Janice Jia Ni Goh 1, Amy Zhan Chuxuan 1,3, Paul Anthony Ongkowidjaya 1, Cheng Shoong Chong 1, Daniel Ng 3, Eric Chan Chun Yong 3, Prakash Arumugam 2
1 Bioinformatics Institute, A*STAR (Singapore, Singapore ), 2 Singapore Institute of Food and Biotechnology Innovation, A*STAR (Singapore, Singapore), 3 Department of Pharmacy, National University of Singapore (Singapore, Singapore)
Introduction: Despite making up more than 60% of the world’s population, Asian genomes are undercharacterized. A large-scale whole genome sequencing initiative in Singapore, SG10K¹, has given us better insight into the pharmacogenomic (PGx) polymorphisms that are enriched in our population. However, the actual clinical impact of these PGx polymorphisms on drug exposure have yet to be well defined. Current Clinical Pharmacogenetics Implementation Consortium (CPIC®) guidance relies mainly on clinical studies to develop strong evidence of PGx polymorphism impact. However, clinical studies are often costly, and polymorphism frequencies vary with different populations and ethnicities, making clinical studies hard to generalize. Furthermore, recruitment can be challenging as common variants can be as low as 1% of the population frequency, making it hard to achieve sufficient numbers for meaningful analyses. This is a problem especially with highly polymorphic enzyme CYP2D6, where more than 186 unique star alleles have been described on the Pharmacogene Variation Consortium (PharmVar)², with more novel star alleles still being described and reported from different populations. CYP2D6 is also responsible for approximately 20% of the metabolism of clinical drugs³, making this a pharmacogenomic variant of interest. To develop a lower cost method that can describe PGx impact on drug exposure, we thus developed a semi mechanistic population PK model with PGx covariates informed by our yeast based microsome assay.
Methods: Known and novel CYP2D6 variants relevant to the Singaporean population were selected based on the reported frequencies in SG10K⁴. Yeast and plasmids containing wild type CYP2D6 were a gift from the Maitreya Dunham lab. Relevant mutations were introduced using polymerase chain reaction (PCR) site directed mutagenesis and the enzymes subsequently expressed in yeast and harvested. Dextromethorphan (3.125, 6.250, 12.50, 25.00, 50.00, 100.0, and 200.0μM) was used as the probe substrate, with major metabolite dextrophan as the main analyte to measure enzyme activity to characterize Michaelis-Menten kinetics.
A base model of dextromethorphan population pharmacokinetics was adapted from Abduljalil et al⁵, where hepatic clearance from CYP2D6 and covariates for CYP2D6*1, *2 and non functional variants were described. A dose of 30mg dextromethorphan was simulated for 24 hours. Intrinsic clearance was calculated as Vmax/Km. The fold change in intrinsic clearance from wild type for each mutant was calculated and used as a covariate for change (scaling factor 1 (SF1) or scaling factor 2 (SF2)) in clearance with each copy of the corresponding star allele (Equation 1).
CL_CYP2D6 = CL2D6_1*SF1 + CL2D6_2*SF2 (Equation 1)
Results: Clinically reported CYP2D6*2/*2 change in dextromethorphan clearance was used to validate the model, with our model predictions Area Under the Curve 0-24h (AUC0-24h) (3.68 µg/L*h, (1.04-12.2 µg/L*h) (95% Confidence Interval (CI)) falling within 2-fold of the simulated output using published covariates for CYP2D6*2/*2 (6.94 µg/L*h, (1.83-24.0µg/L*h)). This gave us confidence to apply the same method to other variants and evaluate their AUC0-24h. Notably, CYP2D6*10 (37.8% allele frequency in SG10K1) had a much larger impact on drug clearance. CYP2D6*10/*10 had an AUC0-24h 9.86 fold higher (29.4 µg/L*h, (7.38-84.2 µg/L*h)) than wild type *1/*1 (2.98 µg/L*h, (0.83-9.63 µg/L*h)), suggesting that for CYP2D6*10 and its novel variants (most of which had lower activity than their CYP2D6*10 backbone), dosing for CYP2D6 substrates could be lowered significantly.
Conclusion: Our yeast-based microsome assay was able to successfully inform a semi mechanistic popPK model with new PGx covariates, allowing us to extrapolate predicted drug exposures to new populations, aside from the ones that the model was previously built from. This allows us a lower cost and faster method of evaluating PGx polymorphism impact on drug clearance without having to conduct a new study, saving time and resources.
References:
1. Chan, S. H. et al. Analysis of clinically relevant variants from ancestrally diverse Asian genomes. Nat. Commun. 13, 6694 (2022).
2. PharmVar. https://www.pharmvar.org/genes.
3. Nahid, N. A. & Johnson, J. A. CYP2D6 pharmacogenetics and phenoconversion in personalized medicine. Expert Opin. Drug Metab. Toxicol. 18, 769–785 (2022).
4. Maulana, Y. et al. The variation landscape of CYP2D6 in a multi-ethnic Asian population. Sci. Rep. 14, 16725 (2024).
5. Abduljalil, K. et al. Assessment of activity levels for CYP2D6*1, CYP2D6*2, and CYP2D6*41 genes by population pharmacokinetics of dextromethorphan. Clin. Pharmacol. Ther. 88, 643–651 (2010).
Reference: PAGE 34 (2026) Abstr 12039 [www.page-meeting.org/?abstract=12039]
Poster: Drug/Disease Modelling - Other Topics