Woan Torng Tang 1, David Tat Wei Yong 2, Sabariah Noor Harun 1, Azhari Rosman 2, Siti Maisharah Sheikh Ghadzi 1
1 School of Pharmaceutical Sciences, Universiti Sains Malaysia (, Malaysia), 2 Department of Cardiology, National Heart Institute (, Malaysia)
OBJECTIVES:
Rivaroxaban is a cornerstone of stroke prevention in atrial fibrillation (AF), yet major bleeding remains a significant clinical concern [1]. Heart failure (HF) presents a unique paradox: while blood stasis necessitates anticoagulation [2], HF is also a potent independent predictor of bleeding [3]. Given that rivaroxaban undergoes dual renal and hepatic elimination [4], it is highly susceptible to HF-induced changes in organ perfusion. However, the distinct influences of left- and right-ventricular function on rivaroxaban pharmacokinetics (PK) and clinical hazard remain poorly characterised. The objective of this study was to develop a population PK of rivaroxaban and its pharmacodynamics (PD) impact on Prothrombin Time (PT), alongside characterizing the longitudinal hazard of major bleeding using Time-to-Event (TTE) modelling. This work performed a comprehensive covariate analysis to evaluate whether HF severity markers, Ejection Fraction (EF) and Tricuspid Annular Plane Systolic Excursion (TAPSE), provide superior predictive value for drug disposition and clinical hazard compared to binary “Yes/No” HF classifications and standard clinical factors.
METHODS:
Population PK/PD Modelling:
A prospective, sparse-sampling PK/PD study was conducted at the National Heart Institute, Malaysia, in adult patients (≥18 years) with AF receiving steady-state rivaroxaban therapy. Three time-matched blood samples were collected per patient, which were pre-dose (<1h), early post-dose (1–3h), and late post-dose (>3–16h) samples. Rivaroxaban plasma concentrations were quantified via a validated liquid chromatography-tandem mass spectrometry (LC-MS/MS) assay, and PD response was assessed through PT.
Nonlinear mixed-effects modelling was performed using NONMEM (v7.5). Population parameters were estimated using the First-Order Conditional Estimation method with Interaction (FOCE-I). Rivaroxaban disposition was characterized via a one-compartment model (ADVAN2 TRANS2), evaluating both Transit Compartment (MTT) and absorption lag-time (ALAG) structures. To stabilize the sparse-sampling design, the $PRIOR NWPRI subroutine was tested with literature-informed priors. Stochastic optimization included comparing diagonal and block OMEGA matrices (BLOCK(3)) for exponential IIV, while residual unexplained variability (RUV) was assessed via additive, proportional, and combined error models.
For PD, a two-stage sequential approach utilized individual PK parameters to drive PT response. Structural models (Linear, Log-linear, Emax, Power) were evaluated using the Bayesian Information Criterion (BIC).
Time-to-Event (TTE) Modelling:
A longitudinal analysis of retrospective data (2012–2022) characterised the hazard of the first major bleeding event (defined by ISTH criteria). Parametric survival models (Exponential, Weibull, Gompertz) were evaluated using the Laplace method to identify the underlying hazard distribution from treatment initiation to event or censoring.
Covariate Analysis and Validation:
A systematic, stepwise covariate search (Forward: ∆OFV > 3.84; Backward: ∆OFV > 6.64) was executed for modelling. Continuous covariates were centred on population medians or clinical reference to enhance stability. For PKPD modelling, model accuracy was assessed via Xpose4 goodness-of-fit. Validation was performed using Visual Predictive Checks (VPC) with 1000 simulations. The TTE model was assessed using Kaplan-Meier VPCs (KM-VPC) with 500 simulations. For both modelling, parameter precision and uncertainty were evaluated via Sampling Importance Resampling (SIR).
RESULTS:
For the population PK/PD modelling, 106 subjects were recruited and divided equally into a development cohort (n=53; 150 samples) and an external validation cohort (n=53; 142 samples). Rivaroxaban plasma concentrations in the development set ranged from 10.92 to 788.38 ng/mL, and no specialized handling of below-limit-of-quantitation (BLOQ) values was required.
Rivaroxaban pharmacokinetics were best characterised by an oral one-compartment structural model. The typical population estimates for apparent clearance (CL/F), apparent volume of distribution (V/F), and the absorption rate constant (Ka) were 4.63 L/h (RSE 4.6%), 43.5 L (RSE 7.1%), and 0.536 h-1 (RSE 10.9%), respectively. Substantial biological diversity was observed in the cohort, with inter-individual variability (IIV) identified for CL/F (60.2%), V/F (75.7%), and Ka (67.8%). The variance-covariance relationship between these random effects was successfully characterized using a block OMEGA structure, while the RUV was captured through a proportional error model of 35.1%. EF was identified as a significant predictor of Ka. Every 1% increase in EF above 45% increased Ka by 2.165%, indicating faster absorption in patients with higher cardiac output. Bilirubin (BIL) was a significant predictor of elimination, where each 1 µmol/L increase above 14 µmol/L resulted in a 0.826% reduction in CL/F.
For the PD analysis, a linear model best described the relationship between rivaroxaban concentration and PT, with a baseline PT of 12.1 seconds. The concentration-response slope was significantly influenced by concomitant medications. Anti-epileptic drug use increased the slope by 3.77-fold, while loop diuretic use increased the slope by 1.40-fold. The final PK/PD model was validated, passing both internal VPC and external validation, confirming its predictive accuracy.
For the TTE modelling, a total of 363 patients with 876 observations was included. Among these patients, 32 patients experienced major bleeding over a median follow-up of 2.64 years (follow up time ranged from 1 day to 10.12 years). The Weibull model best described the bleeding risk and indicated a decreasing hazard, with the highest bleeding risk occurring early in treatment. Higher serum albumin (>35 g/L) and TAPSE >1.8 cm were shown protective against major bleeding, with reduction of bleeding risk by 18.5% (adjusted odd ratio (aHR) 0.81; 95% confidence interval (CI) 0.78–0.86) and 81.4% (aHR 0.19; 95% CI 0.08–0.43), respectively. Conversely, concomitant use of aspirin, dual antiplatelet therapy, clopidogrel, and hypertension significantly increased bleeding risk. The TTE model demonstrated good predictive accuracy. KM-VPC plots showed observed bleeding events remained well within the 95% prediction intervals of the simulated data.
CONCLUSION:
This study is the first to demonstrate that quantitative markers of left- and right-ventricular performance outweigh binary HF classifications and standard covariates (age, weight, renal function etc.) in predicting rivaroxaban PK/PD and bleeding outcomes. The findings reveal a functional differentiation in rivaroxaban outcomes. Left-ventricular function, EF, serves as a primary determinant of drug disposition, likely reflecting the impact of systemic perfusion on intestinal absorption and metabolic flux [5]. Conversely, right-ventricular function, TAPSE, acts as the critical driver of bleeding hazard, potentially capturing the contribution of venous congestion to tissue fragility and bleeding vulnerability [6,7]. These findings suggest that precision anticoagulation in AF patients requires a shift toward quantitative hemodynamic assessment, allowing for more accurate risk stratification and individualized dosing in complex heart failure populations.
References:
1. Patel MR, Mahaffey KW, Garg J, Pan G, Singer DE, Hacke W, et al. Rivaroxaban versus warfarin in nonvalvular atrial fibrillation. N Engl J Med. 2011;365(10):883-91.
2. Circulation. Risk of Thromboembolism in Heart Failure: An Analysis From the Sudden Cardiac Death in Heart Failure Trial (SCD-HeFT). Congest Heart Fail. 2007;13(4):241.
3. Kirchhof P, Haas S, Amarenco P, Hess S, Lambelet M, Van Eickels M, et al. Impact of modifiable bleeding risk factors on major bleeding in patients with atrial fibrillation anticoagulated with rivaroxaban. J Am Heart Assoc. 2020;9(5):e009530.
4. Mueck W, Stampfuss J, Kubitza D, Becka M. Clinical pharmacokinetic and pharmacodynamic profile of rivaroxaban. Clin Pharmacokinet. 2014;53(1):1-16.
5. Ogawa R, Stachnik JM, Echizen H. Clinical pharmacokinetics of drugs in patients with heart failure: an update (part 2, drugs administered orally). Clin Pharmacokinet. 2014 Dec;53(12):1083-114. doi: 10.1007/s40262-014-0189-3.
6. Ghio S, Recusani F, Klersy C, Sebastiani R, Laudisa ML, Campana C, et al. Prognostic usefulness of the tricuspid annular plane systolic excursion in patients with congestive heart failure secondary to idiopathic or ischemic dilated cardiomyopathy. Am J Cardiol. 2000;85(7):837-42. doi: 10.1016/S0002-9149(99)00877-2.
7. Kjaergaard J, Iversen KK, Akkan D, Møller JE, Køber LV, Torp-Pedersen C, et al. Predictors of right ventricular function as measured by tricuspid annular plane systolic excursion in heart failure. Cardiovasc Ultrasound. 2009;7(1):51. doi: 10.1186/1476-7120-7-51.
Reference: PAGE 34 (2026) Abstr 11920 [www.page-meeting.org/?abstract=11920]
Poster: Clinical Applications