Metagenomics: The Test That Unlocked Covid-19 Now Saves Lives Daily

In late December 2019, a team of virologists from three major Chinese cities began analysing swabs taken from patients hospitalised with mysterious cases of severe pneumonia. All had been present at the Huanan Seafood Market in Wuhan before rapidly falling unwell. But when the virologists ran the samples, they drew a blank: the genetic material did not match any previously discovered pathogen.

The doctors turned to a new screening method known as metagenomic testing. Instead of checking for known microbes, the test could reveal all the microorganisms present in the samples, previously identified or not. It enabled them to see that the pathogen in the patients' samples bore a close resemblance to the coronavirus family, a group that included viruses which had previously caused outbreaks of severe pneumonia in humans, such as Middle East Respiratory Syndrome. The scientists had just discovered Covid-19. Knowing what the pathogen was made it possible to develop Covid diagnostics rapidly to try to curb the spread, and kickstarted the frantic search for treatments and vaccines around the globe.

This remains the best-known use of metagenomics to date. But the test is now increasingly being used in hospitals. From quickly diagnosing sepsis to identifying the causes of unexplained ailments, it is already saving lives. Metagenomics, the study of all the DNA and RNA found in a particular environment, is seen as an increasingly important clinical approach. Metagenomic testing involves extracting genetic material from a sample, then using advanced sequencing methods to identify the genetic codes of everything present. These results can be analysed by computers with reference to large databases, allowing doctors to pinpoint the exact organisms present, such as bacteria or viruses.

Whenever a patient arrives in a neurological ward displaying signs of confusion and drowsiness, and complaining of fever, headache and weakness, doctors are acutely aware that it could be a life-or-death emergency. This constellation of symptoms can indicate meningitis or encephalitis, deadly conditions affecting the brain and spinal cord. Yet doctors often face a critical dilemma. The most suitable treatment would depend on whether the patient's condition was due to a virus or bacteria, or an overreaction from the brain's immune system. With limited tests and time, however, doctors are often forced to hedge their bets. You need to make a decision immediately, says Charles Chiu, a professor and infectious diseases expert at the University of California San Francisco Department of Medicine, who has been carrying out new studies into the use of metagenomic tests.

In some ways, treating patients for both infection and autoimmune disease is contradictory, because it involves suppressing the immune system even though the patient might have an infection. But there is often no other choice. Metagenomics is now helping to inform these decisions. Chiu and his colleagues have developed a pioneering metagenomics test for identifying pathogens within cerebrospinal fluid, the liquid that bathes the brain and spinal cord. The test has already flagged cases where, unbeknownst to doctors, a pathogen was driving a patient's symptoms. For example, Chiu encountered a patient who had been diagnosed with meningitis and liver failure, and was scheduled to receive a liver transplant. But the metagenomics test on her spinal fluid revealed that she was actually infected with the hepatitis E virus. While hepatitis E is a known and treatable cause of liver disease, it is not commonly tested for in high-income countries such as the US. She was given an antiviral medication, she got better, and she was spared the transplant.

Until recently, the use of metagenomics has been limited by the length of time required for the analysis. Separating human and microbial DNA, and identifying the signs of a potentially worrisome pathogen, requires highly specialist skills. Doing the sequencing and the analysis can take several days to a week, says Niranjan Nagarajan, an associate professor at the National University of Singapore. The patient might die before a result is obtained. Now, though, machines can sequence DNA within as little as one to two hours, and AI tools can ingest vast datasets to decipher patterns within them. It is finally making metagenomics more widely accessible.

For people with poorly functioning immune systems, any infection can swiftly turn into a potentially deadly threat. Transplant recipients, cancer patients undergoing immune-suppressing treatment and those living with autoimmune diseases must be on constant high alert. Studies have suggested that immunosuppressed patients who develop sepsis or blood poisoning due to an infection are at a 23 per cent higher risk of dying in hospital. However, making a quick diagnosis in these cases is not always easy. Conditions such as pneumonia or sepsis can be caused by hundreds of potential pathogens, which results in a very long diagnostic odyssey in which test after test must be sent. It can be a race against time.

In Singapore, Nagarajan is leading a project to see whether AI-enhanced metagenomic screening can lead to faster, more targeted treatment for immunocompromised individuals. A typical dataset, says Chiu, is littered with spurious information. This includes the seemingly harmless viruses carried by most people and microbes from the lab that have accidentally contaminated the sample. AI can help by decluttering metagenomic datasets, making it quicker and easier for doctors to get an idea of what is going on. A patient might also be carrying multiple co-infections, only one of which is making them ill, or damage might be related to an autoimmune response and not linked to a pathogen at all. AI can help to pick out which pathogens are causing the damage by analysing how the person's genes are being expressed, such as which proteins they are making. This can be used to ascertain whether a patient's immune system is likely fighting a virus or happily co-existing alongside it. It gives doctors the ability to do precision medicine, enabling them to tailor the optimal treatment.

As data rolls in on just how effective metagenomics can be, it could play an ever-increasing role in hospitals. Last year, a pilot study carried out in two London hospitals looked at whether metagenomics could provide additional information on patients hospitalised with respiratory infections. It found that the tests could identify pathogens which are not routinely detected 30 per cent of the time. In one case, the test saved a patient from having to undergo invasive surgery. The individual had tuberculosis rather than what had been assumed to be lung cancer. The test also identified two cases of the respiratory virus parvovirus B19, months before UK and European agencies issued public health alerts about its increasing prevalence, suggesting a role for metagenomics in giving authorities a heads up on emerging epidemics.

Indeed, Nagarajan says that several labs in Singapore are now running metagenomics tests on sewage and air samples. The data is then passed through AI models trained to look for signs of new viruses, as part of efforts to prevent the next pandemic. Proactive monitoring could have similarly given an early warning about Covid-19. In Italy, an analysis of sewage found that traces of Covid-19 showed up in wastewater months before the country's first confirmed cases.