NEW YORK (GenomeWeb) – A research team from the University of Colorado has developed the Integrating Molecular Profiles with Actionable Therapeutics (IMPACT) analysis pipeline that interprets raw whole-exome tumor sequencing data and matches the cancer's unique genetics to US Food and Drug Administration-approved targeted treatments.
In a paper published last week in the Journal of the American Medical Informatics Association, the researchers showed that the pipeline was able to accurately identify single-nucleotide variants and copy number variants in known tumor samples from lung cancer patients and match them with appropriate drugs for treatment.
The IMPACT tool involves four analytical modules: detecting somatic variants, calling copy number alterations, predicting drugs targeting deleterious variants, and analyzing tumor heterogeneity.
As described in the study, the researchers started with data generated by whole-exome sequencing and mapped it to a reference exome using the Burrows-Wheeler Aligner to determine single nucleotide variants. Then, they used VarScan2 to identify copy number alterations. The IMPACT tool generates a list of candidate genes based on the completion of these first two analytical modules.
IMPACT runs through a two-tier system to connect genes with therapies. The first tier follows the National Cancer Institute MATCH and MD Anderson Personalized Cancer Therapy criteria to identify genetic markers that are druggable with FDA-approved therapies, and, using the database DSigDB, identify those that are druggable with FDA-approved kinase inhibitors. The second-tier therapeutics includes all FDA-approved drugs and their gene targets found in DSigDB. Finally, the researchers also used IMPACT to measure tumor clonality and heterogeneity using mutant allele frequency estimates using the Kolmogorov-Smirnov test.
Existing tools such as SAMTools and GATK can analyze whole-exome sequencing data for variants and copy numbers, however, these take additional steps and tools to determine actionable therapeutics.
"Most of the time you get a list of variants back and what [researchers] have to do next is to go through other resources to see if it can inhibited by drug-gene interactions," Aik-Choon Tan, associate professor of bioinformatics at the University of Colorado Anshultz Medical Campus and senior author on the paper, told GenomeWeb. "We have put it together as a package."
This study served as a sanity check for the IMPACT analysis pipeline, Tan said. The first step was to show that it worked with data that had already been published, so the researchers used three matched normal-tumor TCGA lung adenocarcinoma samples from the University of California Santa Cruz Cancer Genomics Hub. They found that IMPACT successfully identified the EGFR mutation as a tumor driver and recommended FDA-approved EGFR inhibitors.
Tan and his colleagues also used the tool, in collaboration with William Robinson's laboratory at the University of Colorado Cancer Center, to retrospectively analyze a series of exome sequences from patients diagnosed with melanoma. They were able to validate the tool's ability to discover a patient's activating mutation and pair it with useful treatment.
For example, they were able to identify a BRAF mutation in one of the patients that responded to BRAF kinase inhibitors initially, Tan said. However, that patient relapsed, and upon analyzing post-treatment data, researchers determined that the patient had acquired an NRAS mutation, and was then treated with a combination of BRAF and NRAS inhibitors. When the patient relapsed again, they were able to analyze post-BRAF and NRAS treatment data, which showed that the patient had developed a deletion of the CDKN2a gene, according to Tan.
With the IMPACT tool, researchers were able to predict future precision medicine options as the patient progressed, as long as there were drugs available.
Unfortunately, once the melanoma patient acquired a CDKN2a gene deletion, there wasn't much to be done. There are currently no FDA-approved inhibitors against the CDK gene family for melanoma. However, the drug palbociclib recently earned FDA approval to treat a subset of breast cancers, and as such, it may be possible to evaluate in a clinical trial if palbociclib is useful in melanoma.
Once the IMPACT tool itself has been validated and approved for broader use, the researchers hope that it might be used as a means to generate support for clinical trial research into novel uses for an existing drug or kinase inhibitor, Tan said.
For now, the researchers have kept close tabs on kinase inhibitor clinical trials that might be useful for exploring precision medicine approaches, and as drugs are approved in new molecularly-defined indications, they will continue to update the database underlying IMPACT on a quarterly basis. They do eventually hope to automatically update the database as drugs become FDA-approved, Tan noted.
Currently, Tan and his colleagues have plans to conduct a prospective, randomized trial to compare how using IMPACT to guide treatment decisions affects patient outcomes against when treatment decisions are made without the tool. However, they need to obtain Institutional Review Board approval first, Tan said.
While IMPACT has only been used in the context of cancer, Tan believes it could one day have uses as a therapeutic matching tool in other diseases.
The IMPACT analysis pipeline is publicly available through the University of Colorado.