Expanding the Spectrum of Canine Diffuse Large B-Cell Lymphoma Genetic Aberrations Through Whole Genome Sequencing Analysis

Abstract

Diffuse large B-cell lymphoma (DLBCL) is one of the most prevalent haematological malignancies in both humans and dogs, characterised in both species by significant clinical heterogeneity and limited prognostic predictability. With the introduction of next-generation sequencing (NGS) technologies in veterinary medicine over the past decade, researchers have begun to elucidate the molecular basis of canine DLBCL (cDLBCL); however, much of the clinical heterogeneity associated with this tumour remains unexplained. In this study, we performed whole genome sequencing on 10 cDLBCL cases, all treated with chemo-immunotherapy, which exhibited similar clinico-pathological features but markedly different outcomes. Cases were classified as "poor" or "good" responders based on whether their lymphoma-specific survival fell below or above the cohort's median. Protein-coding variants and copy number aberrations unique to poor or good responders revealed novel candidate genes not previously identified in cDLBCL studies, while splicing, untranslated regions, and intronic variants were detected in genes already known to be recurrently mutated. In conclusion, our investigation has broadened the spectrum of potentially pathogenic variants implicated in cDLBCL, though further studies with larger cohorts are necessary to validate these findings.

Keywords: DLBCL; diffuse large B‐cell lymphoma; dog; prognosis; whole genome sequencing.

FIGURE 1

Kaplan–Meier (KM) curves for 10 dogs with DLBCL and treated with chemo‐immunotherapy according to clinical outcome. KM curves of TTP (A) and LSS (B) according to clinical outcome. Dogs classified as “poor responders” show both shorter TTP and LSS compared to those classified as “good” responders.

FIGURE 2

Variant distribution across 10 cDLBCL analysed by WGS. Distribution of somatic short variants (SNVs and indels) across 10 cDLBCLs. Each bar represents a single dog, while on the y‐axis is reported the total number of variants identified in each sample. “Poor” and “good” responders are indicated with red and green bars, respectively.

FIGURE 3

Oncoplot of genes carrying protein‐coding variants in cDLBCL. Top 30 mutated genes harbouring somatic protein‐coding SNVs and indels (upper panel) and distribution of nucleotide substitutions (lower panel) identified by WGS in 10 cDLBCLs. Genes are represented in descending order according to the frequency of mutation. Different mutation types are identified with different colours. Each column represents a single dog.

FIGURE 4

Mutational signatures analysis in cDLBCL. Mutational signatures were extracted using the Sigminer R package and confronted against COSMIC signatures. The plot shows the distribution of the six types of substitution (in 96 different trinucleotide contexts) defined by the pyrimidine as inferred from the non‐negative matrix factorisation algorithm. Both in poor (A) and good (B) responders, the analysis of single base substitution signatures (SBS) revealed that the signatures mainly contributing to mutation spectra corresponded to COSMIC SBS6 (defective DNA mismatch repair), SBS5 (clock‐like signature of unknown aetiology), SBS40c (of unknown aetiology) and SBS3 (defective homologous recombination DNA damage repair) in different proportions.

FIGURE 5

Recurrent somatic copy‐number aberrations (SCNAs) identified in cDLBCL. The picture shows statistically significant SCNAs peaks identified by GISTIC in poor (A) and good (B) responders. Amplification peaks are depicted in red, while deletion peaks in blue. Significance threshold is also indicated by a green line.