hypoplasia] in the Danish Rough Collie population, whereas the clinical and genetic diagnosis is in accordance with each other in the Shetland Sheepdog population. Based on these results, it can be concluded that the intronic deletion in NHEJ1 is not the causative mutation but, rather, a marker linked to the locus underlying the trait in some populations but linked to both the wild-type and CH-causing locus in most dogs in the Danish Rough Collie population." Further doubt on the causality of the deletion was raised by the results of Brown et al. (2018) in Nova Scotia Duck Tolling Retrievers. Grosås et al. (2018) reported "A good compliance between the clinical diagnosis and the genetic test results was found" in 103 puppies from a random sample of 16 unrelated Norwegian Border Collie litters. The authors reported that one puppy was clinically affected and one puppy was homozygous for the deletion variant, but intriguingly, readers were not told whether this was the same puppy. Also, even though the parents were genotyped, readers were not told whether the parents of the affected puppy were both heterozygotes.

Pathology

The choroid is the vascular supply to the retina. The primary defect in this disorder is choroidal hypoplasia. Secondary lesions include colobomas around and in the optic nerve head or the adjacent fundus. Retinal detachment is also a secondary lesion.

Prevalence

Prevalence is estimated to be 70-97% in rough and smooth collies in the USA and Great Britain, and 68% in rough collies in Sweden. In Border collies, prevalence is estimated at 2-3% in the USA and Great Britain (Lowe et al., 2003). Mizukami et al. (2016) reported the frequency of the NHEJ1 deletion allele as 0.143 in 500 Border collies in Japan. Grosås et al. (2018) reported the frequency of the deletion variant as 6.3% in Norwegian Border Collies. In a sample of 465 Italian dogs from five breeds, Marelli et al. (2022) reported that the frequency of carriers (healthy heterozygotes) for the deletion variant (OMIA variant 632) was 50% in Shetland Sheepdog, 45% in Nova Scotia Duck Tolling Retriever, 30% in Border Collie, 17% in Australian Shepherd, and 13% in Rough Collies. Donner et al. (2018) reported that the variant (OMIA variant 632) was identified in additional breeds and reported carrier frequencies: Australian Kelpie: 7.5% (6/80), Chinook: 0.7% (1/151), Jack Russell Terrier: 3.8% (4/105), Parson Russell Terrier: 1.2% (3/243), Koolie: 41% (16/39), Lapponian Herder: 5% (1/20), Tamaskan Dog: 3.5% (3/86) Clark et al. (2023) "utilized a large set of genotypes from dogs tested for the progressive rod-cone degeneration–progressive retinal atrophy (prcd-PRA) G>A missense PRCD variant (n = 86,667) and the collie eye anomaly (CEA)-associated NHEJ1 deletion (n = 33,834) ... . ... the NHEJ1 deletion may actually be linked to the true causal mutation, as the deletion does not segregate with coloboma or CH in some cases ... . Regression modeling showed time progression to significantly affect the odds of a dog being homozygous or heterozygous for either [variant], as do variables including breed and breed popularity. This study shows that genetic testing informed breeding decisions to produce fewer affected dogs. However, the presence of dogs homozygous for the disease variant, especially for prcd-PRA, was still observed fourteen years after test availability, potentially due to crosses of unknown carriers."

Inheritance

Colobomas only occur in dogs of four breeds (Collie, Border Collie, Australian Shepherd, and Shetland Sheepdog) that are homozygous for the likely causal mutation (Parker et al., 2007). Some affected dogs in these breeds have been observed to “go normal” later, most of which were homozygous for the likely causal mutation, but a small minority were heterozygous. This suggests other loci may have an effect on CEA expression, i.e. that the disorder is multifactorial.

Genetic testing

For limitations on the utility of testing for the published likely causal variant, see the results of Fredholm et al. (2016) and Brown et al. (2018).

Human analog

OMIA links this condition to the human gene record in OMIM (Mendelian Inheritance in Man), the place to read across to the deeper human literature for the same biology.

Gene · OMIM 611290

Source: OMIA (Nicholas, Tammen & the Sydney Informatics Hub), entry OMIA:000218-9615, doi:10.25910/2AMR-PV70 (CC-BY 4.0).

The evidence

Published references

The peer-reviewed papers behind this disease, curated by OMIA. Starred entries are OMIA-designated landmark papers. Showing 6 of 58.

References curated by OMIA (Nicholas, Tammen & the Sydney Informatics Hub), doi:10.25910/2AMR-PV70 (CC-BY 4.0). Full list at the OMIA entry.

Predict a litter

Set each parent's status for Collie Eye Anomaly (CEA) and see the odds for their puppies. Single recessive variant, exact Mendelian math.

Parent A
Parent B
NNClear
NmCarrier
NmCarrier
mmAffected
Clear25%
Carrier50%
Affected25%

These are the genetic odds for one known variant, not a promise: a real litter varies around them, and penetrance or other genes can change whether the condition ever appears. Use it to avoid pairing two carriers and to keep a line healthy, not to engineer a dog. Inheritance mode per OMIA.
Your breed

See what Collie Eye Anomaly (CEA) looks like in your dog's breed.
Carrier frequency by breed

Top 25 well-sampled breeds (n ≥ 50)

Maximum per breed across variants in the Donner 2023 cohort, with . The list below is split into well-sampled breeds (n ≥ 50 tested) and small-sample breeds (n < 50, where the Wilson CI typically spans more than 20 percentage points and frequencies should not be compared directly to the well-sampled entries). Frequencies are population-level, not per-litter or per-line.

0%38%75%
Collie72.2% · n 1,207
Boykin Spaniel22.4% · n 154
Shetland Sheepdog12.5% · n 945
Border Collie11.1% · n 6,714
Nova Scotia Duck Tolling Retriever7.1% · n 63
Australian Shepherd5.0% · n 2,296
Anatolian Shepherd Dog3.8% · n 66
Australian Kelpie3.4% · n 104
Miniature American Shepherd0.58% · n 1,476
Australian Cattle Dog0.56% · n 982
Parson Russell Terrier0.55% · n 181
Chow Chow0.54% · n 643
Great Pyrenees0.50% · n 1,985
German Shepherd0.40% · n 15,648
Saint Bernard0.35% · n 721
n = 33,185 dogs · Donner et al. 2023 carrier-screening cohort · Sniff Atlas
Each bar is one well-sampled breed; the whisker is its Wilson 95% CI, and fainter bars have wider intervals. Frequencies are population-level, not per-litter. Carrier status for Collie Eye Anomaly (CEA) is measured; phenotype outcome depends on penetrance and modifiers.
▸ Full table with Wilson 95% confidence intervals
Breed Carrier frequency n tested
Collie 72.2% 1,207
Boykin Spaniel 22.4% 154
Shetland Sheepdog 12.5% 945
Border Collie 11.1% 6,714
Nova Scotia Duck Tolling Retriever 7.1% 63
Australian Shepherd 5.0% 2,296
Anatolian Shepherd Dog 3.8% 66
Australian Kelpie 3.4% 104
Miniature American Shepherd 0.58% 1,476
Australian Cattle Dog 0.56% 982
Parson Russell Terrier 0.55% 181
Chow Chow 0.54% 643
Great Pyrenees 0.50% 1,985
German Shepherd 0.40% 15,648
Saint Bernard 0.35% 721
Treeing Walker Coonhound 0.30% 336
Whippet 0.28% 177
Dobermann Pinscher 0.27% 2,219
Schnauzer Giant 0.22% 230
American Staffordshire Terrier 0.21% 42,793
Siberian Husky 0.16% 9,034
Old English Sheepdog 0.12% 423
Labrador Retriever <0.1% 16,856
American Foxhound <0.1% 574
Chihuahua <0.1% 4,273

Top 25 of 36 well-sampled breeds with at least one observed carrier shown.

▸ Also observed in 8 small-sample breeds (n < 50)

Frequencies in this section are statistical estimates with wide Wilson 95% confidence intervals (typically >20 percentage points). Treat these as "carriers observed but the true population frequency is not yet measurable" rather than as comparable to the well-sampled entries above.

Breed Estimate n tested
Bulgarian Shepherd 50.0% 1
Hokkaido Inu 50.0% 4
Mcnab 17.9% 28
Lancashire Heeler 11.8% 17
Lacy Dog 9.4% 32
Bearded Collie 8.3% 12
Maremma Sheepdog 2.7% 37
Lapponian Herder 1.9% 26

222 additional breeds in the Donner 2023 cohort were tested but showed no carriers.

Scope of this record

Scope

This record carries the breed-level carrier frequencies from the Donner 2023 cohort. Penetrance data (the fraction of at-risk dogs that develop the phenotype) is not yet quantified for this disease in the Sniff Atlas v1.0.1. The OMIA entry is the authoritative reference for the clinical phenotype, inheritance pattern, and gene assignment.

Predicted disease relevance at the per-dog level is UNPROVEN. The carrier frequency is measured; phenotype outcome depends on penetrance, environment, and modifier loci. Consult a veterinarian for clinical interpretation.

How to cite this record

Citations

If you use this record in published work, cite the Sniff Atlas (the published dataset that carries the breed-level carrier frequencies) and the upstream sources:

  • Sniff Atlas v1.0.1 for the per-breed carrier frequencies:

    Gehring, M. (2026). Sniff Atlas v1.0.1. Zenodo. https://doi.org/10.5281/zenodo.20566358. CC-BY 4.0.

  • OMIA for the disease definition, inheritance, and gene assignment:

    Nicholas, F. W., & Tammen, I. (2024). OMIA. Sydney Informatics Hub, The University of Sydney. https://doi.org/10.25910/2AMR-PV70. Entry: OMIA:000218-9615.

  • Donner et al. 2023 for the breed × variant carrier-frequency cohort:

    Donner, J., Freyer, J., Davison, S., Anderson, H., Blades, M., Honkanen, L., et al. (2023). Genetic prevalence and clinical relevance of canine Mendelian disease variants in over one million dogs. PLOS Genetics, 19(2), e1010651. https://doi.org/10.1371/journal.pgen.1010651.

Full citation formats (BibTeX, RIS, CITATION.cff) at sniff.world/cite.

Related

Related

Last updated
Sources: Sniff Atlas v1.0.1 · OMIA OMIA:000218-9615 · Donner et al. 2023