Ultrasensitive detection and monitoring of central nervous system tumors from plasma using personalized whole-genome ctDNA profiling.

Ivy Tran,Kristyn Galbraith,Sharon L. Gardner,Jeffrey C. Allen,David Harter,Jeffrey Wisoff,Eveline Teresa Hidalgo,Danielle Afterman,Santiago Gonzalez,James Smadbeck,Tomer Lauterman,Ury Alon,Ravi Kandasamy,Iman Tavassoly,Paz Polak,Boris Oklander,Praveen Raju,Asaf Zviran,Matija Snuderl

DOI: https://doi.org/10.1200/jco.2023.41.16_suppl.2064

IF: 45.3

2023-06-01

Journal of Clinical Oncology

Abstract:2064 Background: Patients with the central nervous system (CNS) tumors are largely followed up by imaging. Current plasma-based liquid biopsy techniques have limited utility in neuro-oncology due to a low circulating cell-free tumor DNA (ctDNA) burden, blood-brain barrier, and low number of mutations in coding regions. Whole genome sequencing (WGS)-derived patient specific mutational signature from a matched tumor-normal WGS can provide a personalized, highly sensitive and specific approach to detect mutations in ctDNA and provide blood-based monitoring in brain tumor patients. Furthermore, it can be performed on lower amount of peripheral blood since WGS requires less sequencing depth compared to targeted ctDNA panels. Methods: We have profiled a cohort of 28 extra- and intra-axial adult and pediatric brain tumors including adult and pediatric low- and high-grade glioma (9), meningiomas (11), medulloblastomas (5), ependymomas (2), neurocytoma (1). Tumor DNA was extracted from archival pathology tissue, normal DNA from unsorted white blood cells, and ctDNA from 1-2 mL of post-surgery plasma. WGS was performed with 40x coverage for Tumor-Normal DNA and 20x for ctDNA. Using WGS of matched Tumor-Normal and plasma samples, we derived a personalized mutational pattern using SNVs, indels, and copy numbers for quantification and ultra-sensitive detection of ctDNA in plasma samples. An AI-based error suppression model was implemented to filter out the noise in the cell-free DNA (cfDNA) while the personalized mutational signature was used to detect the ctDNA in the cfDNA and to amplify the somatic signal to determine the Tumor Fraction at the time of diagnosis, during the therapy or surveillance period. The ctDNA Tumor Fraction (TF) was compared to the clinical status and MRI-based imaging. Results: All subtypes of brain tumors contained enough mutations to derive personalized mutational signatures. Most mutations were distributed in the noncoding DNA. TF correlated with clinical status and with the disease course on imaging at given time points reaching a 10 -4 minimal residual disease detection sensitivity. We were able to detect ctDNA across all WHO grades ranging from WHO 1 meningioma to WHO 4 glioblastoma and medulloblastoma. Furthermore, we were able to detect tumor-specific copy number aberrations such as MYCN amplification in plasma samples and mutational signatures. Conclusions: Here we demonstrate that patient-specific WGS tumor signature in ctDNA from plasma can be used for sensitive monitoring of adults and children with primary low- and high-grade CNS tumors.

oncology