NEW YORK (GenomeWeb) – Foundation Medicine has validated its circulating tumor DNA assay, the firm said at last week's Advances in Genome Biology and Technology meeting in Orlando, Florida. The company plans to launch the test, called FoundationAct, by the end of this quarter.
Geoff Otto, a senior scientist at Foundation Medicine, said in a presentation at the conference that FoundationAct will focus on a 61-gene subset of its tissue-based 300-plus-gene FoundationOne assay that are linked to approved therapies. The test will be able to identify mutations present at a .5 percent allele frequency from a 10 milliliter blood draw.
FoundationAct will be geared toward patients from whom a standard tissue biopsy cannot be taken or in cases where it is not practice to do a tissue biopsy, like in advanced patients with multiple metastases.
The firm plans to use shotgun sequencing with 2x175 paired-end reads and a molecular barcoding and baiting strategy to ensure even coverage and amplification and to detect potential contamination.
Otto said that Foundation Medicine scientists spent a "huge amount of time maximizing the library construction process," in order to capture as much complexity as possible from limited input. The team was able to achieve a 50 percent to 70 percent conversion efficiency of ctDNA input into a sequencing library.
In order to be clinically relevant, the firm wanted an assay that could reliable detect SNVs at a frequency of .5 percent, and indels and copy number variations at a 1 percent frequency with at least a 90 percent sensitivity and a final error rate of no more than .05 percent. Turnaround time should be 10 to 14 days. As such, he said that the firm needed to sequence deeply — to an average of 5,000-fold coverage.
An initial experiment demonstrated that 90 percent of the samples that yielded at least 25 nanograms of ctDNA had a median exon coverage of 5,000-fold and that over 99 percent of the exons were covered 2,500 times or more. That "enabled us to go forward with this assay," he said.
A key component of the assay is a bait strategy that enables the detection of contaminants, Otto said. This was important since the test "push[es] the limits of sensitivity" in trying to call variants at very low frequencies, making it difficult to tell the difference between a very low frequency variant and low-level contamination.
Once the method was developed, the firm validated it on both cell lines and clinical samples. They compared the ctDNA test's ability to detect known variants with its FoundationOne assay and droplet digital PCR.
In a validation test that Otto described, the ctDNA assay was 100 percent concordant with these orthogonal technologies — detecting all 87 somatic alterations, including 47 that were present at frequencies under 5 percent.
Next, the team established sensitivity and positive predictive value for SNVs, indels, and copy number alterations. For SNVs and indels between one and 40 bases, the firm established the ctDNA assay had close to 99 percent sensitivity and positive predictive value for minor allele frequencies of .5 percent and 1 percent, respectively. Sensitivity dropped a bit for copy number alterations, to 93 percent, and positive predictive value was 98 percent.
Otto also stressed the importance of sample quality and collection in the assay's success. For instance, in the case of a sample that had been stored for too long in EDTA, the ctDNA assay was not able to detect a known KRAS mutation due to the "improper preparation of the plasma," Otto said. The plasma mononuclear blood cells had likely been lysed, spilling normal genomic DNA into the mixture with cell-free tumor DNA, he added. By contrast, from a well-prepared sample from the same individual, the KRAS mutation was detected at a frequency of 26 percent.
Otto said that Foundation plans to submit the results of its validation study to a peer-reviewed journal for publication.