With medicine as with food, one man’s meat can be another man’s poison. A doctor might prescribe identical doses of the same drug to treat two patients for the same disease, only to find what works for one patient may be ineffective—or toxic—for the other. This is partly because a drug’s efficacy isn’t just based on what it’s made of, but on how well our bodies process it.
“A single liver enzyme, CYP2D6, critically influences how we metabolise nearly one in five clinically-prescribed medications, from painkillers and heart-rate regulators to anti-cancer drugs and antidepressants.” said Nicolas Bertin, a Group Leader at the A*STAR Genome Institute of Singapore (A*STAR GIS). “Yet as many as 170 different versions of the CYP2D6 gene have been identified in the global population, each affecting medication safety and efficacy to varying degrees in different people.”
Although CYP2D6 stands among the top 20 genes known to have strong gene-drug interactions, the prevalence of its variants remains relatively under-characterised, especially among people of non-European ancestries. Bertin, who leads A*STAR GIS’s Genome Research Informatics and Data Science (GRIDS) Platform, explained that CYP2D6’s complex variation patterns pose challenges when trying to accurately determine—via short-read whole-genome sequencing-based methods—the specific versions of CYP2D6 carried by an individual.
Aiming to examine the distribution of CYP2D6 variants in a predominantly Asian population, Bertin and A*STAR GIS colleagues teamed up with the University of the Witwatersrand, South Africa, and the A*STAR GIS-seeded startup Nalagenetics. They drew on a cohort of 1,850 Singaporean residents of Chinese, Malay and Indian ancestries whose genomes had been sequenced as part of the SG10K_Health dataset, assembled in Phase I of Singapore’s National Precision Medicine (NPM) programme. GRIDS has been tasked to develop the NPM programme’s genome data analytics infrastructure.
To enhance the accuracy of variant identification, the team developed a dedicated bioinformatics workflow that used three different tools to comb through short-read sequencing data, as well as a novel consensus algorithm—developed in-house—to confirm variants when two of the three tools concurred.
The results showcased the unique genetic makeup of East, South and Southeast Asian populations. Normal CYP2D6 metabolisers were the most common, yet made up a smaller proportion of the cohort compared to global studies (53.9 versus 64–68 percent). Poor metabolisers were found in all three ethnic groups, although at a much lower frequency than in Caucasian populations, while the prevalence of ultra-high metabolisers was double that of poor metabolisers.
“Our study also uncovered 14 possible new CYP2D6 variant groups (haplotypes), with seven found in multiple people,” said Bertin. “This suggests there might be more yet-to-be-characterised versions of CYP2D6 prevalent in people of Asian ancestries.”
Critically, the team pinpointed actionable CYP2D6 variants—those with direct implications for treatment selection and drug dosing strategies—in over 80 percent of the cohort when focusing on the 10 most common CYP2D6 haplotypes.
“This research represents the most comprehensive cataloguing of the variation landscape of CYP2D6 in East, South and Southeast Asian populations thus far,” noted Bertin, adding that their findings could help refine diagnostic testing and improve prescription guidelines for patients based on genomic profiles.
The A*STAR-affiliated researchers contributing to this research are from the A*STAR Genome Institute of Singapore (A*STAR GIS).
