Ultrasensitive Detection of Subclinical Lung Cancer by Statistical Analysis of Plasma cfDNA-Derived Whole-Genome Sequencing Data
Abstract
Use of blood-based floating DNA fragments has enormous potential for non-invasive early detection of cancer. Recently, it was demonstrated that cancer-derived DNA shed into the blood was an early indicator of relapse following surgical removal of lung cancer. To date, most blood-based analysis of cancer samples has focused on looking very deeply at a tiny region of the human genome known to be commonly mutated in lung cancer. However, for patients with undetectable cancer, fundamental limitations due to the fact that only a very small amount of floating DNA will be present in the blood mean that this approach cannot work. Because patients with clinically undetectable cancer will typically shed only a few nanograms of floating DNA per tube of blood (the mass equivalent of approximately 1,000 genomes [3.6 ng]), the sensitivity to detect any given mutation cannot be greater than 1 in 1,000, even with perfect DNA sequencing. This is because only 1,000 copies of any given gene will be present in the blood. For an assay to be useful for detection of subclinical cancer, such as for screening or monitoring of recurrence, much greater sensitivity is needed. We propose an entirely new approach that leverages breadth of genomic coverage, rather than depth, to achieve ultra-sensitive detection of cancer. Importantly, this approach can be piloted immediately using standard whole-genome sequencing (WGS) of plasma-based floating DNA obtained from blood collected during the course of clinical care. The key to our proposed innovation is that standard WGS to moderate depth (20-100x) can provide excellent sensitivity to detect the presence of cancer, even though the sensitivity to detect any single cancer mutation is low. This is because the probabilities of various cancer mutations being represented in a sample of floating DNA are statistically independent of one another. Therefore, sensitive detection can be achieved by combining weak evidence from up to millions of cancer mutations. Our preliminary calculations lead us to hypothesize that our approach has the potential to realize a transformational sensitivity increase of up to three orders of magnitude over existing approaches; thus, if successful, it could revolutionize the ability for early detection of a range of cancers.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Oct 29, 2018
- Source ID
- W81XWH1810357
Entities
People
- Scott Carter
Organizations
- Dana–Farber Cancer Institute
- United States Army