NEW YORK – Personalized genomics company MyOme presented data at the recent American Society of Clinical Oncology conference demonstrating its ability to train and validate a cross-ancestry polygenic risk score (caPRS) for breast cancer.
The Menlo Park, California-based company intends to launch a product based on these and other data roughly within the next 12 months.
Premal Shah, the company's CEO, said that the final product will incorporate data that is still being collected, but the data presented at ASCO serves as a proof of concept. Shah anticipates publishing data from ongoing studies related to the caPRS in the next six to nine months.
MyOme conducted a retrospective study of 150,601 women, of whom 31,863 (approximately 21 percent) were of non-European descent. The company obtained its data from multiple cohorts, including the Women's Health Initiative, the Multiethnic Cohort, the ROOT Cohort, and the UK Biobank.
Beyond women of European ancestry, participants identified as South Asian, East Asian, African, and "admixed American," which included Hispanic and Latina.
For each participant, MyOme calculated an ancestry-based PRS derived from the five superpopulations in the 1000 Genomes Project, which it then integrated into a final combined model.
"We then combine those and weight them by fractional ancestry and the effect size of each of those PRS scores," said Kate Im, co-founder and head of research at MyOme.
By building such an ancestry-aware model, explained Akash Kumar, MyOme's cofounder and chief medical officer, MyOme hopes to address the criticism that a SNP may increase the risk in one population yet decrease it in others.
When MyOme does launch its intended caPRS product, it will be joining a small but active commercial space that includes Myriad Genetics, Invitae, and Allelica, among others.
Myriad, for instance, launched its RiskScore breast cancer PRS in 2017 and has since continued to improve its accuracy for women regardless of their ancestry. Similar to MyOme's study population, approximately 22 percent of Myriad's study cohort consisted of women with non-European ancestries.
That proportion has remained relatively stable, said Thomas Slavin, Myriad's chief medical officer.
"However, what's most important for PRS performance is the size of each population used in the development and validation," he explained. "For instance, RiskScore development was based on more than 50,000 patients of African descent and 50,000 patients of Asian descent."
As Myriad continues to evaluate data from its RiskScore assessments, the original 86-SNP PRS has evolved to incorporate 93 SNPs.
In contrast, MyOme asserts that the way in which its approach assesses the whole genome results in a PRS composed of "hundreds of thousands" of SNPs.
Invitae and Allelica also recently partnered to develop their own cross-ancestry breast cancer PRS. The two companies plan to leverage retrospective case-control analyses using both public datasets and genetic and clinical research data generated by Invitae. A breast cancer PRS developed by Allelica and currently available as a preprint on BioRxiv will serve as the foundation for creating the collaboration's cross-ancestry model.
"We are still in the early phases of our development, largely gathering data," said Robert Nussbaum, Invitae's chief medical officer. "We plan to begin data analysis and active model development later this summer."
While MyOme's cross-ancestry PRS for breast cancer is furthest along in development, the company has others in the pipeline. These include neurodevelopmental conditions such as autism, intellectual disability, and general developmental delay.
"We're looking forward to, in the coming months, sharing data on how we've incorporated that cross-ancestry polygenic risk score with the conventional clinical risk predictors that are used in breast cancer," Kumar said.