Blood Test May Detect Multiple Cancer Types: Study

Roxanne Nelson, RN, BSN

March 31, 2020

Editor's note: The headline of this story has been changed to clarify that these are early results from a single study.

A new blood test for cancer was highly accurate in detecting more than 50 types of cancer, and also identified the tissues where the cancer originated, say researchers.

The test (developed by Grail) uses next-generation sequencing to analyze the arrangement of methyl groups on circulating cell-free DNA in a blood sample.

"The results of this study suggest that such assays could be a feasible way of screening people for a wide variety of cancers," co-lead author Geoffrey Oxnard, MD, from Dana-Farber Cancer Institute in Boston, Massachusetts, said in a statement.

The results were published online on March 31 in the Annals of Oncology.

Overall, the test detected cancer with a specificity of 99.3% and a false positive rate of 0.7%.

However, the sensitivity was lower, at 67%, for the 12 most common cancers.  

When cancer was detected, the test correctly identified the tissue where the cancer originated in 90% of cases.

However, a researcher not involved with the study suggests the results may be overly optimistic and show higher sensitivity and specificity than are likely to be seen in follow-up studies.

William Grady, MD, a member of the Clinical Research Division and Public Health Sciences Division at the Fred Hutchinson Cancer Research Center, Seattle, Washington, told Medscape Medical News: "This is a result of the impact of preanalytical variables, confounding factors, inappropriate control population, poor design of validation studies that use a suboptimal validation cohort not relevant to the training cohort," he said.

Could Blood Test Be Used for Cancer Screening?

Senior author Michael Seiden, MD, PhD, president of The US Oncology Network, The Woodlands, Texas, noted the research was not a screening study.

"The test was evaluated in patients with known cancer as well as a few thousand patients thought not to have cancer," he told Medscape Medical News

Clinical studies evaluating the test are ongoing.

"The test is intended to be complementary to, and not replace, existing guideline-recommended screening tests, and might provide new avenues of investigation for cancers that don't currently have screening tests," said Seiden. "We anticipate there will more information around the test later this year."

He added that if the test were to become commercially available, it would require very well-controlled processing at a highly specialized lab for the foreseeable future.

Commenting for Medscape Medical News, Grady did not envisage that the new blood test could be used for cancer screening.

"For a screening test to have clinical utility, it must impact on cancer-related survival, be appropriate for the population it is applied to, have an appropriate natural history for screening, have clinically effective therapies, have confirmatory tests that are safe, have a low burden on the healthcare system, and have a clear management plan in the healthcare system," he explained.

These are different for every cancer type, and the criteria are currently only met in the US general population for breast cancer and colorectal cancer, he pointed out. The likelihood that a pan-cancer screening test will be able to meet current regulatory standards unless designed and used in the fashion used for hereditary cancer gene panels is quite low, he suggested. 

Finally, for a screening test to be used clinically "it must have very low risk because it is applied to a healthy population," he added. "A pan-cancer screening test used in the general population does not meet that standard."

Study Details

The study was conducted in 6689 participants, 2482 with cancer (more than 50 types, at various stages) and 4207 without cancer. They were divided into training and validation sets, and a classifier was developed and validated for cancer detection and tissue of origin.

Performance of the test was consistent in training and validation sets, and cancer detection improved with advancing cancer stage.

Results are reported for a group of 12 prespecified cancers (which account for 63% of all cancers and include anal, bladder, colon/rectum, esophageal, head and neck, liver/bile duct, lung, lymphoma, ovarian, pancreas, plasma cell neoplasm, and stomach). For this group of 12 cancers, the sensitivity of the test was 39% in stage I cancers, 69% in stage II, 83% in stage III, and 92% in stage IV. For all 50 cancer types together, the corresponding rates were 18%, 43%, 81%, and 93%, respectively.

Tissue of origin was predicted in 96% of specimens that had cancer-like signals and was accurate in 93% of the validation set.

How Does it Compare With CancerSEEK?  

How do the current results with the Grail test compare to those reported 2 years ago with CancerSEEK, another cancer diagnostic blood test reported to identify eight common cancers (ovary, liver, esophagus, pancreas, stomach, colorectal, lung, and breast).

The CancerSEEK test measures circulating tumor DNA from 16 genes and eight protein biomarkers, and then uses machine-based learning to analyze the data.

Seiden told Medscape Medical News that even though CancerSeek is also a blood-based test, it evaluates both DNA and protein, and the DNA technique is very different from the one used by the Grail test. In addition, he noted: "our study population is larger, more exhaustive in its overall analysis of circulating DNA, and includes a larger collection of different cancers."

"One additional attribute of our test was the ability to accurately identify the most likely tissue of origin for more than 50 cancer types, which in practice would help provide an efficient strategy for evaluating patients with a positive test," Seiden added.

Anne Marie Lennon, MB, PhD, professor of medicine at Johns Hopkins University, Baltimore, Maryland, and an author of the CancerSEEK study, noted it is "exciting to see that there is now so much interest in developing methods for the earlier detection of cancer."

Both tests are designed for early detection of multiple cancer types, but have distinct differences, she noted. "CancerSEEK is designed to detect mutations in cancer driver genes derived from neoplastic cells," she said. "Mutations are attractive for analysis as they are stable and exquisitely specific for cancer, and signals from noncancerous blood cells can be easily excluded in these assays."

She also pointed out that CancerSEEK includes a small number of highly validated protein biomarkers to complement the DNA analysis, whereas the current article evaluated methylation patterns of DNA rather than mutations. "There is no specific methylated site that is specific for cancer," Lennon explained. "Techniques that rely on methylation therefore require recognition of patterns of changes rather than individual changes, in turn requiring mathematical algorithms to distinguish them from those of normal cells."

Both approaches are promising and the two tests could be complementary, she suggested. "All the patients studied by us, as well as in this study, had previously diagnosed cancers. The critical next step for both approaches will be to perform a prospective, interventional trial on a cohort of average-risk individuals who are not known to have cancer prior to their analysis," she said. "This is the only way to determine whether the benefits of blood tests outweigh their risks."

The study was funded by Grail, manufacturer of the test. Seiden has reported financial relationships with McKesson, Grail, Merck, and Bristol Myers Squibb; several coauthors also reported relationships with industry. Lennon has reported no relevant financial relationships.

Ann Oncol. Published online March 31. Full text 

For more from Medscape Oncology, join us on Twitter and Facebook.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.