NEW YORK – A team from China and Australia has linked mutations in a known bladder cancer driver gene to broader somatic mutation profiles patterns in the tumors, while demonstrating the potential for these mutations to help predict treatment response.
As they reported in Cell Genomics on Friday, the researchers used machine learning and whole-genome sequence data spanning 382 bladder cancer cases assessed with the MSK-IMPACT assay to compare genome-wide mutation patterns present in tumors from individuals with or without mutations in the driver gene ERCC2. The collection included 343 cases with wild-type ERCC2 and 39 ERCC2-mutated cases.
"While ERCC2 mutations are known to influence cisplatin sensitivity, their exact effects on genome instability and prognosis have remained elusive," senior and co-corresponding author Jason Wong, a researcher affiliated with the University of Hong Kong, the University of Hong Kong's Centre for PanorOmic Sciences, and the Hong Kong Science Park, said in an email.
Together, he explained, the bladder cancer tumors suggested that ERCC2 mutations coincide with "substantial" shifts in the broader patterns of somatic mutations present in the tumors.
Wong noted that the team's findings suggest that ERCC2-mutated bladder cancer cases have "a distinct genomic distribution of somatic mutations, signaling a significant departure from normal mutation patterns."
For example, the team found that ERCC2 wild-type tumors tended to have a higher proportion of closed chromatin, coupled with a lower mutational burden, whereas the ERCC2-mutant tumors were marked by mutation hotspots at CTCF-cohesin binding sites, enhanced overall mutation patterns, and altered APOBEC cytosine deaminase enzyme activity.
"While the biological mechanisms underlying ERCC2 mutation-driven mutagenesis require further investigation," the authors wrote, "our results support the role of ERCC2 mutants in compromised DNA repair at open chromatin."
The genomic analyses also pointed to potential overlap between mutations in ERCC2 and decreased expression of the UNG gene, which codes for a uracil-DNA glycosylase enzyme that has been linked to APOBEC-related mutation patterns. Likewise, the researchers saw uracil accumulation at mutation hotspots in the ERCC2-mutated tumors, hinting at ties between the mutations and uracil misincorporation.
Consistent with findings reported in the past, the presence of pathogenic ERCC2 mutations coincided with enhanced response to platinum-based chemotherapy, prompting the team to put together a machine learning model for distinguishing between ERCC2 mutations with predicted driver or passenger roles.
"[U]ntil now, there has been uncertainty about prognosis in tumors harboring ERCC2 mutations that fall outside of known hotspots in the gene," Wong said.
"Our study first confirmed that ERCC2 hotspot mutations are indeed an independent predictor of prognosis in bladder cancer patients," he explained. "Then, we show that by applying machine learning on whole bladder cancer genome sequencing data, it is possible to infer the pathogenicity of non-hotspot ERCC2 mutations and, therefore, infer the prognosis of patients."
Along with the cisplatin chemotherapy response patterns identified, the investigators noted that the current findings offer clues to other potential treatment responses in bladder cancer cases marked by ERCC2 mutations.
Given the apparent overlap between ERCC2 and UNG deficiency, Wong explained, the findings from the study hint at the possible effectiveness of uracil metabolism-targeting drugs such as 5-fluorouracil in bladder cancer tumors containing ERCC2 mutations.
Still, he cautioned that that possibility "is an early-stage hypothesis, and much more preclinical research would need to be performed to determine whether it is a worthy avenue of investigation."