Longitudinal Study Charts Brain Tumor Evolution

Amy Blum, M.A.

Tumor evolution schematic and phylogenetic tree

Creative Commons 4.0 License

© 2016 Beerenwinkel et al.

Glioblastoma is the most common and most deadly type of brain tumor in adults due to the frequent occurrence of tumor relapse. To better understand why glioblastomas are so difficult to treat and how we might improve care in the future, scientists from Columbia University analyzed the genomes of 114 glioblastoma tumors at the time of diagnosis and at the time of recurrence, including recurrent tumors from The Cancer Genome Atlas. The study, published in Nature Genetics in June, 2016, revealed many differences between the initial and recurrent tumors, and found that the relapsed tumors likely arose years before the initial diagnosis.

The majority of the glioblastomas studied contained several mutations in the initial tumor that were absent at relapse. Because gene mutations are unlikely to change back to their original sequence, this means that the recurrent sub-tumor, or clone, that was discovered after treatment had diverged from the initial clone before diagnosis. Selection pressure from drug treatment may have given these clones the opportunity to expand and re-populate the tumor. Based on the frequency of new mutations, the researchers estimated that the average time of divergence between the diagnostic and recurrent clones was more than a decade before the initial tumor was discovered. This suggests that when glioblastoma is diagnosed, it is likely a mixture of different sub-tumors, or clones.

Tracing Tumor Development

Comparing the initial and recurrent tumors enabled the researchers uncover the evolutionary progression of glioblastoma. They found that certain mutations including IDH, PIK3CA, and ATRX represent early events, because they are often found in both tumors. Mutations that were found only in the initial or recurrent tumors, but not both, are more likely to be later events because they occurred after the two lineages diverged. These included TP53, NF1, PTEN, and genes that were mutated predominantly in recurrent tumors, MSH6 and LTBP4.

The authors also observed that in eleven percent of cases the initial and the recurrent tumor contained different mutations in the same known glioblastoma driver gene, such as EGFR, TP53, or PDGFRA. These mutations therefore represent two discrete, convergent events, and because they occurred separately in the initial and relapse clones, they likely occur late in glioblastoma development. In addition, two thirds of the patients’ initial and recurrent tumors differed in molecular subtype, accentuating the extent of divergence between glioblastoma clones.

Learning from History

These findings highlight the profound complexity of glioblastoma, but they also reveal a common order of events that may inform treatment. For example, knowing the early mutational events, which are present in all clones, can help scientists develop treatments that are effective against the whole tumor. The study also found that certain mutations are associated with relapse, such as mutations in the gene LTBP4, which stimulates the TGF-β pathway. TGF-β inhibitors may therefore be effective therapies in glioblastoma to treat relapsed tumors or to prevent relapse before it occurs.

This study reveals new insight into brain tumor biology: glioblastoma is difficult to cure because several different clones may exist in parallel before the disease is discovered and these clones complicate treatment. But now, with a better understanding of glioblastoma’s evolutionary history, researchers can begin to develop approaches to treat this disease more effectively in the future.