NEW YORK (GenomeWeb) – A team from Korea and the US has characterized genomic similarities and distinctions that turn up in different parts of a glioblastoma tumor or at different time points.
As they reported today in Nature Genetics, researchers from Sungkyunkwan University, Columbia University, and elsewhere used exome sequencing, transcriptome profiling, and other approaches to detect somatic mutations in more than 100 samples from 52 individuals with GBM — a set that included samples taken over time and/or from distinct tumor regions from patients at Seoul hospital or from participants in the Cancer Genome Atlas.
The team saw general mutation and expression similarities between samples from a single tumor mass. But differences began to arise between longitudinal samples or samples from multifocal or spatially separated tumors from the same individuals, leading to an apparent dip in drug response reflecting these distinct underlying seed clones.
"We showed that [multifocal-GBMs] are more genetically diverse than locally adjacent tumors and that genetic similarity between multi-region samples is associated with consistent drug response," the authors wrote.
Based on their results, the study's authors argued that it will likely be most effective to target truncal mutation events, when possible, rather than the varied, private mutations that start to occur as GBM tumors spread and evolve.
For their analyses, the researchers tapped new and existing exome sequence and/or transcriptome sequence data, representing 127 multi-region or longitudinal samples from 42 individuals with GBM at the Samsung Medical Center and 10 TCGA participants with GBM. Bulk RNA sequence data was available for 83 tumors from 41 patients, they noted, while 305 cells isolated from seven samples in three patients were subjected to single-cell RNA sequencing.
In general, the genomic mutation and expression patterns identified in the samples were consistent with a so-called "multiverse model" of tumor evolution, the authors explained. In that model, they noted, "tumor samples that are derived from different tumor masses share very few genomic alterations, indicating that tumor clones are geographically segregated at an early stage of evolution and that each clone acquires distinct private alterations, leading to the construction of multiple universes."
While mutations in certain genes, such as IDH1 or PIK3CA tended to truncal and shared across multiple samples from the same individual, for example, many other alterations varied depending on when and where the sample was taken.
When the researchers expanded their analysis to include samples from localized or "solitary" tumors, or from multifocal tumors isolated from 160 more untreated individuals with GBM, they saw an over-representation of PIK3CA mutations in the multifocal tumors.
They suspected that the propensity for PIK3CA-mutated tumors to have multiple foci may explore the relatively poor outcomes associated with mutated PIK3CA in GBM — a notion supported by response patterns detected in its subsequent chemical screen on patient-derived glioma cells from 11 GBM patients.
There, PIK3CA mutations were associated with a range of drug responses. Likewise, the researchers found that patient-derived glioma cells from multifocal GBMs tended to be more diverse than responses from cell lines originating from solitary GBMs.