NEW YORK(GenomeWeb) – Researchers from the Onconetwork consortium, a group of European institutions working with Thermo Fisher Scientific's Life Technologies to develop a targeted sequencing cancer mutation panel for the Ion PGM, have published a description of their development and initial validation of this assay.
The study, which describes the panel's performance in detecting known and unknown mutations in a cohort of 155 samples supplied by the different Onconetwork labs, appeared last week in BMC Cancer.
Thermo Fisher announced in January that it had received a CE mark for its commercial IVD version of the same panel. Last October the company also received a CE mark for the Ion PGM Dx System, the clinical-grade version of its sequencing instrument on which the newly released solid tumor assay is designed to run.
Though designed by the Onconetwork participants primarily for lung and colon cancer, the newly-launched commercial panel, called the Oncomine Solid Tumor DNA kit, is also clinically applicable for other solid tumors including melanoma, Jon Sherlock, senior product manager for oncology at Thermo Fisher, told GenomeWeb in an email.
The company also plans to file for CE marking in the next few weeks for a second clinical sequencing panel that can be run either on its own or as an adjunct to the Oncomine kit, Sherlock added. This will be an RNA-based fusion detection kit covering markers that are currently relevant mainly for NSCLC, but potentially also other cancer types. A research-version of the fusion panel is already available from the company.
Bastiaan Tops, the first author of the Onconetwork study and a pathologist at Radboud University Medical Centre in the Netherlands, told GenomeWeb this week that his group and others became interested in making available a sequencing panel for lung and colorectal cancers soon after the Ion Torrent PGM was launched.
"At the time that Life Tech first marketed the PGMs, there was not a single gene panel you could use. You could buy the machine and sequence anything, but not with a pre-designed panel," he said. "Several clinical labs approached the company, and then they contacted a [number] of groups and said 'Hey lets see if we can come up with a test that meets the needs of most of the people in the group and then try to validate it within the consortium."
According to Tops, settling on a final list of targets was not without difficulty. "For the main genes — KRAS, NRAS, EGFR — everyone agreed on those. And then there are some genes expected to get to the clinic in the near future, which were easy to agree," he said.
"But then you get to the areas where some research institutes have a specific interest in gene A or B and the list got longer and longer," he added.
Also, Tops explained, the group intended to create a one-tube assay, requiring only one PCR, which is very cost effective but limited the list essentially to hotspot genes.
"For a one-tube assay, it only works if you don’t have overlapping amplicons," he said. "So that meant mainly hotspot genes, because when you start including tumor suppressor genes like P53, which is actually partly in there, you can't ever get full gene coverage."
In the end, the group settled on a panel that includes in a single multiplex PCR of 87 hotspot regions in 22 genes, and requires only 10 ng of input DNA.
According to the study authors, the set of targets was chosen to contain "well-known predictive markers in the receptor tyrosine kinase pathway," such as mutations of the EGFR and KRAS genes, but also to include other genes that have less current clinical relevance but might serve as targets in the near future or have other prognostic relevance, like BRAF, AKT1, DDR2, and ERBB2.
"Since a smaller gene panel requires less sequence capacity, more samples can be pooled in a single sequence run resulting in lower costs per sample and an increased throughput compatible with a diagnostic activity," the authors wrote.
"We designed the amplicons together with [Life Technologies]. They provided in silico design and we checked them manually and went back with some things; for example, if some regions didn’t have the right coverage. At some point we had a final design and we started testing it," Tops said.
In testing the agreed-upon panel, the team analyzed a total of 155 FFPE samples containing 112 previously identified mutations in the KRAS, EGFR and BRAF genes. According to the study authors, only one sample failed analysis due to poor DNA quality. All the other samples were successfully and correctly genotyped for the known mutations, even at a frequency as low as 2 percent, provided that coverage was greater than 500x.
The researchers also found that the sequencing panel revealed other mutations that were not previously known.
In a first phase of experiments, the seven consortium labs tested five control samples in an inter-laboratory "ring trial." All labs correctly identified the known mutations present in the samples. In addition all seven labs also identified six new variants in two of the control samples, and confirmed them using Sanger sequencing.
After another ring trial using 10 FFPE tumor samples, in which all labs correctly detected 47 previously identified mutations, the researchers moved on to a study of samples chosen by each participating lab as both typical of the diagnostic clinical setting and problematic to analyze due to the size of biopsies, percentage of cancer cells, or DNA quality.
In total, 29 colorectal and 61 lung cancer samples containing 56 mutations in the KRAS, EGFR and BRAF genes were tested in this experiment. Among this set, the DNA of one sample failed to amplify due to low quality. All the other samples were successfully analyzed and all known variants were identified, the authors wrote.
After excluding two other samples from further analysis due to evidence of over-fixation with formalin, the team identified a total of 92 previously unknown variants in the remaining 87 samples. Most of these new variants were in RTK-signaling genes, including BRAF, EGFR, KRAS, and PIK3CA, as well as in TP53.
These results provide some added evidence of the clinical benefit of sequencing over more targeted molecular diagnostic technologies. "Since drugs targeting these latter molecular alterations are in clinical trials, the availability of this information might significantly improve the possibility of these patients to receive a personalized therapy," the authors wrote.
The team's preliminary testing also showed that up to five samples could be pooled on an Ion PGM 316 chip with more than 500x average read depth. When the group redesigned the panel using an updated primer design algorithm available from Life Tech, they were able to increase this to up to eight samples pooled on a single 316 chip.
When the team retested some of the initial samples using this second version of the panel, they were able to detect all the mutations previously identified by the first version. According to the authors, materials costs using the second version, which allows multiplexing of eight samples, should be between about $150 and $200 per sample. "While this may be higher compared to a single conventional diagnostic test, it is actually a lot cheaper considering that the NGS approach replaces multiple conventional tests," they wrote.
In combination with rapid sample preparation, the team also reported that turnaround time from DNA isolation to results is between 48 and 72 hours. "While this may not be as fast as some conventional methods, it is adequate for most routine clinical applications," they wrote.
According to the Onconetwork group, new amplicons or genes of interest can easily be added to the current panel, but the authors stressed that re-validation is important for any modification.
Tops said that his group in theNetherlandshas been working with a slightly modified version of the panel in a clinical setting for about a year and a half now.
"In the Netherlandsthe rules are not as strict as some other European countries or the US. We are allowed to have non-IVD kits as long as we validate in house," he said. "We downgraded [the panel] a bit compared to what was published and we've been using it now for more than a year in a diagnostic setting."
Other participating labs have adopted the panel as-is, he added. And still others, in regions with stricter rules about clinical diagnostics, have awaited the newly-launched CE-IVD version, which went through additional validation by Life Tech and its research collaborators after the experiments described in the Onconetwork study, Tops said.
Thermo Fisher listed the Ion PGM Dx with the US Food and Drug administration last fall for clinical use as a class II medical device. Company representatives declined to detail any plans for submitting the Oncomine solid tumor kit or upcoming clinical fusion detection kit to the FDA.