From the Journals

New multi-analyte blood test shows promise in screening for several common solid tumors


Key clinical point: New blood test demonstrates ability to identify presence of eight common cancers.

Major finding: CancerSEEK demonstrated a mean sensitivity of 70% for the eight cancer types and a specificity of greater than 99%.

Data source: Retrospective study of 1,005 patients with known malignancy and 812 healthy controls.

Disclosures: The study was funded by multiple sources including grants from the National Institutes of Health. The authors reported several disclosures involving diagnostics and pharmaceutical companies.

Source: Cohen JD et al. Science 2018 Jan 18. doi: 10.1126/science.aar3247.

View on the News

What are the clinically relevant questions answered by this test?

Molecular panels are here to stay – and the GI community will in some shape or form be impacted, be it in performing diagnostic procedures on test-positive patients, or risk-stratifying patients prior to testing.

The conceptual challenge is that it is not about what any given test measures – various panels use separate combination of markers from epigenetics to DNA mutations as well as whole or truncated proteins – but how well a specific test with its somewhat arbitrarily chosen components and cutoffs performs. And, more importantly, what the clinical implications of positive or negative test results are. And no one knows that. At least for now.

A recent report in Science from a group from the Ludwig Center for Cancer Genetics at Johns Hopkins proposes a new cancer blood test based on a very systematic and thoughtful approach to include select mutations in cell-free DNA and circulating proteins associated with various solid organ tumors. For validation, they used healthy and advanced but nonmetastatic cancer cohorts. Through stringent controls and a series of validations, the authors present a range of sensitivities for the various cancer types with an impressive specificity. This is a technically very strong approach with many nifty and thoughtful additions to give this test a very promising first foray – did anybody watch CNN?

While not ready for prime time, which is a tall order for a first report, the authors dutifully point out the need for a prospective real life cohort validation. In the meantime, regardless of the outcome of this particular test, it is a repeated reminder that we need to stay abreast of the advances and the details of each molecular test, especially with a likely very diverse and distinct group of tests to choose from.

Dr. Barbara Jung, the Thomas J. Layden Endowed Professor and chief of the division of gastroenterology and hepatology, University of Chicago
Dr. Barbara Jung

Many of us will be part of interpreting results and determining further management. Just as with hereditary cancer genetic panel testing, our technical ability may have stretched beyond our ability to fully understand the implications. Many questions will arise: What about true false positives? False negatives? Intervals? Can such tests replace other screening? How to choose any given test over the other? Should tests be combined or alternated? The tests will be technically refined and are here to stay – we need to get to work on finding answers to the clinically relevant questions.

Barbara Jung, MD, AGAF, is the Thomas J. Layden Endowed Professor and chief of the division of gastroenterology and hepatology, University of Chicago.



Imagine a single blood test that would cost less than $500 and could screen for at least eight cancer types.

It’s early days for the technology, called CancerSEEK, but the test had a sensitivity of 69%-98%, depending on the cancer type, and a specificity of 99% in a cohort of 1,005 patients with stage I-III cancers and 850 healthy controls, wrote Joshua D. Cohen of the Ludwig Center for Cancer Genetics and Therapeutics at Johns Hopkins University, Baltimore, and his colleagues. The report was published in Science.


CancerSEEK tests for mutations in 2,001 genomic positions and eight proteins. The researchers examined a 61-amplicon panel with each amplicon analyzing an average of 33 base pairs within a gene. They theorized the test could detect between 41% and 95% of the cancers in the Catalog of Somatic Mutations in Cancer dataset. They next used multiplex-PCR techniques to minimize errors associated with large sequencing and identified protein biomarkers for early stage cancers that may not release detectable ctDNA.

The researchers used the technology to examine blood samples from 1,005 patients with stage I (20%), stage II (49%), or stage III (31%) cancers of the ovary, liver, stomach, pancreas, esophagus, colorectum, lung, or breast prior to undergoing neoadjuvant chemotherapy. Participants had a median age of 64 years (range of 22-93 years). The healthy controls did not have a history of cancer, chronic kidney disease, autoimmune disease, or high-grade dysplasia.

Next Article: