But while scientists need the actual virus in order to check a diagnostic test, they only need its genetic code in order to build one. Chinese health
But while scientists need the actual virus in order to check a diagnostic test, they only need its genetic code in order to build one. Chinese health authorities released a draft genome of the virus in early January, which enabled labs like the CDC to develop diagnostic tests based on a standard technology known as reverse-transcriptase polymerase chain reaction, or RT-PCR.
This kind of test requires designing small pieces of DNA that match sections of the viral genome that are distinctive from other coronaviruses—like the ones that cause SARS, MERS, and the common cold—but stable enough that they’re not going to mutate. These small sections, called primers, grab onto any viral DNA that happens to be around. (Coronaviruses are actually made up of RNA, not DNA, so you need the reverse-transcriptase to make the programming language platform compatible.) Then a DNA-building enzyme makes a bunch of copies. To test whether pieces of virus they’re interested in are actually in the sample, technicians add a dye that fluoresces in the presence of DNA. The more viral bits there are, the brighter it glows, creating a pattern of light that tells the technicians whether they’ve found 2019-nCoV.
But to double-check how well the test performs, now the scientists need those actual living specimens of the pathogen. Boehme’s group at FIND is using both viral isolates obtained from the CDC and samples from Europe to evaluate the CDC’s test and others like it. That work is ongoing, she says. They’ll also evaluate any commercial tests that may be in development—all-inclusive kits that can be run at a clinic or hospital and don’t require a big lab filled with highly trained staff, as RT-PCR tests do. So far, more than 30 companies have reached out to FIND about potential products they hope to turn out quickly.
In the US, at least two startups are developing Crispr-based diagnostics that could deliver a diagnosis in less than two hours without any complex instrumentation. (Think paper swabs + paper strip, like a pregnancy test for coronavirus.) Mammoth Biosciences in San Francisco and Sherlock Biosciences in Cambridge, Massachusetts, both confirmed to WIRED they are working on such tests. Mammoth has not yet applied for an emergency use authorization with the FDA and is waiting on samples from the CDC and state public health department to validate the accuracy of its test. A spokesperson for Sherlock declined to comment on whether the company will seek a fast-tracked authorization.
In addition to a test that could detect 2019-nCoV directly, Mammoth is also working on a version that would home in on and identify a patient’s immune responses to the virus. The problem with RT-PCR is that it can pick up viral genetic material only if there’s a lot of it floating around. And early in an infection, before someone starts to feel really sick, there’s often not enough RNA to flag the test. Nancy Messonnier, the CDC’s coronavirus boss, pointed this out this week to reporters, saying that the health agency’s test is accurate among people who look sick. But using it as a screening tool for asymptomatic people is less useful. “A negative test most likely means a person is not infected,” she said. “However, it may mean that an infection has not developed enough to be detected by the test.”
A test like the one Mammoth is developing could potentially pick up on some signature of infection during the period of time when the virus isn’t yet rapidly replicating, but the body’s immune system has already detected it. Charlies Chiu, an infectious disease doctor at the University of San Francisco who is advising Mammoth, says this approach has been shown to work for influenza. He says the company’s goal is to offer a complementary test that might help fill that gap and, by offering a test for people who aren’t visibly ill, potentially cut down on the more draconian quarantine measures the US implemented over the weekend.