10. GENETIC ASSESSMENT OF GLIOBLASTOMA PRIMARY TUMORS AND MATHCED PATIENT-DERIVED PRE-CLINICAL MODELS USING WHOLE EXOME SEQUENCING

Department: Bioengineering
Research Institute Affiliation: Graduate Program in Bioinformatics
Faculty Advisor(s): Kelly Frazer

Primary Student
Name: Shawn E Yost
Email: seyost@ucsd.edu
Phone: 801-554-8822
Grad Year: 2013

Abstract
Glioblastoma Multiform (GBM) was the first cancer type analyzed by the TCGA which revealed a wealth of information about its genetic and epigenetic landscape, resulting in molecular classification, promising prognosis markers and new therapeutic avenues. If the genetics of the primary samples are now well-described, then the genetics of patient-derived pre-clinical models used in research are the next step. In particular, the extent of the genetic changes, due to genetic drift or clonal selection between primary and patient-derived models is not known. These changes need to be assessed before developing targeted and personalized therapies using these models. Here we selected four patients? primary tumors, matched pre-clinical tumor models, and their matched blood for whole exome sequencing. We identified a novel method to remove false positive variants from sequencing data. With this we identify known GBM mutations, PTEN and TP53, as well as known copy number alterations (CNA), loss of chromosome 10 and amplification of chromosome 7, from matched primary tumor and blood. Tumor DNA isolated from xenograft models are contaminated with mouse DNA, we developed a unique filtering method that removes mouse DNA contamination from the sequenced reads. We also implemented a new technique to identify shared vs. unique somatic variants between matched primary and pre-clinical tumors. We showed that 97% of the primary somatic mutations are also identified in the tumor models with very few specific mutations, likely due to sampling noise. Our study suggests that the four GBM samples carry known genetic alterations in GBM, which are conserved in their respective tumor models. Therefore, these patient-derived samples are realistic, genetically authentic pre-clinical models on which drug effects and response can be studied.

« Back to Posters or Search Results