FDA-Led Consortium Details Improvements Needed in Liquid Biopsies

That’s a change in procedure from earlier liquid biopsy studies, the authors say. “These studies are in general performed by the vendors of labs that are performing the test, and they are not using standardized samples,” says first author Ira Deveson, a genomics scholar at the Garvan Institute of Medical Research in Australia.

The effort, led by the US Food and Drug Administration (FDA), tested the accuracy of the candidate liquid biopsies—none are approved for clinical use—at 12 labs in the US, Europe, Asia, and Australia. Consortium members made all of their data publicly available, and say it is the largest such study they are aware of to date. “People can take the data and scrutinize how we do the analysis,” says Joshua Xu, a computer scientist at the FDA’s National Center for Toxicological Research in Jefferson, Arkansas.

The promise of liquid biopsies

A liquid biopsy detects tumor DNA that begins to circulate in blood after some tumor cells die and shed their genomic material. Patients complete a blood draw as part of the procedure, which is less invasive than a traditional biopsy that requires surgically excising cells or tissue to detect whether a cancer exists or is spreading. The ease of the liquid biopsy procedure may sometimes come at the cost of accuracy, according to experts who spoke to The Scientist.

When somebody has a bad infection, a lot of times you’re just monitoring lab assays and changing the antibiotic cocktail appropriately. Oncology has always been missing that type of feedback assay.

—Donald Johann Jr., University of Arkansas Medical Sciences

“Liquid biopsies in their current form are not going to be as sensitive as a tissue biopsy in their current form,” says Ben Ho Park, a professor of medicine at Vanderbilt University Medical Center who was not involved in the analysis. He says that its results confirm what he and others have seen in their own experiments. Park adds that this difference in sensitivity is why standard tissue biopsies are more commonly used than liquid biopsies, either to detect cancer in the first place or to see how a patient is responding to treatment.

Should liquid biopsies improve in accuracy, the results for patient management could be profound, says Donald Johann Jr., a medical oncologist at the University of Arkansas Medical Sciences in Little Rock and one of the leaders of the analysis. Liquid biopsy products are used for limited applications currently, Johann notes, such as detecting whether metastatic lung cancer patients respond to the drug erlotinib. (Roche developed this assay, which is different from the Roche assay used in research only that Johann and his colleagues assessed.) His long-term vision is for liquid biopsies to become routine diagnostic and patient management tools for many types of cancer.

“When somebody has a bad infection, a lot of times you’re just monitoring lab assays and changing the antibiotic cocktail appropriately. Oncology has always been missing that type of feedback assay,” Johann says. This new analysis, in Johann’s view, will serve as a seminal piece of infrastructure for future validation studies of liquid biopsy tools. That’s because it provides a roadmap for how to conduct standardized analyses of liquid biopsies, Johann points out, as well as suggested best practices for analyzing these assays.

Consortium members assessed the detection of cancerous DNA from reference samples derived from 10 different cancer cell lines, and one cancer-free cell line. These were combined in different proportions to simulate blood samples with varying ratios of tumor DNA to healthy DNA, otherwise known as variant allele frequency.

For this study a genetic mutation was defined as small variant, generally fewer than five base pairs in size, on a chromosome. Generally speaking, the higher the allele frequency, the worse the cancer stage.

They tested the accuracy of the five different assays on the cell-line–derived genetic material. For any cancer in which the variant allele frequency was 0.5 percent or greater, every biopsy could successfully detect it. The assays were much less reliable at lower allele frequency levels. An in silico simulation yielded similar results.

The authors note that many advanced cancers are detectable at 0.5 percent variant allele frequency and above, meaning liquid biopsies could be a valuable tool in these instances. The goal would be to detect cancer early, though, when variant allele frequency is not as pronounced. The analysis shows that this is an area in which many leading liquid biopsy tools need to improve.

Tina Lockwood, a geneticist at the University of Washington who was not involved in the study, says she hopes this paper will bring more attention to the analytical challenges inherent in designing liquid biopsy products.

Lockwood notes that improving the accuracy of these assays is not the only challenge. Providing tools for patients and doctors to better understand what the results of these tests mean is also critical.

“We also see a lot of misleading information when it comes to interpretation,” Lockwood cautions. Sometimes what looks like circulating tumor DNA may in fact be an artifact of aging or a result of cancer treatment, she says, meaning that nuance and medical judgment will always be part of the process when interpreting liquid biopsy results.

I.W. Deveson et al., “Evaluating the analytical validity of circulating tumor DNA sequencing assays for precision oncology,” Nat Biotechnol, doi:10.1038/s41587-021-00857-z, 2021. 

Correction (May 20): This article has been corrected to show that the Roche liquid biopsy related to erlotinib is for people with metastatic lung cancer, not breast cancer; to reflect the number of cancer-free cell lines used in the analysis; and to clarify how the researchers defined a genetic mutation. The Scientist regrets the errors.