Introduction/Objective: Thiopurines (azathioprine, 6-mercaptopurine) are pivotal for IBD remission induction but possess a narrow therapeutic index and large PK/PD variability. We assess integrating TPMT and NUDT15 genotyping with pharmacometric models to support genotype-directed, model-informed precision dosing beyond empiric approaches and traditional therapeutic drug monitoring. Methods: We performed a narrative review of pharmacometric and pharmacogenetic research, evaluating (i) genotype–phenotype correlation for TPMT and NUDT15 by ancestry; (ii) population PK/PD and Bayesian or nonlinear mixed-effects modeling with genetic, biochemical, and co-medication covariates; and (iii) biomarker approaches comparing red-cell 6-thioguanine nucleotide assays vs. DNA-thioguanine, and clinical outcomes. Results: NUDT15 R139C alleles impart high early-onset leukopenic risk at nominal doses, particularly in East/South Asians, challenging genotype-blind dosing and conventional 6-TGN–based monitoring. TPMT polymorphisms, despite their traditional prominence, exhibit genotype–phenotype discordance and diminished predictive value outside of primarily Caucasian populations. Bayesian and mixed-effects models of the present day provide individualized dosing in real time; however, widely generalizable, genotype-normalized PK/PD paradigms and routine inclusion of higher-resolution biomarkers like DNA-TG are limited. Discussion: Dependence on historic TDM and TPMT-favoring strategies is inadequate in heterogeneous populations and for NUDT15 variant-associated early toxicity. Incorporation of diplotype-directed MIPD with solid PK/PD modeling and improved biomarkers may change thiopurine treatment from dose-measure-reactive to dose-predictive. Challenges involve ancestry-sensitive algorithms, standardized DNA-TG assays, and EHR-based, validated decision support. Conclusion MIPD combining TPMT/NUDT15 diplotypes, adaptive PK/PD modeling, and high-resolution biomarkers can make thiopurine treatment in IBD a modernized reality. Widespread implementation necessitates ancestry-adapted, clinically established algorithms, decision support, and multicenter testing to provide safe, effective, and equitable dosing.
P et al. (Fri,) studied this question.