(210e) Multiscale Prediction of Patient-Specific Platelet Function Under Flow

Diamond, S. L., University of Pennsylvania
Colace, T., University of Pennsylvania
Chatterjee, M. S., University of Pennsylvania
Jing, H., University of Pennsylvania
Zhou, S., University of Pennsylvania
Jaeger, D., University of Pennsylvania
Brass, L. F., University of Pennsylvania

During thrombotic or hemostatic episodes, platelets bind
collagen and release ADP and thromboxane A2, recruiting additional platelets to a growing deposit that
distorts the flow field. Prediction of clotting function under hemodynamic
conditions for an individual's platelet phenotype remains a challenge. A
platelet signaling phenotype was obtained for 3 healthy donors using Pairwise
Agonist Scanning (PAS), where calcium dyeloaded platelets were exposed to
pairwise combinations of ADP, U46619, and convulxin to activate P2Y1/P2Y12, TP, and GPVI receptors,
respectively, in the presence or absence of the IP receptor agonist, iloprost.
A neural network model was trained on each donor's PAS experiment and then was
embedded into a multiscale Monte Carlo simulation of donor-specific platelet
deposition under flow. The simulations were directly compared to microfluidic
experiments of whole blood flowing over collagen at 200 and 1000 s-1 wall shear rate. The simulations
predicted the ranked order of drug sensitivity for indomethacin, aspirin,
MRS-2179 (P2Y1
inhibitor), and
iloprost. Consistent with measurement and simulation, one donor displayed
larger clots, while another donor presented indomethacin-resistance (revealing
a novel heterozygote TP-V241G mutation). In silico representations of an
individual's platelet phenotype allowed prediction of blood function under
flow, essential to identifying patient-specific cardiovascular risks, drug
responses, and novel genotypes.

Fig. 1. Multiscale model of
combinatorial platelet activation and thrombus formation under flow.
Platelet agonists (blue)
used individually or in pairs to activate GPVI or G-protein coupled receptors
(thromboxane receptor, TP; purinergic receptors P2Y1 and P2Y12; and the
prostacyclin receptor, IP) result in modulation of intracellular calcium (green)
from intracellular stores distal of phospholipase C (PLC) activation or from
store operated calcium entry via Stim1-Orai1 activation. Inhibitors (red)
such as acetylsalicylic acid (ASA) or indomethacin inhibit cyclooxygenase-1
(COX-1). Autocrine pathways include release of TXA2 and ADP (A). A 2-layer,
8-node/4-node neural network (NN) with feedback is trained with 74 measured
calcium traces to predict [Ca]i for each patient-specific platelet Pairwise
Agonist Scan (B). The multiscale simulation of platelet deposition under
flow requires simultaneous solution of the instantaneous velocity field over a
complex and evolving platelet boundary Ω(t)
by Lattice Boltzmann (LB), concentration fields of ADP and TXA2 by finite element
method (FEM), individual intracellular platelet state ([Ca]i) and release
reactions (R) for ADP and TXA2 by neural network, and all platelet positions and
adhesion/detachment by lattice Kinetic Monte Carlo (LKMC) (C,D).