The aim of this study was to investigate the population pharmacokinetic and pharmacodynamic relationship of rivaroxaban in elderly healthy Chinese subjects.
Plasma concentrations, factor Xa activity, activated partial thromboplastin time, prothrombin time and Hep test were determined after single dose of 5, 10, 20, 30 and 40 mg rivaroxaban. Nonlinear mixed-effects modeling with First Order Conditional Estimates (FOCE) was used to build up the population pharmacokinetic and pharmacodynamic relationship of rivaroxaban. Model evaluations included objective function value (OFV), goodness-of-fit (GOF), precision of parameter estimates. The primary tool used for selection between hierarchical models was their difference in OFV. Internal model validation was performed by visual predictive check (VPC).
The population pharmacokinetic of rivaroxaban is described by an oral, two-compartment model with first-order absorption and elimination. Factor Xa activity correlated with rivaroxaban plasma concentrations following an proportional inhibitory Emax model; prothrombin time (PT) and rivaroxaban plasma concentrations correlated with a linear model; activated partial thromboplastin time (aPPT) and rivaroxaban plasma concentrations correlated with Emax model; Hep test and rivaroxaban plasma concentrations also correlated with a Emax model.
Get your grade
or your money back
using our Essay Writing Service!
This population analysis in elderly healthy Chinese subjects demonstrated that rivaroxaban has predictable, dose-dependent pharmacokinetic and pharmacodynamic relationship that was well described by an oral two-compartment model and linear/Emax model with lower inter-individual variability.
Thromboembolic disorders are a major cause of morbidity and mortality . Venous thromboembolism (VTE), a potentially life-threatening disease which includes deep vein thrombosis (DVT) and pulmonary embolism (PE), is a common complication among hospitalized patients. Atrial fibrillation (AF) accounts for approximately one in six ischemic strokes and is a potentially preventable cause of stroke related disability, dementia, and death .
As one of the strategies to decrease the morbidity and mortality of thromboembolic disorders, anticoagulation has been suggested to be the regular use for the prevention and treatment of VTE, the prevention of stroke in patients with AF and the secondary prevention of myocardial infarction in patients with acute coronary syndrome. The prevalence of medical conditions representing a risk for thromboembolic complications, such as AF and VTE, and requiring antithrombotic therapy increases with age [5, 6].
Vitamin K antagonists (VKAs) and heparins have been the most frequently used anticoagulants for the past fifty years , while direct thrombin inhibitors (DTIs) and indirect Xa inhibitors, such as dabigatran and fondaparinux, have been discovered in recent decades. Although having been approved the efficient in treating and reducing the risk, both of these treatments have disadvantages and drawbacks that limit their utilization and satisfactoriness in the clinical setting. Vitamin K antagonists (VKAs), such as warfarin, are orally available and suitable for long-term use but have an unpredictable pharmacokinetic and pharmacodynamic relationship, numerous food and drug interactions, and a narrow therapeutic window requiring frequent monitoring and resulting in individually tailored dosing schemes [8, 9]. Because of its large size, unfractionated heparins (UFHs) is only partially absorbed from subcutaneous tissue, and it has a variable anticoagulant response due to interactions with plasma proteins, macrophages, and endothelial cells. Unlike UFHs, low molecular weight heparins (LMWHs) have only negligible nonspecific binding to plasma proteins, endothelial cells, and monocytes, resulting in an efficient coagulations and predictable dose response which obviates the need for laboratory monitoring even when used in full (therapeutic) dosing. However, LMWHs, DTIs and indirect Xa inhibitors are restricted by the requirement for parenteral administration, making them inconvenient and costly for long term use, particularly outside of the hospital setting.
Having been shown a high level oral bioavailability , rivaroxaban (Bayer Healthcare AG, Wuppertal, Germany), an oral, direct factor Xa (FXa) inhibitor for the prevention and treatment of thromboembolic disorders, is a gifted target for effective anticoagulation. Inhibiting both the intrinsic and extrinsic activation pathways of coagulation at the same time (Figure 1), rivaroxaban provides a more effective anticoagulation than inhibitor of other enzymes in the coagulation cascade, like VKAs. Rivaroxaban inhibits thrombin generation by inhibiting FXa, where one molecule of FXa results in the generation of more than 1000 thrombin molecules, thereby preventing clot formation. This mechanism of action results in the dose dependent prolongation of global clotting tests, such as prothrombin time (PT), activated partial thromboplastin time (aPPT), and Hep test, with rivaroxaban. Therefore, inhibiting FXa activity will inhibit thrombin generation, thereby diminishing thrombin-mediated activation of coagulation and platelets.
Always on Time
Marked to Standard
It has been shown that rivaroxaban had a predictable dose-dependent pharmacokinetic (PK) and pharmacodynamic (PD) relationship in both healthy Caucasian subjects  and patients undergoing major orthopaedic surgery . Thus, without any prerequisite to close monitoring of the anticoagulant effect, rivaroxaban can be administered orally from the beginning of the treatment . Now, it has been approved by several countries for the prevention of VTE after elective hip and knee replacement surgery .
Excretion occurred principally via the renal route (66%) with the 36% of rivaroxaban excreted unchanged in the urine . As renal function decrease progressively with age and clearance determined in the healthy adult subjects may be delayed further in elderly, who stand for a significant percentage of the potential target population, investigation of pharmacokinetic and pharmacodynamic relationship of rivaroxaban in elderly subjects is necessary for their safety and efficacy consideration.
The aim of this study was to investigate the population pharmacokinetic and pharmacodynamic relationship of rivaroxaban in healthy, elderly Chinese subjects.
This was a single-center, randomized, single-blind, placebo-controlled, parallel-group, dose-escalation study. More detailed information about the study design and descriptive statistical results can be found at . Briefly, rivaroxaban was given to six male and six female elderly healthy Chinese subjects (additional two male and two female subjects received placebo at the same time) at each dose level with food, including 5, 10, 20, 30 and 40 mg respectively. The study was conducted in accordance with Good Clinical Practice Guidelines and the Declaration of Helsinki. The study protocol and amendments were approved by the Independent Ethics Committee of Peking Union Medical College Hospital and by the State Food and Drug Administration of China. All subjects participating in the study were required to be more than 60 years old with a BMI of 19-28 kg/m2 and provide written informed consent, etc.
Blood samples collected before (time 0), and 0.5, 1, 2, 3, 4, 6, 8, 12, 15, 24, 36, 48 and 72 h after administration of rivaroxaban for each subject to measure rivaroxaban concentration in plasma, FXa activity, activated partial thromboplastin time, prothrombin time and Hep test were performed according to the study design.
Population PK/PD Modeling
The population PK/PD model building process was performed using nonlinear mixed effects modeling approach implement in the software NLME (Version 1.0 Pharsight Corporation, Cary, NC). First Order Conditional Estimate (FOCE) was used to find the parameter estimates and variability of PK/PD models. Model evaluation included objective function value (OFV), visual inspection of basic goodness-of-fit (GOF) plots, precision of parameter estimates were used. The primary tool used for selection between hierarchical models was their difference in OFV. Internal model validation was performed by Visual Predictive Check (VPC, 1 000 simulations).
The population PK analysis was performed with rivaroxaban plasma concentration data. PK model development was performed stepwise. Structural model was developed including an investigation of different compartment models, including one-, two-, and three-compartment model.
After building up the population pharmacokinetic model and fixing the estimated PK parameters, the PK/PD structural models for FXa activity, PT, aPPT and Hep Test were developed from linear and/or Emax models with or without Hill exponent respectively. As hysteresis phenomena were not observed in any PK/PD relationship by visual inspection of PK/PD plots (not shown), effect compartment was not considered for any of these PK/PD models.
During the development of the base model, inter-individual and residual (unexplained) variability were investigated. An exponential random effect term was used to account for inter-individual variability (IIV) in structural parameters. Residual variability that represents the discrepancy between the observed and the model-predicted data after incorporation of IIV was investigated as being additive, multiplicative or exponential.
To advance the population analysis, a covariate analysis was performed starting from the base model. A two step model building procedure, including forward inclusion and backward exclusion at significance levels of p<0.01 (Î”OFV > 6.635) and p < 0.001 (Î”OFV > 10.828), respectively, was used to identify the covariates in PK/PD models among demographics, including age, body weight, gender.
57 subjects (29 males and 28 females) administered with rivaroxaban were included in this analysis while other subjects administered with placebo were excluded. A total number of 798 samples were collected. Descriptive statistical analysis of demographic characteristics that performed by SAS 9.1.3 (SAS Institute Inc., Cary, NC, USA) for each dose level is presented in Table 1. For all parameters, the range was narrow and the variance was small. For example, coefficient of variation (CV %) of weight was lower than 15% in each dose level, which was difficult for identifying covariate relation. The distribution of gender in this study showed a balance between males and females.
Population Pharmacokinetic Model
This Essay is
a Student's Work
This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.Examples of our work
The plasma rivaroxaban concentration versus time data were well described by an oral, two-compartment model with first order absorption and first order elimination from the central compartment. The distribution of random residual errors was best described by a multiplicative error model. Due to the similarity of the subjects, no covariate influencing the PK profile was discovered.
The population pharmacokinetic parameters estimated for the final model are listed in Table 2 and Goodness of Fit (GOF) graphics are shown in Figure 2 (a, b, c). The typical estimates (CV%) for Ka, CL/F, V/F, CL2/F and V2/F were 0.36 1/hr (5.87%), 9.41 L/hr (3.16%), 17.68 L (9.81%), 1.44 L/hr (10.53%) and 21.13 L (10.03%) respectively and their correspondence inter-individual variability were 0.01% (0.03%), 4.01% (0.22%), 15.80% (1.19%), 31.00% (2.61%) and 27.16% (2.03%). The residual variability was best implemented with a multiplicative model of the model was 38% (3.51%). Visual Predictive Check for the final population pharmacokinetic model was show in Figure2 (d).
Pharmacokinetic/Pharmacodynamic Relationship Model
Factor Xa activity
Plasma rivaroxaban concentration and FXa activity were correlated following a fractional inhibitory Emax model without Hill exponent according to:
in which Cp was the plasma rivaroxaban concentration in the central compartment, Imax was the maximum inhibitory ratio and IC50 was the concentration at 50% inhibitory drug effect.
Additive model was used to describe the residual variability.
The final estimates and their inter-individual variability for FXa activity are listed in Table 3. The Imax for FXa activity in this analysis was 97.17% (5.60%) and IC50 was 267.50 mg/L (12.34%), where the typical baseline value of FXa was 0.71 (2.03%).
The population pharmacodynamic parameters estimated for the factor Xa activity model are listed in Table 3. Visual Predictive Check for the factor Xa activity model was shown in Figure 3 (a).
Activated Partial Thromboplastin Time
Plasma rivaroxaban concentration and aPPT were correlated following an ordinary Emax model with multiplicative residual error.
Additive model was used to describe the residual variability.
The final estimates and their inter-individual variability for activated partial thromboplastin time are listed in Table 3. The Emax for aPPT in this analysis was 53.98s (14.61%) and EC50 was 712.48 mg/L(21.84%), where the typical baseline value of aPPT was 31.07s (1.20%).
The population pharmacodynamic parameters estimated for the activated partial thromboplastin time model are listed in Table 3. Visual Predictive Check for the activated partial thromboplastin time model was shown in Figure 3 (b).
Plasma rivaroxaban concentration and PT were correlated following a linear intercept model rather than Emax model.
The residual error was proportional to the squared rivaroxaban plasma concentration.
The final estimates for prothrombin time are listed in Table 3. The typical baseline value of PT in the study population was 12.34s (0.62%). The slope of the correlation between PT and rivaroxaban plasma concentration was 0.0379s*L/mg (4.16%).
The population pharmacodynamic parameters estimated for the prothrombin time model are listed in Table 3. Visual Predictive Check for the prothrombin time model was shown in Figure3 (c).
Plasma rivaroxaban concentration and Hep test were correlated following an ordinary Emax model.
Multiplicative error was used to describe the residual variability.
The final estimates for Hep test are listed in Table 3. The Emax for Hep Test in this analysis was 54.24 s (8.81%) and EC50 was 471.98 mg/L (14.28%), where the typical baseline value of Hep test was 17.63 s (17.17%).
The population pharmacodynamic parameters estimated for the Hep test model are listed in Table 3. Visual Predictive Check for the Hep test model was shown in Figure 3 (d).
The result of this population pharmacokinetic and pharmacodynamic analysis shows that rivaroxaban has a predictable dose-response relationship in elderly healthy Chinese subjects, which was similar to the previous studies in healthy Caucasian subjects.
Compared to the population pharmacokinetic model in healthy Caucasian subjects, elderly healthy Chinese subjects has a lower absorption rate (Ka) and smaller distribution volume of central compartment (V). However, they have a similar typical value in clearance. Adding a lag time to the two compartment model did not significantly improved the objective function value and the accuracy of model estimations. Thus, a two compartment pharmacokinetic model with multiplicative error was finally chosen to describe the concentration time profile of rivaroxaban in elderly healthy Chinese subjects.
It has been showed that rivaroxaban had a strong FXa inhibition ability in healthy subjects. Studies in healthy Chinese subjects showed that rivaroxaban could inhibit FXa activity up to about 50% (30mg b.i.d.). The estimated absolute maximum inhibition value was -0.86 U/mL while the baseline value was 0.87 U/mL in healthy Caucasians. In this analysis, the 95% confidence interval for Imax of FXa activity was (86.07%, 108.27 %), where the observed maximum inhibition rate in this study was about 65% (40mg single dose), indicating that rivaroxaban may totally inhibit FXa activity at an adequate concentration level. However, the 100% inhibition rate may not be reached by oral administration, because bioavailability in healthy subjects did not increase further with rivaroxaban at 50 mg compared to 40 mg in healthy elderly subjects, suggesting a ceiling effect in absorption.
The final estimate shows that there was a strong linear relationship between rivaroxaban plasma concentration and PT. As suggested in Caucasians, PT might be useful to assess rivaroxaban exposure in the patients, if this was required. However, as the residual error model showed that the variability increases as the rivaroxaban plasma concentration went up higher, the PT result might have an unacceptable bias estimate of rivaroxaban plasma concentration, which requires confirmation in further studies.
PK/PD relationship for both aPPT and Hep test were described by the ordinary Emax model without Hill exponent. As indicated in the Emax model for aPPT, where EC50 was 712.476 mg/L, the Emax value of aPPT may not be actually observed in the clinical trial. Also, Emax could not be reached by rivaroxaban administered orally because of the ceiling effect observed in pharmacokinetic at a higher dosage.
Due to the homogeneity of demographics (Table 1) and standard clotting tests baseline value in these subjects, no covariate that influence the pharmacokinetic and pharmacodynamic profile has been identified in this analysis. 29 male and 28 female subjects are involved in the covariate analysis. No statistical relevant differences between male and female subjects were found in the population pharmacokinetic and pharmacodynamic models, indicating that gender does not influence the PK/PD profile in elderly healthy subjects.
The small inter-individual variability and lack of covariates in this Phase I clinical study demonstrated that rivaroxaban could be administered at a fixed dose regardless of the patients' gender or weight.
In conclusion, this analysis of data from a Phase I study in elderly healthy Chinese subjects showed that the pharmacokinetic of rivaroxaban was well described by an oral, two-compartment model with first-order absorption and first-order elimination from the central compartment without a lag time in absorption phase. The correlation between the pharmacokinetic profile of rivaroxaban and standard clotting tests conformed to conventional Emax models without Hill exponent, where as PT correlating following a simple linear intercept model. Both PK and PD profile have a lower inter-individual variability, showing that rivaroxaban has a consistent dose-response relationship in elderly healthy Chinese subjects.
The authors acknowledge the assistance of the clinical trial personnel from Clinical Pharmacology Research Center, Peking Union Medical College Hospital and the participants in the trial. They also thank Jianyan Zhang for general medical support. Financial disclosure: The clinical trial was sponsored by Bayer Schering Pharma AG and Johnson & Johnson Pharmaceutical Research & Development, L.L.C..
Table1. Demographic characteristics of subjects enrolled in this population pharmacokinetic and pharmacodynamic analysis
Mean (CV %)
Mean (CV %)
Mean (CV %)
Mean (CV %)
Table2. Final two-compartment population pharmacokinetic model parameter estimation and inter-individual variability
Table3. Final population pharmacodynamic model parameter estimation and inter-individual variability
95% Confidence Interval
Figure1. Coagulation cascade with mechanism of rivaroxaban.
Figure2. Goodness of Fit and Visual Predictive Check. (a) Goodness of Fit graph with line of identity. Population prediction compared with observed rivaroxaban plasma concentration (b) Goodness of Fit graph with line of identity. Individual prediction compared with observed rivaroxaban plasma concentration (c) Goodness of Fit graph Population predication vs. Conditional Weighted Residual (d) Visual Predictive Check for PK model.
Figure3 Visual Predictive Check for PD models. Solid lines in each figure represent the upper 5%, median and lower 5% visual prediction, while the dash lines represent the upper 5%, median and lower 5% quantile for the observations (a) Visual predictive check for inhibition of facto Xa activity. (b) Visual predictive check for activated partial thromboplastin time. (c) Visual predictive check for prothrombin time. (d) Visual predictive check for Hep test.