By Monica Heger
Foundation Medicine's sequencing-based cancer test can detect clinically actionable mutations, including ones that would be undetectable with other methods, according to new data the company presented this week at the American Society of Clinical Oncology meeting in Chicago.
The company, which is aiming for CLIA certification this year and has targeted the second quarter of 2012 for the commercial launch of its comprehensive cancer test, demonstrated its performance on 75 cancer tissue samples from non-small cell lung cancer, melanoma, and colon cancer.
The study "gives us great insight" into how "we can take this technology out of the R&D setting and move it into the clinic," Foundation Medicine's CEO Michael Pellini told Clinical Sequencing News.
The company used a targeted sequencing approach with custom baits on Agilent Technologies' SureSelect and paired-end sequencing on the Illumina HiSeq, sequencing 2,574 exons from 176 genes from each sample. Each sample was sequenced to an average coverage of 213-fold.
Importantly, the researchers demonstrated their ability to sequence clinical samples — all were from formalin-fixed, paraffin-embedded tissue samples that had been extracted either by major resections, incisional biopsies, needle core biopsies, or fine needle aspirates. For each sample — comprising 49 colorectal samples, 22 NSCLC samples, and four melanoma samples — they extracted 200 nanograms of DNA from 40 microns of tissue. Pellini added that the company is now running the test using only 50 nanograms of DNA.
The team compared their method to conventional single-gene analysis methods, such as CE sequencing and mass spectrometry, and achieved 100 percent concordance with those tests, which detect known mutations in genes such as KRAS or EGRF.
In total, the team found 214 mutations, including 134 base substitutions, 44 insertions and deletions, 34 copy number alterations, and two structural rearrangements.
Nearly 80 percent of the mutations that they detected would not have been found using the conventional CE and mass spectrometry-based methods, and just over 50 percent of them were "clinically actionable."
Pellini defined three tiers of clinically actionable mutations. In the first, the mutation could be matched to a specific drug that is on the market, although not necessarily for that particular cancer. Tier two includes mutations tied to therapeutics that are in clinical trials. Tier three clinically actionable mutations would be those that "provide insight into other pathways that might be active and might provide additional insight for the oncologist to treat the patient," he said.
In the data presented at ASCO, the team did not break out the percentages of actionable mutations, but in other data generated internally the company found that 34 percent of all mutations were either tier one or two and 18 percent were from tier three, or "plausibly actionable" mutations, Pellini said.
[ pagebreak ]
Notably, Pellini said genes that are known to be associated with one type of cancer were frequently found to be mutated in other cancers as well. This finding demonstrates the importance of using a comprehensive test that screens for mutations in many different genes, as opposed to simply testing single genes that are known to play a role in a specific disease state.
For example, testing melanoma patients for BRAF mutations is common practice today, he said. However, it's not a test that would be given to patients with other types of cancer, such as colon cancer.
"If an oncologist has a patient with colon cancer, he's probably not thinking about using a BRAF test as part of the diagnostic panel," he said. "But our data has found unexpected mutations." Six of 49 colorectal cancer patients harbored a BRAF mutation, for instance.
"The evidence is mounting that these mutations, even though they might have first been discovered in one disease state, are now being found in other disease states, too," Pellini said.
From the 49 colorectal cancers, the team's test identified 155 mutations, about 58 percent of which were base substitutions, 28 percent indels, 12.7 percent copy number alterations, and 1.3 percent rearrangements. Forty-six mutations to the KRAS gene had been previously identified and all 46 were confirmed by the test.
Fifty three mutations were identified from the 22 non-small cell lung carcinoma samples. Sixty four percent were base substitutions, 28 percent copy number variants, and 8 percent were indels. Previous testing of the EGFR gene had identified 17 mutations, all of which were confirmed.
The company did not break out data on the mutations discovered in the four melanoma patients.
Pellini said the company is continuing to optimize the test. For example, it has developed an algorithm that can predict with 95 percent accuracy whether a given sample will yield a result, before the test is run, which will have important implications for patients and pharmaceutical companies using the test in clinical trials.
Moving forward, Pellini said the company is on track to commercially launch the test in the second quarter of 2012 and is targeting a turnaround time of two weeks. The company has not yet decided on a price for the test, but said it would be in the same price range as other molecular tests, but with much more comprehensive data.
Soon, physicians will be running upwards of 10 molecular tests for a lung cancer sample, he said. He noted that the approach "becomes extremely expensive," with an estimated cost of $5,000 to $8,000. By contrast, Foundation Medicine's test will be a "comprehensive genomic profile that will cost roughly half of that. And there will not be a need for additional genomic tests."
Foundation Medicine is also collaborating with pharmaceutical companies Novartis and Celgene (CSN 4/13/2011 and 5/17/2011), and over the next several months, Pellini said the company would announce additional collaborations with pharmaceutical companies and academic researchers.
He said the company is staying mindful of the regulatory climate, developing its processes "with the understanding that [the test] may have to be submitted to the [Food and Drug Administration]." For the time being, however, the company plans to launch the offering as a laboratory-developed test under CLIA regulations.
While the results presented at ASCO were generated with an Illumina HiSeq, the company also has an Ion Torrent PGM in house. Pellini said the company has not yet decided on what platform it would launch the commercial test, and that it is considering not only the platforms it currently has in house but everything else as well.
"I believe, ultimately, we will utilize a few or several platforms," Pellini said. "It's easy to envision that for a pharmaceutical setting we'd use one platform, and for a clinical setting another."
Have topics you'd like to see covered by Clinical Sequencing News? Contact the editor at mheger [at] genomeweb [.] com.