Next-generation sequencing is beginning to change clinical practice, especially for patients with rare diseases and cancer, a number of researchers reported at a recent conference. While issues such as reimbursement, interpretation, and incidental findings have not yet been resolved, laboratories offering NGS-based clinical testing are reporting a growing number of cases.
At last week's Future of Genomic Medicine Conference at the Scripps Translational Science Institute in San Diego, early adopters of clinical sequencing, including from Washington University, Partners HealthCare, Baylor College of Medicine, and the Medical College of Wisconsin provided updates on their clinical sequencing initiatives.
NGS for undiagnosed disease
Exome, and to some extent whole-genome, sequencing has begun to see widespread adoption in cases of rare disease, where it is often used in an attempt to end diagnostic odysseys.
In the early days, clinical exome sequencing made headlines due to a number of initial success stories, where sequencing not only identified the molecular cause of a disorder that had stumped patients' doctors for years, but also pointed toward treatments that made remarkable impacts on the patients' lives — for example, in the case of Nic Volker, exome sequencing by the Medical College of Wisconsin led to a life-saving cord blood transplant.
Now that a number of institutions and companies have launched clinical exome and whole-genome pipelines for diagnosing rare disorders, the diagnostic rate has plateaued at around 25 percent, according to researchers that presented at last week's conference.
Baylor College of Medicine, which launched its exome sequencing test in November 2011, now processes around 200 samples per month, Arthur Beaudet, professor and chair of molecular and human genetics at Baylor, said during a presentation.
Of its first 1,012 cases, 265, or around 26 percent, have been solved, Beaudet said. "In the majority of those [unsolved] cases, the mutation is in the data, but we're not able to identify it." Additionally, one weakness of exome sequencing is that current capture kits do not cover the entire coding region.
While other labs have also reported diagnostic rates of around 25 percent, Beaudet thought that this number would increase as more cases were sequenced. For instance, "of the cases we have solved, about 90 (34 percent) involved a gene/disorder that was first described in 2012," he said. "Each time you identify a new gene/phenotype [relationship], you increase the number of cases that will be solved."
Heidi Rehm, chief laboratory director at Partners HealthCare's Laboratory for Molecular Medicine, also said that around 75 percent of exome sequencing cases are unsolved. One problem, she said, is that "no unified database exists for patients with unsolved Mendelian disease."
As she said previously, the LMM is working in conjunction with Ada Hamosh, professor at the Institute of Genetic Medicine at Johns Hopkins University, to develop a matchmaking system for unsolved exome and whole-genome cases.
Unsolved cases would be submitted into a database containing de-identified genotype and phenotype information, where matching algorithms could connect them with similar cases.
Last year, around 40 people from research consortia and clinical sequencing centers met to discuss the need for such a system and formed two workgroups, she said.
Another problem with diagnostic exome sequencing is that exome capture does not cover all relevant genes. As such, her lab has been working with Emory Genetics Laboratory and the Children's Hospital of Philadelphia to develop a so-called medical exome — a custom-designed kit to ensure complete coverage of just over 4,631 medically relevant genes.
Additionally, the team is collaborating with ClinGen and OMIM to curate all those genes. Thus far, they evaluated around 647 genes and have 3,984 to go, she said.
Howard Jacob, director of the Medical College of Wisconsin's Human and Molecular Genetics Center, remains a proponent of whole-genome sequencing to ensure that relevant mutations are not missed. The laboratory initially offered sequencing internally through a partnership with the Children's Hospital of Wisconsin, but last year began offering both clinical whole-genome sequencing and exome sequencing externally.
"We and others are already reporting a number of cases where whole-exome sequencing would not have uncovered the causal variant," he said. "Some of these are in protein-coding regions."
Jacob added that of 25 patients that had both exome and whole-genome sequencing done at 100x coverage and 40x coverage, respectively, exome sequencing missed 212 actionable variants, while covering 2,762 actionable variants. Whole-genome sequencing only missed three. In addition, there were many more genes that were poorly covered by exome sequencing, he said.
Whole-genome sequencing also improves diagnostic success rate, Jacob said. Whole-genome sequencing is able to make a diagnosis 25 percent more often than exome sequencing, he said. Despite the improved success rate and the fact that the MCW laboratory "prefers whole-genome sequencing," insurance is still more likely to pay for exome sequencing, he said.
Another area where clinical sequencing has taken off is in the case of cancer care. In this area, a wide range of approaches are being used, from small panels of less than 50 genes to comprehensive panels, such as the 236-gene panel offered by Foundation Medicine, and even whole-genome, exome, and full transcriptome sequencing.
Washington University's Genome Institute has been at the forefront of comprehensive sequencing for cancer patients. The institute has developed a pipeline that includes tumor/normal whole-genome and exome sequencing, as well as whole-transcriptome sequencing of the tumor genome.
Whole-genome sequencing enables the identification of SNVs, copy number variants, structural variants, and indels, while exome sequencing can both validate those findings as well as give a more in-depth analysis of clonality due to higher coverage. The transcriptome sequencing component provides an analysis of genes that are over expressed regardless of whether they contain point mutations, an analysis of which SNVs are being expressed, and identifies gene fusions.
Transcriptome sequencing provides "additional information that I find compelling in every case," Elaine Mardis, co-director of the Genome Institute, said at the conference.
For instance, she said, sequencing may identify many mutations at the DNA level, but often around 50 percent of those genes are not being expressed. "This is critically important in culling the list of what to look at in the context of druggability."
The institute has also developed a database, the Drug-Gene Interaction Database, DGIdb, of drug-gene interactions to facilitate matching cancer patients' sequencing results with appropriate therapies.
Mardis presented two case examples of how comprehensive sequencing has changed patient management. The first case, a 10-year-old girl with low-grade glioma was actively progressing on chemotherapy and had no other treatment options in sight. Sequencing was performed on the Illumina HiSeq 2500, which identified 33 tier 1 somatic point mutations and four indel mutations. The initial analysis pointed to no obvious drugs or pediatric clinical trials.
Further analysis found that the 3 bp insertion in the BRAF gene was being expressed in the patient's tumor. Based on that finding, the patient was enrolled in a pediatric clinical trial of a MEK inhibitor. Mardis said that the patient was enrolled in the trial last October, and in December, she received a message from the oncologist saying that for the first time, the tumor was regressing.
In a second example, the Genome Institute sequenced a 5-year-old girl with high-grade glioma on the HiSeq 2500. She was sequenced because oncologists could not agree on the diagnosis of her surgically resected biopsy and whether the tumor was aggressive or indolent. So, sequencing was done to help with the diagnosis and potentially identify a druggable target.
The researchers identified three SNVs, none of which were supported by the RNA-seq data, and no indels. Transcriptome sequencing also found four genes that were over expressed, but there were no strong clues that any of those genes were drivers, Mardis said.
As a result, the oncologists made the decision to treat the patient with low-dose pinpoint beam radiation, rather than to treat the tumor aggressively with high-dose total beam irradiation. The initial low-dose radiation was followed by a low dose of chemotherapy and the patient is being monitored via imaging to detect recurrence. While Mardis said it is too early to declare success, the patient is currently doing well with no sign of recurrence.
While the Genome Institute has been pushing forward with comprehensive sequencing, others have adopted more targeted approaches in cancer management.
For instance, Robert Nussbaum, chief of the division of genomic medicine at the University of California, San Francisco, said that expanded gene sequencing has been a "wonderfully disruptive technology" that has "affected my practice."
For instance, he said, in cases of patients with a family history of cancer, it was previously necessary to sift through family pedigrees and go through Boolean algebra criteria to figure out which genes to test. Now, he said, he can simply do a 50-gene hereditary cancer panel.
He presented one example of a 49-year-old patient with high-grade serous ovarian cancer and three small adenomatous colonic polyps. Her 55-year-old sister was diagnosed with ductal breast cancer at age 49 and she had one grandmother who passed away from breast cancer in her 70s. While an initial analysis of the family pedigree pointed to testing BRCA1 and BRCA2 genes, the cancer panel found that those genes were negative, but did find two mutations involved in DNA repair, one in the PMS2 gene and one in PALB2.
In another case, next-gen sequencing helped elucidate the diagnosis of a 54-year-old woman who was initially diagnosed with pancreatic cancer. A commercial gene panel found no mutations in the KRAS gene, which would have been very unusual had it truly been pancreatic cancer, said Nussbaum. However, the panel did identify a frameshift mutation in the BRCA2 gene. The woman's tumor was re-analyzed and she was given the diagnosis of ovarian carcinoma.
However, Nussbaum said that such expanded gene testing is "both a blessing and curse." Currently, there are many variants of unknown significance and penetrance of even well-studied gene mutations can be highly variable, making it difficult to communicate risk to patients.
For instance, said Nussbaum, even in the case of BRCA1 and BRCA2 genes, at a population-level, mutations in those genes can increase risk of developing breast cancer by 40 percent to 50 percent. But within a specific family pedigree, that risk can be as high as 70 percent to 80 percent.
Penetrance estimates are generally derived from families that are studied because of a clear-cut Mendelian inheritance, he said. But as more people undergo expanded testing, these variants will also be found in people who wouldn't have otherwise been tested and it is unclear what their actual risk will be.
Moving forward, Nussbaum echoed the sentiments of the other speakers in making a call for more data sharing. "We will not harvest the fruits of genomics unless we pool our data to understand pathogenicity and penetrance," he said.