NEW YORK – A SNP in a genomic region used to target SARS viruses can potentially impact non-diagnostic results from some commercial molecular diagnostic tests for COVID-19 infection. The issue highlights the need for multi-target tests and raises concerns about the industry practice of keeping primer and probe chemistries under wraps.
As described in the Journal of Clinical Microbiology this month, researchers at the University of Liège in Belgium discovered a C-to-U transition at position 26,340 of the SARS-CoV-2 genome that was associated with failure of the Roche Cobas SARS-CoV-2 E-gene qRT-PCR in eight patients.
The researchers discovered the SNP is in the envelope, or E gene. They presume the SNP caused the E gene detection failures, but primer and probe information is considered proprietary across the industry, so they can't test this experimentally.
The E gene is common to all Sarbecoviruses, and Roche's test targets the E gene as well as an open reading frame gene, ORF1ab, that is specific to SARS-CoV-2. Only the E gene target failed in patients infected with SNP-containing virus, and since this is the pan-SARS target the failure did not impact the diagnostic results specific for SARS-CoV-2.
In addition to the Roche assay, an analysis of the US Food and Drug website shows that a total of 25 Emergency Use Authorized tests incorporate primers and probes to detect the E gene. It is not known whether the E gene SNP would lead to failures in other commercial tests targeting the region. The researchers found that the SNP did not impact E gene results for four samples retested with a non-commercial assay that includes an E gene target.
The SARS-CoV-2 genome is small, with a few distinct genes and non-coding regions, and primers and probes designed for qPCR of the reverse-transcribed virus tend to target a few specific genomic regions.
The nucleoplasmid gene, or N gene, was targeted by the original EUA test from the US Centers for Disease Control, and many assays focus on that gene.
The pan-SARS E gene target was incorporated into the first assay for the SARS-CoV-2 virus — a test sometimes referred to as the Charité test, or Corman et al. — that was quickly developed by an international team of researchers and published in late January. The E gene is common to other coronaviruses as well, and using a pan-Sarbecovirus approach was thought to offer some advantages. Meanwhile, the first published clinical evaluation of a SARS-CoV-2 diagnostic used an assay from researchers in Hong Kong incorporating the N and ORF1ab genes.
Since these foundational assays debuted, others have chosen to target a gene for RNA-dependent RNA polymerase, or RdRp, and a few assays target the gene for the SARS-CoV-2 spike protein, otherwise known as the S gene. Some assays also target multiple regions within a single genomic area, for example the N1, N2, and N3 targets of the original CDC assay.
Overall, assays with EUAs tend to incorporate primers and probes targeting N, ORF1ab, RdRp, S, and E regions, either singly or in combinations of two or three.
At of the end of July, the FDA's website listed 118 individual EUAs for SARS-CoV-2 molecular diagnostic assays that are not collection kits or sequencing-based tests.
A search of the instructions for use of these tests reveals that 61 of them target a single genomic region of the virus — 43 of these assays target the N gene alone, while 12 target the ORF1ab alone.
Three assays — the Abbott IDNow, an assay from Co-Diagnostics, and one from Access Genetics — target RdRp alone. Assays from Quidel target a region of the SARS-CoV-2 genome the company calls the non-structural polyprotein pp1ab, while a test from Applied DNA Sciences exclusively targets the S gene.
Of the remaining 57 tests targeting multiple genomic regions, 39 target two regions and 18 target three regions.
Most multi-region tests are combinations of the previously described genomic regions, with the exception of a test from BioFire that also targets a region the company calls the membrane protein, or M gene, in addition to the S gene.
Because it is a pan-SARS target, the E gene target is never used singly, but rather in combination with SARS-CoV-2-specific targets like N, ORF1ab, RdRp, or S genes.
In total, 25 assays include an E gene target.
Specifically, among the two-target tests with E gene targets, three assays also target the N gene of SARS-CoV-2, three target ORF1ab, seven aim to detect the RdRp gene, and one test — from Altona Diagnostics — targets the S gene in addition to the E gene.
There are also 11 three-target assays with an E gene component. Six target N and ORF1ab in addition to E, while five target the N and RdRp genes as well as the E gene.
The 25 assays with E gene targets that could be potentially be affected by a SNP in the gene include two-target tests from Cephied and SD Biosensor, and three-target tests from Color Genomics, DiaCarta, Luminex, and Seegene.
Proprietary pandemic probes
The precise formulation of primer and probe sets in molecular diagnostics is almost always proprietary, so whether the E gene SNP of concern is contained in the region targeted by other company's reagents is hard to know until failures are reported.
At least in Cepheid's case, David Persing, chief medical and science officer, said the firm has already tested a limited number of samples containing the E gene SNP and has not seen an impact on test sensitivity. "Both the E and N2 genes are detected at expected levels," he said.
Still, experts are beginning to raise concerns, and wonder if it could be worth reconsidering keeping target sequence information confidential during a viral pandemic.
Keith Durkin, a researcher and co-author on the JCM study, said the team searched viral genomes deposited on the sequencing database GISAID and found the particular mutation has arisen independently at least four times.
Durkin said his team has been rapidly sequencing SARS-CoV-2 viral samples using the Oxford Nanopore MinIon and the Artic Network pipeline. When Marie-Pierre Hayette, the head of clinical microbiology at the University Hospital of Liège, noticed samples negative for the E gene qPCR using the Roche test, these were prioritized for inclusion in the sequencing work, Durkin said.
In the global context, this particular E gene SNP seems to be rare, with a prevalence of 0.09 percent, although there appear to be some local clusters where the frequency is higher, Durkin said. He and his coauthors suggest the study results demonstrate the need for two regions of the SARS-CoV-2 viral genome to be targeted by assays, in order to avoid false negatives due to mutations.
The lab reached out to Roche to confirm that the SNP it uncovered was part of the target for the firm's E gene primers and probes, but the company was not able to disclose the specific sequence information, Durkin said.
Representatives at Roche have emphasized that the assay performed as designed and yielded accurate results, and the firm's dual-target design managed mutations as it was designed to do.
Dmitriy Kosarikov, director of development for Roche Molecular Diagnostics, said the JCM analysis "demonstrates the power and necessity of dual target design."
Alexandra Valsamakis, chief medical officer at Roche Molecular Diagnostics, said the company has not received any reports of false-negative SARS-CoV-2 diagnoses due to the presence of SNPs that could have impaired viral RNA detectability with the Cobas SARS-CoV-2 assay.
"The test performed exactly as intended due to its dual-target design," she said, adding, "The final results for all patients were accurate: SARS-CoV-2 RNA detected."
As such, there are no adaptations of the test or change in use that would be necessary, but Valsamakis suggested end users need to have an understanding of the result interpretation table in the assay's instructions for use.
Valsamakis also confirmed that the JCM study authors communicated their findings to Roche prior to publication, but emphasized that Roche does not divulge proprietary oligonucleotide designs.
Indeed, Durkin said that Roche was "quite cooperative and relatively open," but it would not release the precise position of the primers and probes.
Generally, the components that go into diagnostic tests might be a bit of a "black box" due to commercial considerations. "In the current context, I think this is unfortunate," Durkin said, specifically because there is an unprecedented SARS-CoV-2 sequencing effort going on worldwide, with more than 75,000 SARS-CoV-2 genomes available on GISAID.
"With a knowledge of the primer/probes in commonly used assays, it would be possible to more rapidly identify potentially problematic mutations," Durkin argued. This is particularly true in the context of rapidly available sequence information, he also said. Using the MinIon, his lab could likely get a consensus genome sequence in as little as one day.
If primer and probe sequences are available, it is indeed possible to do the sleuthing.
As part of an attempt to understand differences in sensitivity between primer-probe sets of RT-qPCR assays that were developed early on during the pandemic, researchers at Yale University characterized four common non-commercial diagnostic assays. They also calculated the accumulated genetic diversity from 992 available SARS-CoV-2 genomes, and compared that to the primer and probe binding regions.
In a study published in MedRixv in April, they reported 12 primer-probe nucleotide mismatches that had occurred in at least two of the SARS-CoV-2 genomes assessed.
Chantal Vogels, the lead author on the study commented in an email that that mismatches can definitely have a negative impact on sensitivity of molecular assays.
"It is very important to monitor SNPs in primer and probe binding regions as these can indeed reduce the performance of the assay," she said, adding that mismatches in the 3' end can be particularly detrimental.
"GISAID is performing in silico analyses for primer and probe mismatches on a regular basis, which is very important," Vogel said. In addition, clinical investigations of these SNPs, as was described in the JCM paper, are important to determine what the effect of a mismatch is on the outcome of the assay.
The fact that the positions of the primers and probes used in the Roche Cobas assay have not been disclosed is "problematic," Vogels said, as it is important to monitor for possible SNPs that can lead to decreased performance of assays.
Alex Greninger, a virology researcher at the University of Washington, concurred in an email that while it is rare for SNPs to impact sensitivity, "it definitely happens." For example, a study he co-authored in February showed herpes simplex virus mistyping due to HSV-1 and HSV-2 interspecies recombination. The team found 14 FDA-approved commercial HSV typing assays, with information on the target genes available for 11. Seven of these targeted at least one HSV gene where the HSV-1 x HSV-2 recombination events occurred.
Although the standard protocol is for developers to track sequence changes for diagnostics and report to the FDA, having "a distributed method, like academia" focus on the problem would be much better, Greninger argued.