In mapping the landscape of clustered somatic mutations, researchers have found that such clusters are enriched on driver oncogenes and associated with differential survival. They also reported the prevalence of a particular cluster pattern, kataegis, on extrachromosomal DNA, describing it as kyklonas, potentially a new mode of cancer evolution.
Using artificial intelligence, researchers at the University of California, San Diego, have begun mapping the landscape of somatic mutations that aren't randomly scattered but cluster at hotspots in cancer genomes. These clusters were enriched in driver genes, and their presence or absence could potentially predict patient outcomes (Nature 2022;602:510–7).
Clustered mutations occur in various patterns, including kataegis and omikli (Greek for thunderstorm and fog). With kataegis, large streaks of mutations are close together, resembling a storm; omikli looks like small bursts of kataegis, or a “fog-like clustering,” explains Erik Bergstrom, the study's first author.
Led by Ludmil Alexandrov, PhD, the team developed algorithms to detect and characterize every mutation in the sequenced genomes of 2,583 patients, encompassing 30 different cancers. They found, for instance, that omikli accounted for a large proportion (50.5%) of all clustered base substitutions, which occurred more often within oncogenes than tumor suppressors and at higher frequency in certain drivers—such as BTG1 and NOTCH2—compared with others. In all, “we determined that approximately 10% of known driver mutations are clustered,” Alexandrov says.
The researchers noted, intriguingly, that clustered rather than nonclustered BRAF mutations were associated with better overall survival. The opposite was true for TP53 and EGFR, however. “All we did was separate samples based on whether or not clustered mutations were detected within known driver genes,” Bergstrom remarks. “From that alone, we saw clear differences in survival, but we don't yet know why.”
Diving deeper into kataegis specifically, the team reported its prevalence on extrachromosomal DNA (ecDNA), which—unlike linear DNA—passes along genetic information unevenly and fuels aggressive cancers by being a powerhouse of oncogene amplification. Playing on ecDNA's circular motif, they coined a new name for this clustered pattern, kyklonas (Greek for cyclone). In three tumor types—61 sarcomas, 280 lung cancers, and 186 esophageal cancers—widespread, recurrent kyklonic events were seen in 45%, 28%, and 46% of samples harboring mutated ecDNA, respectively.
The main culprit causing kyklonas, an enzyme family called APOBEC3, “normally modulates innate immunity by restricting viruses, many of which have circular genomes,” Bergstrom explains. “Our theory is that APOBEC3 may regard ecDNA as a similar foreign body that needs to be cut up.”
“We saw repeated APOBEC3 attacks on ecDNA circles after they formed; it wasn't a one-off occurrence,” Alexandrov adds. “This isn't unlike what's been seen on viral genomes in HPV [human papillomavirus]-driven cancers. That's why we think APOBEC3 is mistaking ecDNA for a virus, creating kyklonas along the way and, because ecDNA often carries oncogenes, inadvertently accelerating cancer development.”
Roel Verhaak, PhD, of The Jackson Laboratory for Genomic Medicine in Farmington, CT, observes that “ecDNA has of late been revitalized as an important topic.” Not involved in this study, he was most intrigued by the hypothesis as to why kyklonas occurs. “It suggests a targeted, proactive role for APOBEC3 in trying to prevent the genesis of ecDNA,” he says. “I thought this was quite provocative, and I always appreciate it when authors go out on a limb in their study discussion.”
“There's still much to unpack with our data,” Alexandrov says, including validating kyklonas as a novel mode of cancer evolution. Meanwhile, he and Bergstrom are further exploring survival differences seen with clustered mutations and their clinical utility.
“More patients are undergoing routine genomic testing, and the information—whether or not mutations are clustered—is already in the results,” Alexandrov notes. “It just hasn't been used to predict outcomes, but that's readily implementable. As well, are there particular therapies to which a patient with clustered mutations responds, and vice versa? It's another question to pursue to better understand the prognostic power of our findings.” –Alissa Poh
For more news on cancer research, visit Cancer Discovery online at http://cancerdiscovery.aacrjournals.org/CDNews.