Integration analysis of metabolites and single nucleotide polymorphisms improves the prediction of drug response of celecoxib.

Introduction: Pharmacogenetics and pharmacometabolomics are the common methods for personalized medicine, either genetic or metabolic biomarkers have limited predictive power for drug response.

Objectives: In order to better predict drug response, the study attempted to integrate genetic and metabolic biomarkers for drug pharmacokinetics prediction.

Methods: The study chose celecoxib as study object, the pharmacokinetic behavior of celecoxib was assessed in 48 healthy volunteers based on UPLC-MS/MS platform, and celecoxib related single nucleotide polymorphisms (SNPs) were also detected. Three mathematic models were constructed for celecoxib pharmacokinetics prediction, the first one was mainly based on celecoxib-related SNPs; the second was based on the metabolites selected from a pharmacometabolomic analysis by using GC-MS/MS method, the last model was based on the combination of the celecoxib-related SNPs and metabolites above.

Results: The result proved that the last model showed an improved prediction power, the integration model could explain 71.0% AUC variation and predict 62.3% AUC variation. To facilitate clinical application, ten potential celecoxib-related biomarkers were further screened, which could explain 68.3% and predict 54.6% AUC variation, the predicted AUC was well correlated with the measured values (r = 0.838).

Conclusion: This study provides a new route for personalized medicine, the integration of genetic and metabolic biomarkers can predict drug response with a higher accuracy.