The lab is utilizing the power of next generation sequencing technologies such as DNA, RNA sequencing, ATAC-seq, and Hi-C technologies to identify novel genetic and epigenetic features of brain cancer in brain tumor tissues, immune cells, as well as the plasma component in blood. We utilize machine learning methodologies with the eventual goal of developing and validating the technology developed for the early and non-invasive detection of brain tumors.
Glioblastoma Treatment
Glioblastoma is the most challenging tumor to treat in the body. Despite many advances in treatments of systemic cancers, few improvements have been made in the treatment of glioblastoma. Our goal in the lab is to identify ways to diagnose brain tumors, including high grade gliomas at an earlier phase when the tumor is smaller, deeming surgery and other treatments more effective and less risky. To do this, we utilize the newest DNA and RNA sequencing technologies to detect the presence of circulating molecules in the blood of these patients. To enhance the tumor specific signal recognized by these technologies, we combine new genetic and epigenetic approaches with novel surgical tools such as MRI focused ultrasound and Laser Interstitial Thermal Therapy (LITT). The hope is to obtain a more sensitive and specific signal present in individuals that are asymptomatic and have smaller tumors.
Glioblastoma Long Term Survivors
Glioblastoma is a tumor that is very resistant to any treatment. As a result, only a small percentage of patients live past the two year mark from diagnosis. However, up to 15% of patients will respond very well to treatment and 5% of patients will live past the five year mark. In this work, we conducted a whole genome methylation analysis of several long term and short term glioblastoma survivors from a multi-institutional cohort of individuals across three different continents (North America, Asia, Oceania). We found that unlike the short term survivors (patients with shorter than average survival), the long term survivors (patients surviving>3 years) exhibited significant CpG island hypermethylation in genes involved in oncogenic pathways, suppressing tumor growth. However, hypomethylation was observed in long term survivors in regions away from the CpG islands, thus decreasing the mutation rate associated with global genome methylation.
Early Detection of Liver Cancer
Liver cancer is a major source of cancer mortality around the world. There is currently no approved screening test for the diagnoses of liver cancer at an early stage. In this manuscript, my colleagues and I have implemented DELFi – a plasma based cfDNA assay that uses various fragmentation based on cancer specific features that in combination with machine learning showed very high accuracy in diagnosis of liver cancerin patients at risk, as well as in the general population. We additionally showed that cfDNA in liver cancer patients is a combination of DNA released from the white blood cells and liver cancer cells.
Early Detection of Lung Cancer
Lung cancer is the leading cause of cancer death to date. The current approved screening test of low dose helical CT, while sensitive and specific for screening of at-risk individuals, is severely underutilized. As a result, only up to 6% of patients have their annual screening and therefore many patients are diagnosed with lung cancer when the cancer has become metastatic. In this multi-disciplinary, international effort, my colleagues and I have implemented DELFI-a plasma based cfDNA assay that uses various fragmentation based, cancer specific features that, in combination with machine learning, showed a robust performance in detection of localized disease at a stage where a cure could be obtained. Additionally, in this effort we showed that a link exists between cfDNA fragmentation, tumor aggressiveness and patient outcome after adjusting for any known factors of survival. Importantly, we were able to make a molecular distinction between small cell and non-small cell lung cancer non-invasively, by utilizing insights derived from cfDNA fragmentation in areas of the genome known to be bound by ASCL1, a master regular of small cell lung cancer development.
