By Monica Heger
Foundation Medicine has demonstrated that its targeted sequencing cancer diagnostic test can identify clinically actionable mutations, including novel and potentially actionable gene fusions — a finding that helps validate the test ahead of its broad commercial launch this summer.
While the company has been accepting its first clinical samples since October and recently received CLIA certification for its laboratory in Cambridge, Mass., it is planning to launch the test broadly in June at the annual American Society of Clinical Oncology meeting.
Additionally, Foundation plans to expand its test for use in hematology tumors and, in a subsequent iteration of the test, plans to include transcriptome sequencing of the tumor to complement the targeted sequencing test and better detect gene fusions and expression levels.
Foundation will launch the test on the Illumina HiSeq and is aiming for a 14-day turnaround time. Maureen Cronin, Foundation's senior vice president of research and product development told Clinical Sequencing News that the company is also "doing experiments with newer platforms," but said that there will always be a gap between testing a technology in research and doing clinical validation work "to understand the limitations … and make sure everything performs."
In a paper published last week in Nature Medicine, researchers from the company and the Dana-Farber Cancer Institute detailed results of the test from 64 formalin-fixed, paraffin-embedded tumor samples, demonstrating that it can identify mutations for which known therapies exist and also novel gene fusions that are potentially targetable.
The test analyzed 2,574 coding exons from 145 genes that are associated with cancer pathways, targeted therapy, or prognosis; plus 37 introns from 14 genes that are frequently rearranged in cancer. Sequencing was done on the Illumina platform to an average 229-fold coverage.
In 40 colorectal cancer samples, 125 alterations were identified in 21 genes, with 39 out of 40 samples having at least one mutation. Eighty percent of the samples had mutations to TP53, a known tumor suppressor that is mutated in a broad range of cancers, and 67.5 percent of the samples had mutations to APC, a tumor suppressor gene that regulates the growth of polyps.
Twenty one samples had at least one mutation that has been linked to a clinical treatment or is currently being studied in a clinical trial of targeted therapy.
For example, mutations were found in KRAS and BRAF that predict resistance to Eli Lilly's and Bristol-Myers Squibb's Erbitux (cetuximab) or Amgen's Vectibix (panitumumab); and in FBXW7 that predict anti-tubulin resistance.
Additionally, the test found mutations in BRCA2, for which there are clinical trials for PARP inhibitors; in GNAS, for which there are MEK- and ERK-inhibitor trials; and in PIK3CA, for which there are mTOR-inhibitor trials.
The team also detected a novel gene fusion involving ALK, suggesting that ALK inhibitors such as Pfizer's Xalkori (crizotinib) may be effective.
Among the 24 non-small cell lung cancer samples, 72 percent had at least one alteration associated with a current clinical treatment or targeted therapy trial. There were mutations to KRAS, some of which point to EGFR kinase-inhibitor resistance and others that suggest eligibility for PI3K- and MEK-inhibitor trials. Mutations in BRAS suggest eligibility for BRAF-inhibitor trials, and mutations in EGFR suggest that EGFR-inhibitors such as AstraZeneca's Iressa (gefitinib) or Genentech's Tarceva (erlotinib) may be effective.
Two novel mutations were also found. JAK2 contained a mutation at low frequency that is commonly found in myelodysplastic syndromes, but has not previously been described in solid tumors and suggests that patients with this mutation may be sensitive to JAK2 inhibitors.
Additionally, a novel gene fusion involving RET was identified in a 44-year old 'never smoker.' The team next looked for the fusion in lung cancer samples from 121 Caucasian and 405 Asian patient samples, finding it in 0.8 percent and 2 percent of the samples, respectively. Notably, the samples with the fusion did not have any known driver mutations, suggesting that the fusion may be driving the cancer.
A different RET gene fusion is frequently found in thyroid cancers, and is sensitive to RET inhibitors. A subsequent in vitro analysis found that cells with the novel fusion were also sensitive to a number of RET kinase inhibitors, suggesting that those drugs "should be tested in prospective clinical trials for therapeutic benefit in individuals with NSCLC" with the novel fusion, the authors wrote.
In fact, Pasi Jänne, a co-author and associate professor of medicine at Harvard Medical School and Dana-Farber, said that Dana-Farber researchers now plan to initiate a clinical trial for a RET inhibitor in non-small cell lung cancer patients to determine whether it will be effective.
Jänne added that Dana-Farber researchers are also testing out their own versions of targeted sequencing tests for cancer patients, including a whole-exome sequencing test. The team plans to initially test the approach on colon and lung cancer patients to try and determine whether it is better than doing a more focused sequencing test. Jänne said the team will look both at what new, relevant information exome sequencing can uncover as well as "what to do with the additional information" such as germline variants that may be related to other diseases like diabetes, Alzheimer's, or cardiovascular disease.
While many think that the long-term future of sequencing-based diagnostics will involve more comprehensive approaches like whole exome, transcriptome, and whole-genome sequencing, Cronin said that the targeted approach still offers a number of advantages.
Even though sequencing prices are dropping, it is still cost-prohibitive to sequence cancer samples to a great enough depth to detect rare mutations. Foundation's test has over 200-fold coverage, which enables the company to detect very rare mutations — below 5 percent frequency and even down to 1 percent frequency — from samples that are extremely heterogeneous, said Cronin.
"One of the challenges with cancer samples is that biopsies are not pure tumor," she said, and instead are often contaminated with normal cells and contain multiple clones.
Foundation has validated that its test can call base substitutions at greater than 99 percent sensitivity from samples that contain only 10 percent tumor and call indels at greater than 95 percent sensitivity from samples that are only 20 percent tumor content, with a false discovery rate of less than 1 percent.
With whole-genome sequencing, "because you have to sequence so many more positions, you can't do it at the depth that we're doing while keeping the costs and interpretation under control," she said.
Arul Chinnaiyan, who heads the University of Michigan's comprehensive cancer sequencing program Mi-OncoSeq, said this ability to determine sensitivity is one advantage of a targeted approach.
"For each mutation, [Foundation Medicine] has an a priori assessment of how sensitive they are going to be to detect that mutation," he said. "But, imagine doing that for the whole genome. Having a similar sort of bar for more comprehensive or systematic approaches becomes difficult," he said.
"That will be one of the major issues in rolling out a clinical test," he said. "This focused approach has the advantage of being able to define sensitivity and specificity for each type of mutation that it's looking for."
Additionally, he said, for now targeted approaches are more amenable to degraded samples such as those from FFPE tissue, although there are many groups, including his own, that are working on optimizing DNA extraction so that it can be used in more comprehensive sequencing approaches.
On the other hand, targeted approaches may miss important mutations or not provide the complete picture.
For instance, even if the test does identify a potential driver mutation, there could be other mutations that counteract that mutation.
With a more comprehensive approach, "you can examine resistance pathways that might be activated, or synergistic mutations that might occur, or competitory mutations," said Chinnaiyan.
Even though Foundation's test showed it could identify actionable mutations in around half of the samples it tested, it did not identify mutations in the other half of patients.
This suggests that the test is not picking up all the relevant cancer-related pathways, or that there are other levels of regulation that are important, such as gene expression levels, said Cronin.
Transcriptome sequencing in particular has the potential to illuminate important changes in gene expression and gene fusions, she said. "If you see something at the DNA level that is unusual and not been seen before, you need to know that it's being expressed," Cronin said. Also, "there are some genes that are not mutated, but because of changes in expression have an impact on cancer."
Cronin said that the company eventually plans to incorporate RNA-seq into the test, although she did not provide a timeline, and said that it would be offered as an add-on to the existing test.
Chinnaiyan, who does transcriptome sequencing of patients in the Mi-OncoSeq protocol, agreed that RNA-seq has been particularly informative for identifying potentially actionable mutations, such as gene fusions, as well as highlighting important pathways that are either over-expressed or under-expressed.
"Many groups are now finding rare subsets of common solid tumors, like colon and lung cancer, that harbor these rare subsets of potentially driving actionable fusions," he added.
Have topics you'd like to see covered by Clinical Sequencing News? Contact the editor at mheger [at] genomeweb [.] com.