Molecular landscape of canine soft tissue sarcoma.

Canine soft tissue sarcomas (STS) are a heterogenous group of malignant tumors arising from mesenchymal cells of soft tissues. This simplified collective of tumors most commonly arise from subcutaneous tissues, are treated similar clinically, and conventionally exclude other sarcomas with more definitive anatomical, histological, or biological features including hemangiosarcoma, lymphangiosarcoma, histiocytic sarcoma, synovial sarcoma, leiomyosarcoma, and rhabdomyosarcoma. Histologically, canine STS sub-types are difficult to discern at the light microscopic level due to their overlapping microscopic features. Thus, genomic, and transcriptomic profiling of canine STS may prove more valuable in differentiating the diverse sub-types of mesenchymal neoplasms within this grouping. To this purpose we sought to characterize the mutation and expression profiles of canine STS. In this study we used whole exome capture (Agilent Sure Select Canine V2) to sequence 29 tumors, grade I-III, from non-visceral locations, and matched normal samples, along with RNAseq of tumor samples. For the WES data, 40-170 million 150 bp paired-end reads were obtained from Illumina sequencing which were mapped against CanFam3.1 genome via BWA and short variants were called and annotated Mutect2 and VEP tools, respectively. The median depth of sequencing for normals and tumors were 97X (range: 35X-128X) and 110X (range: 22X-133X), respectively. Additionally, RNAseq reads were sequenced from 29 tumors and mapped against CanFam3.1 genome via STAR and gene count data was generated by HTSeq-count tool. The total number of somatic variants identified across 29 samples ranged from 275 to 4,196, of which, 6% to 31% were annotated as cDNA variants. The protein coding mutations per callable megabase ranged from 0.3 to 37.3. Known cancer genes (COSMIC database) with recurrent mutations were KMT2D(21% of the samples), TP53 (21%), CNTNAP2(14%) genes. Using the RNAseq data, we identified three samples carrying driver fusions of platelet derived growth factor B (PDGFB) and collagen genes. These gene fusions were confirmed through Sanger sequencing of amplified cDNA. To delineate the transcriptomic sub-types, we used hierarchical and K-means clustering methods. This resulted in 4 groups with similar expression profiles comprising of 11, 4, 8 and 6 samples in each. Further, genes set variation analysis (GSVA) was used to profile the distinct pathways across all 4 clusters. Preliminary data indicates that the 11-sample and 6-sample clusters had up-regulated non-homologous end joining pathway and immune response pathways, respectively. Further study of genes and pathways that are enriched in sample clusters will help in categorizing the canine STS tumors for diagnosis and treatment.