Photographer: AFP PHOTO/Peter PARKS

Make Way for Better Germ Tests

Steven Salzberg is the Bloomberg Distinguished Professor of Biomedical Engineering, Computer Science and Biostatistics, and the director of the Center for Computational Biology at Johns Hopkins University.
Read More.
a | A

Every year, millions of people see their doctors for relief of some kind of infection -- a cold, a fever, food poisoning or something more serious -- but doctors can rarely pin down the exact cause. In many cases, it’s hard to tell even whether the bug is a bacteria or a virus. That’s a huge difference, akin to not knowing whether the food on your plate is animal or vegetable.

The standard diagnostic technology for identifying microbes, decades old, is simply too slow and imprecise to help most patients. The good news is that scientists have developed far better technology, capable of diagnosing thousands of different infections quickly, and with remarkable precision. But doctors won’t be able to use it until federal regulators grant approval, and unfortunately, this process is about to get much harder, if the U.S. Food and Drug Administration stays on its current course.

People don’t always need to know precisely what’s making them sick. Usually, someone who catches a cold or suffers mild food poisoning doesn’t even visit a doctor, and the problem goes away on its own. But when someone has an infection that the body can’t easily fight off, the results can be deadly. In such cases, to get the right treatment, patients urgently need to identify the pathogen.

A doctor can do this by taking a sample -- a tissue swab, a blood sample or a biopsy, depending on where the infection is located -- and sending it to a lab for testing. Most tests today involve culturing the specimen, which means growing it on a petri dish under controlled conditions, and then testing whatever grows. Pathologists have tests for most of the bacteria and viruses that infect humans, but each of these is designed for one and only one pathogen. Thus doctors often must “guess first, then test.” Might the patient have E. coli? How about strep throat (a Streptococcus bacteria infection) or influenza? Each test takes time and money, and dozens might be needed before the right one is found. This is why doctors rarely bother, unless the patient is gravely ill.

Two dramatic technology breakthroughs of the past decade are set to make things far easier, cheaper and faster, by testing for thousands of infectious organisms at once.

First, there’s DNA sequencing, which has been growing steadily faster. Today’s automated sequencers can generate up to 400 billion bases (the As, Cs, Gs and Ts that make up DNA) per day -- 400,000 times faster than 15 years ago and 75,000 times more cheaply. This means it takes only a few hours to sequence all the DNA in a tissue swab, at a cost of perhaps $100.

The second advance is in the scope of our genome databases. The very first bacterial genome was sequenced only 20 years ago, but already we have more than 30,000 bacterial genomes and 80,000 viruses, plus the genomes of hundreds of other pathogens (such as malaria) caused by microbes that are neither bacterial nor viral. These include every major known human pathogen, with hundreds of different strains for the most common ones.

This means that once we sequence that swab, we can look in our database and figure out precisely what’s in it. In other words, the DNA in a single test can be used to diagnose any of the thousands of possible infections.

In a study published this year, for example, my colleagues and I demonstrated that we could diagnose brain infections in critically ill patients by sequencing DNA directly from biopsies. Even though more than 99 percent of the DNA in these samples was human DNA, the remaining 1 percent provided enough sequence data to implicate a particular organism about one-third of the time. In one case, we found clear evidence of a tuberculosis infection and, after being put on anti-tuberculosis medication, the patient rapidly improved, eventually recovering fully from an illness that would likely have been fatal. Sequencing and diagnosis in this instance took about two days; if this were part of a routine protocol, it could be even faster.

Before DNA-based diagnostic tests can become routine, however, they need to meet the so-called CLIA standards, set by the 1988 Clinical and Laboratory Improvement Amendment and enforced by the Centers for Medicare and Medicaid Services and the FDA. To get this approval, a lab must demonstrate precisely how its test works, and show that it is accurate and reproducible. CLIA certification is time-consuming and expensive, and so far, no major hospital or company has produced an approved protocol using metagenomic sequencing for diagnosis, though their efforts continue.

The trouble is, the FDA is proposing to make the process even more difficult. The agency recently announced that it wants to regulate DNA-sequencing diagnostic tests, something it has never done before. And getting FDA approval is much more onerous than getting a CLIA approval.

The FDA’s proposal comes with other problems, as well, including an old-school approach to diagnosis, in which every pathogen has to have its own separately approved test. That would delay the technology, perhaps by years, and make it more expensive. Five of the largest organizations of infectious disease doctors, pathologists, virologists and microbiologists have recently argued that this would be a mistake. 

Even more alarming is the FDA’s proposal that all labs be required to use the FDA’s own ARGOS database of genome sequences for testing. ARGOS has only a few hundred genomes. My lab’s database, in contrast, has more than 30,000 bacteria and 80,000 viruses. Why ignore the thousands of genomes that the scientific community has made freely available?

If the FDA will refrain from setting up unnecessary obstacles, scientists in both academia and industry can move ahead to develop CLIA-approved tests for DNA diagnostics. Of course, doctors and hospitals would then need to demonstrate to insurance companies that the new tests can save money. Accurate diagnosis of infections has the potential to save thousands of lives per year, but until we have an effective system in widespread use, it will be hard to collect the data to prove that this will also save money.

Meanwhile, though, hundreds of thousands of patients annually suffer from hospital-acquired infections, millions more get infected at home and at work, and only a tiny percentage will ever know what they have. The next time you or someone you care about gets sick, think about how much it would be worth to diagnose exactly what bug is to blame. We have the technology to solve this problem right now.

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

To contact the author of this story:
Steven Salzberg at

To contact the editor responsible for this story:
Mary Duenwald at