A Genomic Portrait of Tumor Progression Using Second Generation Sequencing

Genetic alterations are thought to enable cancers to proliferate and survive more effectively, or to resist cytotoxic therapies. However, the molecular basis of tumorigenesis and progression is not fully understood. In order to understand the genetic aberrations underlying tumor progression in refractory neuroblastoma, we performed next-generation sequencing on the whole exomes of three sequential tumor samples taken from a high-risk stage 4 neuroblastoma patient at diagnosis, after cytotoxic therapies, and at death, paired with a germline DNA sample from the same patient. Each genomic DNA sample was first partitioned using SureSelect whole human exome kits targeting ~38Mb of genomic regions, then subjected to sequencing. Each exome sequencing experiment yielded approximately 240 million mappable reads representing 12 billion base pairs. We used DiBayes to identify single nucleotide variants (SNVs). After filtering away the low-quality SNVs, we identified approximately 30,000 high-quality SNVs in each sample, of which ~98% were also found in germline samples, i.e., only 2% could be confirmed as somatic SNVs. Using the frequencies of the SNVs derived from the sequencing experiments, we discovered an allelic imbalance (LOH and triploidy) signature at the chromosome level among all three tumors indicating they originate from a common cancer progenitor. Overlaying RNA-Seq data with the variants found in exome data allowed us to detect 5 SNVs located in highly expressed genes and therefore suitable as therapeutic target candidates. Two of these variants were detected in all three tumor samples indicating their importance as they were present already at diagnosis and persisted in the tumor throughout therapy. These variants are currently being validated using Sanger sequencing in the same tumor samples to investigate for pathway disruption that may lead to chemo-refractory disease.

Speaker Bio: In 2006 I obtained my Ph.D. in Theoretical Physics at Lund University, Sweden. The focus of the thesis was to investigate how one can use machine learning such as support vector machines to identify patterns in biological high dimensional data from, e.g., expression arrays. I received a scholarship in 2006 to visit Queensland Institute of Medical Research in order continue my work on how a mutation in the BRAF gene effects the downstream expression profile. In 2007 I received a Cancer Research Training Award and have since then been working in the Oncogenomics Group in the Pediatric Oncology Branch, National Cancer Institute. Initially I continued my work combining machine learning with expression arrays to reveal differences between pediatric cancers types. Lately, I have engaged more in analysis of data from second generation sequencing and the challenges that arise when going from billions of short sequencing reads to learn some new cancer biology.

Presentation Slides: PDF

Note: Visitors from outside NIST must contact Robin Bickel; (301) 975-3668; at least 24 hours in advance.