PROJECT SUMMARY/ABSTRACT Glioblastoma (GBM) is the most common and most aggressive adult primary brain tumor. Regardless of therapy, the median survival time is less than 15 months as GBM nearly always recurs. Diagnosing GBM currently depends on acquiring tumor tissue for histologic examination to differentiate GBM from other types of brain lesions with a similar imaging appearance. Detecting GBM recurrence is also challenging as treatment effects such as pseudoprogression and radiation necrosis mimic recurrent disease. Thus, risk-associated diagnostic invasive procedures remain a critical aspect of GBM patient care even though surgically acquired neurologic deficits may reduce an already short survival time. Therefore, a non-invasive approach to diagnose GBM and identify true recurrence has the potential to directly impact patient care throughout the entire disease course. DNA released from cells during programmed cell death, termed cell-free DNA, is an emerging biomarker for diagnosing human cancers and monitoring response to therapy. Detection of tumor-derived cell- free DNA in plasma, also known as circulating tumor DNA (ctDNA), has previously been largely unsuccessful in humans with GBM due to an inability to find the ctDNA amongst the abundant background of normally occurring cell-free DNA. The large intra- and inter-tumor genetic heterogeneity of GBM has hindered searches for specific tumor variants in cell-free DNA, while the rarity of GBM to metastasize beyond the central nervous system has considerably restricted the quantity of ctDNA present. However, our uncovering of ctDNA in a xenograft brain tumor model using human GBM stem-like cells strongly supports the feasibility of detecting GBM-derived ctDNA in humans. Moreover, our previous success in the animal model provides direction for the human translation – reducing or eliminating noise associated with next-generation sequencing (NGS) that interferes with the detection of very low frequency variants is necessary to allow searches for ctDNA that are not dependent on a priori knowledge of solid tumor variants. In this proposal, we attenuate NGS-related noise and demonstrate the unbiased detection of GBM-derived ctDNA at time of initial diagnosis and at recurrence. We also show that identification of GBM solid tumor DNA variants in ctDNA is enhanced by reducing errors associated with NGS to improve sampling of the solid tumor genetic heterogeneity typical of GBM. Therefore, this proposal seeks to translate plasma cell-free DNA as a biomarker to detect GBM in humans by suppressing errors associated with NGS to improve variant detection at very low allele frequencies. The successful application of ctDNA biomarkers to non-invasively assess GBM will directly affect patient care by enabling the optimization of clinical management prior to or instead of risk-associated invasive diagnostic procedures.

PROJECT NARRATIVE The most common and most aggressive adult primary brain tumor is glioblastoma (GBM). In this study, we seek to develop tumor-derived cell-free DNA present in the circulation as a non-invasive biomarker to diagnose GBM and detect recurrent disease. The non-invasive assessment of GBM has the potential to directly affect patient care by enabling the optimization of clinical management prior to or instead of risk-associated invasive diagnostic procedures.