Speedy Raman Technique Identifies Bacteria Responsible For Infections

When a patient arrives at a hospital with symptoms of a urinary tract infection, a doctor will likely hand her a cup and ask for a urine sample. But after that sample goes to the lab, it will be days before her doctor knows which bacteria are causing the problem. Now a team of German researchers has used Raman spectroscopy to identify the offending microbes in just a couple of hours (Anal. Chem. 2013, DOI: 10.1021/ac401806f). The technique could lead to a quicker clinical test, which would help doctors make more effective decisions about the best antibiotic to prescribe.

The standard laboratory test for diagnosing a UTI involves spreading urine samples out on bacterial growth plates and giving the bacteria time to grow. Technicians then count the bacterial colonies and identify the species.

But culturing takes time—at least 24 hours for slow-growing bacteria. Doctors usually don’t wait for the results and will prescribe a broad-spectrum antibiotic to the patient to stop the infection from spreading. If doctors could more quickly identify the pathogen responsible, says Jürgen Popp of the University of Jena, in Germany, they could choose a more specific antibiotic and limit the spread of antibiotic-resistant bacteria.

Popp developed the UTI test based on his previous work on methods to identify many types of cells using Raman microspectroscopy. With a microscope, he shines a laser beam on a single cell and collects its Raman spectrum—a composite of signals from every molecular vibration in the cell. That spectrum serves as a cell’s fingerprint.

To identify a bacterial species based on that fingerprint, Popp’s team had to train a computer algorithm to detect subtle differences in Raman spectra between species. The researchers built a database of spectra from 11 common UTI pathogens, including Escherichia coli and Enterococcus faecalis, grown under various conditions. They then inputted thousands of spectra into the algorithm, training it to identify the pathogens. After training, their model could correctly identify a bacterial species from a set of known test samples 95% of the time.

The team then tested their model on urine samples from 10 patients. Half of each sample went to a lab for a typical cell culture diagnostic test. Popp’s lab took the other half, isolated the bacterial cells with a centrifuge, washed them, and collected their Raman spectra. In two hours, they determined that seven urine samples contained E. coli and three had E. faecalis. Twenty hours later, the independent laboratory found the same results.

Identifying bacteria by shining laser light on them is an attractively simple approach, says Joseph C. Liao of Stanford University. Yet, he adds, the test currently is not practical for a clinical setting or doctor’s office. The equipment needs to be smaller, and the results need to be in a form that doctors can easily interpret. Popp’s group is actively working on miniaturizing and automating the test.

Guido Schmiemann, a public health researcher at the University of Bremen, in Germany, says the study is a nice first step, but points out that physicians also need to know if a bacterial strain is resistant to certain antibiotics. One of Popp’s colleagues at Jena University Hospital is now developing a separate Raman spectroscopic method to spot these resistant strains.