Multiscale simulation of thrombus growth and vessel occlusion triggered by collagen/tissue factor using a data-driven model of combinatorial platelet signalling

Decoding thrombosis through code: a review of computational models

From the molecular level up to a blood vessel, thrombosis and hemostasis involves many interconnected biochemical and biophysical processes over a wide range of length and time scales. Computational modeling has gained eminence in offering insights into these processes beyond what can be obtained from in vitro or in vivo experiments, or clinical measurements. The multiscale and multiphysics nature of thrombosis has inspired a wide range of modeling approaches that aim to address how a thrombus forms and dismantles. Here, we review recent advances in computational modeling with a focus on platelet-based thrombosis. We attempt to summarize the diverse range of modeling efforts straddling the wide-spectrum of physical phenomena, length scales, and time scales; highlighting key advancements and insights from existing studies. Potential information gleaned from models is discussed, ranging from identification of thrombus-prone regions in patient-specific vasculature to modeling thrombus deformation and embolization in response to fluid forces. Furthermore, we highlight several limitations of current models, future directions in the field, and opportunities for clinical translation, to illustrate the state-of-the-art. There are a plethora of opportunity areas for which models can be expanded, ranging from topics of thromboinflammation to platelet production and clearance. Through successes demonstrated in existing studies described here, as well as continued advancements in computational methodologies and computer processing speeds and memory, in silico investigations in thrombosis are poised to bring about significant knowledge growth in the years to come.

Recent Advances in Computational Modeling of Biomechanics and Biorheology of Red Blood Cells in Diabetes

Diabetes mellitus, a metabolic disease characterized by chronically elevated blood glucose levels, affects about 29 million Americans and more than 422 million adults all over the world. Particularly, type 2 diabetes mellitus (T2DM) accounts for 90-95% of the cases of vascular disease and its prevalence is increasing due to the rising obesity rates in modern societies. Although multiple factors associated with diabetes, such as reduced red blood cell (RBC) deformability, enhanced RBC aggregation and adhesion to the endothelium, as well as elevated blood viscosity are thought to contribute to the hemodynamic impairment and vascular occlusion, clinical or experimental studies cannot directly quantify the contributions of these factors to the abnormal hematology in T2DM. Recently, computational modeling has been employed to dissect the impacts of the aberrant biomechanics of diabetic RBCs and their adverse effects on microcirculation. In this review, we summarize the recent advances in the developments and applications of computational models in investigating the abnormal properties of diabetic blood from the cellular level to the vascular level. We expect that this review will motivate and steer the development of new models in this area and shift the attention of the community from conventional laboratory studies to combined experimental and computational investigations, aiming to provide new inspirations for the development of advanced tools to improve our understanding of the pathogenesis and pathology of T2DM.

A three-dimensional multiscale model for the prediction of thrombus growth under flow with single-platelet resolution

Modeling thrombus growth in pathological flows allows evaluation of risk under patient-specific pharmacological, hematological, and hemodynamical conditions. We have developed a 3D multiscale framework for the prediction of thrombus growth under flow on a spatially resolved surface presenting collagen and tissue factor (TF). The multiscale framework is composed of four coupled modules: a Neural Network (NN) that accounts for platelet signaling, a Lattice Kinetic Monte Carlo (LKMC) simulation for tracking platelet positions, a Finite Volume Method (FVM) simulator for solving convection-diffusion-reaction equations describing agonist release and transport, and a Lattice Boltzmann (LB) flow solver for computing the blood flow field over the growing thrombus. A reduced model of the coagulation cascade was embedded into the framework to account for TF-driven thrombin production. The 3D model was first tested against in vitro microfluidics experiments of whole blood perfusion with various antiplatelet agents targeting COX-1, P₂Y₁, or the IP receptor. The model was able to accurately capture the evolution and morphology of the growing thrombus. Certain problems of 2D models for thrombus growth (artifactual dendritic growth) were naturally avoided with realistic trajectories of platelets in 3D flow. The generalizability of the 3D multiscale solver enabled simulations of important clinical situations, such as cylindrical blood vessels and acute flow narrowing (stenosis). Enhanced platelet-platelet bonding at pathologically high shear rates (e.g., von Willebrand factor unfolding) was required for accurately describing thrombus growth in stenotic flows. Overall, the approach allows consideration of patient-specific platelet signaling and vascular geometry for the prediction of thrombotic episodes.

The excessive formation of blood clots under flow within the circulatory system (thrombosis) is known to initiate heart attacks and strokes. Therefore, obtaining insights into the formation and progression of these clots will be useful in evaluating pharmacological options. To this end, we have developed a 3D computational model that tracks the growth of a blood clot under flow from initial platelet deposition to full vessel occlusion in the presence of soluble platelet agonists. We first validated the model against experimental predictions of blood clots formed in vitro. Due to the construction of the model in 3D, we were able to carry out simulations of clot formation under important clinical situations, namely cylindrical blood vessels and acute flow narrowings (stenoses). To our knowledge, our model is the first of its kind that can account for patient-specific platelet phenotypes to perform robust 3D simulations of thrombus growth in geometries of clinical relevance.

Sensitivity analysis of a reduced model of thrombosis under flow: Roles of Factor IX, Factor XI, and γ'-Fibrin

A highly reduced extrinsic pathway coagulation model (8 ODEs) under flow considered a thin 15-micron platelet layer where transport limitations were largely negligible (except for fibrinogen) and where cofactors (FVIIa, FV, FVIII) were not rate-limiting. By including thrombin feedback activation of FXI and the antithrombin-I activities of fibrin, the model accurately simulated measured fibrin formation and thrombin fluxes. Using this reduced model, we conducted 10,000 Monte Carlo (MC) simulations for ±50% variation of 5 plasma zymogens and 2 fibrin binding sites for thrombin. A sensitivity analysis of zymogen concentrations indicated that FIX activity most influenced thrombin generation, a result expected from hemophilia A and B. Averaging all MC simulations confirmed both the mean and standard deviation of measured fibrin generation on 1 tissue factor (TF) molecule per μm2. Across all simulations, free thrombin in the layer ranged from 20 to 300 nM (mean: 50 nM). The top 2% of simulations that produced maximal fibrin were dominated by conditions with low antithrombin-I activity (decreased weak and strong sites) and high FIX concentration. In contrast, the bottom 2% of simulations that produced minimal fibrin were dominated by low FIX and FX. The percent reduction of fibrin by an ideal FXIa inhibitor (FXI = 0) ranged from 71% fibrin reduction in the top 2% of MC simulations to only 34% fibrin reduction in the bottom 2% of MC simulations. Thus, the antithrombotic potency of FXIa inhibitors may vary depending on normal ranges of zymogen concentrations. This reduced model allowed efficient multivariable sensitivity analysis.

A 1D-3D Hybrid Model of Patient-Specific Coronary Hemodynamics

PURPOSE

Coronary flow is affected by evolving events such as atherosclerotic plaque formation, rupture, and thrombosis, resulting in myocardial ischemia and infarction. Highly resolved 3D hemodynamic data at the stenosis is essential to model shear-sensitive thrombotic events in coronary artery disease.

METHODS

We developed a hybrid 1D-3D simulation framework to compute patient-specific coronary hemodynamics efficiently. A 1D model of the coronary flow is coupled to an image-based 3D model of the region of interest. This framework affords the advantages of reduced-order modeling, decreasing the global computational cost, without sacrificing the accuracy of the quantities of interest.

RESULTS

We validated our 1D-3D model against full 3D coronary simulations in healthy and diseased conditions. Our results showed good agreement between the 3D and the 1D-3D models while reducing the computational cost by 40-fold compared to the 3D simulation. The 1D-3D model predicted left/right coronary flow distribution within 3% and provided an accurate estimation of fractional flow reserve and wall shear stress distribution at the stenosis comparable to the 3D simulation.

CONCLUSION

Savings in computational cost may be significant in situations with changing geometry, such as growing thrombosis. Also, this approach would allow quantifying the time-dependent effect of thrombotic growth and occlusion on the global coronary circulation.

Computational modeling of blood component transport related to coronary artery thrombosis in Kawasaki disease

Coronary artery thrombosis is the major risk associated with Kawasaki disease (KD). Long-term management of KD patients with persistent aneurysms requires a thrombotic risk assessment and clinical decisions regarding the administration of anticoagulation therapy. Computational fluid dynamics has demonstrated that abnormal KD coronary artery hemodynamics can be associated with thrombosis. However, the underlying mechanisms of clot formation are not yet fully understood. Here we present a new model incorporating data from patient-specific simulated velocity fields to track platelet activation and accumulation. We use a system of Reaction-Advection-Diffusion equations solved with a stabilized finite element method to describe the evolution of non-activated platelets and activated platelet concentrations [AP], local concentrations of adenosine diphosphate (ADP) and poly-phosphate (PolyP). The activation of platelets is modeled as a function of shear-rate exposure and local concentration of agonists. We compared the distribution of activated platelets in a healthy coronary case and six cases with coronary artery aneurysms caused by KD, including three with confirmed thrombosis. Results show spatial correlation between regions of higher concentration of activated platelets and the reported location of the clot, suggesting predictive capabilities of this model towards identifying regions at high risk for thrombosis. Also, the concentration levels of ADP and PolyP in cases with confirmed thrombosis are higher than the reported critical values associated with platelet aggregation (ADP) and activation of the intrinsic coagulation pathway (PolyP). These findings suggest the potential initiation of a coagulation pathway even in the absence of an extrinsic factor. Finally, computational simulations show that in regions of flow stagnation, biochemical activation, as a result of local agonist concentration, is dominant. Identifying the leading factors to a pro-coagulant environment in each case—mechanical or biochemical—could help define improved strategies for thrombosis prevention tailored for each patient.

Computational studies aiming to model thrombosis often rely on an arterial wall injury. Collagen and other extracellular matrix components are exposed to the bloodstream, which facilitates platelet adhesion to the wall and subsequent clot formation. However, these models are not adequate to explain thrombosis in other settings where even in the absence of a focal lesion, clots may still form under certain flow conditions. Coronary artery aneurysm thrombosis following KD is an example of the need to understand the mechanisms of thrombus initiation in the absence of an extrinsic factor. This study provides a new framework to investigate thrombus initiation in KD from a patient-specific perspective, which integrates fluid mechanics and biochemistry and which could help quantify the pro-coagulant environment induced by the aneurysm and become a predictive tool. The work presented here has broad relevance to other clinical situations where flow stagnation and transport are driving factors in thrombus formation.

Reduced model to predict thrombin and fibrin during thrombosis on collagen/tissue factor under venous flow: Roles of γ'-fibrin and factor XIa

During thrombosis, thrombin generates fibrin, however fibrin reversibly binds thrombin with low affinity E-domain sites (K_D = 2.8 μM) and high affinity γ’-fibrin sites (K_D = 0.1 μM). For blood clotting on collagen/tissue factor (1 TF-molecule/μm²) at 200 s^-1 wall shear rate, high μM-levels of intraclot thrombin suggest robust prothrombin penetration into clots. Setting intraclot zymogen concentrations to plasma levels (and neglecting cofactor rate limitations) allowed the linearization of 7 Michaelis-Menton reactions between 6 species to simulate intraclot generation of: Factors FXa (via TF/VIIa or FIXa), FIXa (via TF/FVIIa or FXIa), thrombin, fibrin, and FXIa. This reduced model [7 rates, 2 K_D’s, enzyme half-lives~1 min] predicted the measured clot elution rate of thrombin-antithrombin (TAT) and fragment F1.2 in the presence and absence of the fibrin inhibitor Gly-Pro-Arg-Pro. To predict intraclot fibrin reaching 30 mg/mL by 15 min, the model required fibrinogen penetration into the clot to be strongly diffusion-limited (actual rate/ideal rate = 0.05). The model required free thrombin in the clot (~100 nM) to have an elution half-life of ~2 sec, consistent with measured albumin elution, with most thrombin (>99%) being fibrin-bound. Thrombin-feedback activation of FXIa became prominent and reached 5 pM FXIa at >500 sec in the simulation, consistent with anti-FXIa experiments. In predicting intrathrombus thrombin and fibrin during 15-min microfluidic experiments, the model revealed “cascade amplification” from 30 pM levels of intrinsic tenase to 15 nM prothrombinase to 15 μM thrombin to 90 μM fibrin. Especially useful for multiscale simulation, this reduced model predicts thrombin and fibrin co-regulation during thrombosis under flow.

During blood clotting events, a complex series of reaction are involved. Simulation gives insights to the concentration of different enzymes which are at too low of concentration to be detected. However, the models are often large and difficult to solve for clotting under flow conditions. With a thin film approximation, we were able to simplify clotting under flow with parameters from literature, with only 3 adjusted in order to fit the experimental data. This model gave insights into the dynamics of the species involved, and the roles of γ’-fibrin and thrombin feedback activation. This reduced model may be useful in further multiscale simulations.

Coagulopathy implications using a multiscale model of traumatic bleeding matching macro- and microcirculation

Quantifying the relationship between vascular injury and the dynamic bleeding rate requires a multiscale model that accounts for changing and coupled hemodynamics between the global and microvascular levels. A lumped, global hemodynamic model of the human cardiovascular system with baroreflex control was coupled to a local 24-level bifurcating vascular network that spanned diameters from the muscular artery scale (0.1-1.3 mm) to capillaries (5-10 μm) via conservation of momentum and conservation of mass boundary conditions. For defined injuries of severing all vessels at each nth-level, the changing pressures and flowrates were calculated using prescribed shear-dependent hemostatic clot growth rates (normal or coagulopathic). Key results were as follows: 1) the upstream vascular network rapidly depressurizes to reduce blood loss; 2) wall shear rates at the hemorrhaging wound exit are sufficiently high (~10,000 s^-1) to drive von Willebrand factor unfolding; 3) full coagulopathy results in >2-liter blood loss in 2 h for severing all vessels of 0.13- to 0.005-mm diameter within the bifurcating network, whereas full hemostasis limits blood loss to <100 ml within 2 min; and 4) hemodilution from transcapillary refill increases blood loss and could be implicated in trauma-induced coagulopathy. A sensitivity analysis on length-to-diameter ratio and branching exponent demonstrated that bleeding was strongly dependent on these tissue-dependent network parameters. This is the first bleeding model that prescribes the geometry of the injury to calculate the rate of pressure-driven blood loss and local wall shear rate in the presence or absence of coagulopathic blood. NEW & NOTEWORTHY We developed a multiscale model that couples a lumped, global hemodynamic model of a patient to resolved, single-vessel wounds ranging from the small artery to capillary scale. The model is able to quantify wall shear rates, seal rates, and blood loss rates in the presence and absence of baroreflex control and hemodilution.

Review of quantitative systems pharmacological modeling in thrombosis

Hemostasis and thrombosis are often thought as two sides of the same clotting mechanism whereas hemostasis is a natural protective mechanism to prevent bleeding and thrombosis is a blood clot abnormally formulated inside a blood vessel, blocking the normal blood flow. The evidence to date suggests that at least arterial thrombosis results from the same critical pathways of hemostasis. Analysis of these complex processes and pathways using quantitative systems pharmacological model-based approach can facilitate the delineation of the causal pathways that lead to the emergence of thrombosis. In this paper, we provide an overview of the main molecular and physiological mechanisms associated with hemostasis and thrombosis, and review the models and quantitative system pharmacological modeling approaches that are relevant in characterizing the interplay among the multiple factors and pathways of thrombosis. An emphasis is given to computational models for drug development. Future trends are discussed.

Microfluidics contrasted to thrombelastography: perplexities in defining hypercoagulability

BACKGROUND

Elevated clot strength (maximum amplitude [MA]) measured by thrombelastography (TEG) is associated with thrombotic complications. However, it remains unclear how MA translates to thrombotic risks, as this measurement is independent of time, blood flow, and clot degradation. We hypothesize that under flow conditions, increased clot strength correlates to time-dependent measurements of coagulation and resistance to fibrinolysis.

MATERIALS AND METHODS

Surgical patients at high risk of thrombotic complications were analyzed with TEG and total thrombus-formation analysis system (T-TAS). TEG hypercoagulability was defined as an r <10.2 min, angle >59, MA >66 or LY30 <0.2% (based off of healthy control data, n = 141). The T-TAS AR and PL chips were used to measure clotting at arterial shear rates. T-TAS measurements include occlusion start time, occlusion time (OT), occlusion speed (OSp), and total clot generation (area under the curve). These measurements were correlated to TEG indices (R time, angle, MA, and LY30). Both T-TAS and TEG assays were challenged with tissue plasminogen activator (t-PA) to assess clot resistance to fibrinolysis.

RESULTS

Thirty subjects were analyzed, including five controls. TEG-defined hypercoagulability by MA was detected in 52% of the inflammatory bowel disease/cancer patients; 0% was detected in the controls. There were no TEG measurements that significantly correlated with T-TAS AR and PL chip. However, in the presence of t-PA, T-TAS AR determined OSp to have an inverse relationship with TEG angle (-0.477, P = 0.012) and LY30 (-0.449, P = 0.019), and a positive correlation with R time (0.441 P = 0.021). In hypercoagulability determined by TEG MA, T-TAS PL had a significantly reduced OT (4:07 versus 6:27 min, P = 0.043). In hypercoagulability defined by TEG LY30, T-TAS PL had discordant findings, with a significantly prolonged OT (6:36 versus 4:30 min, P = 0.044) and a slower OSp (10.5 versus 19.0 kPa/min, P = 0.030).

CONCLUSIONS

Microfluidic coagulation assessment with T-TAS has an overall poor correlation with most TEG measurements in a predominantly hypercoagulable patient population, except in the presence of t-PA. The one anticipated finding was an elevated MA having a shorter time to platelet-mediated microfluidic occlusion, supporting the role of platelets and hypercoagulability. However, hypercoagulability defined by LY30 had opposing results in which a low LY30 was associated with a longer PL time to occlusion and slower OSp. These discordant findings warrant ongoing investigation into the relationship between clot strength and fibrinolysis under different flow conditions.

Establishing the Transient Mass Balance of Thrombosis: From Tissue Factor to Thrombin to Fibrin Under Venous Flow

Supplemental Digital Content is available in the text.

Objective—

We investigated the coregulation of thrombin and fibrin as blood flows over a procoagulant surface.

Approach and Results—

Using microfluidic perfusion of factor XIIa-inhibited human whole blood (200 s⁻¹ wall shear rate) over a 250-μm long patch of collagen/TF (tissue factor; ≈1 molecule per μm²) and immunoassays of the effluent for F1.2 (prothrombin fragment 1.2), TAT (thrombin–antithrombin complex), and D-dimer (post–end point plasmin digest), we sought to establish the transient mass balance for clotting under venous flow. F1.2 (but almost no free thrombin detected via TAT assay) continually eluted from clots when fibrin was allowed to form. Low-dose fluorescein-Phe-Pro-Arg-chloromethylketone stained fibrin-bound thrombin—a staining ablated by anti–γ′-fibrinogen or the fibrin inhibitor glypro-arg-pro but highly resistant to 7-minute buffer rinse, demonstrating tight binding of thrombin to γ′-fibrin. With fibrin polymerizing for 500 seconds, 92 000 thrombin molecules and 203 000 clot-associated fibrin monomer equivalents were generated per TF molecule (or per μm²). Fibrin reached 15 mg/mL in the pore space (porosity ≈0.5) of a 15-μm-thick thrombus core by 500 seconds and 30 mg/mL by 800 seconds. For a known rate of ≈60 FPA (fibrinopeptide-A) per thrombin per second, each thrombin molecule generated only 3 fibrin monomer equivalents during 500 seconds, indicating an intraclot thrombin half-life of ≈70 ms, much shorter than its diffusional escape time (≈10 seconds). By 800 seconds, gly-pro-arg-pro allowed 4-fold more F1.2 generation, consistent with gly-pro-arg-pro ablating fibrin’s antithrombin-I activity and facilitating thrombin-mediated FXIa activation.

Conclusions—

Under flow, fibrinogen continually penetrates the clot, and γ′-fibrin regulates thrombin.

A parallel fluid-solid coupling model using LAMMPS and Palabos based on the immersed boundary method

The study of viscous fluid flow coupled with rigid or deformable solids has many applications in biological and engineering problems, e.g., blood cell transport, drug delivery, and particulate flow. We developed a partitioned approach to solve this coupled Multiphysics problem. The fluid motion was solved by Palabos (Parallel Lattice Boltzmann Solver), while the solid displacement and deformation was simulated by LAMMPS (Large-scale Atomic/Molecular Massively Parallel Simulator). The coupling was achieved through the immersed boundary method (IBM). The code modeled both rigid and deformable solids exposed to flow. The code was validated with the Jeffery orbits of an ellipsoid particle in shear flow, red blood cell stretching test, and effective blood viscosity flowing in tubes. It demonstrated essentially linear scaling from 512 to 8192 cores for both strong and weak scaling cases. The computing time for the coupling increased with the solid fraction. An example of the fluid-solid coupling was given for flexible filaments (drug carriers) transport in a flowing blood cell suspensions, highlighting the advantages and capabilities of the developed code.

Microfluidic whole blood testing of platelet response to pharmacological agents

Platelets present a number of intracellular and transmembrane targets subject to pharmacological modulation, either for cardiovascular disease reduction or as an unintended drug response. Microfluidic devices allow human blood to clot on a defined surface under controlled hemodynamic and pharmacological conditions. The potencies of a number of antiplatelet and anticancer drugs have been tested with respect to platelet deposition on collagen under flow. Inhibitors of cyclooxygenase-1 (COX-1) reduce platelet deposition, either when added ex vivo to blood or ingested orally by patients prior to testing. Some individuals display a functional "aspirin-insensitivity" in microfluidic assay. When certain nonsteroidal anti-inflammatory drugs (NSAIDs) are taken orally, they block COX-1 acetylation by aspirin with concomitant reduction of aspirin efficacy against platelets in microfluidic assay. Both P2Y₁ and P2Y₁₂ inhibitors reduce platelet deposition under flow, as do NO donors and iloprost that target the guanylate cyclase and the prostacyclin receptor, respectively. In a microfluidic assay of 37 kinase inhibitors, dasatinib had potent antiplatelet activity, while bosutinib was less potent. Dasatinib and bosutinib have known profiles against numerous kinases, revealing overlapping and nonoverlapping activities relevant to their unique actions against platelets. Also, dasatinib caused a marked and specific inhibition of GPVI signaling induced by convulxin, consistent with a dasatinib-associated bleeding risk. Microfluidic devices facilitate drug library screening, dose-response testing, and drug-drug interaction studies. Kinase inhibitors developed as anticancer agents may present antiplatelet activities that are detectable by microfluidic assay and potentially linked to bleeding risks.