Abstract

8519 Background: Lung cancer incidence and mortality are increasing worldwide despite more effective treatments. This is primarily due to the late stage of diagnosis when treatments are less effective. Although large randomized trials have demonstrated a significant decrease in lung cancer mortality through screening of high-risk individuals with chest low dose computed tomography (LDCT), LDCT has made little impact in the community, mainly due to lack of accessibility. There is therefore an unmet clinical need for development of cost-effective and easily implemented tests for early lung cancer detection. Methods: We have previously shown that altered genome-wide fragmentation of cell free DNA (cfDNA) is a common characteristic of many cancers. In this study, we leverage this knowledge to increase the sensitivity of lung cancer detection by interrogating characteristics of the size distribution of cfDNA fragments across the genome using machine learning methods. The approach we present, called DELFI (DNA evaluation of fragments for early interception) generates a score that reflects the presence of tumor-derived DNA in plasma based on a multi-feature genomic analysis that assesses millions of cfDNA fragments for tumor-derived genomic and epigenomic changes in a small amount of blood (2-4 mls) via inexpensive low coverage (1-2x) whole genome sequencing. We applied this methodology in a prospectively collected cohort of 365 individuals under investigation for lung cancer and we prospectively validated it in a separate case-control cohort of patients with newly diagnosed early stage lung cancer as well as individuals without cancer (n=427). Results: These analyses revealed high performance for detection of early and late stage disease (Table). When DELFI was used as a prescreen for LDCT it increased specificity from 58% with CT imaging alone to 80% using the combined approach. The DELFI score was significantly associated with T and N stage in lung cancer cases (p<0.0001) as well as with overall survival (p=0.003). In a multivariable analysis including age, histology and stage, DELFI score was an independent prognostic factor of overall survival (HR=2.53; p=0.0003). Finally, we determined that genome-wide fragmentation profiles can be used to distinguish small cell lung cancer from non-small cell lung cancer with high accuracy (AUC 0.98). Conclusions: These findings provide key insights into cfDNA fragmentation in patients with cancer and a new and easily accessible avenue for non-invasive diagnosis and molecular profiling of lung cancer.[Table: see text]

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call