Elicit: Comparative Mechanisms of Direct Oral Anticoagulants

Skip to main content

Comparative Mechanisms of Direct Oral Anticoagulants

Research reportView only

Create alertChat

May 6, 2026

Comparative mechanism: rivaroxaban vs apixaban vs edoxaban

Rivaroxaban binds factor Xa dramatically faster than apixaban but fluctuates more in plasma concentration, apixaban binds slower but maintains stable levels throughout the dosing interval, and edoxaban demonstrates intermediate properties, with assay-dependent potency reversals across all three drugs indicating mechanistic complexity beyond direct factor Xa inhibition alone.

Abstract

Rivaroxaban, apixaban, and edoxaban share the same therapeutic target—factor Xa—but differ substantially in their molecular mechanisms and pharmacodynamic profiles. Rivaroxaban binds factor Xa up to 4-fold faster than apixaban in kinetic assays, with this difference amplifying to 1,193-fold when factor Xa is incorporated into the prothrombinase complex. These faster association kinetics directly explain rivaroxaban’s 3-fold greater effects on prothrombin time and thrombin generation compared to apixaban, despite similar equilibrium binding affinities (Ki 0.4-0.74 nM). Pharmacokinetic profiles diverge primarily due to dosing frequency, with apixaban’s twice-daily regimen producing 3.6-fold lower peak-to-trough fluctuation than once-daily rivaroxaban, translating to more stable anticoagulation throughout the dosing interval. In renal impairment, rivaroxaban and edoxaban accumulate preferentially in severe versus moderate dysfunction, while apixaban maintains consistent activity across renal function levels. Drug potency rankings reverse across assay types—edoxaban demonstrates strongest effects in global coagulation tests while apixaban shows strongest thrombin generation inhibition—indicating that anti-factor Xa activity alone incompletely captures these drugs’ biological effects. Despite substantial differences in binding kinetics, pharmacokinetics, and assay-specific potencies, all three drugs maintain fibrin monomer complex within normal ranges throughout dosing intervals, suggesting trough concentrations suffice for sustained thrombin suppression in most patients.

Methods

We analyzed 10 sources from an initial pool of 200, using 9 screening criteria. Each paper was reviewed for 8 key aspects that mattered most to the research question. More on methods

Records from Elicit search

n = 200

Papers screened using: DOAC Focus and Comparison, Mechanistic Study Type, Factor Xa Mechanism Focus, Drug Processing Mechanisms, Adult Population, Appropriate Study Design, Mechanistic Data Inclusion, Target DOAC Inclusion, Study Quality and Detail

n = 200

Papers screened out

n = 190

Papers included for extraction

n = 10

Press enter or space to select a node.You can then use the arrow keys to move the node around. Press delete to remove it and escape to cancel.

Press enter or space to select an edge. You can then press delete to remove it or escape to cancel.

Paper search

We performed a semantic search across over 138 million academic papers from the Elicit search engine, which includes all of Semantic Scholar and OpenAlex.

We ran this query: “Comparative mechanism: rivaroxaban vs apixaban vs edoxaban”

The search returned 200 total results from Elicit.

We retrieved 200 papers most relevant to the query for screening.

Screening

We screened in sources based on their abstracts that met these criteria:

We considered all screening questions together and made a holistic judgement about whether to screen in each paper.

Data extraction

We asked a large language model to extract each data column below from each paper. We gave the model the extraction instructions shown below for each column.

Extract which factor Xa inhibitors were studied (rivaroxaban, apixaban, edoxaban), including:

Extract study population details that may affect drug mechanisms, including:

Extract all pharmacokinetic measurements comparing rivaroxaban, apixaban, and edoxaban, including:

Extract anti-factor Xa activity measurements for mechanism comparison, including:

Extract results from coagulation assays that reflect mechanistic differences, including:

Extract data on molecular-level mechanisms comparing the three drugs, including:

Extract direct comparative statements and conclusions about mechanistic differences between rivaroxaban, apixaban, and edoxaban, including:

Extract mechanistic data in special populations or conditions that may reveal differential drug behavior, including:

Results

Characteristics of included studies

Ten studies compared the mechanisms of rivaroxaban, apixaban, and edoxaban. The studies varied in design from in vitro biochemical analyses to clinical pharmacokinetic studies, with six studies available only as abstracts.

Study

Full text retrieved?

Study type

Drugs compared

Population/Design

Key focus

Goto et al., 2019

No

Clinical PK study

Rivaroxaban 15mg QD, Apixaban 5mg/2.5mg BID, Edoxaban 60mg QD

329 patients with NVAF or VTE, 3.6 year follow-up

Population PK modeling and anti-FXa assay sensitivity

Perzborn et al., 2009

No

In vitro study

Rivaroxaban, Apixaban

Cell-free and plasma-based assays

Binding kinetics and plasma-based potency

Frost et al., 2014

Yes

Randomized crossover

Apixaban 2.5mg BID, Rivaroxaban 10mg QD

14 healthy volunteers, 4 days each drug

Direct PK/PD comparison

Kim et al., 2018

Yes

In vitro kinetic study

Rivaroxaban, Apixaban

Purified systems and plasma

Association/dissociation kinetics

Suwa et al., 2024

Yes

Clinical observational

Rivaroxaban 15/10mg QD, Apixaban 5/2.5mg BID, Edoxaban 60/30mg QD

268 NVAF patients, 6-12 months

Drug concentration and coagulation biomarkers

Siddiqui et al., 2019

Yes

In vitro profiling

Apixaban, Betrixaban, Edoxaban, Rivaroxaban

Healthy donor blood samples

Comprehensive coagulation assay panel

Jourdi et al., 2015

No

In vitro kinetic modeling

Rivaroxaban, Apixaban

Simulation and plasma validation

Association rates and PT sensitivity

Tobe et al., 2020

No

Clinical observational

Rivaroxaban, Apixaban, Edoxaban

284 Japanese NVAF patients with renal impairment

Anti-Xa activity in renal dysfunction

Gosselin et al., 2016

No

In vitro comparison

Apixaban, Edoxaban, Rivaroxaban

Plasma-enriched samples

Coagulation assay effects

Siddiqui et al., 2018

No

In vitro profiling

Apixaban, Betrixaban, Edoxaban, Rivaroxaban

Human pooled plasma

Anticoagulant and biochemical profiles

toof

Pageof

Among the ten studies, four provided full-text data while six were available only as abstracts. Five studies were conducted entirely in vitro using purified systems or plasma-based assays, while five involved clinical populations. The clinical studies focused primarily on patients with nonvalvular atrial fibrillation, with sample sizes ranging from 14 healthy volunteers to 329 patients. Studies employed diverse methodologies including kinetic modeling, chromogenic assays, thromboelastography, and thrombin generation measurements to characterize mechanistic differences among the drugs.

Pharmacokinetic profiles

Direct comparisons revealed substantial differences in pharmacokinetic behavior among the three factor Xa inhibitors, particularly between once-daily and twice-daily regimens.

Drug

Daily exposure (AUC0-24)

Peak concentration

Trough concentration

Peak-to-trough ratio

Half-life

Coefficient of variation

Rivaroxaban 10mg QD

1,094 ng·h/mL

171 ng/mL

10 ng/mL

16.9

7.9 hours

29-46%

Apixaban 2.5mg BID

935 ng·h/mL

81 ng/mL

17 ng/mL

4.7

8.7 hours

20-24%

Rivaroxaban 15mg QD

2,710 ng·h/mL

Not reported

Not reported

Not reported

Not reported

Not reported

Apixaban 5mg BID

4,550 ng·h/mL

Not reported

Higher than rivaroxaban

Not reported

Not reported

Not reported

Edoxaban 60mg QD

1,290 ng·h/mL

Not reported

Not reported

Not reported

Not reported

Not reported

toof

Pageof

Apixaban demonstrated 3.6-fold lower peak-to-trough fluctuation compared to rivaroxaban at equipotent doses, resulting from its twice-daily dosing regimen. This translated to more stable plasma concentrations throughout the dosing interval, with apixaban maintaining 70% higher trough levels than rivaroxaban despite similar total daily exposure. The coefficient of variation for exposure parameters was substantially lower for apixaban (20-24%) compared to rivaroxaban (29-46%), indicating more consistent intersubject pharmacokinetics.

At higher therapeutic doses, apixaban 5mg BID achieved the greatest overall exposure (AUC 4,550 ng·h/mL) compared to rivaroxaban 15mg QD (2,710 ng·h/mL) and edoxaban 60mg QD (1,290 ng·h/mL). Apixaban also demonstrated higher trough concentrations than both rivaroxaban and edoxaban. Both drugs reached similar time to peak concentration (Tmax) of 2 hours, indicating comparable absorption rates.

In clinical practice, trough plasma concentrations varied substantially across drugs in patients with NVAF. Apixaban maintained median trough concentrations of 80.9-88.4 ng/mL at standard doses, significantly higher than rivaroxaban (13.0-25.1 ng/mL) and edoxaban (15.4-21.8 ng/mL). Drug accumulation patterns also differed, with rivaroxaban showing elevation in peak concentrations over time in 50.9% of patients compared to 31.7% for edoxaban.

Molecular mechanisms and binding kinetics

Despite all three drugs targeting factor Xa, substantial differences emerged in their molecular interactions with the enzyme. Kinetic analyses revealed that rivaroxaban associates with factor Xa 4-fold faster than apixaban in equilibrium binding studies.

Parameter

Rivaroxaban

Apixaban

Edoxaban

Ki for free FXa

0.4-0.7 nM

0.6-0.74 nM

Not reported

Association rate (kon) for free FXa

1.7-2.9 × 10^7 M^-1s^-1

0.88-7.3 × 10^6 M^-1s^-1

Not reported

Dissociation rate (koff)

5 × 10^-3 s^-1

2.4 × 10^-3 s^-1

Not reported

Association rate for prothrombinase-bound FXa

1,193-fold faster than apixaban

Baseline

Not reported

Ki for prothrombinase complex

0.7 nM

2.9 nM

Not reported

toof

Pageof

The most striking difference occurred in the association rate with prothrombinase-bound factor Xa, where rivaroxaban bound 1,193-fold faster than apixaban. For free factor Xa, rivaroxaban’s association rate was 10-fold faster. While dissociation rates were approximately 3-fold faster for rivaroxaban, the net effect of these kinetic differences was 4-fold greater potency for rivaroxaban in inhibiting prothrombinase-induced thrombin generation.

Equilibrium binding affinities (Ki values) were similar between rivaroxaban (0.4-0.7 nM) and apixaban (0.6-0.74 nM), indicating comparable thermodynamic binding strength. Both drugs inhibited prothrombinase-bound factor Xa with similar potency in equilibrium conditions (rivaroxaban 2.1 nM, apixaban 2.7 nM). However, these equilibrium measurements failed to predict the substantial kinetic differences observed in plasma-based systems.

All factor Xa inhibitors functioned through competitive reversible inhibition, but their pharmacodynamic profiles suggested additional mechanisms beyond simple factor Xa blockade.

Anti-factor Xa activity

Measurements of anti-factor Xa activity revealed discordance between drug concentration and functional inhibition, with concentration-activity relationships varying substantially among the three agents.

Drug

Peak anti-Xa activity

Trough anti-Xa activity

Anti-Xa AUC0-24

Peak-to-trough fluctuation

Concentration-activity slope

Rivaroxaban 10mg QD

2.82 IU/mL

0.17 IU/mL

17.8 IU·h/mL

16.5

0.0172 IU/ng

Apixaban 2.5mg BID

1.12 IU/mL

0.24 IU/mL

13.3 IU·h/mL

4.7

0.0134 IU/ng

toof

Pageof

Rivaroxaban exhibited 2.5-fold higher peak anti-Xa activity than apixaban, but 30% lower trough activity. The peak-to-trough fluctuation in anti-Xa activity was 3.5-fold greater for rivaroxaban, mirroring its plasma concentration fluctuations. Despite similar total daily exposure at the plasma concentration level, rivaroxaban achieved 34% higher anti-Xa AUC than apixaban.

The concentration-activity relationship differed between drugs, with rivaroxaban displaying a steeper slope (0.0172 IU/ng) compared to apixaban (0.0134 IU/ng). This meant that for each ng/mL increase in plasma concentration, rivaroxaban generated greater anti-Xa activity than apixaban.

IC50 values for direct factor Xa inhibition showed drug-specific patterns. In amidolytic assays, edoxaban demonstrated the strongest inhibition (IC50 340-430 ng/mL), followed by apixaban (400-490 ng/mL), rivaroxaban (670-840 ng/mL), and betrixaban (>1,000-1,300 ng/mL). However, this rank order based on isolated factor Xa inhibition did not predict effects in more complex coagulation assays.

In patients with renal impairment, anti-Xa activity patterns diverged. For rivaroxaban, peak anti-Xa values increased significantly in severe versus moderate renal impairment. Edoxaban showed significantly elevated trough anti-Xa values in severe renal impairment, with a trend toward higher peaks. In contrast, apixaban demonstrated no significant differences in peak or trough anti-Xa activity between moderate and severe renal impairment. The peak-to-trough ratio was significantly lower for apixaban compared to both rivaroxaban and edoxaban in renally impaired patients.

Effects on coagulation assays

The three factor Xa inhibitors produced markedly different effects across multiple coagulation assays, with the magnitude of differences exceeding what would be predicted from anti-factor Xa potency alone.

Assay

Relative potency ranking

Key findings

Prothrombin time

Edoxaban > Rivaroxaban ≈ Betrixaban > Apixaban

Apixaban minimally affects PT; 3-8 fold higher concentrations needed vs rivaroxaban

aPTT

Betrixaban > Edoxaban > Rivaroxaban ≈ Apixaban

3-8 fold concentration difference between drugs to double aPTT

Activated clotting time

Edoxaban > Rivaroxaban > Betrixaban >> Apixaban

Apixaban showed minimal effects at 2.5 μg/mL

Thromboelastography

Betrixaban > Edoxaban ≈ Rivaroxaban > Apixaban

Edoxaban exhibited strongest effects on R-time, K-time, alpha, MA

Thrombin generation

Apixaban > Edoxaban ≈ Betrixaban > Rivaroxaban

Rivaroxaban IC50 (100 ng/mL) vs apixaban IC50 (50 ng/mL)

Fibrin formation

Rivaroxaban (67%) > Edoxaban (42%) > Apixaban (32%) > Betrixaban (12%)

Percentage inhibition at 1 μg/mL

toof

Pageof

Prothrombin time sensitivity varied dramatically, with rivaroxaban producing significantly greater PT prolongation than apixaban at equivalent concentrations. This differential sensitivity persisted across concentration ranges, with PT showing minimal responsiveness to clinically relevant apixaban concentrations. Concentrations of apixaban 3-8 fold higher than rivaroxaban were required to achieve comparable PT prolongation.

In thrombin generation assays using plasma triggered with 20 pM tissue factor, 100 ng/mL rivaroxaban increased lag time 3.0-fold compared to 1.8-fold for apixaban. Time to peak increased 2.0-fold with rivaroxaban versus 1.4-fold with apixaban. Rivaroxaban reduced peak thrombin 3.1-fold and endogenous thrombin potential 3.5-fold, compared to 1.2-fold and 1.7-fold reductions with apixaban. These differences aligned with predictions from kinetic modeling based on association rate constants.

However, in isolated thrombin generation inhibition assays measuring IC50 values, apixaban demonstrated the strongest activity (IC50 50 ng/mL), followed closely by edoxaban (58 ng/mL) and betrixaban (60 ng/mL), with rivaroxaban showing relatively weaker activity (IC50 100 ng/mL). This reversal of potency order from global coagulation tests to isolated thrombin generation assays highlighted the assay-dependent nature of these drugs’ anticoagulant profiles.

Fibrin formation kinetics revealed another dimension of differential activity. Rivaroxaban produced the strongest inhibition of fibrin formation (67% at 1 μg/mL), followed by edoxaban (42%), apixaban (32%), and betrixaban (12%). This suggested potential effects on fibrin polymerization beyond factor Xa inhibition.

In clinical coagulation monitoring, fibrin monomer complex (FMC) levels remained within normal ranges (<6.1 μg/mL) for all three drugs despite wide variations in plasma concentration from peak to trough. Peak FMC values were 3.1 μg/mL for rivaroxaban, 2.3 μg/mL for apixaban, and 2.1 μg/mL for edoxaban, with no significant differences among drugs. This indicated sustained thrombin suppression throughout the dosing interval even for once-daily regimens with substantial peak-to-trough fluctuations.

Effects in special populations

Renal impairment revealed distinct mechanistic differences in how the three drugs accumulated and maintained anticoagulant activity.

Drug

Severe vs moderate RI: Peak AXA

Severe vs moderate RI: Trough AXA

Peak-to-trough ratio in RI

Accumulation pattern

Rivaroxaban

Significantly higher in severe RI

Not significantly different

Higher than apixaban

50.9% showed elevation over time

Apixaban

No significant difference

No significant difference

Lowest among three drugs

Not reported

Edoxaban

Tended to be higher in severe RI

Significantly higher in severe RI

Higher than apixaban

31.7% showed elevation over time

toof

Pageof

In patients with severe renal impairment (creatinine clearance 15-29 mL/min) compared to moderate impairment (30-49 mL/min), rivaroxaban demonstrated significantly elevated peak anti-Xa activity. Edoxaban showed significantly higher trough anti-Xa values in severe renal impairment, with peak values trending higher. Apixaban uniquely maintained similar peak and trough anti-Xa values across both degrees of renal impairment, suggesting more predictable anticoagulation in this population.

The peak-to-trough ratio of anti-Xa activity was significantly lower for apixaban compared to both rivaroxaban and edoxaban in renally impaired patients. This translated to more stable anticoagulant effects throughout the dosing interval despite fluctuating renal function.

Drug accumulation over time differed substantially. Rivaroxaban demonstrated temporal elevation in peak plasma concentrations in 50.9% of patients, significantly more than edoxaban (31.7%). This accumulation pattern aligned with rivaroxaban’s higher bleeding incidence in clinical practice. Apixaban maintained higher trough concentrations than rivaroxaban and edoxaban across all renal function levels, but without the progressive accumulation seen with rivaroxaban.

Despite wide variations in plasma concentrations and renal function, fibrin monomer complex levels remained stable within normal ranges for all three drugs. This suggested that once-daily factor Xa inhibitors maintained adequate thrombin suppression even at trough levels in patients with renal impairment.

Synthesis

The observed mechanistic differences among rivaroxaban, apixaban, and edoxaban reflect three distinct but interconnected properties: binding kinetics, pharmacokinetic profiles, and assay-specific potencies. These differences produce clinically relevant variations in anticoagulant stability and safety profiles.

Binding kinetics explain differential potency in plasma-based systems

Despite similar equilibrium binding affinities (Ki values 0.4-0.74 nM), rivaroxaban inhibits factor Xa up to 4-fold faster than apixaban in kinetic assays. This difference becomes dramatically amplified when factor Xa is incorporated into the prothrombinase complex, where rivaroxaban associates 1,193-fold faster than apixaban. The mechanistic basis for this difference lies in the drugs’ association rate constants: rivaroxaban’s kon of 1.7-2.9 × 10^7 M^-1s^-1 versus apixaban’s 0.88-7.3 × 10^6 M^-1s^-1.

These kinetic differences directly predict the divergent effects on global coagulation tests. Numerical integration models based solely on association rate constants accurately reproduce the observed 3-fold greater sensitivity of PT and thrombin generation assays to rivaroxaban compared to apixaban. The faster binding kinetics allow rivaroxaban to more rapidly intercept factor Xa as it forms thrombin, resulting in greater prolongation of lag time and suppression of peak thrombin generation.

However, this mechanistic explanation applies only to rivaroxaban versus apixaban comparisons. Edoxaban demonstrates the strongest effects in PT, aPTT, and ACT assays, yet kinetic data for edoxaban binding to factor Xa were not reported in any study. The rank order of drugs varies by assay type, suggesting that factors beyond simple factor Xa binding kinetics—such as plasma protein interactions or off-target effects—contribute to the overall anticoagulant profile.

Pharmacokinetic profiles determine anticoagulation stability

The 3.6-fold difference in peak-to-trough plasma concentration ratios between rivaroxaban (16.9) and apixaban (4.7) translates directly into comparable fluctuation in anti-Xa activity (16.5 vs 4.7). This pharmacokinetic difference stems primarily from dosing frequency—once-daily versus twice-daily administration—rather than inherent drug properties.

However, the clinical consequences of these fluctuations appear limited. Despite rivaroxaban’s substantial peak-to-trough variations, fibrin monomer complex levels remain consistently below 6.1 μg/mL throughout the dosing interval, indicating sustained thrombin suppression even at trough concentrations. This suggests that trough levels of all three drugs exceed the threshold needed for effective anticoagulation in most patients.

The situation differs in renal impairment. Rivaroxaban and edoxaban both accumulate preferentially in severe versus moderate renal dysfunction, whereas apixaban maintains consistent activity across renal function levels. This difference likely reflects apixaban’s lower renal clearance (27% renal excretion) compared to edoxaban (50% renal) and rivaroxaban (33% renal), though these clearance data were not directly reported in the included studies. The progressive accumulation of rivaroxaban over time in 50.9% of patients may explain its association with higher bleeding rates, as accumulation would push peak concentrations into ranges associated with hemorrhagic events.

Assay-dependent potencies reveal mechanistic complexity

The complete reversal of potency rankings between different assay types—with edoxaban strongest in global coagulation tests but apixaban strongest in thrombin generation inhibition—indicates that anti-factor Xa activity incompletely captures these drugs’ biological effects. Several mechanisms may contribute:

First, the drugs differ in their effects on fibrin formation, with rivaroxaban inhibiting fibrin polymerization 67% at concentrations where apixaban achieves only 32% inhibition. This cannot be explained by factor Xa inhibition alone and suggests interaction with other components of the coagulation cascade or direct effects on fibrin structure.

Second, plasma protein binding varies (rivaroxaban 92-95%, apixaban 87%), yet apixaban requires higher concentrations to achieve equivalent effects despite lower protein binding. This paradox suggests that the free drug concentration—typically considered the pharmacologically active moiety—does not fully predict anticoagulant potency in plasma.

Third, the concentration-activity relationship differs between drugs, with rivaroxaban generating 28% more anti-Xa activity per ng/mL than apixaban. Combined with rivaroxaban’s steeper PT response, this indicates greater pharmacodynamic sensitivity independent of plasma concentration.

Implications for clinical use

These mechanistic differences rationalize several clinical observations. Apixaban’s lower peak-to-trough fluctuation and consistent activity across renal function levels provide more predictable anticoagulation, potentially explaining its favorable safety profile. Rivaroxaban’s faster kinetics and greater PT sensitivity may offer advantages in situations requiring rapid anticoagulation but increase monitoring complexity. Edoxaban’s intermediate properties—stronger than apixaban in global tests but less accumulation than rivaroxaban—position it between these extremes.

The finding that all three drugs maintain normal fibrin monomer complex levels throughout dosing intervals suggests that differences in peak concentrations matter less for efficacy than for safety, as trough levels suffice for thrombin suppression. This explains why clinical trials demonstrate similar efficacy across these agents despite substantial mechanistic differences. The safety distinctions likely emerge from peak concentrations and accumulation patterns rather than antithrombotic potency.

References

Eri Goto, S. Horinaka, T. Ishimitsu, Toru Kato\ (2019).Factor Xa inhibitors in clinical practice: Comparison of pharmacokinetic profiles. Drug Metabolism and Pharmacokinetics

E. Perzborn, A. Tersteegen, M. Harwardt, U. Lange\ (2009).Different Characteristics of Direct Factor Xa Inhibitors: In Vitro Comparative Studies of Rivaroxaban and Apixaban. Blood

Robert C. Gosselin, R. P. Grant, D. Adcock\ (2016).Comparison of the effect of the anti‐Xa direct oral anticoagulants apixaban, edoxaban, and rivaroxaban on coagulation assays. International Journal of Laboratory Hematology

C. Frost, Yan Song, Y. Barrett, Jessie Wang, J. Pursley, and 2 more\ (2014).A randomized direct comparison of the pharmacokinetics and pharmacodynamics of apixaban and rivaroxaban. Clinical Pharmacology : Advances and Applications

P. Kim, C. Yeh, B. Dale, B. Leslie, A. Stafford, and 3 more\ (2018).Mechanistic Basis for the Differential Effects of Rivaroxaban and Apixaban on Global Tests of Coagulation. TH Open

Fakiha Siddiqui, S. Mehrotra, Vishnu Venkitasubramony, Rithik Raina, Lorenzo Storino Ramacciotti, and 5 more\ (2018).Factor Xa Inhibitory Profile of Apixaban, Betrixaban, Edoxaban and Rivaroxaban Does Not Fully Reflect Their Biologic Spectrum. Blood

Michihiro E. Suwa, Isao Morii, Masaya Kino, Y. Matsui, Masahiro Yoshinaga, and 11 more\ (2024).Efficacy and Safety of Rivaroxaban, Apixaban, and Edoxaban for Nonvalvular Atrial Fibrillation Based on Blood Coagulation Activity and Drug Plasma Concentration: SETtsu and North Osaka Multicenter Direct Oral AntiCoagulant (SET DOAC) Registry. Pharmaceuticals

Fakiha Siddiqui, D. Hoppensteadt, W. Jeske, O. Iqbal, A. Tafur, and 1 more\ (2019).Factor Xa Inhibitory Profile of Apixaban, Betrixaban, Edoxaban, and Rivaroxaban Does Not Fully Reflect Their Biologic Spectrum. Clinical and applied thrombosis/hemostasis

Georges Jourdi, V. Siguret, Anne-Céline Martin, J. Golmard, A. Godier, and 4 more\ (2015).Association rate constants rationalise the pharmacodynamics of apixaban and rivaroxaban. Thrombosis and Haemostasis

Akihiro Tobe, H. Osanai, A. Tanaka, Teruhiro Sakaguchi, T. Kambara, and 5 more\ (2020).Comparison of Anti-factor Xa Activity Among Three Different Factor Xa Inhibitors in Non-valvular Atrial Fibrillation Patients with Renal Impairment. Clinical drug investigation

Download BIBDownload RISDownload TXT

Report

Status

Gather sources

200 sources found

Details

Screen sources

10 sources included

Details

Extract data

80 data points extracted

Details

Generate report

Save PDF

BIBLaTeX, ZoteroRISZotero, MendeleyTXTAPA BibliographyPDFPDFDOCXMicrosoft Word

Chat

Got some follow-up questions?

Sign up or sign in to chat with this report.

Back

Mechanistic Basis for the Differential Effects of Rivaroxaban and Apixaban on Global Tests of Coagulation

P. Kim, C. Yeh, B. Dale, B. Leslie, A. Stafford, J. Fredenburgh, J. Hirsh, J. Weitz

TH Open·

2018·

23 citations

SourceDOI

Plain textPDF

Searching for PDF

Unable to find PDF from source

Back

Study Drugs

- Specific drugs compared: Rivaroxaban and Apixaban - Doses used for each drug: Concentrations up to 2,530 and 2,350 µM (1,100 ng/mL) for rivaroxaban and apixaban, respectively - Dosing frequency: Not mentioned (in vitro study) - Duration of treatment/observation: Not mentioned (in vitro study) - Any dose adjustments or criteria used: Not mentioned

Population Characteristics

- Sample size for each drug group: Not mentioned - Age and demographics: Not mentioned - Clinical condition: Healthy volunteers - Renal function status: Not mentioned - Other relevant comorbidities or characteristics: Not mentioned

Pharmacokinetic Data

Not mentioned (the paper does not provide specific pharmacokinetic data for rivaroxaban, apixaban, or edoxaban)

Anti-Xa Activity

- IC50 values for factor Xa inhibition: Rivaroxaban (0.7 nM), Apixaban (2.9 nM) - Association rates: Rivaroxaban is 10- and 1,193-fold faster than apixaban for free factor Xa and prothrombinase complex, respectively - Dissociation rates: Rivaroxaban is about 3-fold faster than apixaban - Concentration-activity relationships: Rivaroxaban and apixaban inhibit factor Xa in a concentration-dependent manner - No data on peak and trough anti-Xa activity values, anti-Xa AUC data, or peak-to-trough fluctuation in anti-Xa activity - No assay-specific differences in anti-Xa measurements between drugs mentioned

Coagulation Tests

- Prothrombin time (PT) effects: Rivaroxaban has a significantly greater effect on PT compared to apixaban. - Activated partial thromboplastin time (aPTT) effects: Rivaroxaban has a significantly greater effect on aPTT compared to apixaban. - Thrombin generation assay results: Rivaroxaban prolongs the lag time and suppresses endogenous thrombin potential more than apixaban.

Molecular Mechanisms

- Binding kinetics: Rivaroxaban inhibits factor Xa up to 4-fold faster than apixaban. Association rates for rivaroxaban are 10- and 1,193-fold faster than apixaban for free factor Xa and prothrombinase-bound factor Xa, respectively. Dissociation rates are about 3-fold faster for rivaroxaban. - Equilibrium binding constants (Ki values): Similar for both drugs, but rivaroxaban is more potent in prothrombinase-induced thrombin generation (Ki values of 0.7 ± 0.3 and 2.9 ± 0.5 nM for rivaroxaban and apixaban, respectively). - Selectivity for factor Xa vs other targets: Not mentioned. - Effects on free factor Xa vs prothrombinase-bound factor Xa: Rivaroxaban inhibits both forms more rapidly than apixaban. - Inhibition models: Mixed inhibition model. - Structural or biochemical differences: Implied by differences in association and dissociation rates, but specific details not provided. - Off-target effects or additional mechanisms: Not mentioned.

Comparative Findings

- Rivaroxaban inhibits factor Xa up to 4-fold faster than apixaban. - Rivaroxaban is 4-fold more potent than apixaban in inhibiting prothrombinase-induced thrombin generation. - Rivaroxaban has a greater effect on global tests of coagulation, prolonging lag time and reducing endogenous thrombin potential more than apixaban. - The faster association rates of rivaroxaban with factor Xa are a key mechanistic difference. - Rivaroxaban is a more potent inhibitor of prothrombin activation due to its rapid inhibition of factor Xa in prothrombinase. - No comparative findings are available for edoxaban as it is not mentioned in the paper.

Special Conditions

Not mentioned (the paper does not provide specific mechanistic data for special populations or conditions like renal impairment)

Abstract Rivaroxaban and apixaban are both small molecules that reversibly inhibit factor Xa. Compared with rivaroxaban, apixaban has minimal effects on the prothrombin time and activated partial thromboplastin time. To investigate this phenomenon, we used a factor Xa-directed substrate in a buffer system. Although rivaroxaban and apixaban inhibited factor Xa with similar Ki values at equilibrium, kinetic measurements revealed that rivaroxaban inhibited factor Xa up to 4-fold faster than apixaban (p < 0.001). Using a discontinuous chromogenic assay to monitor thrombin production by prothrombinase in a purified system, rivaroxaban was 4-fold more potent than apixaban (Ki values of 0.7 ± 0.3 and 2.9 ± 0.5 nM, respectively; p = 0.02). Likewise, in thrombin generation assays in plasma, rivaroxaban prolonged the lag time and suppressed endogenous thrombin potential to a greater extent than apixaban. To characterize how the two inhibitors differ in recognizing factor Xa, inhibition of prothrombinase was monitored in real-time using a fluorescent probe for thrombin. The data were fit using a mixed-inhibition model and the individual association and dissociation rate constants were determined. The association rates for the binding of rivaroxaban to either free factor Xa or factor Xa incorporated into the prothrombinase complex were 10- and 1,193-fold faster than those for apixaban, respectively, whereas dissociation rates were about 3-fold faster. Collectively, these findings suggest that rivaroxaban and apixaban differ in their capacity to inhibit factor Xa and provide a plausible explanation for the observation that rivaroxaban has a greater effect on global tests of coagulation than apixaban.

Introduction

Rivaroxaban and apixaban are oral factor Xa inhibitors that were developed as alternatives to warfarin for the prevention and treatment of arterial and venous thrombosis. 1 These direct, non-vitamin K antagonist oral anticoagulants (DOACs) are licensed in the United States and Europe for stroke prevention in patients with atrial fibrillation 2,3 and for the treatment of venous thromboembolism, 1,[4][5][6] and have experienced rapid adoption in clinical practice. 7,8 he oral factor Xa inhibitors exhibit a similar mechanism of action. As small molecules that bind reversibly to the active site of factor Xa, rivaroxaban, and apixaban inhibit the enzyme with high affinity as evidenced by their sub-nanomolar inhibition constant (K i ) values. 9, 10 In addition to inhibiting free factor Xa, these agents inhibit factor Xa incorporated within prothrombinase, the complex of factor Xa and factor Va that assembles on activated platelets and converts prothrombin to thrombin. 9, 10 Prothrombinase is the central effector of clotting because its assembly induces structural changes in factor Xa that increase the catalytic efficiency of prothrombin activation by over 100,000-fold. 11 rothrombinase propagates coagulation by rapidly generating thrombin at sites of vascular injury. Therefore, the anticoagulant activity of rivaroxaban and apixaban reflects their rapid association with factor Xa incorporated within the prothrombinase complex.

3][14][15][16][17][18] A recent study by Jourdi et al reported that rivaroxaban binds free factor Xa with a 4-fold higher on-rate than apixaban and a 1.5-fold lower K d , and modeling data suggested that this phenomenon explains the greater effect of rivaroxaban on the PT. 19 This difference is thought to contribute to the superiority of rivaroxaban in the thrombin generation assay. 19,20 However, the DOACs may have different inhibitory effects on factor Xa when it is incorporated into the prothrombinase complex and its substrate is prothrombin rather than a low-molecular-weight substrate. 9,10 Therefore, we compared rivaroxaban and apixaban in terms of their affinities for free factor Xa and for factor Xa incorporated into the prothrombinase complex and their rates of inhibition of prothrombinase-induced thrombin generation and we related these effects to those on the PT, aPTT, and thrombin generation assay.

Materials

Human prothrombin, factor Va, and dansylarginine-N-(3ethyl-1,5-pentanediyl)amide (DAPA) were purchased from Haematologic Technologies (Essex Junction, Vermont, United States), whereas factor Xa and thrombin were purchased from Enzyme Research Laboratories (South Bend, IN). A variant prothrombin molecule cleaved solely at Arg320 (R155A, R271A, R284A, rMZ) was used to generate a stable recombinant form of meizothrombin, which was expressed in BHK cells and isolated as described previously. 21 The factor Xa-directed chromogenic substrate, Z-D-Arg-Gly-Arg-pnitroaniline (S-2765), was purchased from Chromogenix (Milano, Italy), whereas the thrombin-directed substrate, Tos-Gly-Pro-Arg-pNA (Chromozym-thrombin [Chz-Th]), was from Hyphen BioMed (Neuville-sur-Oise, France). Fluorogenic factor Xa substrate Pefafluor Xa was from Pentapharm (Basel, Switzerland). Z-Gly-Gly-Arg-7-amino-4methylcoumarin (GGR-AMC) and hirudin were from Bachem Bioscience, Inc. (Philadelphia, Pennsylvania, United States). Phospholipid vesicles were prepared in a 3:1 ratio of phos-phatidylcholine and phosphatidylserine (PCPS) and stored in 10% sucrose at À80°C as described previously. 22,23 Recom-biPlasTin 2G, which contains recombinant tissue factor at a concentration of 0.3 µg/mL, 24 was from Instrumentation Laboratory (Bedford, Massachusetts, United States). Prionex was from Pentapharm (Basel, Switzerland).

Rivaroxaban and apixaban, obtained from Suzhou Howsine Biological Technology Company (Suzhou, China), exhibited single and distinct peaks by HPLC analysis. After dissolving the agents in 100% dimethyl sulfoxide (DMSO) to a concentration of 10 mg/mL, they were stored in aliquots at À80°C. Molar concentrations were calculated using molecular weights of 435.9 and 459.5 for rivaroxaban and apixaban, respectively. Concentrations of rivaroxaban and apixaban were confirmed with the Rotochrom chromogenic anti-Xa assay (Diagnostica Stago, Parsippany, New Jersey, United States), which was performed using a ST4 coagulometer (Diagnostica Stago) with commercial rivaroxaban or apixaban calibrators (Technoclone, Vienna, Austria).

To prepare pooled normal human plasma (PNP), 50 mL of blood was collected from the antecubital veins of 10 to 15 healthy volunteers into a syringe containing 0.105 M sodium citrate. Platelet-poor plasma was obtained as described previously, with the exception that 2,500 Â g was used for centrifugation. 24

Prothrombin Time and Activated Partial Thromboplastin Time

Stock solutions of rivaroxaban and apixaban were diluted in PNP to 500 ng/mL (1,150 and 1,068 μM, respectively) and then serially diluted with PNP to a final concentration of 31.25 ng/mL (72 and 68 nM, respectively). The ranges of inhibitor concentrations were chosen to encompass the peak and trough plasma levels measured in patients treated with these agents. 25,26 The PT was performed using HemosIL HS PLUS (Instrumentation Laboratory, Lexington, Massachusetts, United States) on an ACL 7000 coagulometer (Instrumentation Laboratory) according to the manufacturer's instructions. The aPTT was performed using Actin FSL reagent (Siemens, Marburg, Germany) on a BCS XP analyzer (Siemens) according to the manufacturer's instructions. Assays were performed three times in duplicate at 37°C. These PT and aPTT reagents were chosen because of their wide availability and their reported sensitivities to rivaroxaban and apixaban. 15,27 Molecular Devices, Sunnyvale, California, United States) at excitation and emission at 360 and 460 nm, respectively, with the cutoff filter at 455 nm. All samples and controls were tested in quadruplicate.

Determination of Inhibition Constants of Prothrombinase-Induced Thrombin Generation

To determine their inhibitory effects on prothrombinaseinduced thrombin generation, rivaroxaban or apixaban was preincubated with prothrombin at 25°C prior to addition to a solution containing factor Xa, factor Va, PCPS, and calcium. Reactions were performed in clear or black flat-bottom 96well polystyrene plates (Corning Inc., Corning, New York, United States). The final activation solution consisted of 20 µL of 0 to 200 nM inhibitor, 0.5 nM factor Xa, 5.5 nM factor Va, 50 µM PCPS, 500 nM prothrombin, and 5 mM CaCl 2 in 20 mM HEPES, 150 mM NaCl, pH 7.4 (HBS) containing 0.01% Tween-80 (HBST). Reactions were stopped after 20 seconds by addition of 20 µL of 10 mM EDTA, and thrombin activity was measured by adding 60 µL of 500 µM Chz-Th and monitoring its hydrolysis at 405 nm using a plate reader. The rates of Chz-Th hydrolysis were converted to concentrations of thrombin generated using a specific activity of 17.2 mOD•min À1 •nM À1 , which was determined in a separate experiment. Thrombin concentrations were divided by reaction time to obtain the rate of thrombin formation (nM/s) and divided by the factor Xa concentration to obtain the turnover number (k cat ). The turnover number was then plotted against the concentration of inhibitor, and the data were fit to a rectangular hyperbola equation by nonlinear regression analysis using TableCurve version 4.04 (Jandel Scientific):

where V 0 and V min are the turnover numbers of thrombin generation in the absence and presence of saturating concentrations of inhibitor, respectively; IC 50 is the concentration of inhibitor required to produce half-maximal reduction; and [I] is the inhibitor concentration. K i values were then obtained from the IC 50 value using the Cheng-Prusoff equation:

where [S] is the prothrombin concentration and K M is the Michaelis constant of prothrombinase for prothrombin (157.4 nM) determined in a separate experiment as described previously. 21

Determination of Steady-State Inhibition Constants of Factor Xa

To measure the affinity of the inhibitors for factor Xa at equilibrium, 10 µL aliquots containing 0 to 200 nM rivaroxaban or apixaban were added to wells containing 10 µL of 50 nM factor Xa. Residual factor Xa activity was then measured by adding S-2765 and CaCl 2 to a final volume of 100 µL, yielding final concentrations of 400 µM and 5 mM, respec-tively. Absorbance was monitored at 405 nm using a Spec-traMax Plus plate reader (Molecular Devices, Sunnyvale, California, United States) and the rates of S-2765 hydrolysis were converted to factor Xa concentrations using a specific activity of 25.7 mOD•min À1 •nM À1 , which was determined in a separate experiment (not shown). Residual concentrations of factor Xa were then plotted against anticoagulant concentrations, and the data were analyzed as above to obtain the K i values. All inhibition experiments were conducted at 25°C in HBST containing 0.1% Prionex and 5 mM CaCl 2 (HBSTC). Stock solutions of inhibitors were diluted in HBST, such that the final concentration of DMSO was < 0.01% (v/v); a concentration that, in control experiments, did not influence the hydrolysis of S-2765 by factor Xa.

Rate Constants of Inhibition of Factor Xa

The rates of association and dissociation of the inhibitors with factor Xa were determined simultaneously in HBSTC buffer under pseudo first-order conditions using Molecular Devices Flex Station 3 plate reader. In microplate wells, 100-µL aliquots of a solution containing the inhibitor and Pefafluor Xa were added to 100-µL aliquots of factor Xa, such that final concentrations were 2.5 nM, 100 µM, and 0.25 nM, respectively. Samples were rapidly mixed by the instrument and fluorescence was monitored at 2-second intervals at 37°C

with excitation and emission wavelengths of 342 and 440 nm, respectively, and a 435-nm cutoff filter. To convert relative fluorescence units to nominal concentrations of 7-amino 4-methylcoumarin (AMC), a standard curve was generated from intensity values at 440 nm (I 440 ) of varying concentrations of fully hydrolyzed Pefafluor Xa.

For analysis of the results, a mixed inhibition model 10 was fit to each time course of AMC production (►Scheme 1) using equations (see ►Supplementary Materials) formulated to determine the best-fit individual rate constants of association and dissociation by numerical integration using Berkeley Madonna Software (University of California, Berkeley, California, United States). 21 In this model, FXa is the free factor Xa concentration, S is the concentration of Pefafluor Xa, I is the concentration of inhibitor (apixaban or rivaroxaban), FXa•S is the factor Xa-Pefafluor Xa complex, FXa•I is the factor Xa-inhibitor complex, and FXa•S•I is the ternary Scheme 1 Mixed inhibition of factor Xa by apixaban or rivaroxaban. This model depicts two ways in which factor Xa may bind its inhibitors. FXa, factor Xa; S, substrate (Pefafluor Xa); I, inhibitor (apixaban or rivaroxaban).

factor Xa-Pefafluor Xa-inhibitor complex. The best-fit rate constants were first determined in the absence of inhibitors, then the theoretical dissociation constants for each inhibitor were calculated for free factor Xa (K Xa1 ¼ j 3 /k 3 ) and for the factor Xa-substrate complex (K Xa2 ¼ j 4 /k 4 ).

Rate Constants of Inhibition of Factor Xa in Prothrombinase

The individual rate constants for association and dissociation of the inhibitors with factor Xa incorporated into the prothrombinase complex were determined by measuring thrombin generation to infer residual factor Xa activity in real time. Thrombin generation was quantified using DAPA, a probe that specifically inhibits thrombin and exhibits increased fluorescence on doing so. 29 For the prothrombinase reaction, 60 µL of a 2Â stock solution (solution A) containing 1 nM factor Xa, 11 nM factor Va, 50 µM PCPS, and 10 µM DAPA was added to wells of a 96-well UV-compatible microtiter plate (Corning). In a separate low protein-binding 96-well plate (Thermo), 80 µL of a 2Â solution (solution B) containing 1 µM prothrombin and varying concentrations of either apixaban or rivaroxaban (0, 2, 10, or 30 nM) was added. All reagents were dissolved in 20 mM Tris-HCl, 0.15 M NaCl, pH 7.4, and 5 mM CaCl 2 . Both plates were then placed in a FlexStation 3 (Molecular Devices) and allowed to equilibrate at 37°C for 5 minutes and fluorescence was monitored by reading from the bottom of the wells at 2second intervals using excitation and emission wavelengths of 280 and 545 nm, respectively, and a 530 nm cutoff filter. After determining baseline fluorescence for 30 seconds, 60-µL aliquots of solution B were then transferred to the respective wells of the first plate, and samples were mixed by instrument trituration. Experiments were performed at least three times. To convert the fluorescence signal to nominal thrombin con-centration, a standard curve was generated using known amounts of thrombin in the presence of DAPA.

For analysis, data were fit using a mixed inhibition model 10 (►Scheme 2), wherein all forms of factor Xa (E), both free or within prothrombinase, could interact with the inhibitor (I). Prothrombin (P) activation by prothrombinase was limited to the meizothrombin (M) pathway. This approach is justified because (1) meizothrombin is the predominant intermediate generated when prothrombin is activated by factor Xa in the presence of factor Va; (2) there was minimal prethrombin-2 accumulation by SDS-PAGE analysis (data not shown); and (3) unlike meizothrombin or thrombin, prethrombin-2 does not possess a functional active site. 30 Therefore, the change in fluorescence signal can be attributed to thrombin or meizothrombin. Ratcheting of meizothrombin, a phenomenon whereby meizothrombin alters its conformation and renders the second cleavage by prothrombinase more effectively, was included (k 5 , j 5 in ►Scheme 2) to be consistent with current models of prothrombinase. 31 The inhibition rate constants for the binding of the inhibitors to prothrombinase when it is in complex with either form of meizothrombin were kept the same. In addition, the fluorescence quantum yield of meizothrombin was taken to be 1.5 times that of thrombin. 32,33 Based on this model, 13 differential equations for the reactants and product were formulated (see ►Supplementary Materials) to determine the best-fit rate constants as described earlier.

Briefly, the rate constants of the main reaction that do not involve the inhibitor (k x and j x ) were determined first using similar step-wise restrictions imposed over rounds of fitting (►Supplementary Table S1 ). Once the best-fit k x and j x values for each set of data were obtained, they were fixed in order to determine the best-fit ki x and ji x values for each inhibitor. The Scheme 2 Mixed inhibition of factor Xa in prothrombinase by apixaban or rivaroxaban during prothrombin activation. This model depicts three ways in which prothrombinase may bind its inhibitors. Ratcheting is included in this model, which signifies a conformational change of the intermediate, meizothrombin, so that it is able to be further processed to thrombin. E, prothrombinase; P, prothrombin; M, meizothrombin; M2, ratcheted meizothrombin; T, thrombin; I, inhibitors (apixaban or rivaroxaban).

theoretical dissociation constants for each inhibitor were then calculated for prothrombinase without prothrombin (K Pase1 ¼ ji 1 /ki 1 ), prothrombinase in complex with prothrombin (K Pase2 ¼ ji 2 /ki 2 ), and for prothrombinase in complex with either form of meizothrombin (K Pase3 ¼ ji 3 /ki 3 ) using the individual on-and off-rates.

To determine whether the divergent effects of rivaroxaban and apixaban on the coagulation assays are due to differential inhibition of factor Xa leading to differences in substrate recognition, a mutated form of prothrombin rMZ, which can only be cleaved at Arg320, was used as the substrate so that initial cleavage at this site could be examined in isolation. 34 MZ activation by prothrombinase was monitored and the data were analyzed under conditions identical to those used with wild-type prothrombin. The model shown in ►Scheme 1 was used to describe this reaction, where the substrate (S) was replaced with rMZ and the product (P) was replaced with rMZa. The theoretical dissociation constants for rivaroxaban or apixaban were again calculated using the best-fit rate constants determined as described earlier.

Statistical Analyses

All experiments were performed at least in triplicate and data are expressed as mean AE SD. Inhibition constants (K i ) and inhibition rates (k 2 ) for rivaroxaban and apixaban were compared by Student's t-tests using Microsoft Office Excel 2010. PT, aPTT, and thrombin generation data were compared by two-way analysis of variance using GraphPad Prism 6 (San Diego, California, United States). For all analyses, pvalues < 0.05 were considered statistically significant.

Effect of Inhibitors on the PT and aPTT

The effects of rivaroxaban and apixaban on the PT and aPTT were examined using concentrations up to 2,530 and 2,350 µM (1,100 ng/mL), respectively, well beyond the peak plasma levels obtained with the doses of the drugs that are used clinically. 26,35 Although both agents prolonged the PT and aPTT in a concentration-dependent manner (►Fig. 1), rivaroxaban had a significantly (p < 0.0001) greater effect on both tests than apixaban, consistent with previous reports. 13,27 Therefore, rivaroxaban has a greater effect on global tests of coagulation than apixaban, regardless of whether coagulation is activated via the extrinsic or intrinsic pathway.

Plasma-Based Thrombin Generation Assay

The inhibitory effects of rivaroxaban and apixaban on thrombin generation induced by 3 pM tissue factor in plasma were compared using a calibrated thrombin generation assay. Both agents prolonged the lag time, reduced peak thrombin concentration, and decreased endogenous thrombin potential (ETP) in a concentration-dependent manner (►Fig. 2). Rivaroxaban significantly prolonged the lag time and decreased ETP compared with apixaban, consistent with previous reports. 19,20 The agents had similar effects on other parameters of thrombin generation.

Inhibition of Factor Xa

To begin to investigate the mechanism responsible for the divergent effects of rivaroxaban and apixaban, we first measured their steady-state binding affinities for free factor Xa and for factor Xa assembled in the prothrombinase complex using a chromogenic assay for factor Xa. Rivaroxaban and apixaban inhibited the chromogenic activity of free factor Xa in a concentration-dependent manner (►Fig. 3) ith K i values of 2.6 AE 0.4 nM and 2.0 AE 0.1 nM, respectively, values comparable with those previously reported. 9,10 At a concentration of 20 nM, both agents inhibited the chromogenic activity of 5 nM factor Xa by $85%.

Studies of factor Xa inhibition were then repeated in the presence of factor Va, PCPS, and prothrombin. Hirudin was added to prevent substrate hydrolysis by the generated thrombin. As observed with factor Xa alone, in the presence of prothrombinase components, rivaroxaban and apixaban inhibited over 95% of factor Xa chromogenic activity in the same dose-dependent manner (not shown) with K i values of 1.0 AE 0.6 nM and 0.9 AE 0.8 nM, respectively, values not significantly different from those determined with free factor Xa. Together, these data confirm that rivaroxaban and apixaban bind to free factor Xa and factor Xa incorporated into the prothrombinase complex with comparable affinities. 9,10 Therefore, differences in steady-state binding affinities do not explain the divergent effects of rivaroxaban and apixaban on the PT, aPTT, and thrombin generation.

Inhibition of Prothrombinase-Induced Thrombin Generation

The previous experiments examined the effect of the inhibitors on factor Xa chromogenic activity, an interaction that excludes those employed by prothrombin at sites remote from the active site. 36 To examine the effect of the inhibitors when prothrombin is the substrate of prothrombinase, we quantified thrombin generation. Although both inhibitors reduced prothrombinase-induced thrombin production in a concentration-dependent manner (►Fig. 4), the K i value for rivaroxaban was 4-fold lower than that for apixaban (0.7 AE 0.3 nM and 2.9 AE 0.5 nM, respectively; p < 0.02). Therefore, rivaroxaban is a more potent inhibitor of prothrombinase-induced thrombin formation than apixaban.

Inhibition Rate Constants: Factor Xa Binding

To better understand why rivaroxaban and apixaban have different effects on prothrombinase-induced thrombin generation, we next compared the binding kinetics of the inhibitors with factor Xa using a mixed inhibition model (►Scheme 1). 10 Inhibition was quantified by continuously monitoring factor Xa fluorogenic activity in the presence of rivaroxaban or apixaban. Although the time courses revealed gradual reduction of factor Xa fluorogenic activity, as evidenced by declining rates of fluorophore release with both inhibitors (►Fig. 5), plots of residual factor Xa activity revealed more rapid inhibition with rivaroxaban than with apixaban (►Fig. 5, inset). Plots of product versus time were analyzed by global fitting and the best-fit values for k 1 , j 1 , and k 2 were determined to be 7.05 AE 0.30 nM À1 •s À1 , 1.74 AE 0.09 s À1 , and 187 AE 14 s À1 , respectively. With these values fixed, the best-fit values for the rate constants for inhibition were determined (►Table 1). The on-and off-rate constants for binding factor Xa alone (k 3 and j 3 , respectively) for rivaroxaban were, on average, 21-fold faster than those for apixaban. A similar trend was observed for inhibitor binding to the factor Xa-fluorogenic substrate complex (k 4 and j 4 ), whereby rivaroxaban bound, on average, 25-fold faster than apixaban. Therefore, contrary to a previous report, 19 our data suggest that a faster on-rate is not the sole explanation for the greater effect of rivaroxaban on tests of coagulation than apixaban.

Inhibition Rate Constants: Prothrombin Activation

Because of the differences observed in the prothrombinase endpoint assay (►Fig. 4), we used a continuous assay to determine the kinetic rate constants for inhibition of factor Xa assembled in prothrombinase. Thrombin generation was quantified by continuously monitoring the increase in fluorescence that accompanies binding of DAPA, a fluorophore that binds exclusively to the active site of meizothrombin or thrombin. 29 Plots of thrombin concentration versus time in the presence of varying concentrations of the inhibitors were analyzed using the mixed inhibition model (►Scheme 2), wherein the inhibitor is allowed to bind to prothrombinasebound factor Xa in the absence or presence of prothrombin or meizothrombin. Concentrations of thrombin and meizothrombin over time were calculated by numerical integration. Individual rate constant parameter values were optimized using the Simplex procedure 37 as described previously. 21 Best-fit association and dissociation rate constants, and the calculated K Pase value, representing the dissociation constant for inhibitor with factor Xa in prothrombinase, are illustrated in ►Table 2. The rate constants associated with prothrombin activation by prothrombinase in the absence of inhibitor are shown in ►Supplementary Table S1 . Using the best-fit rate constants, regression lines for the progress curves in the presence of 2 to 30 nM apixaban (►Fig. 6A) or rivaroxaban (►Fig. 6B) were generated. On-rate constants ki 1 and ki 2 for rivaroxaban were 1,193-fold and 3fold higher than those for apixaban, respectively (►Table 2). The higher ki 1 and ki 2 values with rivaroxaban are comparable with the 10-fold higher k 3 and 45-fold higher k 4 values observed with rivaroxaban relative to apixaban in the factor Xa chromogenic assay (►Table 1). Interestingly, the on-rate for rivaroxaban with the prothrombinase-meizothrombin complex, ki 3 , was 3-fold slower than that for apixaban. Therefore, rivaroxaban and apixaban have different off-rates as well as different on-rates. The greatest difference in off:on (ji x /ki x ) rate ratios of these constants (K Pase ) was with free enzyme inhibition (K Pase1 ), where the value for rivaroxaban was 454-fold lower than that for apixaban, mostly due to the very low ki 1 value for apixaban (►Table 2). Closer inspection reveals that only the on-rate of apixaban is altered by the assembly of factor Xa into prothrombinase, suggesting that rivaroxaban and apixaban interact with factor Xa differently.

Inhibition Rate Constants: rMZ Activation

Because prothrombin activation requires two cleavages, kinetic analyses are complicated by the requirement to account for parallel reaction pathways. To simplify the analysis, we used rMZ, a recombinant prothrombin derivative that can only be cleaved at Arg320. 34 Activation of rMZ was monitored using DAPA and kinetic analyses were performed as outlined earlier. Best-fit rate constants (►Table 3) were used to generate regression lines (►Fig. 7) . The off:on (ji x /ki x ) ratio of free prothrombinase (K Pase1 ) was 16-fold lower with rivaroxaban than with apixaban, whereas K Pase2 , representing inhibition of the prothrombinase-rMZ complex, was 3-fold lower with rivaroxaban than with apixaban. The difference in the effect between rivaroxaban and apixaban on rMZ activation is most evident in the initial stages of rMZ activation (►Fig. 7, inset), similar to what was observed with wild-type prothrombin (►Fig. 6). Therefore, these data suggest that rivaroxaban and apixaban have different effects on the initial interaction of factor Xa in the prothrombinase complex with prothrombin, and that this difference occurs upon the initial presentation and cleavage of prothrombin at Arg320.

Discussion

14]17 To determine whether the differences between the on-rates of the inhibitors for free factor Xa 19 are maintained when factor Xa is incorporated into prothrombinase and to correlate those with global tests of coagulation, we compared the effects of rivaroxaban and apixaban in three different systems. First, we compared their effects in plasma on the PT, aPTT, and thrombin generation assay. Second, we determined their K i values for free factor Xa and factor Xa incorporated into prothrombinase. Third, we measured the rate constants for inhibition of factor Xa alone and when assembled in prothrombinase.

As previously reported, rivaroxaban and apixaban inhibited factor Xa with similar K i values, confirming that the agents bind factor Xa with similar affinities at equilibrium. 9, 10,19 3][14] In the thrombin generation assay, rivaroxaban and apixaban influenced all parameters in a concentration-dependent manner, although rivaroxaban prolonged the lag time and reduced ETP to a significantly greater extent than apixaban. Because clot formation typically occurs shortly after the lag time, the bulk of thrombin is generated after the clot has formed. The fact that rivaroxaban prolongs the lag time more than apixaban is consistent with its greater effects in clot-based assays such as PT and aPTT. The effects of apixaban and rivaroxaban on the rates of prothrombin activation were examined. These experiments differ from those for determination of K i because all the components of prothrombinase are present, and we quantified thrombin generation rather than factor Xa activity. 38 ivaroxaban was 2-fold more potent than apixaban at inhibiting prothrombinase-induced thrombin generation. This was further investigated by determining the association and dissociation rate constants for inhibition of factor Xa. Using a mixed-type inhibition model, on-and off-rates for rivaroxaban were higher than those for apixaban. These findings are consistent with the results from the prothrombinase and clotting assays and they suggest that there are mechanistic differences in the way that rivaroxaban and apixaban interact with factor Xa incorporated within the prothrombinase complex.

rMZ was used to shed further light on the molecular mechanism responsible for the divergent inhibitory effects of rivaroxaban and apixaban on factor Xa incorporated into the prothrombinase complex. Like the results with native prothrombin, rivaroxaban is more potent than apixaban at inhibiting prothrombinase-induced activation of rMZ (►Figs. 6 and 7). Together, these data suggest that the differences between the inhibitors can mainly be attributed to their divergent inhibitory effects on the initial activation cleavage of prothrombin at Arg320 by factor Xa, which is the major pathway of prothrombin activation in the presence of factor Va, and the only available cleavage site in rMZ. The off: on ratios for rivaroxaban were 1.5-fold (►Table 2) and 2.8fold (►Table 3) lower than those for apixaban with native prothrombin and rMZ, respectively. Interestingly, incorporation of factor Xa into prothrombinase appears to have more profound effects on the inhibitory activity of apixaban than rivaroxaban, suggesting that there may be fundamental differences in the way that the two agents interact with factor Xa when it assembles into its activation complex.

A recent study suggested that the differential effects of rivaroxaban and apixaban on the PT reflected a 4-fold faster association rate constant of rivaroxaban for free factor Xa. 19 We have extended this concept by examining factor Xa assembled into the prothrombinase complex. Thus, we have shown that the association rate of rivaroxaban is faster than that of apixaban not only with free factor Xa, but also with factor Xa assembled into the prothrombinase complex, and with prothrombinase activation of prothrombin. Furthermore, using the rMZ variant, we demonstrate that the initial cleavage of prothrombin at Arg320 is faster with rivaroxaban than with apixaban. Biochemical modeling of enzyme-substrate-inhibitor interaction confirms that the difference between the inhibitors is not limited to the association rates with the enzyme (either factor Xa or prothrombinase). Instead, differences in both the association and dissociation rate constants are responsible for the greater activity of rivaroxaban relative to apixaban.

Although rivaroxaban and apixaban were both developed as small molecule factor Xa inhibitors, differences in their kinetic parameters and functional potencies are evident. This could arise from differences in sites of contact with the enzyme, as well as how the substrate binds. 1][42][43] These differences may account for the disparate association rate constants and their divergent effects on prothrombin activation.

Our study has some limitations. First, the kinetic analyses were conducted in purified systems, whereas the clotting and thrombin generation assay were performed in plasma. Therefore, translation of findings from one system to the other can be problematic. For example, binding to proteins such as albumin attenuates the anticoagulant effects in plasma. However, this phenomenon does not explain why rivaroxaban is more potent than apixaban in the plasmabased tests because apixaban exhibits less protein binding than rivaroxaban. Second, in the assays in purified systems, the binding of rivaroxaban and apixaban to factor Xa may differentially alter the kinetics of chromogenic substrate hydrolysis. 44 However, the consistent differences between rivaroxaban and apixaban in all of the tests render this possibility unlikely. Therefore, there appear to be fundamental biochemical differences in the way that rivaroxaban and apixaban interact with factor Xa incorporated within the prothrombinase complex. In conclusion, we show that rivaroxaban is a more potent inhibitor of prothrombin activation because it inhibits factor Xa incorporated into prothrombinase more rapidly than apixaban. This difference provides an explanation for the observation that rivaroxaban has a greater effect on global tests of coagulation than apixaban.

annex

Conflicts of Interest J.I.W. has served as a consultant and has received honoraria from Bristol-Myers Squibb, Pfizer, Bayer, and Janssen. None of the other authors declare conflicts of interest.

Funding

Funding C.H.Y. is supported by a doctoral scholarship from the Canadian Institutes of Health Research.B.J.D is supported by a Scientific Education Grant from the Australasian Society of Thrombosis and Haemostasis.P.Y.K. is supported by the Hamilton Health Sciences Early Career Award and the New Investigator Fund.J.I.W. holds the Canada Research Chair (Tier I) in Thrombosis and the Heart and Stroke Foundation J. Fraser Mustard Chair in Cardiovascular Research at McMaster University.

Failed to load PDF:

StripeM-Inner

Paper sources

Abstract screening pilot

Abstract screening results

Extraction pilot

Extraction results

Research report

Modify setup