SAN FRANCISCO (GenomeWeb) – The Children's Hospital of Los Angeles plans to begin offering a next-generation sequencing panel at the end of the month as a laboratory-developed test specifically tailored toward pediatric cancers.
Timothy Triche, co-director of the Center for Personalized Medicine at CHLA, said in a presentation at the recent Global AMP conference in Berlin and in a subsequent interview that the group could not simply modify available panels for adult cancers because the genomic alterations seen in pediatric cancers are so different.
For instance, he said, pediatric cancers are dominated by gene fusions and typically have very few point mutations, while adult cancers tend to have lots of point mutations and very few gene fusions. In addition, Triche said, the group wanted a panel that would cover the full spectrum of childhood cancers, including leukemias, lymphomas, sarcomas, and brain tumors.
The panel, dubbed OncoKids, will be run on Thermo Fisher Scientific's equipment, including the Ion Chef for library prep, the Ion S5 for sequencing, and the Ion Reporter and Ion Torrent Suite for analysis. In addition, the CHLA team has designed a custom informatics pipeline, called integrated curation environment (ICE) for clinical interpretation and report generation.
The panel will analyze both DNA and RNA requiring 20 nanograms of input each. Samples can include nucleic acids extracted from formalin fixed-paraffin-embedded tissue, fresh frozen tissue, or bone marrow. OncoKids evaluates 187 genes, including 82 hotspot regions and 44 full-length genes, as well as 24 copy number variants and 78 gene fusions.
Turnaround time will likely be two weeks, although Triche said that it would be possible to run it in under one week for urgent cases. The panel will likely have a list price between $3,000 and $4,000, and CHLA aims to eventually secure reimbursement for it. Triche also said the group hopes that the panel will be used in clinical trials.
A Thermo spokesperson said that it plans to commercialize the panel reagents as a research-use only kit in the fall.
Triche said that the team has so far run 503 samples representing 237 unique tumors. They have also validated the panel against synthetic controls where they achieved 5,000x average coverage for DNA variants and could detect SNVs at an allele frequency of 5 percent and indels down to 10 percent allele frequency. On the synthetic control samples the panel achieved 100 percent sensitivity and greater than 99 percent specificity for SNVs. And, the ICE pipeline was able to boost sensitivity for indels to 100 percent from 63 percent using standard variant calling methods, and also achieved 100 percent specificity.
"We spent a lot of time cleaning up the call accuracy for indels," Triche said. He added that while the panel showed 100 percent accuracy on the control materials, they've analyzed clinical samples with 18-mer homopolymer runs with a mutation at the end. "Those are very hard to call, but we're getting close," he said.
In putting together the list of genomic alterations that would make up the panel, Triche said his team collaborated with many stakeholders and experts, inviting them to submit a "wish list" of targets. Triche solicited feedback from pediatric oncologists at CHLA, since they would be the target user of the panel, as well as the Children's Oncology Group. Ultimately, the panel included 49 of the 51 targets identified by COG. Triche said the only two the panel does not include are the BRCA1 and BRCA2 genes because CHLA is also developing a germline panel that will include those genes.
The OncoKids panel will be used to molecularly profile patients' tumors and in some cases will enable physicians to identify potential treatment strategies for those who fail first-line therapy. But, just as important will be the ability of the panel to do better risk stratification, Triche said. In pediatric cancers, knowing whether a patient is high- or low-risk is extremely important, since high-risk patients tend to progress rapidly and need aggressive treatment. "We've developed risk-stratified therapies, but you need to stratify patients correctly," Triche said. The current approach is to do single-gene tests, but that often results in missing something important for stratification or will add to the time and cost of obtaining a correct classification.
For example, Triche said, neuroblastoma patients with MYC amplification have a poor prognosis. But, overexpression of cMYC also results in a poor prognosis for those patients. "So, you need both the DNA and RNA features in one test," he said.
Risk stratification is also important for those patients with good prognoses so physicians can avoid overtreating them. Pediatric cancer patients who have been treated with chemotherapy often have long-term side effects, even when they are cured of their cancer. Patients with brain cancer, for instance, can suffer cognitive development issues. High dosages of chemotherapy can also cause osteonecrosis, in which blood vessels going to the bone are damaged causing the bones to weaken.
Another important reason for analyzing both DNA and RNA is that gene fusions and overexpression play major roles in pediatric tumor progression. Being able to look for those is extremely important, Triche said. While the panel targets 78 fusions, it also enables the discovery of novel fusions because of its ability to find fusions with one of the target genes and an unknown partner. As such, those 78 gene fusion targets actually correspond to around 1,400 potential fusions that the panel can detect.
The spectrum of mutations in pediatric tumors seems to be nearly opposite of what is found in adult tumors, with fusions dominating and SNVs being very infrequent. Researchers are still working out exactly why and what the implications are, but Triche said that one theory is that in adults, cancer is typically a disease of age. It results after a lifetime of exposures that cause changes to the DNA, as well as mistakes in DNA replication that build up over time. Adults accumulate many mutations throughout their life, but kids don't have that lifetime of acquired mutations.
Another difference between pediatric and adult cancers is how quickly tumors grow, on average. Tumors tend to grow much faster in pediatric cancers, Triche said, and one theory is that the faster growth rate is a direct effect of the gene fusions. A gene fusion is like "throwing gas on a fire," Triche said.
However, there are specific cases where SNVs are seen in pediatric cancers and are important for understanding the patient's disease and therapy regimen, Triche said, and that's when chemotherapy-induced mutations begin to play a role in resistance.
Triche described one such patient example: a 13-year old boy who was diagnosed with T-acute lymphoblastic leukemia. He continued to progress after standard of care treatment and was analyzed via cytogenetics, a microarray test, and the OncoKids panel. The OncoKids panel identified two different fusions, each of which was singularly identified by the cytogenetics and microarray test. One fusion involving the PDGFRA gene indicated that imatinib would potentially be effective. The patient was started on the drug and initially responded. Eventually though, he relapsed. A second analysis identified mutations in the PDGFRA gene known to confer resistance to imatinib.
"That's an example of a mutation occurring in a childhood tumor that's commonly seen in adults treated with imatinib," he said.
In addition, he said, mutations to the TP53 gene, which are very common in adult cancers, are rare in pediatric cancers, but are much more frequent in patients after they have undergone therapy.
Long term, Triche said he hopes that the OncoKids panel will help spur research into these areas of pediatric cancers. Currently, the knowledge around the genomic landscape of pediatric tumors lags far behind that of adult tumors, he said.
And, if the panel proves to be successful at CHLA and there is demand for it outside of the institution, Triche said the lab would consider making the ICE pipeline available in the cloud. That would allow other hospitals to develop and run their own assays but use the CHLA-developed interpretation pipeline.