Tracking Tumor Evolution through Mathematics

Although a tumor can accumulate thousands of mutations as it grows, a new study suggests that the key mutations driving malignancy are often present when the tumor forms. Subsequent mutations are mere passengers that provide no benefit to the tumor's cells, the results imply.

Researchers have long assumed that competition among a tumor's different subclones is fierce, with natural selection favoring mutations that increase the ability of particular subclones to survive, proliferate, and out-compete their rivals. This assumption remains difficult to verify, despite the vast amount of genomic data available. For example, tracking the appearance of mutations over time often isn't possible because researchers don't usually have access to sequential biopsies.

Trevor Graham, PhD, of Queen Mary University of London, UK, Andrea Sottoriva, PhD, of the Institute of Cancer Research, UK, and colleagues developed a mathematical model to overcome these obstacles. They used it to predict mutation patterns in different tumors, and by applying the model to large data sets such as The Cancer Genome Atlas (TCGA), they sought to determine if a given tumor's evolutionary dynamics were driven by selection or were largely neutral.

The scientists tested their model on 108 colorectal tumors, including 101 TCGA specimens. “We assumed that selection would be important,” Graham says. “It's a surprise that that's not what we saw.” Instead, they calculated that 35.1% of the tumors were evolving neutrally.

The researchers then analyzed 819 TCGA tumors comprising 14 cancer types, including the 101 colorectal tumors they'd used before. Overall, 31.6% of the tumors displayed neutral evolution. This pattern was particularly common for certain tumors, including stomach, lung, cervical, and bladder cancers. The model didn't fit as well for other malignancies, such as melanoma, glioblastoma, and pancreatic cancer, indicating that in these cancers, the evolutionary dynamics typically were not neutral.

Treatment would change the situation for neutrally evolving tumors, the researchers note, potentially creating a new selective pressure and allowing some subclones to expand. In other words, certain mutations accrued by tumor cells during their neutral evolution would become nonneutral and might help the cells adapt to treatment, Graham explains.

Nicholas Navin, PhD, of The University of Texas MD Anderson Cancer Center in Houston, agrees that “there are certainly periods of neutral evolution” in a tumor's history. However, he doesn't find the new analysis convincing. He and his colleagues have shown that although copy-number changes appear early on, tumors continue to accumulate mutations over time that undergo positive selection. “I'm a strong believer that selection is ongoing and that you can have additional driver mutations later in tumor evolution,” he says.

Moritz Gerstung, PhD, of the European Bioinformatics Institute in Hinxton, UK, is also a skeptic. He contends that the researchers may have overestimated mutation rates by up to 100-fold due to undetected DNA copy-number changes. This study is “a first step toward understanding how tumor composition and development are related, but in order to get there, we have to account for many other factors that could substantially change the interpretation of the data,” he says.

Graham responds that his paper's results don't clash with studies that suggest ongoing selection in tumors because “we found evidence for neutral evolution in only about 30% of cancers, and the other studies may have involved the majority (70%) of tumors that are evolving nonneutrally.” As for the issue of mutation rates, Graham point outs that “there are no previous direct measurements of the mutation rate per cell division in human cancer, but nevertheless the rates we measured are in line with the previous estimates and are also invariant to copy-number change.” –Mitch Leslie