PLINK breed data showing SLAMF1 is associated with canine atopic dermatitis

Canine atopic dermatitis (CAD) is a common inflammatory skin condition in dogs. It is a life-long problem which poses a significant welfare issue due to the chronic skin discomfort and pruritus (itch) experienced. Excessive scratching, licking and chewing cause self-trauma to the skin and increased risk of secondary infections. Several breeds are predisposed including the Labrador Retriever, Boxer and French Bulldog, suggesting a genetic component to the condition. Our access to a sizeable population of dogs genotyped on a medium density single-nucleotide polymorphism (SNP) array along with linked clinical record information allows for large-scale and highly powered genome-wide association study (GWAS) analysis. In this study, over 28 thousand dogs were used to look for genetic changes associated with CAD. We identified a statistically significant signal on canine chromosome 38, with a particularly strong association in the French Bulldog breed. Genome resequencing revealed a provocative splice donor variant in SLAMF1 (Signaling lymphocytic activation molecule 1), an immune system associated gene. Analysis of further genome sequences and RNA samples from the Mars Petcare Biobank confirmed the SLAMF1 variant as a strong candidate for causing an increased risk of atopic dermatitis.   

Methods
Samples

DNA samples were collected via commercial testing of Wisdom Panel™ Premium, Wisdom Panel™ Essential, Wisdom Panel™ Health and Optimal Selection™ retail products, and genetic testing performed as a part of Optimal Wellness Plans® for puppies, through Banfield Pet Hospital branches (Vancouver, WA, USA) and as part of the MARS PETCARE BIOBANK™. Samples were either collected through non-invasive cheek swabbing by dog owners or veterinary professionals or through blood sampling by a veterinary professional at a Banfield Pet Hospital in line with regulations governing diagnostic testing. Consent for use of DNA data in research was obtained through the client’s agreement with terms and conditions of DNA testing through Wisdom Panel. Analysis and sequencing of cDNA were performed using samples collected from dogs participating in the Mars Petcare Biobank. All samples originated from the United States or Mexico. 

Analysis

An initial GWAS was performed using 14,378 single breed cases and 14,633 breed matched controls to tightly control against potential population stratification. Within breed GWAS was performed where at least 200 cases and controls were available for a given breed.

Inclusion criteria 

CAD cases and controls were categorized based on labelling provided by general veterinary practitioners in the Banfield EMR and through criteria set by a board-certified veterinary dermatologist. Given the retrospective nature of the data, this is a diagnosis of atopic dermatitis in the broad sense which may include dogs with food allergic dermatitis. 

CAD cases were in the age range of 0.3 to 20 years and must have received at least three months of ectoparasite control. CAD cases had also been diagnosed with recurrent (more than once) clinical signs of atopic dermatitis including: Dermatitis, Atopic Allergic; Otitis; Pododermatitis; Pruritus; Pyoderma or Malassezia. In addition, CAD cases could also have been prescribed recurrent systemic anti-inflammatories, antihistamines or medicated ear drops. 

Controls were in the age range of 3 to 20 years and must not have been diagnosed with of any of the following recurrent symptoms: Acne; Alopecia Undetermined; Dermatitis (of any variety), Folliculitis; Lichenification; Malassezia; Otitis (of any variety); Paronychia; Pododermatitis; Pruritus; Pyoderma (of any variety).  Controls also must not have been prescribed recurrent systemic or topical anti-inflammatories, antihistamines or medicated ear drops.  

For breed assignment single breed dogs were defined as dogs with greater or equal to 80% single breed ancestry as defined by the Wisdom Panel breed classification algorithm (BCSYS).

Genotype Analysis 

Genome-wide association study analysis was performed using a linear mixed-model approach in the software package GEMMA v0.98.5 including a centered relatedness matrix. PLINK v1.9 was used to apply the following quality control: Samples with an overall genotyping success rate of lower than 95% across all tested SNPs were also filtered out; then, variants below 1% minor allele frequency or lower than 95% genotyping success rate were filtered out.