CAP TODAY Pathology/Laboratory Medicine/Laboratory Management
Dr. Shadt
In addition to preparing a 300-gene panel for clinical use, Dr. Eric Schadt at Mount Sinai has used PacBio sequencers in several medical discovery investigations. In one, he was part of a group that showed that the oncogene FLT3 was a key driver of AML and that the lack of efficacy observed in clinical trials for FLT3 inhibitors was a consequence of mutations in FLT3 that were selected for as a result of the treatment (Smith CC, et al. Nature. 2012;485:260–263). In that case, Dr. Schadt says, the PacBio’s ability to identify phasing mutations (which of the mutations are co-occuring on the same chromosome or along the same stretch of a given gene) was a key advantage. “In other papers we have demonstrated PacBio’s ability to detect structural variations,” Dr. Schadt adds (Bashir A, et al. Nat Biotechnol. 2012;30:701–707; Rasko DA, et al. N Engl J Med. 2011; 365:709–717).
NGS for determining the vaginal microbiome in clinical samples
While next-generation sequencing is often used to detect oncogenes in cancer and genetic variants in inherited disorders, it can also provide clinically valuable microbiological information. Ulf Gyllensten, MD, professor of immunology, genetics, and pathology at Uppsala University in Sweden, and his colleagues have used both the Ion Proton (Life Technologies) and the Pacific Bioscience (PacBio) RSII instruments to study the vaginal microbiome in Swedish and South African women, with focus on human papillomavirus genotypes.
“Our results ended up being something other than we thought they would be,” Dr. Gyllensten tells CAP TODAY. They found not only HPV but also many co-infections.
“Both of these instruments enable high-throughput screening of vaginal biofilm,” Dr. Gyllensten said at the Advances in Genome Biology and Technology conference. In particular, he emphasized the utility of the PacBio RSII, which has not been used much in this setting. “The PacBio generated single reads of entire viral genomes, providing rapid and unambiguous pathogen identification,” Dr. Gyllensten said. “The rapid turnaround time makes this methodology suitable for high-throughput pathogen screening in clinical samples.”
Dr. Gyllensten and his colleagues do NGS through the National Genomics Infrastructure, a Swedish facility established as a joint activity of Uppsala and Stockholm universities. Initially, they performed whole genome sequencing of samples from hospitalized patients with unknown multiresistant bacteria, where results are needed rapidly to treat the patient and prevent nosocomial spread. With Ion Proton sequencing, medically actionable results can be provided by day four, Dr. Gyllensten says. For instance, they identified an organism that was not resistant to methicillin but had the mecA gene missing. Also, they found plasmids carrying beta-lactamase genes.
“We are doing sequencing and submitting these results to clinicians,” Dr. Gyllensten says. In Sweden, “If I set up a test and validate it, I can use it,” he explains. “I don’t need CLIA-type certification.”
When they introduced the PacBio platform into this clinical setting, it provided even more rapid, three-day delivery of genome sequences, Dr. Gyllensten says. “Its simplified bioinformatics allow complete bacterial sequence assembly in one contig with a complete genome.”
Next they used NGS to analyze vaginal biofilms. Screening for HPV by real-time PCR was introduced in Sweden for some segments of the population in 2011, but the assay delineates only a limited number of HPV genotypes. To get a more detailed picture of the vaginal biofilm, the Swedish investigators used NGS to analyze the vaginal flora of Swedish women and HIV-positive South African women on antiretroviral therapy. “We need information about co-infections,” Dr. Gyllensten said. This is particularly true for HIV-positive women because infection with HIV increases the risk of acquiring secondary viral and bacterial infections, and methods are needed to determine the spectrum of co-infections for proper treatment. Sequencing of the entire vaginal microbiome met this need.
After identifying known viruses and bacteria by comparison to reference databases, remaining unmapped reads revealed many novel HPV genotypes in these patient samples, as well as plasmids. About twice as many HPV types were identified by NGS as by rtPCR, including two novel types (Ameur A, et al. Sci Rep. 2014;4:4398). “The pattern of co-infections varied dramatically, with each woman having a unique spectrum of viral, bacterial, and parasitic co-infections,” Dr. Gyllensten said.
Looking specifically at the performance of the PacBio instrument, Dr. Gyllensten noted that it provides complete full-length HPV genomes in a single read, which aids in genome assembly and annotation. In addition, it allows visualization of the evolution of recombination between viral genotypes. Its main drawback is that its throughput is much smaller than that of Ion Proton and Illumina sequencers.
In general, NGS is “quite an effective tool” for screening the vaginal flora, Dr. Gyllensten says.
He is thinking now of using NGS to identify organisms in blood infections, particularly sepsis. It would provide a more rapid turnaround time than even current molecular methods. “That would be really satisfying,” he says.
—William Check, PhD
Dr. Schadt disputes the belief of some that PacBio machines don’t have sufficient throughput for clinical work. A targeted panel with a few hundred genes, where mainly exons are being sequenced, covers about a megabase of sequence. “We are getting between 500 megabases to one gigabase of sequence data per PacBio SMRT Cell,” Dr. Schadt says, “so enough to cover that megabase of DNA about 1,000 times, enough to reliably detect somatic variants.” With his laboratory’s workflow, Dr. Schadt estimates that with an appropriate bar-coding strategy, he could run as many as 80 samples a day per machine for a panel in the range of 10 genes, or 16 patient samples a day for a panel with a few hundred genes. “The PacBio SMRT technology does not achieve the throughput at this point that an Illumina HiSeq 2500 does,” Dr. Schadt acknowledges, “but that mainly limits its utility in sequencing large genomes, such as whole genome sequencing in humans.”
At Memorial Sloan Kettering, Dr. Marc Ladanyi has introduced the panel called MSK-IMPACT, for Integrated Mutational Profiling of Actionable Cancer Targets. It is based on hybrid capture followed by sequencing on Illumina HiSeq 2500s. “We launched IMPACT in the clinical laboratory as a research assay in January,” Dr. Ladanyi says. “Recently we received conditional approval from the New York State Department of Health to run it as a clinical assay.” He says it is now in “soft launch” mode.
Evaluating accurately the yield of these large panels is difficult, Dr. Ladanyi says. “How hard do you look before you go to an NGS panel? Do you send all cases of colorectal cancer or lung cancer that you would otherwise study? Or only a subset that is negative for all the common mutations?”
Also important to point out, he adds, is that some cancers, such as lung, require multiple different assays on the same small tissue sample. Doing separate assays for mutations, copy number alterations, and gene fusions can take a lot of time and use a lot of tissue. In this context, “it becomes attractive to run a single NGS-based assay that can pick up all those changes at once,” he says.
Dr. Ladanyi
Dr. Ladanyi raises two other issues. First, which genes have clear therapeutic implications, either FDA approved or in process? “Obviously that number is very small, maybe 10 to 20 genes that are immediately and widely actionable. When you have a panel with hundreds of genes, the majority are of exploratory use,” he says.
Second, large panels are efficient from a laboratory workflow perspective. “We can have one assay that includes every gene you might possibly want to know about in virtually any solid tumor, so you don’t have to have multiple different separate assays for every histology of cancer,” Dr. Ladanyi says. “Of course it is a very complex assay,” he adds. “But it does allow you to route any cancer into one common workflow.”
Vanderbilt continues to use several small tumor-specific SNaPshot panels—lung, melanoma, breast, colorectal cancer, hematology, and brain. Panels range from three to 11 or 12 genes, Dr. Mia Levy says, with many variants. “We do track the percent of patients with and without mutations,” she adds. In a recent followup of 200 melanoma patients, 64 percent had mutations detected and went on to gene-directed therapy.
Vanderbilt-Ingram’s panels are limited in size for two reasons. “That is what the [SNaPshot] technology allows us,” Dr. Levy says. And “we do bill for these tests. We want to make sure we are billing for clinically actionable results.”
She draws a distinction between two types of institutions. “Do you have a business model that requires you to be financially self-sustaining from day one? Or do you have a large pool of money where testing is done on a research basis and not billed to insurance? Because those large panels are not reimbursable.
“Over the next 12 to 18 months we will see how this will shake out,” she predicts. “When you have to pay on the order of $5,000 for 200-gene panels for each patient, you can’t [be financially self-sustaining]. What people [in that situation] are banking on is that eventually reimbursement will come, and they are trying to ride the wave until that comes. But it is a gamble.”
Dr. Morrison’s approach to oncogene testing at Roswell Park is similar to Dr. Levy’s. “The size of our [NGS] panel is determined by the knowledge database. Our panel today is 23 genes. That is what the databases support. It is like the old lawyer’s axiom: Don’t ask a question if you don’t know the answer.” In pathology parlance: “Don’t test for something if you don’t know what to do with it,” Dr. Morrison translates. “That goes not just for genes, but for specific variants within a gene.”
A somewhat different question, he says, is what payers are going to reimburse for. “Our decision here is to choose variants that are therapeutic because we believe that is what payers will reimburse for.”
Dr. Ladanyi offers these thoughts about reimbursement. “Right now reimbursement is not for the panel; it’s for the individual genes that are useful in a given patient. So the size of the panel doesn’t matter to the payer as long as it provides information on the few genes that are critical in that patient.
“I think next year there will be a change in CPT codes, so there could be a CPT code specific for panel testing. My understanding is that CPT codes will cover smaller panels first.
“All this is evolving and hard to predict,” he says.
What he sees happening in the future is a reduced version of these panels becoming more accessible to labs, so instead of 300 or 400 genes it might be 100 genes. “And it may be possible to run in easier ways. That might meet the vast majority of a laboratory’s needs,” he says, “and if that doesn’t work, it might be an indication to send out to a reference lab.”
[hr]
William Check is a writer in Ft. Lauderdale, Fla.