LOS ANGELES – At the annual meeting of the American College of Medical Genetics and Genomics held last week, researchers showed how the simultaneous detection of somatic mutations and alternate splice variants could be used to identify mechanisms of cancer drug resistance and may facilitate more personalized cancer therapy in the future.
Colette Felton, a postdoctoral researcher at the University of California, Santa Cruz, presented data on FLAIR3, a computational tool used to detect somatic variants, gene fusions, and gene fusion isoforms from long-read RNA-seq data.
Felton said in her talk that one of the drawbacks to short-read RNA-seq is that it is not the optimal technology for getting a full picture of a person's mutational landscape. She likened its use in detecting cancer gene alterations to a Buddhist parable of many blind men trying to describe an elephant.
"If everyone is touching a different part of the elephant," she said, "then everyone thinks they're seeing a different thing." Long-read sequencing, she continued, facilitates a more complete picture, enabling researchers "to see the whole elephant."
Long-read sequencing has attracted considerable commercial and research interest. The Human Genome Structural Variation Consortium, for example, recently combined both short- and long-read sequencing information to release a variant database, and late last year, Pacific Biosciences released a lower-throughput benchtop HiFi sequencer.
The Brooks Lab at UCSC, where Felton works, has been developing computational tools for analyzing long-read sequencing data for some years now. The lab's first FLAIR algorithm was designed to narrow RNA-sequencing reads down to a few high-confidence isoforms, and FLAIR2 further improved splice detection by incorporating RNA-editing data.
Felton said that one key aspect of FLAIR3 was that by combining single-nucleotide variants with gene fusions and their isoforms, the algorithm returns a "more nuanced functional interpretation of cancer variants." She demonstrated FLAIR3's capabilities with examples from studies done in osteosarcoma and lung cancer patients.
Osteosarcoma provided a natural application of FLAIR3, Felton said, as it is a cancer primarily driven by structural variants.
Felton and her colleagues investigated three osteosarcoma patients who had P10 loss, CCNE1 amplification, and MYC and CDK4 amplifications, respectively, and were sequenced via Pacific Bioscience's HiFi long-read RNA-sequencing platform.
"The question we are asking with this study is, what new alterations in druggable cancer genes can we identify through long-read RNA sequencing," she said.
Felton showed that the patient with the CCNE1 amplification was actually found to have a fusion in that gene's three prime untranslated region. While this doesn't affect the gene's protein-coding region, Felton said that it enables them to see that only one haplotype is being expanded in the pericentromeric region, which provides a more detailed understanding of its mechanism of action.
In the MYC amplification, the team identified a splicing change leading to an altered protein product. This same patient also showed a 24 base pair deletion in the KEAP1 gene that led to a 10 amino acid deletion in the KEAP1 protein.
"The really crazy thing about this," Felton said, "is that this is completely undetectable by short-read sequencing. A deletion of this size actually causes short reads to fail to align to the locus, while long reads align around it and allow you to detect this variant."
In collaboration with researchers at the Fred Hutchinson Cancer Research Center, Felton and her colleagues are also currently studying alterations in lung cancer patients. She presented findings from one such patient who had altered splicing in the BRAF gene, causing increased usage of one particular isoform of that gene called BRAF-220. The investigators discovered that a cancer-driving BRAF variant was only expressed on the BRAF-220 isoform.
The ability to identify allele-specific expression, Felton said, should facilitate treatment.
"BRAF-220 is the canonical isoform for BRAF," she said. "So in this case we can be relatively confident in recommending a BRAF inhibitor as a treatment and not worrying about there being inherent resistance through the alternative splicing of this gene."
Felton is currently preparing a manuscript and told GenomeWeb that she is now planning a study of lung cancer variants in a larger cohort of roughly 60 patients. Although it is too early to make concrete plans, she added that her lab does hope to carry these studies forward to the clinical stage.