Abstract 1354: Whole transcriptome analysis of primary and metastatic osteosarcoma in the canine model

Canine osteosarcoma is an excellent model of the human disease. Understanding the molecular events leading to metastatic disease is foundational for developing effective treatments. The objective of this study was to characterize single nucleotide variations (SNVs) acquired during metastasis. Primary and metastatic tumor samples were obtained from a dog with spontaneously occurring osteosarcoma, treated with surgery and adjuvant chemotherapy. Tumor total RNA was isolated and processed using standard techniques. Created libraries were used for paired-end sequencing (2 lanes of 2x80 per sample), in the Illumina Platform. Alignments to canFam2 reference sequence were performed using the TopHat splice aligner and quantified expression levels for all canFam2 ENSEMBL gene models using CuffLinks with GC-correction and upper-quartile normalization. After SNV discovery via custom pipeline, quantification of their abundance was performed by Benjamini-Hochberg corrected Fisher exact tests, on SNVs read counts. Thesequence data were visualized using Integrated Genome Viewer, while Polyphen2 was used to assess the potential impact of SNVs in protein function. Tumor samples from a 6-year-old neutered male Golden Retriever, diagnosed with spontaneously-occurringosteosarcoma, treated with surgery and adjuvant cheemotherapy were used. Lung metastases were confirmed to be metastatic osteosarcoma. Tumor RNA of adequate quantity and quality was isolated from primary and metastatic lesions. Seventy-four and 89 million pair-end reads were generated from the primary and the metastatic site, respectively. There were 13,292 heterologous and 19,780 homologous SNVs identified in the primary tumor, while 48,453 heterologous and 72,737 homologous SNVs identified in the metastatic tumor. At 20% false discovery rate, we identified 190 SNVs with increased abundance in the metastatic site compared to the primary site. The majority of these SNVs (>95%) were nearly or fully homozygous in the metastatic sample and either heterozygous or low abundance in the primary tumor sample. These results suggest frequent and potentially convergent loss-of-heterozygosity in the metastasis compared to the primary. More specifically, a missense SNV with high probability of affecting protein function was identified in p53, only in the metastatic sample. A putativealternative splicing event was identified in BRCA1, skipping exons 4, 5, and 6. A missense SNV was identified in BRCA2, with 1/9 reads having a T>