Beyond every binary is a more complex reality. And so it is with driver and passenger mutations.

The separation of tumor mutations into drivers and passengers underpins much progress in the development of targeted therapies.

By looking at passenger mutations more carefully, though, researchers at Yale University have shown that passenger mutations, too, played a role in how tumors progressed.

Analyzing roughly 2,500 cancer genomes, senior author Mark Gerstein told BioWorld, the researchers found that in addition to the known driver mutations, which exert strong effects, the tumors they looked at had additional “weak drivers” that could improve the prediction of tumor phenotype by slightly more than 10%.

The results, which Gerstein and his colleagues reported in the Feb. 20, 2020, issue of Cell, is part of the Pan-Cancer Project, a decades-long project that sequenced the whole genomes of nearly 40 different tumor types.

The main body of the Pan-Cancer Project was published in more than two dozen papers earlier in February.

The work now published in Cell “has a particular place relative to those papers,” said Gerstein, who is the Albert L Williams Professor of Biomedical Informatics and professor of molecular biophysics and biochemistry, and of computer science at Yale University, as well as the co-director of the Yale Program in Computational Biology & Bioinformatics.

The bulk of papers published with the Pan-Cancer project “represent the dominant viewpoint or headset in cancer genomics,” he said.

That dominant viewpoint is that driver mutations will largely be found in the protein-coding parts of the genome.

Gerstein’s laboratory, though, is “keen on the noncoding genome,” he said.

To date, he noted, sticking to sequencing the protein-coding parts of the genome has been a successful strategy in cancer, as well as in rare diseases.

“You can get a lot of value, for considerably less money, by just sequencing exomes,” Gerstein acknowledged.

However, he said, “my lab’s view is very much that the noncoding genome is important.”

To date, the noncoding genome has emerged as important in germline genomics of common diseases, but less so in rare diseases and cancer genomics.

Passengers drive, too

To test whether there are important functions of the noncoding genome to be found in cancer genomics as well, the team “repurposed an idea from germline genomics,” he said.

There are a number of traits such as height that are clearly highly heritable. Parental height accounts for 80% of the variability in their offspring’s height.

However, early genomewide association studies were hard-pressed to find the underlying genetic variants, leading to the concept of missing heritability.

Identifying the genes underlying common traits made progress by looking for additive effects of different genes, Gerstein said. “We took that headset to cancer genomics.”

In doing so, Gerstein and his colleagues were able to detect weak effects of many genes that have been classified as passengers to date. On the average, tumors had eight such “weak drivers” that contributed to their fate.

One unexpected finding was that Gerstein and his colleagues identified not only weak drivers that cooperated with strong drivers, but also some that were deleterious to tumors, impeding their progress.

“Many in the cancer community would be very surprised that the passengers have any effect at all,” Gerstein said. That they might be beneficial to patients, though, is “even more surprising.”

One interesting possibility to arise from the work is that the effects of classical driver genes could be context-specific, at least to a degree.

Whether that is the case is not yet discernible from the analyses that Gerstein and his team have conducted.

But, he said, “one could ask whether they manifest in one context and not in another,” or manifest differently in one context than in another.

“This is a little different, and we know it’s different,” he added. But “we think it’s a good message to get out.”