Finding Genes and Pathways for Canine Cancers

Finding Genes and Pathways for Canine Cancers
Dr Ingegerd Elvers, Kerstin Lindblad-Toh lab
Broad Institute, MA, USA, and SciLife Lab, Uppsala University, Uppsala, Sweden
Dogs and humans share many diseases where genetic background and environment
play a role, such as several forms of cancer. The dog population structure, with
distinct breeds derived from small founder populations, makes trait mapping
considerably easier than in humans. Power calculations and proof of principle
studies have shown that 100-300 cases and 100-300 controls can suffice to map risk
factors contributing a 2-5 fold increased risk in dogs (Lindblad-Toh et al, Nature
2005). Dogs have been used successfully to map complex diseases including
systemic lupus erythematosus, obsessive-compulsive disorder and osteosarcoma
(Wilbe et al, Nat Genet 2010; Dodman et al, Mol Psych 2010; Karlsson et al, Genome
Biol 2013).
In our research, we have focused on cancers where the genetic complexity
and/or relative rarity limit studies in humans, such as osteosarcoma, lymphoma,
and hemangiosarcoma. Our research shows several important principles for canine
cancer that may be extrapolated to human cancers:
A. Common risk factors for two diseases: B-cell lymphoma and
hemangiosarcoma are two highly aggressive hematopoietic cancers, both common
in golden retrievers. Using genome-wide association (GWA) we find two unlinked
loci on canine chromosome 5 that explain ~20 % of total risk in the US golden
retriever population to develop both canine B-cell lymphoma and
hemangiosarcoma. We identify several, partially overlapping haplotypes in these
two loci. Using gene expression data from RNA-Seq of tumors, the effect of the
different haplotypes on gene expression can be compared. Genes alternatively
regulated by the risk haplotype implicates local microenvironment and T-cell
activation together with B-cell growth and angiogenesis.
B. Alternative pathways in different breeds with osteosarcoma:
Osteosarcoma is a relatively rare and poorly understood cancer characterized by
early metastasis and high mortality. Global tumor gene expression signature
indistinguishable from tumors from human pediatric patients and, while age of
onset is higher in dogs, the clinical progression is remarkably similar. Both human
and canine osteosarcomas most commonly arise at the ends of the long bones of the
limbs and metastasize readily, usually to the lungs (F. Mueller, B. Fuchs, B. KaserHotz, Comparative biology of human and canine osteosarcoma. Anticancer research
27, 155 (Jan-Feb, 2007)).Canine studies indicate that hormonal factors and size
plays a role, but distinct breed differences indicate genetic factors. GWA studies
using up to 100 cases and controls per breed identified 33 associated loci, with no
overlap between breeds. We identify 6-15 loci per breed explaining 40-80% of the
phenotypic variation. Interestingly, the greyhound top associated locus is near the
p15/p16 genes and further analysis of this locus has shown that rottweilers and
Irish wolfhounds are fixed for the greyhound risk haplotype. Serial luciferase assays
also identifies an enhancer that harbors the best candidate mutation on the risk
haplotype. Pathway analysis of GWAs regions, regions of fixed regions in each breed
together with CGH data identifies both common and breed specific disease
Notably, the top candidate mutations in all three cancers fall outside proteincoding sequence.
In addition to looking for predisposing regions and mutations, we have
performed exome sequencing of paired tumor-normal samples from over a hundred
dogs with B- or T-cell lymphoma. Using three different dog breeds with different
predisposition to B- and T-cell lymphoma, we can look at differences and
similarities in affected genes and pathways. The breeds constitute patient groups
with different genetic backgrounds. Breed comparison shows that the B-cell
lymphoma predisposed breeds have more overlap in somatic mutations compared
to the T-cell lymphoma predisposed breeds. Somatic copy number alterations and
coding mutations partially implicate the same genes.
Of the significantly mutated genes in this comparison of B- and T-cell
lymphoma predisposed breeds, about half overlap have previously been implicated
in human lymphoma or leukemia. We hope that the novel genes could enable new
lymphoma treatment options.
The use of multiple dog breeds and functional assays together with the
extensive genome annotation available on the human genome can help decipher the
regulatory mechanisms underlying disease. We conclude that pet dogs offer an
outstanding opportunity to help dissect complex cancers.