Heritability and segregation analysis of osteosarcoma in the Scottish deerhound

Abstract

Osteosarcoma is the most common malignant bone tumor in dogs and, like its human orthologue, is characterized by aggressive local behavior and high metastatic rates. The Scottish deerhound is a breed of dog with a >15% incidence of osteosarcoma and represents an excellent spontaneously occurring large-animal model of the human disease. We modeled the transmission of the osteosarcoma phenotype in a population of over 1000 related deerhounds ascertained as part of a prospective health study. Variance component analysis, segregation analysis, and linear modeling were performed to evaluate heritability, to infer the presumptive transmission model, and to identify covariate effects for this phenotype within the breed, respectively. Based on variance component analysis, heritability (h2) was estimated to be 0.69. Six transmission models were analyzed by segregation analysis; based on Akaike's information criteria, the most parsimonious model was the Mendelian major gene model with dominant expression. Linear modeling identified gender and genotype as significant predictors of disease outcome. Importantly, duration of gonadal hormone exposure, weight, and height at maturity were not significant predictors of outcome. Inheritance of the putative high-risk allele was thus associated with >75% risk of disease occurrence compared to the <5% baseline risk. These results support the hypothesis that a major gene with a dominant effect explains most of the osteosarcoma phenotype within the Scottish deerhound.

Osteosarcoma inheritance in two families of Scottish deerhounds

“…There is an important difference between the results reported for Deerhounds in Cohort T in our study and for a large Deerhound population in an earlier study by Phillips et al [5] and the results reported for Greyhounds, Rottweilers, and Irish Wolfhounds by Karlsson et al [7]. In Deerhounds, our analysis of Cohort T and the analysis by Phillips et al suggest that osteosarcoma risk is governed largely by a single genetic variant that lies on an autosome; i.e., a single genetic variant is rather strongly associated with the osteosarcoma phenotype, and the nature of the association is cause and effect.…”

“…In contrast, our analysis strongly suggests that osteosarcoma risk in Cohort T was governed largely by a single dominant risk factor with high penetrance, as has been reported previously for Deerhounds [5]. …”

“…If a genetic risk factor is dominantly expressed but has a relatively low penetrance, then many individuals that carry the risk factor will not develop the disease, and disease risk will appear to be random rather than heritable [8]. Osteosarcoma risk in our two Deerhound cohorts does not fit that description; instead, risk appears clearly heritable in each cohort, as was previously reported for Deerhounds [5]. Consequently, we ruled out the possibility that osteosarcoma risk was dominant with low penetrance and instead analyzed for the possibility of dominant expression with high (but incomplete) penetrance.…”

“…We believed that such an analysis could be used to test these conclusions from the 2007 study by Phillips et al [5]:Osteosarcoma risk in Scottish Deerhounds is governed primarily by a single genetic variant, which we will call a genetic risk factor in this paper,The genetic risk factor lies on an autosome, andThe genetic risk factor has a dominant mode of inheritance. …”

“…A 2007 study [5] used pedigrees and phenotype information provided by owners and breeders for a population of over 1,200 related Deerhounds to model the inheritance of the osteosarcoma phenotype. The model was then queried using variance component analysis to estimate the heritability, and segregation analysis to infer the presumptive mode of inheritance.…”