Two brain tumors, two evolutionary playbooks, one landmark study

Deep inside the genomes of two closely related brain cancers, a team of researchers has found evidence that evolution does not follow a single script. In a sweeping longitudinal study published in Nature, the Glioma Longitudinal Analysis Consortium, known as GLASS, has shown that IDH-mutant gliomas, the most common primary malignant brain tumors in adults under the age of 50, adapt to the pressures of tumor progression and treatment in fundamentally different ways depending on whether they are astrocytomas or oligodendrogliomas. The finding, drawn from whole-genome and whole-exome sequencing of longitudinally collected tumor specimens from 183 patients across 21 institutions on five continents, offers the most detailed picture yet of how these cancers resist therapy, and it could reshape how clinicians think about treating them.

IDH-mutant gliomas take their name from gain-of-function mutations in the metabolic enzyme genes IDH1 and IDH2. The 2021 World Health Organization classification splits them into two clinically and molecularly distinct types. Oligodendrogliomas, graded 2 or 3, are defined by the combined loss of chromosome arms 1p and 19q, along with mutations in the TERT promoter. Astrocytomas, graded 2 through 4, lack that codeletion and instead prominently feature mutations in TP53 and ATRX. Standard treatment combines surgical debulking, alkylating chemotherapy, most often temozolomide, and ionizing radiation. Yet virtually all of these tumors eventually develop resistance and recur, and because patients are typically young to middle-aged, the ubiquitous emergence of treatment resistance translates into a disproportionate number of life years lost.

To trace how resistance emerges, the GLASS Consortium assembled a global cohort with extensive clinical annotation, contributed by institutions in Australia, Brazil, China, France, Germany, Japan, Luxembourg, South Korea, the United Kingdom and the United States. Patients had undergone at least two time-separated surgical procedures, allowing the researchers to compare tumor genomes before and after treatment. After rigorous quality filtering, samples from 183 patients, 94 astrocytomas and 89 oligodendrogliomas, were retained for analysis. Most initial tumors were histologically grade 2, and 74 percent of patients received alkylating chemotherapy, with temozolomide being the main agent used. Radiotherapy treated roughly 68 to 69 percent of patients across both tumor types. The median follow-up period was 6.5 years for astrocytoma patients and 11.4 years for oligodendroglioma patients, considerably longer than in historical reference datasets such as The Cancer Genome Atlas.

The most striking result is a dichotomy in how the two tumor types acquire new driver alterations at recurrence. In oligodendrogliomas, the researchers discovered significant enrichment of somatic single-nucleotide variants in NOTCH1 and the PI3K catalytic subunit PIK3CA, along with a trend toward enrichment of the regulatory subunit PIK3R1. These newly acquired mutations appeared regardless of previous treatment and showed mutual exclusivity, meaning tumors tended to carry one or the other but not both. In leukaemias, NOTCH1 has been implicated as a regulator of PTEN with downstream effects on PI3K signalling, which could explain why these mutations are mutually exclusive in oligodendrogliomas. NOTCH1 mutations have also been associated with reduced cellular differentiation and resistance to IDH inhibitors, suggesting they may serve as biomarkers of resistance to a promising class of targeted drugs.

Astrocytomas, by contrast, progressed through a completely different mechanism. Rather than acquiring new point mutations in driver genes, recurrent astrocytomas showed significant enrichment of focal DNA copy-number alterations. Homozygous deletions of CDKN2A/B, a well-known tumor suppressor locus encoding cell-cycle regulators, were significantly enriched at recurrence in non-hypermutant astrocytomas but not in oligodendrogliomas. Focal high-level oncogene amplifications targeting chromosomal bands containing CCND2, CDK4, CDK6, EGFR, MYC, MYCN, PDGFRA and MET-PTPRZ1 were also detected at significantly increased frequency in recurrent astrocytomas. In total, 38 of 94 recurrent astrocytomas contained at least one focal oncogene amplification, compared with 21 of 94 initial astrocytomas. A minority of these amplifications, all found in astrocytomas, were classified as extrachromosomal DNA, a particularly aggressive form of oncogene activation.

Clonality analysis using PyClone-VI reinforced this dichotomy. Acquired PIK3CA, NOTCH1 and PIK3R1 mutations were predominantly clonal in oligodendrogliomas, meaning they appeared in nearly all tumor cells, indicating strong selection. Acquired CDKN2A deletions were mostly clonal in astrocytomas, suggesting a similarly powerful selection effect on cells carrying that homozygous deletion. The researchers also confirmed the previously reported phenomenon of temozolomide-induced hypermutation, in which a subset of both tumor types, roughly one in five, recurred with more than 10 mutations per megabase and more than 25 percent of recurrence-specific mutations attributable to the SBS11 mutational signature associated with alkylating agents. In hypermutated oligodendrogliomas, newly acquired mutations in MSH6, a mismatch repair gene, and PIK3CA were observed, consistent with the idea that defective DNA repair drives the hypermutant phenotype.

The prognostic implications of these divergent trajectories are substantial. Hypermutation at recurrence was confirmed as an unfavourable prognostic marker in both tumor types. CDKN2A/B homozygous deletions were associated with poor post-recurrence survival, with a median of 1.1 years versus 3.2 years, validating the WHO’s 2021 classification of this alteration as a grade-4 marker in recurrent astrocytomas. Focal oncogene amplifications also indicated significantly shorter post-recurrence survival, with a median of 1.2 years versus 3.8 years. Combining CDKN2A/B homozygous deletions and focal oncogene amplifications yielded an even starker contrast in non-hypermutant astrocytomas, with median post-recurrence survival of 1.5 years versus 5.3 years. External validation with the AACR GENIE cohort confirmed the poor prognostic effect of these driver copy-number alterations. PI3K- and NOTCH1-mutant oligodendrogliomas also associated with worse overall survival in the GENIE validation set, though further validation is needed.

Beyond driver genes, the researchers examined genome-wide patterns of mutational signatures, which reflect the underlying mutational processes in a tumor and can be associated with external exposures or internal biological defects. They identified 28 SBS or DBS signatures and 12 ID signatures across the cohort. Several signatures were differentially observed between oligodendrogliomas and astrocytomas. The C>T SBS97 signature, previously associated with mismatch repair deficiency, was found only in astrocytomas, as were SBS37 and SBS101 of unknown aetiology. By contrast, the C>A-dominated SBS107 signature was detected in oligodendrogliomas but not in astrocytomas. The temozolomide-associated SBS11 was detected in 39 percent of temozolomide-treated astrocytomas and 52 percent of temozolomide-treated oligodendrogliomas. A newly characterized signature, SBS119, enriched for T>C substitutions and resembling microsatellite-instability-associated patterns, was detected in non-hypermutated temozolomide-treated tumors of both types, and its proportion correlated positively with the number of temozolomide cycles administered. Radiation-treated astrocytomas showed significant enrichment of the ID8 signature, which was also associated with inferior post-recurrence outcomes in both tumor types.

The researchers also uncovered a progressive increase in chromosomal instability that was specific to astrocytomas. The fraction of genome altered, a measure of aneuploidy, increased significantly at recurrence in astrocytomas and in hypermutant but not in non-hypermutant oligodendrogliomas. Whole-genome doubling frequency increased significantly in recurrent astrocytomas, from 7 percent to 20 percent, but not in recurrent oligodendrogliomas. Chromothripsis, a catastrophic single event that shatters and rearranges chromosomes, was already more frequent in initial astrocytomas relative to initial oligodendrogliomas and continued to be significantly more frequent at recurrence. Its presence was significantly associated with previous radiotherapy in astrocytomas but not in oligodendrogliomas, linking treatment exposure to tumor-type-specific patterns of acquired chromosomal instability. Kataegis, focal hypermutation bursts driven by APOBEC mutagenesis, was also significantly higher in recurrent astrocytomas than in recurrent oligodendrogliomas, further supporting the idea that chromosomal instability processes are specific to tumor type.

The authors suggest that these divergent resistance mechanisms may reflect early, type-specific evolutionary trajectories. The near-ubiquitous presence of TP53 mutations in astrocytoma might contribute to progression mechanisms such as mitotic errors and chromosomal missegregation, while ATRX deficiency has been repeatedly associated with replication stress, DNA damage and abnormalities in chromosomal cohesion. Together, these alterations could create a genome prone to acquiring new copy-number alterations that enable disease progression. Oligodendrogliomas, by contrast, arise from a genomic background defined by 1p/19q codeletion with intact p53 and ATRX pathways, and appear to favour point mutations as their route to resistance. The study acknowledges an inherent selection bias in longitudinal multi-resection cohorts, since inclusion requires survival to at least a second surgery. Still, the findings provide a key framework for understanding the molecular changes that take place after gliomas become resistant to treatment with IDH inhibitors, including the recently approved drug vorasidenib, and point toward the development of type-stratified and evolution-informed treatment strategies that could finally account for the two very different ways these cancers fight back.

Subject of Research: Divergent genomic evolution in astrocytomas and oligodendrogliomas

Article Title: Divergent genomic evolution in astrocytomas and oligodendrogliomas

Article References: Divergent genomic evolution in astrocytomas and oligodendrogliomas. (n.d.). https://doi.org/10.1038/s41586-026-11150-x

Image Credits: AI Generated

DOI: 10.1038/s41586-026-11150-x

Keywords: Divergent, genomic, evolution, astrocytomas, oligodendrogliomas, scientific research