NEW YORK – In new data presented at the annual meeting of the American Association for Cancer Research this week, liquid biopsy firm Guardant Health has demonstrated that an add-on analysis module developed for its circulating tumor DNA testing data could detect and deeply characterize secondary reversion mutations in homologous recombination repair genes in patients who developed resistance to PARP-inhibitor treatment.
The study involved a novel informatic analysis tool that company researchers developed for this specific niche and represents the type of exploratory and experimental research that Guardant has targted with the launch of its clinical-genomic database service, GuardantInform, last summer.
Inactivating mutations in HRR genes, which include BCRA1/2, ATM, CDK12 and others, are prevalent in tumor types such as breast, ovarian, prostate and pancreatic cancers, and can be used to identify patients likely to respond to PARP inhibitors and platinum-based chemotherapy.
Unfortunately, in the context of such treatments, secondary reversion mutations in the same HRR genes can also emerge, with the potential to drive resistance by restoring protein function.
Describing the study in a presentation at the meeting being held online this week, Guardant bioinformaticist Yu Fu said that he and his coauthors had hoped to use Guardant's large database of clinically tested patients to try to characterize the prevalence and spectrum of these reversions across a broad population.
According to Yu, lipid biopsy offers a better tool for analyzing reversions than tissue sequencing because of its ability to capture DNA from multiple tumor clones over multiple time points. "A tumor biopsy is limited in that it can only capture a local part of the tumor due to tumor heterogeneity, and thus it cannot easily capture all the resistance mutations in different tumor clones," he said.
Working with collaborators at the Institute of Cancer Research, London, Fu and his team analyzed cell-free DNA samples from more than 75,000 late-stage cancer patients who had been tested as part of routine clinical care using Guardant's clinical cancer genotyping assay, Guardant360.
Using a bespoke HRR deficiency (HRD) Module designed to annotate SNVs and indels and call multi-exon deletions and copy number loss, the group catalogued deleterious germline and somatic mutations in BRCA1, BRCA2, ATM, and CDK12. For the purposes of the analysis, the group defined deleterious alterations as frame shift, nonsense, splice site and CLINVAR pathogenic missense mutations.
According to the authors, deleterious BRCA1 mutations were present at a rate of 2.7 percent in the more than 20,000 breast, ovarian, prostate and pancreatic cancer patients in the database. BRCA2 mutations were more common, at 4.8 percent, ATM alterations appeared in 2.9 percent of the cohort, and CDK12 mutations in 1.7 percent.
When the team looked for reversions, they found that most (8.7 percent) occurred in BRCA1 and BRCA-2 mutants compared to less than 1 percent of CDK12 mutants and none in those with deleterious germline or somatic ATM mutations.
Reversions also showed selective gene and allele representation. In breast cancer, for example, reversions in BRCA2-mutated patients were twice as frequent as those in BRCA1 mutants (12.9 percent versus 5.5 percent), while the near opposite was true for ovarian cancers.
In pancreatic cancer patients, about 7,5000 individuals in the cohort, HRR gene reversions presented exclusively in BRCA2 mutants. CDK12 reversions were relatively rare with only one incidence detected among more than 5,000 prostate cancer patients.
"We could see that there are some very obvious pathogenic mutation hotspots," Fu said. "For BRCA2, we can see that there are at least two very obvious reversion peaks … [which] combined can explain more than 10 percent of all reversions." There were also some clear reversion cold spots, he added. For example, "pathogenic mutations in the last third of BRCA1/2 proteins rarely reverted."
Finally, the team also examined the distances from the reversion mutations to the initial pathogenic mutations, finding that the majority of BRCA1/2 reversions were caused by small variants that were "very, very close to the initial deleterious sites."
For example, for BRCA1, the median distance was 17.5 base pairs, and for BRCA2 the median distance was just 15 base pairs," Fu said.
Regardless of the gene, the majority of all the reversions Fu and his colleagues detected were small indels — 65 percent were caused by small deletions and 8 percent were caused by small insertions. Another 17 percent represented larger intergenic deletions.
Overall, 16 percent of BRCA1/2 mutant breast, ovarian, pancreatic, prostate patients in the cohort reverted, which he said was consistent with previous reports.
The ultimate hope, according to Fu and his colleagues, is that understanding the details and intricacies of these resistance patterns may aid in the treatment selection for progressing patients, although achieving that goal will require significant further research to test hypotheses generated by the trends seen in studies like this, he said.
In a discussion following the session, Alexandra Snyder, associate vice president of translational oncology at Merck, said it is "tantalizing to think about how these findings can be applied clinically in the space of therapies that target such alterations and are clearly impacted by reversions."
Snyder added that the results illustrate the impressive scale of Guardant's database and the opportunity that the company's ctDNA technology provides for exploratory and discovery-oriented studies of things like drug resistance, which may not be possible in the context of more static or spatially restricted data from tumor tissue.