Comparative oncology / research surface

Lymphoma: the dog as a natural model of human B-cell lymphoma.

Canine B-cell lymphoma, especially diffuse large B-cell lymphoma, is one of the most-studied natural models of human DLBCL. The canine driver landscape is unusually well characterized across five cohorts, and it shares its core with the human disease.

These are somatic tumor alterations, not a germline carrier status. Every number here is a cohort frequency, the fraction of sequenced tumors somatically altered in a gene, reported by a published study. It is not a variant a dog inherits or carries, and it is not a prediction about any individual dog. Cross-species labels (concordant, divergent, canine-enriched) come from a commensurability-gated concordance map (INV-81), not a coarse shared flag.

This is the molecular side. For how often these cancers strike goldens over a lifetime, see the Golden Retriever Lifetime Study cohort → · all cancers →

The conserved core

concordant · 2 genes

Drivers where dog and human agree on status for this comparable cancer (commensurability-gated). That agreement is the evidence the dog models the human disease here.

TP53 concordant

tumor suppressor; a shared driver of both diseases

Dog

14%

of 86 tumors · SNV

Coyle et al. 2022

Human

25%

of 574 tumors · mutation

Schmitz et al. 2018

fused signature constraint LOEUF 0.449 ortholog dog↔human high-corroborated germline 11 ClinVar syndromes

TRAF3 concordant

NF-kB pathway regulator; the flagship conserved driver

Dog

45%

of 86 tumors · inactivating SNV/indel

Coyle et al. 2022

Human

—

rate not tabled (see basis)

Human basis: ~9% locus loss in human DLBCL (Blood 2015).

The flagship canine-model finding: TRAF3 is somatically inactivated in ~45% of canine B-cell lymphoma versus ~9% (locus loss) in human DLBCL (Blood 2015) -- the dog dramatically amplifies a conserved, human-relevant NF-kB driver.

fused signature constraint LOEUF 0.321 ortholog dog↔human high-corroborated

Canine-enriched drivers

darkness · 4

Recurrent in the canine cohort; the human side is not established as a recurrent driver here or is unquantified. Coverage darkness, not a measured disagreement (INV-81 / INV-77).

FBXW7 canine-enriched

20% of 86

ubiquitin-ligase tumor suppressor · SNV · Coyle et al. 2022

Its recurrent codon is shared across human cancers, but FBXW7 is not a recurrent human-DLBCL driver -- canine-enriched.

POT1 canine-enriched

15% of 86

shelterin / telomere-protection · SNV · Coyle et al. 2022

A germline predisposition gene in humans, not a recurrent somatic DLBCL driver -- a canine-enriched somatic event.

DDX3X canine-enriched

20% of 86

RNA helicase · SNV · Coyle et al. 2022

Recurrent in human Burkitt lymphoma; lower in DLBCL, so not encoded on the human side here.

SETD2 canine-enriched

13% of 86

histone H3K36 methyltransferase · SNV · Coyle et al. 2022

Present in human DLBCL at lower frequency; a shared chromatin-regulator pathway, exact human cohort % not extracted.

Not directly comparable

Same gene, different lesion class. Excluded from dissonance (INV-81 forbids apples-to-oranges).

MYC not comparable

oncogene

Same gene, DIFFERENT lesion class (point mutation vs translocation). Not comparable, so excluded from the dissonance map (the apples-to-oranges INV-81 forbids).

In human DLBCL MYC acts predominantly through translocation / double-hit -- a different lesion class than the canine point mutations, so not encoded as a comparable human cell.

The human landscape

Beyond the shared core, human DLBCL is classified by drivers the canine cohorts do not carry prominently: KMT2D (~25%), the MYD88 / CD79B axis (the MCD/ABC subtype), EZH2 (~6% overall, ~22% in GCB), and CREBBP (~11%). Exact per-cohort frequencies live in Schmitz 2018 and the Bakhshi 2020 review; we cite them here rather than restate a number we could not extract from the primary table.