Researchers testing a prototype of a Roche next-gen sequencing assay to measure drug resistance in HIV patients recently reported that the test's performance was comparable to Siemen's TruGene assay, which uses real-time PCR and Sanger sequencing and is approved by the US Food and Drug Administration.
Writing in the Journal of Clinical Microbiology, researchers from the Tel-Aviv Sourasky Medical Center in Israel said that Roche's prototype assay "performs at least as well as TruGene and has the advantage of detecting low-abundance drug-resistance mutations undetected by TruGene."
According to Thomas Schinecker, head of Roche's sequencing solutions business, the company is continuing to develop the assay, obtaining "feedback from our many early-access customers around the world." Roche expects to receive CE-IVD approval for the test in mid-2014, which will allow it to market it as a diagnostic in the European Union. He did not comment on plans for FDA approval.
The study "highlights the fact that a lot of the new sequencing technologies have the possibility of disrupting some of the established methods in clinical lab diagnostics for drug resistance testing," Dave O'Connor, an associate professor of pathology and laboratory medicine at the University of Wisconsin-Madison who has also published on an NGS-based HIV drug resistance assay, told Clinical Sequencing News.
The Tel-Aviv researchers tested the assay using Roche's 454 GS Junior system. They retrospectively analyzed samples from 20 HIV patients, nine of whom had been previously diagnosed by TruGene as failing antiretroviral treatment due to the presence of at least one drug-resistant mutation. The remaining 11 patients had also been previously genotyped by TruGene but were treatment naïve.
The Roche assay encompasses around 1 kilobase of sequencing on the HIV protease and reverse transcriptase gene regions — areas where there are known mutations that confer resistance to standard antiretroviral therapy and that are also tested by the TruGene assay. Four amplicons were generated for each sample.
The team used drug-resistant mutation databases from Stanford University and the International Antiviral Society-USA to classify variants as conferring drug resistance.
A total of 179 drug-resistant mutations were detected, 129 of which were found by both assays. The next-gen sequencing assay detected an additional 50 mutations not found by TruGene.
The NGS assay detected all the mutations found by the TruGene assay, all of which were considered to "high-abundance DRMs" because they were present above a threshold of 20 percent.
The 50 mutations detected by only the NGS assay were all present at levels below 20 percent and therefore out of the range of detection of Sanger sequencing.
These low-abundance mutations were detected in all nine drug-treated individuals and in six of the 11 drug-naïve patients.
While the assay performed well, in that it was able to find al the mutations found by TruGene, the main unknown is whether the low-abundance mutations are clinically relevant. Because these mutations are out of the range of detection of all currently approved tests for HIV resistance, there is no evidence about how abundant a mutation has to be before it impacts drug resistance.
"What you do with that additional information is a clinical setting is not clear," said Ron Kagan, director of bioinformatics at Quest Diagnostics. Kagan has helped develop an HIV tropism test that combines Sanger sequencing with next-gen sequencing (CSN 10/10/2012).
"To bring this into broader use, you need a larger study that includes clinical outcome data comparing the next-gen sequencing test with Sanger," he added.
He added that the drug-resistance test is very different from the HIV tropism test that Quest is developing with next-gen sequencing. That test first uses the standard Sanger-based test to evaluate for the presence of X4 virus, of which 30 percent to 40 percent of patients harbor. These patients are then not eligible for the drug Selzentry, because it is only effective in patients with the R5 virus. Patients who are predicted by Sanger sequencing to be R5 tropic then go on to be confirmed with 454 sequencing, since it is more sensitive at detecting minor non-R5 variants.
The tropism test is a "very different assay," said Kagan, "because you're asking a single question." With a drug-resistance assay, on the other hand, there are many possibilities for mutations that confer resistance to different classes of drugs.
He said that while Quest and other diagnostic companies are very interested in next-gen sequencing technology for HIV drug resistance testing, "no one's converted yet because of the question of clinical relevance."
"Until larger studies are done, it might be awhile before anybody adopts this outside of research purposes," he added.
O'Connor agreed that the question of clinical significance is still the main hurdle toward adopting an NGS test. "The existing assays are well-validated and clinicians know how to interpret them," he said.
His group published a study in PLoS One last year, evaluating an NGS assay for HIV drug resistance on the GS Junior.
O'Connor's assay was different from the Roche assay, although conceptually similar, and he said the results were very concordant, in that the assays were able to detect the same mutations found by Sanger sequencing as well as lower-frequency mutations.
O'Connor said that, aside from determining the clinical relevance of these lower-frequency variants, it will also be necessary to determine the "believability" of these variants. Because they can't be confirmed with Sanger sequencing, some other confirmatory technology will be needed to make sure the low-frequency variants aren't simply sequencing errors or other artifacts.
One potential problem with the Roche assay is that the 454 platform is error prone in homopolymeric regions and there are known drug-resistance sites located in homopolymer tracts. He said for this reason, his team is now mostly working on the Illumina MiSeq platform.
The Israeli team also acknowledged that calling variants in homopolymeric regions could be problematic, although in the study they did not assess the error rate of the assay.
For instance, there is a known mutation in the reverse transcriptase gene K65R that is located in a homopolymer region of subtype C viruses. A recent study found that 454 sequencing had an error rate of detection of up to 1.3 percent for this mutation, while others have reported error rates for the same mutation of up to 3.8 percent, the authors wrote.
"We did not take into consideration potential error rates which might be derived from drug resistance mutations located at homopolymer regions of different HIV-1 subtypes, as demonstrated for the K65R mutation in subtype C viruses," they noted. "This issue should be further investigated."
Another consideration, said O'Connor, is how such tests would be adopted in countries with fewer resources than the US and Europe but with a much higher burden of HIV. For instance, he said that he has been working with partners in South Africa and Brazil, and aside from clinical relevance cost is a major factor.
In his PLoS One study, he found that the 454-based assay could be administered at around $20 per sample when 48 samples were multiplexed, or around three to five times less than the cost of Sanger-based tests.
In parts of the world where there is a major HIV burden, there's a "huge uptake of drugs, but not access to resistance testing," he said. It's possible that these countries "will jump over the Sanger-based technology and begin with next-gen sequencing," as long as the bioinformatics challenges of interpretation can be worked out and the clinical utility of low frequency variants can be determined, he said.