LISBON, Portugal – Cepheid is developing an assay for its flagship GeneXpert platform to differentiate bacterial from viral infections in whole blood samples, company scientists and clinical collaborators said at the European Congress of Clinical Microbiology and Infectious Diseases (ECCMID) held here this week.
The prototype assay integrates sample preparation and RT-PCR to measure a six-mRNA host signature comprising three mRNAs upregulated in bacterial infections, two upregulated in viral infections, and one control.
To develop the assay, Cepheid scientists culled through more than 30 peer-reviewed publications and evaluated potential biomarkers based on their proposed intended use and validation status, then winnowed down that list to a subset of 21 mRNA targets based on their association with infection status in published microarray datasets.
The researchers then validated this smaller list of targets in a training cohort of 230 whole-blood samples collected from April 2017 to February 2020 from six different sites. Samples were collected from patients who had a fever or were otherwise suspected of harboring infections, as well as from healthy controls.
Using further machine learning methods and statistical analyses, Cepheid developed the six-mRNA signature and designed a prototype GeneXpert assay to measure the targets in EDTA blood, both venous and capillary.
As part of the diagnostic microbiology track at ECCMID, Matthew Faron, an assistant professor of pathology at the Medical College of Wisconsin and a practicing pathologist at Froedtert Hospital in Milwaukee, presented data from an early evaluation of the assay, noting the dire need for commercial diagnostic tests that can quickly and accurately discriminate between viral and bacterial infections.
Faron said that patients currently presenting with fever of an unknown origin are a dilemma for physicians, as the risk of not quickly treating with antibiotics is huge. Currently, it takes 24 to 48 hours for rapid microbial identification and antimicrobial susceptibility testing, but if a patient has sepsis, for instance, mortality rates increase exponentially over time, often hours. "We're still relying on blood culture-positive even with newer rapid tests," Faron said.
On the other hand, treating everyone, even people with a viral infection, with broad-spectrum antibiotics runs the risk of contributing to the massive antimicrobial resistance problem.
To evaluate the prototype assay, Faron's group retrospectively tested 314 samples from patients with known viral or bacterial infections, generating 307 valid test results consisting of 204 bacterial infections, 100 viral infections, one fungal infection, and two co-infections. Unsurprisingly, a large percentage of the viral infections were SARS-CoV-2. Comparison diagnoses were made using a variety of test types including blood culture, BioMérieux's BioFire Diagnostics respiratory panel testing, COVID-19/flu PCR testing, and urine culture.
The test uses a similar workflow to other commercial Xpert tests: capillary or venous blood is collected in EDTA tubes for anticoagulation, and 200 μL of the blood is mixed with the Xpert lysis reagent. Then, 1 mL of blood/lysis mix is placed into an Xpert cartridge, which is scanned and placed in a GeneXpert system. The total assay time from sample to result is approximately 35 minutes.
Faron said that his group was initially concerned about the potential degradation of the mRNA biomarkers in blood samples in the time it takes to collect a sample and prepare it for analysis on the GeneXpert. However, in a separate study the group confirmed that RNA showed high stability in EDTA blood.
Faron reported that the assay demonstrated an area under the receiver operating characteristic curve of 0.88 for distinguishing bacterial from non-bacterial and viral from non-viral infections, which in general is considered excellent performance for a diagnostic test. However, Faron also stressed that the data is very early and has caveats.
"The goal was to get some strong data that Cepheid could work with," Faron noted, adding that one of the early limitations of their study is that they cherry-picked a lot of known infections.
Nevertheless, Cepheid has already begun a multi-site study with the intent of increasing the sample size and variability to include samples from consenting patients presenting with noninfectious febrile illness, Faron said.
An audience member at Faron's talk asked what the presentation of results might look like for a future commercial test, with Faron again noting that it's a bit early to consider that. He added that it might be possible to create infection severity scores, but that some physicians might prefer to just see a yes/no answer as to whether an infection is bacterial or viral. "This will probably change by [the] time it’s ready to go into clinical trials," Faron said.
Cepheid is far from the first company to develop a blood test to try and distinguish between bacterial and viral infections — Israeli company MeMed has a CE-marked and US Food and Drug Administration-approved immune-response assay called MeMed BV that uses three soluble proteins to determine the nature of a patient's immune response to infection. In fact, Faron also presented a poster at ECCMID validating the MeMed test, demonstrating that it had strong accuracy and reproducibility when used on frozen contrived samples provided by MeMed. This test is also fast, taking only about 15 minutes to perform.
Regarding the potential strengths and weaknesses of a protein test like MeMed's and mRNA test like Cepheid's, Faron said in a post-presentation interview that there isn't enough solid data yet on either to make such a comparison. "They're two unique approaches and more data is probably necessary to understand in which avenues a protein approach may be more beneficial and when an RNA approach might be more useful," Faron said. "There's honestly space for both of them."
Companies like Inflammatix, Cytovale, and Presymptom Health are also prepping molecular host-response tests for market, with a much more singular focus on early detection of sepsis.
But the allure of a test from a company like Cepheid is that it may be much easier to implement in hospital labs — assuming its development stays on course and it gets regulatory approval — because GeneXpert platforms are widely used. Parent company Danaher reported in January that Cepheid placed approximately 10,000 new GeneXpert instruments in 2021, reaching a total installed base of more than 40,000 instruments.