So far, next-generation sequencers have mostly served as research tools. But could low-cost, high-throughput sequencing enter diagnostics someday?
454 Life Sciences, in collaboration with outside researchers, has been exploring using its sequencing technology to predict viral drug resistance and determine genetic predisposition to cancers.
Scientists from Yale University and Myriad Genetics presented initial results from their separate studies with 454 at the recent Advances in Genome Biology and Technology conference in Marco Island, Fla.
One of these collaborators is Michael Kozal, an associate professor of medicine in the AIDS program at Yale University School of Medicine, who with 454 is testing the ability of the company’s sequencing technology to discover low-level drug-resistance mutations in the AIDS virus, and to predict treatment outcome.
According to Kozal, whether 454 sequencing will eventually become a diagnostic assay is unclear. “It’s really driven by the field,” he told In Sequence this week. “What’s the easiest [assay] to use and gives the best results? And also, cost comes into play.”
Gene sequencing, he said, might improve prediction of HIV drug resistance. HIV mutates frequently, and sometimes a point mutation is enough to create a drug-resistant strain. More than 80 HIV mutations are known to be associated with resistance, Kozal said.
Fifteen percent of HIV-infected patients that were never treated with anti-retroviral drugs harbor some drug-resistant strains, a study by researchers from the US Centers for Disease Control and Prevention has shown.
“There is very strong evidence within the field that if you can detect these minor variants, they do predict subsequent treatment response,” Kozal said. “But most of the studies have only looked for a few mutations, and not sequenced the whole genes.”
Several genotyping methods, including capillary electrophoresis Sanger sequencing, the TruGene HIV-1 test from Bayer Diagnostics, and ViroSeq from Celera Diagnostics and Abbott Laboratories, can detect resistance mutations. But they fail to “find the needle in the haystack” — mutations that are only present in a small percentage of the viruses a patient carries, Kozal said. Many diagnostic companies are using methods like allele-specific RT-PCR, for example, to develop assays that can, he said.
However, these assays, which test for specific mutations, might be difficult to scale up to cover more than 80 mutations. “You might be able to, but it would be pretty laborious,” Kozal said.
The ideal assay would detect all mutations with high sensitivity. “The best thing for a patient would be to detect all the resistant variants that are there so we can actually tailor the therapy, so it can suppress all the variants,” Kozal said. “That’s the goal.”
In its study with Kozal, 454 Life Sciences has been sequencing the protease and reverse transcriptase genes of HIV-1 in patient samples provided by Kozal and his team. So far, the researchers have sequenced 92 out of 400 samples, taken from HIV-infected patients before they started antiretroviral treatment. They sequenced eight PCR amplicons in both directions at 3X coverage and compared the sequences to known resistance mutations.
The researchers were able to pick up mutations that were only present in 1 percent of the virus population, but the sensitivity of the assay might increase with greater coverage, he said.
454’s technology also detected significantly more resistance mutations than Sanger sequencing: While the Genome Sequencer detected such mutations in about a third of the samples, a Sanger-based study found mutations in only about 10 percent of samples, Kozal said. Based on a different database that includes more resistance mutations, 454 found such mutations in almost half the samples, he added.
Once the sequencing is done, the researchers will correlate the results with patient data to see if knowing the resistance mutations can predict clinical outcome. The researchers hope to complete their study by early April.
It’s too early to say, though, how 454’s technology will fare compared to other assays. “We are very optimistic because of the preliminary data, but we have to see what [the study] shows,” Kozal said. “We don’t know until the results are in.”
If the study is successful, the next step will be to apply the technology to assay patients that have failed therapy, Kozal said. There might be other applications for it, he added, for example for assessing drug-resistance mutations in hepatitis B and hepatitis C virus.
In a second study, researchers from 454 Life Sciences and Myriad Genetics have evaluated 454’s Genome Sequencer FLX for determining a predisposition to genetic disease. In particular, they asked whether the technology is both reliable and cost-effective enough for diagnostic sequencing.
Myriad has a long history in cancer predisposition testing: The company has been offering sequence-based tests since 1996 for hereditary forms of melanoma, breast, ovarian, colon, pancreatic, and endometrial cancers, and colon polyps. Those tests all use capillary-based Sanger sequencing.
“The best thing for a patient would be to detect all the resistant variants that are there so we can actually tailor the therapy, so it can suppress all the variants.”
To assess 454’s technology, the researchers compared it to Myriad’s Colaris test for hereditary colorectal and endometrial cancer, which involves sequencing of three genes, MLH1, MSH2, and MSH6, which are involved in the majority of hereditary nonpolyposis colorectal cancers.
Mark Skolnick, chief scientific officer at Myriad Genetics, presented the results of the study at the AGBT conference last month.
The researchers chose 16 DNA samples from individuals who had been characterized by Myriad’s Colaris test and were known to contain about 50 single-base substitutions as well as a small number of insertions and deletions.
They PCR-amplified 61 amplicons in each of these samples using Myriad’s primers with 454’s adaptors. After pooling all PCR products from one sample, they subjected them to emulsion PCR and sequenced them on 454’s Genome Sequencer FLX.
In the first four samples, the instrument correctly identified all positions of the known sequence, he said, including a C-T change and a four-base repeat. Two errors, which occurred only once each in an example Skolnick showed, resulted from the PCR reaction. He called these “isolated events” that were “probably related to the 454 process” and “not hard to pick out.”
454’s known inaccuracies in sequencing homopolymer regions does not represent a major problem if researchers oversample sufficiently and use statistics, Skolnick explained. For example, in a region with a C8 repeat, two-thirds of the reads gave the correct answer, while the remainder called the stretch as a C7, C9, or C10 repeat.
The researchers also assessed whether the technology could detect heterozygous changes in the sequence and found it could not achieve this at 20X coverage with 100 percent certainty but will require approximately 50X coverage.
Skolnick concluded that the accuracy of the FLX data means it “could be used for diagnostic testing.”
However, Myriad has not yet determined whether to use 454’s technology to that end, Skolnick told In Sequence by e-mail. The technology “could be in a year” cheaper than CE-based sequencing, he wrote, though the ability to bar-code samples “needs development.”
Myriad is in the process of purchasing a 454 instrument, but testing Illumina’s and ABI’s next-gen sequencing platforms is “under discussion,” he wrote.