This article was originally posted Dec. 23.
Using real-time reverse transcriptase PCR to detect influenza A and its subtypes may be more complicated than previously thought, according to two papers published online ahead of print this week in the Journal of Clinical Microbiology.
One study, led by Xiaotian Zheng at Children's Memorial Hospital in Chicago, describes the discovery of a variant of the 2009 H1N1 influenza virus that the ProFlu+ test marketed by Gen-Probe's Prodesse business failed to flag as positive for influenza A.
The other study, led by Bobbi Pritt at the Mayo Clinic, found an "unexpected" amount of mutability in the matrix gene of the 2009 H1N1 virus, which led to "loss of viral subtype discriminatory ability using [rRT-PCR-based melting temperature] analysis within just three months of the pandemic."
Currently, rRT-PCR-based influenza tests are considered the gold standard for flu screening. The US Centers for Disease Control and Prevention has recommended rRT-PCR-based tests over rapid antigen testing for 2009 H1N1, largely because the antigen-based tests pose a higher risk of false negatives.
The recent findings published in JCM, however, indicate that clinicians may need to take a bit more care when interpreting the results of rRT-PCR-based tests.
"Our rules are that we treat the PCR results as final, or more or less confirmed," Zheng told PCR Insider. "They overrule all other results, including rapid tests, which are known to be much less accurate than culture or PCR."
Nevertheless, his team's work "underscores the need for cautious interpretation and additional testing when a negative RT-PCR result does not seem to fit clinical presentation," according to the paper.
Zheng's study, conducted in collaboration with researchers from the Cleveland Clinic, Evanston Hospital, and Gen-Probe's Prodesse group, was sparked when a sample from a 4-year-old girl tested "strongly positive" for influenza A on a rapid antigen test, but negative on the ProFlu+ test, an assay cleared by the Food and Drug Administration to detect and differentiate influenzas A and B and respiratory syncytial virus.
After two other rapid antigen tests yielded positive results, the researchers ran the girl's sample on Prodesse's ProFlu-ST — a sub-typing rRT-PCR test that differentiates the 2009 H1N1, seasonal H1, and seasonal H3 viruses — and found that the sample was positive for 2009 H1N1.
This discrepancy poses a potential problem for clinical use, the authors noted in the paper, because many labs perform a subtyping assay only when the ProFlu+ test is positive for influenza A. "Therefore, a negative result by this assay would likely miss the diagnosis of 2009 H1N1 influenza unless other assays, such as virus culture, are performed."
After confirming the results with culture, as well as a second run with both ProFlu+ and ProFlu-ST, the researchers ran the sample on flu assays from other manufacturers, including a PCR-based test from EraGen that is available for research use only; the FDA-approved Luminex xTag Respiratory Viral Panel; and the CDC's RT-PCR assay. The sample tested positive for influenza A on all other platforms, indicating that the issue was only with the ProFlu+ test.
Zheng said that he and his colleagues hypothesized that "maybe the organism sequence changed … in the primer- and probe-binding regions." This could have affected the results of the ProFlu+ test, but not those of the other providers, since "different assays designed by different manufacturers target slightly different regions, all within the matrix gene."
Zheng said that scientists from Prodesse and Gen-Probe "worked very closely with us and were very supportive and very open" in looking further into the discrepancy.
The Prodesse researchers sequenced the target gene, "and it turned out that there is a single nucleotide difference within the probe-binding region that might have caused the total abolishment of the reaction," Zheng said.
Steve Visuri, director of R&D for Prodesse products at Gen-Probe, told PCR Insider via e-mail that the company "determined that there was indeed a rare mutation in the matrix gene of this single patient specimen."
He noted that influenza A is known to mutate actively, but said that based on the company's findings so far, "the specific mutation identified appears to be very rare and well-confined."
[ pagebreak ]
Zheng stressed that the prevalence of the variant is still unknown, and that his group plans to expand the study by retesting patient samples that previously tested negative with the ProFlu+ assay. "Hopefully, we can get a better idea by screening those with ProFlu-ST and see if any of them were missed," he said.
It's likely that the variant is rare, but the clinical implications could be considerable if it's found to be widespread, Zheng said. For instance, he said that his lab tested around 20 samples per day with the ProFlu+ during the peak of the flu season in the fall, but noted that larger groups like the Cleveland Clinic use the test to screen between 100 and 200 samples per day.
Visuri noted that among "several hundred thousand patients tested with our ProFlu+ product, this is the first example of a 'false negative' resulting from an underlying mutation." He said that this may be because the virus matrix gene, the target of the assay, "is highly conserved and less subject to evolutionary pressure than other genes."
For now, Zheng said, physicians should be aware that for patient samples screened using this assay, "if the result is negative there may be a need to verify with a different assay." He cited the CDC RT-PCR assay, the Luminex test, and the Eragen test as valid options.
He said that his lab is also considering Nanosphere's Verigene SP system, which received FDA clearance in October for a test for influenzas A and B and RSV.
"Our plan is to test that assay and see if it can pick this up or not," Zheng said.
Visuri said that Gen-Probe is working with Zheng and his team to further investigate the finding, "as we do with all our customers, to ensure that our products remain robust in the face of a constantly evolving virus."
He added that because RT-PCR assays "are increasingly seen as the gold standard for highly sensitive flu detection, developers of these tests should stay in close contact with their customers, especially during a pandemic."
Matrix Gene Mutability
The Mayo Clinic, meantime, found that its ability to subtype the H1N1 virus via melting temperature analysis was stymied due to its high mutability only three months into the pandemic.
In the spring, Pritt and her colleagues determined that they could use their internally developed rRT-PCR-based influenza A test — which targets the matrix gene with primers and fluorescent resonance energy transfer probes — to distinguish 2009 H1N1, seasonal H1N1, and seasonal H3N2. Each subtype had "discrete, reproducible [melting temperature] ranges," making them easy to identify based on their melting temperatures.
Pritt told PCR Insider that until August, "all of our novel H1N1s had a melting temperature of 51 degrees. And it was within a very tight, predefined temperature range."
Beginning Aug. 14, however, "we started seeing some that dropped significantly — they went from 51 degrees down to 47 degrees, or even 44 degrees, so it was almost a 10-degree drop." All of these samples with problematic melting temperatures were later confirmed by the CDC PCR test to be positive for 2009 H1N1.
Seeking to explain the phenomenon, Pritt and her colleagues sequenced the amplicon for 12 H1N1 isolates and found "multiple point mutations" when compared to the original H1N1 sequence published by the World Health Organization in April.
According to Pritt, all of these mutations were in the matrix gene, which was unexpected because that gene is under less selective pressure than the hemagglutinin and neuraminidase genes, which are targeted by the immune system. By comparison, the matrix gene is "supposed to be relatively conserved," she said.
Another finding that the team considered to be surprising was "how quickly the virus has mutated," Pritt said. "For a three-month time period, we were seeing pretty much the same strain of the virus, and then it changed very quickly, just within three months."
Pritt stressed that the Mayo Flu A assay is still effective for detecting influenza A, but the lab can no longer use it for subtyping. As a result, the group is currently evaluating commercial assays, such as Prodesse's ProFlu-ST, for that purpose.
She added that because the Mayo assay relies on FRET probes, which are relatively long, it should be able to detect influenza A even when there are mutations in the target region.
"I would actually be more concerned about technologies that are not as forgiving as ours, like some of the TaqMan-based assays," Pritt said. "They are typically designed to use shorter probes, and if there is a mutation present, the probes may not anneal, and you might not get a signal at all."
Pritt noted that with melting curve analysis, "we still get a positive signal, but then we see our melting temperature shifted, which indicates that there might be a mutation."
With other assays, she said, "you'd never know that it really was positive and it had mutated to the point where you could no longer detect it."