Abstract PURPOSE: Screening for lung cancer using a biomarker blood test has the potential to overcome the currently low adherence rates in the US, offering patients a convenient and accessible option. Integrating routine blood draws into lung cancer screening could boost participation, mirroring the success of non-invasive tests in colorectal screening. Numerous proof-of-concept studies indicate that analyzing genome-wide cfDNA fragmentation patterns can effectively distinguish between individuals with and without lung cancer. We conducted a multi-institutional case-control study to train and validate a fragmentome based cfDNA classifier for lung cancer early detection. METHODS: The DELFI-L101 study (NCT04825834) is a prospective, multi-site, observational case-control study designed to develop and validate an early detection test for lung cancer. The DELFI (DNA Evaluation of Fragments for Early Interception) method utilizes low-pass whole genome sequencing, a cost-effective approach that examines cell-free DNA (cfDNA) fragmentation patterns associated with lung cancer. Enrolled participants were aged ≥50 y, smoking history of ≥20 pack-years, and had undergone or had planned chest CT imaging. Medical histories were documented, and blood samples were collected for DELFI analysis. A classifier, developed through supervised machine learning, was created to distinguish between samples from individuals with lung cancer and those without. Performance was assessed through repeated 10-fold cross-validation in the training set and then subsequently assessed in an independent split-study clinical validation set. The classifier was optimized for the sensitivity detection of lung cancer in the screening-eligible population. Performance was evaluated across patient and tumor characteristics, adjusting the stage distribution to match that of LDCT screening populations. RESULTS: In an analysis of 958 participants from both the lung cancer and non-cancer control groups, the sensitivity of the test remained consistent across various participant demographics, comorbidities, and tumor characteristics, increasing with higher group, T, and N stages. The stage weighted sensitivity for the screening population was 80% (95% CI: 75%-86%), with a specificity of 53% (95% CI: 43%-64%). At a prevalence of 0.7%, the negative predictive value was 99.7%. CONCLUSIONS: Clinical validation of the DELFI-based blood test demonstrates its potential to detect lung cancer within the screening eligible population. A negative result is associated with a high negative predictive value. Given the limited adoption of low-dose computed tomography (LDCT) screening, especially among socio-economically disadvantaged individuals, this clinically validated, blood-based cfDNA test could significantly increase LDCT screening participation. Citation Format: Peter Mazzone, Peter B. Bach, Lindsey B. Cotton, Caitlin A. Schonewolf. Clinical validation: A blood-based biomarker for early lung cancer detection based on circulating DNA fragmentomics [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 1062.
Read full abstract