Somatic Structural and Rare Germline Variation in Childhood Cancers. D. I. Ritter1,2, B. Powell2,3, H. Cheung2, R. Gibbs1,2, D. A. Wheeler1,2, S. Plon2,3 1) Baylor College of Medicine, Human Genome Sequencing Center, Houston, TX; 2) Baylor College of Medicine, Department of Molecular and Human Genetics, Houston, TX; 3) Baylor College of Medicine, Texas Children's Cancer Center, Department of Pediatrics, Houston, TX.

   We have taken a comprehensive approach with whole genome and whole exome sequencing (WES) to understand the genetic basis of cancer susceptibility in childhood cancer patients with either a family history of cancer, second primary malignancies, congenital anomalies or learning disabilities. The sample set includes multiple expected inheritance patterns: de novo, autosomal dominant, autosomal recessive and X-linked, and patients with the following cancers: Wilms tumor, rhabdomyosarcoma, acute lymphoblastic leukemia, lymphoma, heptaoblastoma and neuroblastoma. Analysis of both somatic structural variation (SV) and loss of heterozygosity (LOH) in concert with rare germline single nucleotide variants (SNVs) and insertions/deletions (INDELS) are performed. We apply an empirical annotative filtering approach to prioritize SV in whole genome sequencing of tumor/normal pairs, combined with somatic LOH and SNVs/INDELs from WES data. To identify SV, we use complementary algorithms of insert-size (Breakdancer) and soft-clip stacking (CREST) first applied to SV seen by karyotype or arrayCGH (somatic t(9;22) and germline t(6;12), and chr17 microdeletion) as well as previously validated events from the published Lupski genome. From this, we derived an empirical filter, combined with downstream annotative filtering and clustering to identify both germline and somatic SV events. Our annotation includes: Refseq genes and gene events (intron/utr/cds), cytobands, centromere/telomere, repeats, transcription factor binding site clusters, frequency in Database of Genomic Variants and noncoding experimentally tested enhancers. Additionally, we use LOHcate, a program for tumor/normal whole-exome loss of heterozygosity, CARNAC, a somatic variant calling pipeline, and Pindel to further analyze the sequencing datasets. We pinpoint breakpoints for somatic SV, such as a complex translocation in POLR1A/UPS22/RPS9 t(2;17;19) from a Wilms tumor sample, TMEM120B/RET t(12;10) translocation in rhabdomyosarcoma with amplifications in 7q31 and 10q11.21, and a translocation in CDK12/noncoding (~25kb upstream of DHRS7) t(17;14) in neuroblastoma. We have identified multiple regions of somatic LOH and are currently preparing rare germline events within these regions for validation. With this analysis, we aim to uncover germline events associated with childhood cancer susceptibility and create a more comprehensive understanding of the etiology of cancer.

You may contact the first author (during and after the meeting) at