Skip to main content
Premium Trial:

Request an Annual Quote

Team Describes Early Detection Test for Nasopharyngeal Carcinoma; Grail to Commercialize This Year


SAN FRANCISCO (GenomeWeb) – A cell-free DNA sequencing approach targeting the Epstein Barr virus can identify individuals likely to have nasopharyngeal carcinoma, according to a new study by researchers from the Chinese University of Hong Kong. The team, led by Dennis Lo, described the method in a study published this week in the Proceedings of the National Academy of the Sciences.

The work improves on a previous study published by Lo's group that used PCR to detect Epstein Barr viral DNA to refer asymptomatic individuals for further testing, and the researchers are collaborating with cancer diagnostics startup Grail to commercialize the test.

Charlotte Arnold, head of corporate communications at Grail, said that the company is currently validating its version of the assay, which will use both qPCR and NGS to make sure that the performance is comparable to what the CUHK team reported in the PNAS study. As it previously said, Grail plans to launch the test in Hong Kong this year and will then expand to other countries in Southeast Asia that also have similar prevalence rates for nasopharyngeal carcinoma. Arnold did not disclose details about Grail's plans for commercializing the test, such as the anticipated cost and whether the company would launch it as a general population screening test or would initially target individuals thought to be at a higher risk of disease.  She said it would provide more details closer to the launch.

The prevalence of nasopharyngeal carcinoma is higher in Southeast Asia than other parts of the world, around 35 cases per 100,000. Five-year survival rate of the disease is as high as 95 percent when it is detected early but drops to 60 percent if detected at later stage. However, most patients are asymptomatic, and so 80 percent are diagnosed with advanced disease.

Lo's team published a 20,000-patient study last year in the New England Journal of Medicine describing a PCR-based assay that screened asymptomatic individuals for circulating DNA from the Epstein Barr virus, which is associated with nasopharyngeal carcinoma, and detected cancer in 34 individuals.

However, that version of the test had two main problems: it had a low positive predictive value and it required individuals who were positive for the virus to return for a second screen four weeks later.

Approximately 5 percent of the general population will have detectable amounts of Epstein Barr viral DNA in their plasma, Lo said, even though not all of those individuals also have nasopharyngeal carcinoma. Doing a second screen about a month later helps to weed out the individuals for whom viral DNA is only transiently detectable, reducing false positives around fourfold, Lo said. Only those who are still positive on a second test are referred for additional diagnostic testing such as MRI or endoscopy.

But the requirement for a second round of testing "created a number of logistical difficulties for large-scale implementation," Lo said.

"If you are trying to use this approach in actual clinical practice, and you tell a person that their blood test is positive and to come back four weeks later, that person will be very worried and many will prefer to do the endoscopy anyway," Lo said. "That defeats the purpose."

As such, he said, since describing the PCR technique, the team had been looking for ways to improve the test's specificity and eliminate the requirement for follow-up testing. Previously, Lo's group and other research groups had demonstrated that there are differences in cell-free DNA in plasma based on its origins and whether or not it is cancer-derived. For instance, Lo's team previously found that circulating tumor DNA in individuals with hepatocellular carcinoma is shorter than circulating cell-free DNA in those without cancer and has found that size differences can help distinguish fetal-derived circulating cell-free DNA from maternal DNA. Others have also found similar size differences between ctDNA and background cell-free DNA. University of Washington spinout Bellwether Bio is working on commercializing technology for cancer diagnostics based on fragmentation patterns in cell-free DNA that the team had used to determine tissue of origin. 

Based on these previous studies, the CUHK team hypothesized that there could be molecular differences between viral DNA from individuals with nasopharyngeal carcinoma and those without.

In this week's study, the researchers made use of the previously analyzed 20,000 samples. Looking first at the individuals for whom the PCR-based assay identified Epstein Barr virus on either both blood tests or just the first screen, the researchers calculated that those who were ultimately diagnosed with cancer had a higher concentration of plasma viral DNA than those who did not have cancer.

Next, the researchers looked to see whether there were differences in the molecular profiles of the viral DNA between those with and without cancer. To do this, they first analyzed 10 patients with nasopharyngeal carcinoma and 40 cancer-free individuals who were positive for viral DNA after the first or both tests.

They used a targeted sequencing approach to target the Epstein Barr virus, generating around 70 million mapped reads per sample. Sequencing confirmed that those with cancer had a higher proportion of viral DNA than those without. In addition, they found size differences between the viral DNA in individuals with and without cancer.

Most circulating autosomal DNA fragments tend to be around 160 base pairs long. Previous studies evaluating circulating tumor DNA has found that the average length is slightly shorter. In this study, the researchers found that circulating viral DNA in cancer-free individuals was even shorter, while the viral DNA fragments from individuals with cancer was slightly shorter than autosomal DNA, but longer than viral DNA from cancer-free individuals.

The researchers hypothesized that the viral DNA detected in individuals with cancer originated from cancer cells within a tumor. In those cases, the viral DNA had likely integrated with the genome, and so was bound to a histone.

Based on previous work, Lo said that circulating DNA has an average size of around 160 to 170 bases because a piece of DNA wrapped around a histone core is around 140 bases, with another 20-base linker region. In the previous liver cancer study, the team demonstrated that ctDNA did not have that 20-base linker region.

By contrast, viral DNA in cancer-free individuals had not integrated into the genome. Lo said that in these cases, plasma viral DNA is "likely released from viral replication."

The team used those observed size differences to come up with a ratio that compared the proportion of viral DNA that fell within a certain size range to the proportion of autosomal DNA that fell within that size range. They then validated their analysis on the remaining individuals who did have cancer, 232 individuals without cancer but who had circulating viral DNA, and an additional 31 individuals with nasopharyngeal carcinoma from a different cohort.

Overall, when they combined the count-based analysis to determine the concentration of viral DNA and the sized-based analysis, the targeted sequencing test had a sensitivity of 97.1 percent, a specificity of 99.3 percent, and a PPV of 19.6 percent, an improvement over the PCR assay which had a PPV of 11 percent and required sampling at two time points. The PCR assay when used at a single time point had a PPV of just 3.1 percent.

However, one advantage of a PCR-based approach is its low cost. In the NEJM study, for instance, researchers reported that the cost of testing was just $30, as compared to $80 and $1,000 for endoscopic tests and MRI, respectively.

For a commercial test, one way to keep costs down is to combine PCR analysis and NGS in a two-tier test that can still be performed on one blood draw at a single time point. The initial PCR test can be used to screen out nearly 95 percent of the tested population. And then only the 5.5 percent of blood samples for which an initial PCR test detects EBV will go on to be sequenced. This is the strategy that Grail plans to use for the commercial test.

Lo said that detecting circulating viral DNA could have implications for other virus-associated cancers. For instance, he said, his team is also studying cancers associated with human papilloma virus, like cervical cancer and head and neck cancer. "It's early days," he said of that work, but "basically we are looking at a similar approach to profile HPV in plasma."