Radboud University's Medical Center in Nijmegen, The Netherlands, is developing a hereditary breast cancer screening test on Life Technologies' Ion Torrent PGM that it plans to begin offering to patients in March.
By the end of the year, the lab plans to offer all cancer mutational testing on the PGM.
Marcel Nelen, a clinical molecular geneticist and head of the Central Genome Analysis Laboratory at the university, reported on the lab's work converting its BRCA 1/2 test from Sanger sequencing to next-generation sequencing on the PGM at an Ion Torrent user meeting at last week's Advances in Genome Biology and Technology meeting in Marco Island, Fla.
In developing the test, the Dutch team collaborated with Jose Luis Costa's group at the Institute of Molecular Pathology and Immunology of the University of Porto (IPATIMUP) in Portugal.
The Radboud team has been offering a BRCA screening test that uses both Sanger sequencing and multiplex ligation-dependent probe amplification (MLPA) to measure exon deletions. Nelen said they were looking to design a test with a faster turnaround time and lower cost.
For the Sanger-based test, the lab was analyzing 15.8 kilobases of sequence including 5.6 kilobases on the BRCA1 gene and 10.3 kilobases on the BRCA 2 gene. "The genes are highly polymorphic," Nelen said during a presentation, and "the mutations are scattered, so there are no hotspots."
Using Sanger sequencing, the lab has been analyzing between 50 and 60 samples per month, he said, with a turnaround time of three to five weeks. "Although efficient, it's still expensive," he said, and reimbursement just barely covers the cost of the test.
By converting to the PGM, he expects to cut the current $3,000 price tag in half, possibly more if the MLPA step can also be cut, and to reduce turnaround time to three to five days.
In considering a next-gen based test, Nelen said the goal was 100 percent coverage of all coding exons as well as at least 10 bases and ideally 20 bases into the intron regions on the intron-exon boundaries. Amplicons should cover all exons and should overlap.
The lab currently has two of PGMs and designed a custom AmpliSeq panel to target the BRCA 1 and 2 genes.
In conjunction with IPATIMUP, the two labs validated the assay. Initially, each lab tested 10 known samples and compared results. They then tested 15 known samples each in a blinded fashion, meaning the IPATIMUP lab tested 15 samples that it knew, then sent those to Radboud for testing and vice versa.
The results yielded a total of 65 pathogenic mutations, none of which were missed by either lab, Nelen said. The mutations included deletions and insertions, point mutations, exon deletions, and mutations in homopolymer stretches, he said.
The lab tested samples with known mutations as well as those that had not previously been characterized.
Nelen said that he is now collaborating with a number of other labs within the OncoNetwork Consortium — a Life Tech coordinated effort to develop cancer panels using its AmpliSeq technology.
The group is working to develop a lung and colon cancer panel for the PGM.
Aside from Radboud University Medical Center, the other members of the OncoNetwork Consortium include the Cancer Research Center of Mercogliano in Italy; the genetics and molecular biology division of pharmaceutical company Viollier in Switzerland; the University of Warwick in the UK; the Institut Gustave Roussy in France; the University Paris Descartes Center in France; and St James' Hospital in Dublin.
Separately, Radboud University Medical Center is also working on a test for hereditary paragangliomas.
The colon and lung cancer panel comprises 90 amplicons covering hotspots within 22 genes, and requires 10 nanograms of input DNA, which can be taken from formalin-fixed paraffin-embedded tissue.
The initial validation was completed by six different labs within the OncoNetwork Consortium on 155 samples.
In stage 1, the labs sequenced five control samples from FFPE tissue, including two from a KRAS cell line, one lung adenocarcinoma, and two xenograft samples. All known mutations were called by each lab.
In stage 2, each lab sequenced 10 different samples from colon and lung cancer, which were known to that lab, but not the others. After the initial lab sequenced the samples, they were sent to the other six labs, so that in total 60 samples were sequenced by all six labs. Again, each lab correctly called all mutations, as well as all wild type variants.
In stage 3, samples were introduced with "large variations of tumor content," Nelen said, which caused a few problems in calling variants. For example, a mutation was missed in one sample because there was not enough DNA. Even when the researchers tried to confirm the mutation with Sanger sequencing, it was "also impossible to get the proper sequence," Nelen said. A second sample was initially missed because the team had set an allele frequency threshold for 5 percent, but the mutant was present at a 2 percent frequency. However, by adjusting the parameters on the software, the variant was called.
Nelen said the group has begun validating a second, optimized version of the chip. The goal is to achieve 500-fold coverage, which enables mutations to be detected down to at least a 5 percent frequency, and potentially as low as 2 percent frequency. However, in the initial version, some amplicons did not have enough coverage, so sensitivity was lower. But in other amplicons, coverage was too high, reducing chip capacity.
The second version includes an optimized primer set to achieve more even coverage of the amplions and eight samples are multiplexed per chip on the 316, instead of five.
In the second validation, the labs initially tested eight FFPE samples with known mutations to make sure the test could detect the mutations. The labs then scaled up and have sequenced a total of 98 samples.
"For molecular tests, you need robust performance, highly accurate, clinically relevant sensitivity for hotspot mutations and loss of function mutations, with results available in days," Nelen said.
Nelen said the lab is continuing to optimize and validate the test, including the primer design and software. Additionally, the entire workflow needs to be implemented in the lab's LIMS system, he said.
The goal is to launch the BRCA1/2 test in March, with all other cancer panels offered by the end of the year. The lab is also in the process of automating its sample-prep and multiplex PCR steps, which it also hopes to complete by the end of the year.