Skip to Main Content

Gut Check is a periodic look at health claims made by studies, newsmakers, or conventional wisdom. We ask: Should you believe this?

The claim:

A blood test can detect DNA associated with nasopharyngeal cancer in seemingly healthy people, leading to earlier diagnosis and saving lives, researchers in Hong Kong reported on Wednesday in the New England Journal of Medicine.

Tell me more:

Diagnosing tumors via a simple blood test has become a holy grail (a company developing such “liquid biopsies” is even named Grail). Much of the research is aimed at figuring out how to detect and analyze “circulating tumor DNA,” genetic material that’s released from dead cancer cells, something scientists first noticed in 1977. Circulating tumor DNA is rare compared to other DNA in the blood, however, so although there are intense efforts underway to detect ever-lower concentrations of it, the Hong Kong scientists looked for something else: DNA from Epstein-Barr, a virus that can cause nasopharyngeal cancer.

That improved their odds. Every nasopharyngeal tumor cell carries 50 copies of the Epstein-Barr genome, and the specific sequence the researchers fished out and amplified occurs about 10 times in every viral genome. That meant 500 targets per tumor cell, compared to one for the human DNA in most cancer cells.


Chemical engineer Dennis Lo of Chinese University of Hong Kong (he also co-founded Cirina, a company that merged with Grail earlier this year) and his colleagues gave the blood test to 20,174 Chinese men, ages 40 to 62. (This demographic group has a higher incidence of the cancer than women or younger men.) The scientists found Epstein-Barr DNA in 1,112 of the men on the first test, which cost $30. Since the virus is common (and usually does not cause cancer), the researchers screened this group again four weeks later. This time 309 had the viral DNA.

Examining the men with an endoscope (at a cost of $80) and a $1,000 MRI, the researchers determined that 34 had nasopharyngeal cancer. That’s 11 percent of those who tested positive. (In contrast, about 30 percent of women in whom mammography finds what seems to be breast cancer in fact have it.) Only one man who tested negative on the DNA test wound up with nasopharyngeal cancer during the nearly three years the participants were followed.


Sixteen of the 34 cancers were very early-stage, which contrasts with the usual high percentage of nasopharyngeal tumors that are detected only once they’ve advanced. (This cancer produces no symptoms in its early stages.)

That made a difference: After three years, all but one of the patients in the study were alive with no sign of their cancer advancing, after undergoing standard radiation therapy. That compares to 70 percent of nasopharyngeal cancer patients in general who survive at least that long.

“Lives have been saved because of this screening,” Dr. Richard Ambinder of Johns Hopkins University, a specialist in Epstein-Barr cancers, wrote in an editorial accompanying the study.

To detect one case of nasopharyngeal cancer, Lo and his colleagues calculated, 593 people would have to be screened at a cost of $28,600. As cancer screenings go, that’s pretty good. (About 1,000 men have to undergo annual PSA screening for 10 years for one life to be saved from prostate cancer, for instance.)


Less good is that the study “had a high rate of false positives,” Ambinder told STAT: Only 34 of 309 men who twice were found to have “cancer” DNA actually had cancer. “Most people with a positive test aren’t going to have cancer,” he said.

Worse, the Hong Kong results might understate the number of such false positives if this blood test were used elsewhere. Nasopharyngeal cancer is relatively common in southern China, at up to 35 cases per 100,000 people. The U.S. has less than one case per 100,000. The lower the incidence of nasopharyngeal cancer, Ambinder explained, the more likely it is that the Epstein-Barr DNA being detected “isn’t related to a tumor at all, but to other conditions” such as a plain EBV infection.

The new study “made a compelling case that you can detect early-stage nasopharyngeal cancer in a high-risk population in a relatively inexpensive way,” Ambinder said. “On the other hand, it’s not something that is going to be used in most of the world.”

Nor is it clear if the men who tested negative for virus DNA were really free of nasopharyngeal cancer, pointed out Dr. H. Gilbert Welch of the Dartmouth Institute for Health Policy and Clinical Practice. Only those who tested positive got endoscopies or MRI, and this cancer can stay asymptomatic for years. By not testing men whose blood tests seemed to put them in the clear, he said, the scientists don’t really know if the test missed some cancers.

Another concern is that this study might not reveal much about the benefits of detecting tumor DNA from cancers not associated with viruses — which is most of them. Doing that requires not only detecting those very rare bits of tumor DNA in blood (500 times rarer than the virus DNA in nasopharyngeal cancer) but also figuring out what it means. Viral DNA is viral DNA. Human DNA might look like it comes from a tumor even when it doesn’t. “Most cancers aren’t caused by viruses,” said Welch. “I don’t think we can generalize much from this.”

The verdict:

Detecting nasopharyngeal cancer by detecting virus DNA is the low-hanging fruit of liquid biopsies, so while it’s good news that it worked in Chinese men, these results cannot be extrapolated to other cancers and other populations. 

  • Yes, well said: The key is the notion of “Low Hanging Fruit”. The blood biomarker fetishists among us cling on “smoking guns” such a VIRAL DNA because that way one eliminates the big big source of false positive findings due to DNA from normal non-cancerous cells (which can also be mutated).
    Of course this works only if the virus is highly efficient in causing a tumor otherwise you will also have also false positives –e.g. in the tumor free carriers in which the virus has not triggered a neoplasia. This is appears to be the case for this EBV/NPC connection. Note that EBV and nasopharyngeal cancer is one of the “best” scenarios for demonstrating proof of principle of liquid biopsies. Virally induced cancer are relatively rare but quite much emphasized for historical reasons (Varmus having been a leader of NCI, hype of immunotherapy, etc).

    But a problem for liquid biopsy that is much wider and not addressed here is the inherently low rate at which the tumor sheds revealing (mutated) DNA traces into the blood. At best, we have the smoking gun phenomenon: Yes, IF there is unambiguous tumor material in the blood, great – that almost proves the presence of a tumor. (But even EBV DNA was not enough!) But if not, we CANNOT exclude anything. Smoking guns are rare. So liquid biopsies will tend to have a high positive predictive value (in the best case of high prevalence populations), but almost always very low negative predictive value (absence of evidence is not evidence of absence).

Comments are closed.