麻豆传媒

THEY were some of the most vulnerable patients in the hospital: newborn babies, many of them premature, or with other problems that meant they needed to stay in the special care baby unit at Addenbrooke鈥檚 Hospital in Cambridge, UK. The infants were hooked up to machines via a tangle of tubes and wires that supplied oxygen, food and fluids and monitored their vital signs to keep them safe. Yet now they faced a new hazard: hospital superbug MRSA.

This form of the bacterium, Staphylococcus aureus, can be a deadly menace thanks to a mutation that gives it resistance to the main antibiotic used to treat it, methicillin. In mid-2011, MRSA was found on the skin of three infants in the special care baby unit at Addenbrooke鈥檚. While its presence wasn鈥檛 yet making the babies ill, the bacterium could find its way into their bodies and trigger a life-threatening infection.

The staff had to take action. The babies were lifted from their cribs and their bodies carefully washed all over. Even the insides of their nostrils were gently dabbed with antimicrobial lotion.

While this was going on, the ward was closed for several days, and everything in it thoroughly disinfected, from the ventilation machines down to the thermometers. 鈥淎 deep clean is very disruptive,鈥� says Est茅e T枚r枚k, an infectious disease specialist at Addenbrooke鈥檚. 鈥淚t鈥檚 not something we take lightly.鈥�

Four days later, the doctors learned that their efforts had been in vain: the superbug had been found on yet another baby鈥檚 skin. Where could it be coming from? If they didn鈥檛 find out soon, babies鈥� lives would be in jeopardy.

When details of the cases landed on the desk of several miles down the road, he felt he could help. Harris鈥檚 team had recently reported on a way that genetic testing could be used to investigate just this kind of outbreak ().

Their paper had described how the team investigated an MRSA outbreak in a hospital in Thailand, several years earlier, using stored samples from patients. By sequencing bacterial DNA from the samples, they had tracked how the infection had spread through the hospital, from patient to patient.

Harris had pointed out that, given the staggering improvements in the speed and cost of the technology involved, it was now possible to a sequence a bacterium鈥檚 entire genome 鈥� its full set of genes 鈥� overnight. Genome sequencing could be used as a tool to investigate a live outbreak, he argued, and maybe even halt it.

Now the team had the chance to put their plan into action at a hospital on their own doorstep. Armed with a desktop DNA sequencing machine, Harris was about to test his mettle as a germ detective.

The first time microbial genetics was used to investigate the spread of disease was in 1990. Kimberly Bergalis, a 21-year-old from Florida had developed AIDS although she鈥檇 never had sex or injected drugs. When HIV experts at the US Centers for Disease Control (CDC) heard about the case, they were alarmed that there might be a new way for the virus to spread so they dispatched two investigators to her house to interview her. Yet it was Bergalis鈥檚 mother, a local healthcare worker, who came up with the best lead: she discovered that her daughter鈥檚 dentist had pneumonia, which can be associated with AIDS. It turned out that he, too, was HIV positive.

Of course, that was no proof he was the source of Bergalis鈥檚 infection. To investigate further, the CDC used DNA sequencing, then in its infancy, to compare a single HIV gene in samples taken from Bergalis, her dentist, and other infected people in the US.

Viruses are constantly mutating and evolving as they pass from person to person, which provides a way to track their spread: the more genetically related two people鈥檚 viruses are, the more likely it is that one person caught the virus from the other. In this case, the samples from Bergalis and her dentist were more closely related than those from the general population, suggesting that the dentist had given Bergalis HIV, or vice versa. For the CDC that discovery confirmed that there was no new route for transmitting what was then an inevitably fatal infection.

Genetic testing has since been used in many other cases of HIV transmission, often where the transmitter is accused of acting recklessly. The ethics of prosecuting such cases may be debatable 鈥� critics say this approach could put people off getting tested for HIV. But as the technology has advanced, genetic testing of microbes has found wider uses.

While most technologies get cheaper and better over time, the rate of progress of DNA sequencing has outstripped even advances in computing (see diagram). The first draft of the human genome, for instance, took 13 years to complete and cost $3 billion. Today, anyone can have their personal genome sequenced in 24 hours, for a few thousand dollars.

Viral and bacterial genomes, which have far less DNA than our own, are even cheaper to sequence than those of humans. That makes it feasible to start sequencing samples from many patients and getting the results back overnight. And that, Harris had realised, could allow real-time tracking of how an infection spreads.

Disease outbreaks are normally investigated by sending out staff to the place where people are getting sick and grilling patients face to face about where they鈥檝e been and who they鈥檝e been in contact with. For this reason, it鈥檚 often called shoe-leather epidemiology. Doctors are looking for links in time and place between people infected by the microbe in question. 鈥淚t鈥檚 hugely laborious and expensive,鈥� says , who studies viral evolution at the University of Oxford.

Genome sequencing, on the other hand, should be faster as well as more accurate. After all, connecting two cases because both people ate at the same restaurant one night is what police call circumstantial evidence; connecting them because their microbes are genetically related provides more certainty.

When the first cases of swine flu in people broke out in Mexico in early 2009, Pybus was one of a team of virologists that used genomic sequencing to track the evolution of the virus. 鈥淚t was the first major outbreak of the post-genomic era,鈥� he says. 鈥淟arge numbers of viral genomes were generated and shared online in real time as the pandemic unfolded.鈥�

In the case of swine flu, the disease鈥檚 origins were already known: it had crossed over to humans from pigs and was now being transmitted between people. But genomic sequencing could also be used to work out how fast flu was spreading between people, and so whether it had the potential to trigger a global flu pandemic. This was vital information for doctors around the world.

鈥淕enomic sequencing was used to work out whether the new strain of flu had the potential to trigger a global pandemic鈥�

The team calculated that each case of swine flu was being , a relatively low figure compared with the 1918 flu pandemic, say. And the figure fitted broadly with the estimates of the shoe-leather epidemiologists: they had come up with a rate of between 1.4 and 1.6 transmissions per case. 鈥淭hey were independent sources of info, and they corroborated each other, which is strong evidence you鈥檝e got it right,鈥� Pybus says. 鈥淲e proved it was possible to do the genetic stuff in real time alongside the traditional analysis.鈥�

But as the outbreak of MRSA at Addenbrooke鈥檚 was about to show, genome sequencing sometimes comes up with different answers to traditional methods.

Compared with viruses, bacteria are much bigger beasts, both physically and in terms of the amount of DNA in their genomes. HIV, for example, has nine genes, while MRSA has about 2800. But in the past couple of years, overnight sequencing of even a bacterial genome has come within our grasp.

It only costs about 拢95 to sequence the complete genome of one person鈥檚 MRSA sample, for instance. Even if dozens of samples need testing, that is small change compared with the cost of an outbreak that might keep patients in hospital for weeks, perhaps needing intensive care.

Widening the net

When Harris was called in to help, the first thing he did was to ask for samples from any previous cases of MRSA that had cropped up in the special care baby unit over the past six months. There had been 17 such cases, appearing in three distinct clusters that the hospital鈥檚 microbiologists had concluded were unconnected. But all they鈥檇 had to go on was the standard method of taking samples from patients, growing them in the lab, and testing them to see which antibiotics they are resistant to. The pattern of antibiotic resistance can indicate how related different samples are, but it is not infallible.

Indeed, as it turned out, in this case the standard method had given a wrong steer. The more accurate genome-sequencing method showed that all but three of the cases were related, Harris was able to tell the doctors.

But there was a problem: the timing of the infections didn鈥檛 make sense. 鈥淵ou鈥檇 expect to have an infected person who overlaps with another infected person,鈥� says Harris. 鈥淏ut here we had three clusters, separated by 17 and 33 days. A month is quite a long time to have no cases and for it to reappear.鈥�

Something else had to be going on. So the team widened its net, searching for other cases of MRSA within the hospital and local community, and found another 10 cases that Harris showed were genetically related.

Suspiciously, all had a link to the special babies ward at the hospital. Five of the cases were new mums with abscesses in their breasts. One needed surgery because the infection was out of control. Others had turned up at their local doctor鈥檚 surgery with painful pustules on their legs or ears. A bit of digging 鈥� back to shoe-leather epidemiology again 鈥� showed that almost all had some connection to the special babies unit, for instance, having been in the maternity ward at the same time as parents of babies in the unit.

For a while it all went quiet, with no further cases of MRSA at Addenbrooke鈥檚 for 64 days. But then alarm bells rang again: another baby in the special babies unit tested positive. The investigation had transformed from a retrospective study into a real-time one.

鈥淎larm bells rang when another baby tested positive for MRSA, changing the study into a live investigation鈥�

Bacteria taken from the baby were sequenced overnight. 鈥淏y 10 am the following morning I had analysed the data and could confirm to the hospital that the new case was linked to the outbreak,鈥� says Harris.

As it seemed unlikely that the bacteria had survived the deep clean and then persisted on hospital equipment for two months, the evidence pointed to the same MRSA strain being repeatedly introduced into the hospital by a healthy carrier of the bacteria. That was plausible as nearly three people in every 100 have MRSA on their skin without suffering ill effects.

If a healthcare worker was carrying the bug, the situation would have to be handled sensitively to avoid apportioning blame. But with the genetic evidence so strong, Addenbrooke鈥檚 now felt able to ask its staff if they would volunteer for testing. Of the 152 staff who came forward, one tested positive for MRSA that was genetically related to the other cases. Sure enough, it was someone who frequently visited the special baby unit.

Fortunately this was a simple problem to address. The staff member, who has remained anonymous, was prescribed an antibacterial body wash and antimicrobial cream to apply inside their nostrils. Soon they were given the all clear and could resume their duties.

That was in 2012 and since then, no one else at Addenbrooke鈥檚 has tested positive for that MRSA strain again. 鈥淣ot only did genome sequencing demonstrate an outbreak more effectively than standard infection control 鈥� it identified a carrier,鈥� says Sharon Peacock of the University of Cambridge, another member of the team. 鈥淚t was actionable information. If you can detect it early you can nip it in the bud and prevent further infections.鈥�

Deadly pneumonia

Other groups are starting to use bacterial genome sequencing too. Around the same time as the Addenbrooke鈥檚 investigation, a similar approach was being used to stop a deadly outbreak of antibiotic-resistant pneumonia at a hospital in Bethesda, Maryland, which killed 11 people.

The technique is also being used to investigate the emergence of new pathogens from animals. It has shown, for instance, that Danish farmers infected with a new form of MRSA were catching it from their sheep (). 鈥淭he advantage is that you can identify the bacteria or virus down to the individual,鈥� says at the University of Cambridge, who led that research. 鈥淚t鈥檚 the fine detail.鈥� It may even be able to help clarify whether badgers in the UK really are a major source of tuberculosis infection in cattle.

Genome sequencing is unlikely to replace shoe-leather epidemiology as a way of investigating novel disease outbreaks, but it will increasingly be applied alongside it. The technique might even help predict the emergence of dangerous new pathogens before they start killing people. 鈥淪hould we be scanning for outbreaks by sequencing water from a sewage plant?鈥� Pybus wonders. 鈥淵ou鈥檝e got the viral population of a city being excreted into wastewater.鈥�

That鈥檚 something for public health doctors to consider in future. In the meantime, microbes continue their silent evolution within our bodies, biding their time until the opportunity to jump to another person presents itself. Inevitably it will, triggering another outbreak of disease.

This time though, the germ detectives will be waiting for them.

This article appeared in print under the headline 鈥淭he germ detectives鈥�