Key result
Adding NONMEM-predicted PRU to clinical and ticagrelor exposure models did not improve 90-day outcome discrimination in 731 patients (all C-index and Brier-score bootstrap intervals crossed zero).
Why the study?
Ticagrelor reduces ischemic events in ACS but may increase bleeding in susceptible patients, motivating an evaluation of whether population PK/PD model-predicted platelet reactivity unit (PRU) dynamics improve patient-level prediction of subsequent clinical outcomes.
Does adding model-predicted platelet reactivity units (PRU) from a NONMEM-based ticagrelor PK/PD framework improve the prediction of subsequent ischemic, bleeding, and all-cause death outcomes in a UK Biobank cohort?
Cohort (n=731)
Does adding model-predicted platelet reactivity units (PRU) from a NONMEM-based ticagrelor PK/PD framework improve the prediction of subsequent ischemic, bleeding, and all-cause death outcomes in a UK Biobank cohort?
Although a NONMEM-based PK/PD framework successfully modeled ticagrelor exposure and PRU dynamics, adding model-predicted PRU derived from prescription records did not improve clinical outcome prediction in a UK Biobank cohort.
Answers are grounded in this paper and cite their sources.
Model-predicted PRU may refine outcome prediction with ticagrelor; leaves open clinical utility pending prospective validation.
Background/Objectives: Ticagrelor reduces ischemic events in acute coronary syndrome (ACS) but may increase bleeding in susceptible patients. We used a population PK and PK/PD model to describe ticagrelor exposure and platelet reactivity unit (PRU) dynamics and assessed whether model-predicted PRU improved prediction of subsequent outcomes in a patient-level analysis. Methods: We collated published evidence from 39 ticagrelor clinical studies (172 group-level datasets). The final population PK and indirect-response PK/PD models for ticagrelor, AR-C124910XX, and PRU were implemented in NONMEM with a two-compartment structure, three transit absorption compartments, and prespecified covariate relationships. Model-predicted PRU from recorded prescription exposure was linked to subsequent outcomes in a de-identified UK Biobank cohort using a prespecified 90-day landmark analysis. Cox models with clinical variables, recorded medication exposure, and model-predicted PRU were compared; a 30-day landmark analysis was used for sensitivity analysis. Results: The final model retained a two-compartment structure with three transit absorption compartments and quantified age, disease status, concomitant aspirin, and fasting effects on exposure. The indirect-response model linked combined parent-metabolite molar concentration to continuous PRU dynamics. In 731 UK Biobank participants, the 90-day ischemic, bleeding, and all-cause death analyses included 112, 48, and 129 events, respectively. Adding model-predicted PRU did not show stable incremental discrimination; all paired-bootstrap intervals for C-index and Brier-score increments crossed zero. Conclusions: We established a NONMEM-based framework linking ticagrelor dosing, exposure, and PRU dynamics to patient-level outcome prediction assessment. Although incremental predictive benefit was not established in the current cohort, the framework provides a basis for further evaluation using paired measured PK, PRU, and clinical outcome data before clinical application.
No takes yet. Share an insight, caveat, or question.
Du et al. (2026) conducted a cohort in Acute coronary syndrome / ticagrelor exposure (n=731). Model-predicted platelet reactivity unit (PRU) added to Cox models with clinical variables and recorded ticagrelor exposure vs. Cox models with clinical variables and recorded medication exposure alone was evaluated on 90-day ischemic events, bleeding, and all-cause death. Adding NONMEM-predicted PRU to clinical and ticagrelor exposure models did not improve 90-day outcome discrimination in 731 patients (all C-index and Brier-score bootstrap intervals crossed zero).
Synapse has enriched 10 related papers. Consider them for comparative context: