Skip to main content
Premium Trial:

Request an Annual Quote

Duke-Led Team Validates Gene Expression Assay for Host Viral Infection Response


A team led by scientists from Duke University School of Medicine has developed and validated an RT-PCR-based assay that can gauge a specific host gene expression signature from a patient blood sample to determine whether a respiratory infection is caused by a virus or other infectious organism.

The researchers are also working on a similar assay to detect bacterial infections. As such, they hope to eventually develop a clinical assay that physicians could use as soon as a patient presents with symptoms of a respiratory infection in order to determine whether the culprit is viral or bacterial in nature and thus prescribe an appropriate treatment.

"What we envision from a platform perspective is having all the probes for both viral infection and bacterial infection on the same platform, so you get a readout telling you it's no infection, co-infection, or one or the other, viral or bacterial," Geoffrey Ginsburg, a co-principal investigator on the study and director of genomic medicine at the Duke Institute for Genome Sciences and Policy, told PCR Insider this week. "I'm quite confident we'll be able to get there."

The researchers originally outlined their host gene expression approach to diagnosing infectious disease in a pair of papers published in January in PLOS One, work that was covered by GenomeWeb Daily News.

In one study, the researchers used microarray analysis to identify a blood-based transcription signature in patients with influenza H1N1 or H3N2 infections; while in the other they profiled gene expression patterns in blood samples from mice and humans infected with Staphylococcus aureus bacteria.

The researchers had also previously shown that microarray profiling could accurately classify individuals with respect to several other respiratory infectious agents, such as rhinovirus and respiratory syncytial virus — findings that have been reproduced and validated by other groups.

The goal of the most recent study was to develop an RT-PCR assay based on what the researchers dubbed the "acute respiratory viral factor" and develop and validate a predictive algorithm that would allow one to estimate the likelihood that a symptomatic patient has a respiratory viral infection.

In order to do this, Ginsburg and colleagues chose to develop their assay on Life Technologies' TaqMan low-density (384-well) arrays running on an ABI 7900 HT Fast Real-Time PCR system, work they described in a paper published this week in Science Translational Medicine.

In their study, the researchers tested eight samples per array card. To select genes for representation on the card, they compared top-ranking genes from their previously defined acute respiratory viral factor to a list of existing exonic primer pairs, of which 29 were available.

Their final array card comprised 48 genes: the aforementioned 29 genes; three control genes; seven genes that were shown to be temporally downregulated in time course analysis of their H3N2 gene expression data; and nine additional genes randomly selected from the original H3N2 data set that displayed no differential expression as possible additional controls.

They validated their test on 102 individuals arriving with symptoms of respiratory infection at Duke University Medical Center and Henry Ford Medical Center in Detroit, as part of the Community Acquired Pneumonia & Sepsis Outcome Diagnostics (CAPSOD) study, a prospective trial funded by the National Institutes of Health to develop novel diagnostic and prognostic tests for severe sepsis and community-acquired pneumonia.

Of the 102 patients, 28 had a viral infection, 39 had a bacterial infection, and 35 were healthy controls. The TaqMan array-based assay was able to correctly classify these patients with a sensitivity of 89 percent and a specificity of 94 percent.

"We view this as sort of the proof of concept that we can take the data from an array, configure an assay on a platform that could be used in the clinic — maybe not exactly this configuration — and begin to model what that would look like," Ginsburg said. "We took samples that came from emergency room patients, and we were … able to detect viral infection as opposed to other causes of infection — primarily bacterial."

Ginsburg added that the group had been using the ABI 7900 HT Fast system and low-density array cards for several years, but realized that it was a good proxy for modeling a potential clinical assay.

"Right now our RT-PCR device is not set up in a CLIA environment, it's a research lab, but we treat it as if it were in a [CLIA-approved] environment," he said. "We wanted to begin to replicate what it would be like to do a clinical assay from sample acquisition to return of results. The reason for the 384-well configuration is that we wanted to be able to move as many samples as possible through the machine, and just try to optimize cost of goods, et cetera."

The ABI 7900 HT Fast system is not approved for in vitro diagnostic use, but a successor to that system, the ABI 7500 Fast Dx Real-Time PCR instrument, does have FDA clearance for IVD use.

Ginsburg said that the group now plans to conduct additional validation studies of its test in other populations and settings, eventually leading to a clinical utility study "where we present the results to a provider [who makes] a treatment decision based on those results, and then [assesses] the outcome. That study would be critical for ultimate approval or adoption by the clinical community, although I think some doctors might use [the assay] right away even without that study."

However, in order to create such an assay for clinical use, especially in an acute care setting, the scientists believe that they first need to identify a testing platform with a faster turnaround time than PCR-based systems such as the ABI 7900 HT Fast, Ginsburg said.

"I don't think that this is ultimately going to be a PCR-based test, at least in all settings," Ginsburg said. "One can imagine in an emergency department, where a turnaround time of an hour is desirable, this platform is not going to work, particularly with some of the times required for sample prep and RNA purification."

However, Ginsburg noted that a platform similar to the one used in their study could work in certain scenarios that don't demand such a fast turnaround time — for instance, a physician's office, where "you see your doc, they run this test, give you a prescription for antibiotics, but tell you not to fill it until results come back tomorrow."

Further, Ginsburg said that the group envisions their test being used to help track potential pandemics, or any other situation where large-scale screening of a population might be appropriate.

"I'm sure people are working on one-hour PCRs that could give rapid turnaround time in the emergency room setting," Ginsburg said. "I'm not eliminating PCR as a method of choice, it's just that right now there are limitations to that. Publishing papers like this should motivate the technology developers to move in our direction."

In fact, he noted that Duke has a strong bioengineering program in which several groups are working on point-of-care nucleic acid testing. "And we've been working very closely with a number of groups on new technology to measure RNA that aren't actually transcriptase-based, and actually may not even require labeling," he added.

Duke has applied for a patent on the respiratory viral gene signature. Ginsburg said that the patent includes claims specifying that a variety of platforms could be used to run the assay, as well as claims detailing the notion of a "viral detection score, something along those lines, like some of the RT-qPCR clinical assays like [Genomic Health's] Oncotype DX. We'll try to come up with some sort of visualization tool for clinicians so they can make better decisions about what the PCR results mean."