Abstract
Ca-ssis / Getty Images
Imagine an annual physical exam that, along with measuring blood pressure and cholesterol, also screened for multiple cancers with a single blood draw. Several companies, including Grail, Guardant Health, and Freenome have received millions of dollars to finance the development of blood-based screening tests for early cancer detection based on circulating tumor DNA (ctDNA) technology.
However, according to Olivier Elemento, Ph.D., director of the Caryl and Israel Englander Institute for Precision Medicine at Weill Cornell Medicine, “There is a misconception in the field that [early] detection should be easy… The bottom line is that if you run the numbers it’s very unlikely that early detection can be performed with a reasonable amount of blood.”
In a manuscript recently posted on bioRxiv, entitled “Challenges in Using ctDNA to Achieve Early Detection of Cancer,” Elemento and co-author Imran Haque, Ph.D., chief scientific officer of Freenome assert that “statistical and physiological limitations suggest that a ctDNA-based mutational assay for early detection would be neither commercially nor biologically viable.”
Firstly, the authors argue that the amount of ctDNA present in the bloodstream—especially for early-stage cancer and premalignancies—is exceedingly small. In addition, the ctDNA needs to carry a cancer-driving mutation that the screening test recognizes, and the extensive heterogeneity in tumors means that not all ctDNA carries the same mutations. Secondly, somatic mutations in normal tissue and blood cells introduce a lot of noise—essentially burying ctDNA signals and decreasing both detection sensitivity and specificity.
The manuscript’s punchline, said Elemento, is that researchers will need to integrate ctDNA with other signals to develop successful screening platforms. Promising results from CancerSEEK, a multi-analyte, blood-based diagnostic recently published in Science, support the manuscript’s argument. “[The authors] realized that if you look at the mutations, you can’t detect cancer very accurately. By combining protein level and mutations you can achieve better results,” said Elemento.
However, protein levels are just the tip of the iceberg. DNA methylation, and DNA fragmentation patterns can distinguish between ctDNA and cell-free DNA from normal cells, which could improve detection by increasing the ctDNA signal. Other potentially useful biomarkers include metabolites and miRNA.
Elemento also emphasized the importance of longitudinal measurements for tracking tumor evolution and preventing over-treatment: “We need to not only have better tests, but also to use those tests as a way to see prospectively the evolution of a disease. If the disease is progressing—that’s when you would want to intervene.”
Weill Cornell Medicine’s Olivier Elemento, Ph.D.
