Multiscale computational model of fluid flow and matrix deformation in decellularized liver

The high risk for metastasis, poor prognosis, and low overall survival rates in triple-negative breast cancer (TNBC) underline the crucial need for more accurate prediction and optimization of the therapeutic response. The goal of this work is to combine patient-specific imaging with a multi-scale mathematical model to predict and optimize the treatment response of TNBC on an individual basis. This study employed a cohort of 50 TNBC patients from the ARTEMIS trial (NCT02276443) who received standard-of-care neoadjuvant chemotherapy that consisted of four cycles of Adriamycin/cyclophosphamide (A/C) and 12 cycles of paclitaxel. We established a cell-scale model that describes drug transport via vascular delivery, tissue distribution, and cellular uptake and coupled it to a tissue-scale model of tumor size and cellularity to predict treatment response. Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) data were used to inform patient-specific pharmacokinetic parameters to estimate drug concentration in tissue, and in-vitro drug uptake data were used to model cellular drug concentration. The tissue-scale model incorporates cell movement, proliferation, and treatment-induced death and was calibrated using DCE-MRI and diffusion-weighted MRI data from pre- and post-two cycles of A/C treatment. Once calibrated on a patient, the model ran forward in time to predict tumor status at the end of A/C. Over the cohort, the concordance correlation coefficient between the predicted and measured changes in total tumor cellularity and volume were 0.96 and 0.91, respectively.We also evaluated hypothetically alternative treatment protocols for each patient by varying the total dose from 180-300 mg/m2 and the number of cycles from four to eight within a 56-day A/C treatment period. For the standard-of-care total dose of 240 mg/m2, eight cycles of A/C treatment resulted in a median (range) of 10% (7.12%-14.99%) reduction in total tumor cell count compared to the standard-of-care four cycles. Further, increasing the total dose improved the patients' response. Within the cohort, a median (range) of 36.83% (29.33%-47.41%) reduction of total tumor cellularity was observed for the 300 mg/m2 total dose over eight treatment cycles.To the best of our knowledge, this is the first multi-scale model calibrated with patient data to generate a testable prediction of treatment response. The methods in this study accurately captured the spatio-temporal dynamics of drug distribution and the response of TNBC to neoadjuvant A/C. Our ongoing work focuses on using this approach to assess the effects of increasing or decreasing the intensity of the treatment on non-responding and responding patients, respectively. Integrating MRI data with pharmacokinetic and mechanistic models has the potential to accurately predict treatment response in TNBC and bring tangible benefits to the patients. Citation Format: Mahesh Nandyala, Chengyue Wu, Casey E. Stowers, Reshmi Patel, Clinton Yam, Jingfei Ma, David Fuentes, Gaiane M. Rauch, Thomas E. Yankeelov. Predicting and optimizing the response of breast cancer patients to neoadjuvant therapy using a multiscale computational model [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 6245.

In-stent restenosis in superficial femoral arteries (SFAs) is a complex, multi-factorial and multiscale vascular adaptation process whose thorough understanding is still lacking. Multiscale computational agent-based modelling has recently emerged as a promising approach to decipher mechanobiological mechanisms driving the arterial response to the endovascular intervention. However, the long-term arterial response has never been investigated with this approach, although being of fundamental relevance. In this context, this study investigates the 1-year post-operative arterial wall remodelling in three patient-specific stented SFA lesions through a fully coupled multiscale agent-based modelling framework. The framework integrates the effects of local haemodynamics and monocyte gene expression data on cellular dynamics through a bi-directional coupling of computational fluid dynamics simulations with an agent-based model of cellular activities. The framework was calibrated on the follow-up data at 1 month and 6 months of one stented SFA lesion and then applied to the other two lesions. The calibrated framework successfully captured (i) the high lumen area reduction occurring within the first post-operative month and (ii) the stabilization of the median lumen area from 1-month to 1-year follow-ups in all the stented lesions, demonstrating the potentialities of the proposed approach for investigating patient-specific short- and long-term responses to endovascular interventions.

In osteoarthritis, articular cartilage degenerates and eventually wears out, resulting in pain and disability. It is a major challenge in health care to prevent osteoarthritis or slow down the progression of the disease. The onset of osteoarthritis may result from an altered stress and strain state in cartilage, for example, due to an injury of ligament, cartilage, or meniscus. However, the disease progression is patient specific and can hardly be predicted. In order to assess the articular cartilage function and possible failure sites in joints, and to evaluate the onset and progression of osteoarthritis patient-specifically, computational models may become useful tools. Although there has been significant progress over the last years, these models need to be further developed before they may eventually become of direct clinical value. For clinical, patient-specific application of a computer model, a realistic description of joint geometry, joint kinematics, and tissue loading is ultimately required. Geometrical data can be obtained from imaging modalities, such as MRI, which become increasingly more detailed. Still, meshing of such information and running computational simulations of such meshes is challenging. Simulation analysis of joint kinematics is computationally expensive and requires knowledge of the kinematic constraints as a consequence of muscles and ligaments, including, for instance, pre stresses and attachment sites of ligaments. In many cases, 3D and/or kinematic simulations are still more feasible in combination with simpler material models for cartilage. More advanced material descriptions are required for a more detailed investigation of mechanical conditions inside cartilage. However, these are essential to move forward in our understanding of the relationship between mechanical loading, cartilage damage, and osteoarthritis. A wide variety of cartilage models have been developed and employed to evaluate the mechanical behavior at the cell, tissue, and joint level. They have been used to evaluate static and dynamic tissue behavior, to explore effects of mechanical and biochemical loading, and even to predict tissue adaptation at the long time scale. The expectation is that such models may be taken to the next level, where patient-specific characteristics are incorporated in 3D kinematic joint models. Given the composite structure of cartilage with the collagen fibril network, proteoglycan matrix, and fluid, this special issue has a focus on the development and application of fibril-reinforced models at different scales. A review of the fibril-reinforced tissue models and their application in the cell level is presented (P. Julkunen et al.). Practical considerations in modeling applications and model limitations are further discussed. A thorough review of recent advances in computational modeling of human knee joints is also presented (M. Kazemi et al.). Creation of knee joint models starting from imaging data (such as MRI) is summarized, clinical applications of the models are discussed, and examples are given for patient-specific evaluation of knee joint mechanics. Furthermore, influence of the altered depth-dependent properties of cartilage in osteoarthritis on fluid pressurization in a knee joint is demonstrated (Y. Dabiri and L. P. Li). Since generation of computational finite element models of knee joints is time-consuming and model simulation times can be relatively long, a semiautomated method is presented to study tibiofemoral contact stresses of adult subjects (D. D. Anderson et al.). Combination of experimental measurements and modeling is reviewed and model validation in cell, tissue, and joint level is discussed (P. Julkunen et al. and M. Kazemi et al.). Specifically, experimental results on the depth-dependent microstructural response of cartilage under directly loaded and unloaded regions are presented (A. Thambyah and N. Broom). Loading-related changes in chondron aspect ratios at different zones are also shown. These results can be used for the validation and development of computational models incorporating higher degrees of structural realism. Further, experimentally measured chondrocyte deformation behavior under mechanical loading of the tissue is presented for normal and osteoarthritic cartilage (P. Tanska et al.). Computational multiscale modeling is applied to explain differences in the cell behavior between the aforementioned groups. Finally, potential future directions in the application of multiscale models for the evaluation of cell responses and disease progression in knee joints are addressed (P. Julkunen et al.). Recent advances in the development of computational models of articular cartilage, as partially demonstrated in this special issue, and models of joints with realistic patient-specific material descriptions and joint kinematics have brought clinical application closer. More advanced adaptation models, likely combinations of models addressing different length scales, should ultimately lead to the prediction of altered tissue properties and wear, which strongly correlate to the development of osteoarthritis. Ultimately, computational modeling may become a clinical tool for identifying or optimizing patient-specific treatment strategies, such as rehabilitation and surgical interventions. Rami K. Korhonen Petro Julkunen LePing Li Corrinus C. van Donkelaar

Integrative computational approaches in pharmaceuticals: Driving innovation in discovery and delivery.

Failure in articular cartilage: Finite element predictions of stress, strain, and pressure under micro-indentation induced fracture

Funding Acknowledgements Type of funding sources: Public grant(s) – National budget only. Main funding source(s): CVON eDETECT 2015-12 Background Patients with genetic cardiomyopathies mostly remain asymptomatic until the occurrence of life-threatening arrhythmias and sudden cardiac death. The responsible pro-arrhythmic mechanisms, as well as the indicators for pro-arrhythmia, often remain unclear. Current research approaches are commonly based on experimental animal models or clinical data obtained from mutation carriers. The translation (and bridging) of results obtained from experimental models to humans and scale integration of these findings remains challenging. As a consequence, clinical relevance is often disputed. In light of these challenges, computer modelling shows strong potential for its ability to capture the complex dynamics of the cardiovascular system across different scales. Purpose We therefore aim to develop a multiscale electromechanical computer model coupling electrophysiology, sarcomere dynamics (contractility), whole-heart function and hemodynamics to investigate mechanisms (resulting from genetic predisposition to cardiomyopathy) of arrhythmogenicity and cardiac dysfunction in these patients. Methods We coupled our previously published model of cellular electromechanics to the CircAdapt computer model of the heart and circulation. Electromechanical coupling was performed through the amount of calcium bound to troponin (as described in the cellular model). The same electrophysiological properties were assumed in all cardiac walls. Diffusion of calcium from the sarcolemmal space to the membrane was described phenomenologically through a resistance model. Results Simulated hemodynamics properties were in range with control values with LVEDV = 110.8mL, LVESV = 38.7mL, RVEDV = 99.2mL, mLAP = 10.2mmHg, mRAP = 4.0mmHg. At the cellular level, sarcolemmal calcium transients showed the effect of mechanical change as illustrated by the bump in CaT. This effect was attenuated at the membrane due to the effect of diffusion. Alterations in the functionality of molecular entities underlying calcium homeostasis recapitulated cellular experimental findings as expected illustrated by implementing changes in adrenergic signaling. Conclusions We have established a new multiscale electromechanical computer model describing electrophysiology, sarcomere mechanics and hemodynamics. The model simulates normal values in the control situation. Current work focusses on parameter sensitivity analysis and validation. Feeding this new model with experimentally obtained data from patient-specific models (ranging from individual cells to e.g. transgenic mouse models based on relevant mutations) therefore has potential to bridge the gap between pathogenic experimental models and patient data. This will add to the understanding of underlying mechanisms responsible for disease onset and progression, and furthermore could develop into a tool to identify patients/mutation carriers at risk for severe disease outcome. Abstract Figure.

Repetitive magnetic stimulation induces synaptic plasticity through cooperative pre- and postsynaptic activation

Biologic substrates, prepared by decellularizing and solubilizing tissues, have been of great interest in the tissue engineering field because of the preservation of complex biochemical constituents found in the native extracellular matrix (ECM). The integrity of the ECM is critical for cell behavior, adhesion, migration, differentiation, and proliferation that in turn affect homeostasis and tissue regeneration. Previous studies have shown that various processing methods have a distinctive way of affecting the composition of the decellularized ECM. In this study, we developed a bioactive substrate for hepatocytes in vitro, made of decellularized and solubilized liver tissue. The present work is a comparative approach of 2 different methods. First, we decellularized porcine liver tissue with ammonium hydroxide versus a sodium deoxycholate method, then characterized the decellularized tissue using various methods including double stranded DNA (dsDNA) content, DNA size, immunogenicity, and mass spectrometry. Second, we solubilized the decellularized porcine liver with hydrochloric acid versus acetic acid (AA) and characterized the resultant solubilized tissues using relevant methodologies including protein yield, immunogenicity, and bioactivity. Finally, we isolated primary porcine hepatocytes, cultured, and evaluated their bioactivity on the optimized decellularized–solubilized liver substrate. The decellularized porcine liver ECM processed by the ammonium hydroxide method and solubilized with AA displayed higher ECM integrity, low dsDNA, no evidence of intact nuclei, low human monocyte chemoattraction, and the presence of key molecules typically found in the native liver, a very important element for normal cell function. In addition, primary porcine hepatocytes showed enhanced functionality including albumin and urea production and bile canaliculi formation when cultured on the developed liver substrate compared to type I collagen.

Load-generated intracortical fluid flow is proposed to be an important mediator for regulating bone mass and morphology [1]. Although the mechanism of cellular response to induced flow parameters, i.e., fluid pressure, pressure gradient, velocity, and fluid shear stress, are not yet clear, interstitial fluid flow driven by loading may be necessary to explain the adaptive response of bone, which is either coupled with load-induced strain magnitude or independent with matrix strain per se [2]. It has been demonstrated that load-induced intracortical fluid flow is contributed by both bone matrix deformation and induced intramedullary (IM) pressure [3]. To examine the hypothesis of fluid flow generated adaptation, it is necessary to test the mechanism under the circumstances of solely fluid induced bone adaptation in the absence of matrix deformation. While our previous data has demonstrated that bone fluid flow and its associated streaming potential product can be influenced by the dynamic IM pressure quantitatively [4], the objective of this study was to evaluate fluid induced bone adaptation in an avian ulna model using oscillatory IM fluid pressure loading in the absence of bone matrix strain. The potential fluid pathway was measured in the model.

<strong class="journal-contentHeaderColor">Abstract.</strong> Despite a rich history of studies investigating transport over bedforms and dunes in rivers, the spatiotemporal patterns of sub-bedform bedload transport remain poorly understood. Previous experiments assessing the effects of flow separation on downstream fluid turbulent structures and bedload transport suggest that localized, intermittent, high-magnitude transport events (i.e., permeable splat events) play an important role in both downstream and cross-stream bedload transport near flow reattachment. Here, we report results from flume experiments that assess the combined effects of flow separation/reattachment and flow re-acceleration over fixed, two-dimensional bedforms (1.7 cm high; 30 cm long). A high-speed camera observed bedload transport along the entirety of the bedform at 250 f/sec. Grain trajectories, grain velocities, and grain transport direction were acquired from bedload images using semi-automated particle tracking techniques. Downstream and vertical fluid velocity was measured 3 mm above the bed using Laser Doppler Velocimetry (LDV) at 15 distances along the bedform profile. Mean downstream fluid velocity increases nonlinearly with increasing distance along the bedform. However, observed bedload transport increases linearly with increasing distance along the bedform, except at the crest of the bedform, where both mean downstream fluid velocity and bedload transport decrease substantially. Bedload transport time series and manual particle tracking data show a zone of high-magnitude, cross-stream transport near flow reattachment, suggesting that permeable splat events play an essential role in the region downstream of flow-reattachment.

Integrating single cell sequencing with a spatial quantitative systems pharmacology model spQSP for personalized prediction of triple-negative breast cancer immunotherapy response

Modeling study of the impact of actomyosin contractility on cell proliferation in the Drosophila wing imaginal disc.

Concrete is a composite material with heterogeneities across multiple length scales. Degradation of concrete due to external loadings starts with diffuse microcracking, followed by damage localization that eventually leads to structural failure. Identification of damage at an early stage of degradation reduces the costs associated with maintenance of the structure. Weak changes can be detected using diffuse ultrasonic waves (so‐called Coda waves). In this contribution, a virtual testing environment for the assessment of concrete damage using coda waves is presented. The virtual test environment combines multiscale computational modeling of concrete damage, modeling of ultrasonic wave propagation, and supervised learning. At the scale of mortar material, microcrack growth is modelled using a combination of continuum micromechanics and linear elastic fracture mechanics. The micromechanics model is incorporated into a reduced‐order Lippmann‐Schwinger based mesomodel for concrete. Synthetic concrete specimens at various damage levels are generated using the multiscale damage model and subsequently these specimens are subjected to wave propagation analysis using the rotated staggered‐grid‐finite‐difference scheme. A convolutional neural network (CNN) based supervised learning framework is further employed to classify damage given the coda signals.

Understanding the constraints and limitations of various potential hull structure materials and armor is paramount in design considerations of future combat vehicles. Developing and applying theoretical and computational models that guide the development of design criteria and fabrication processes of high ballistic-resistant material are essential. The ultimate objective is to develop and field a contingency armor that is thin and lightweight, but with a very high level of overpressure protection system that provides low penetration depths. Therefore, performing multiscale computational modeling and simulation of the ballistic response of fighting vehicles made of high performance materials under hypervelocity projectile impact is invaluable. However, as soon as material failure dominates a deformation process, the material increasingly displays strain softening (localization) and the finite element computations are considerably affected by the mesh size and alignment and gives non-physical descriptions of the damaged regions and failure of solids. Therefore, the effective use of existing Finite Element Codes in the direct simulation of hypervelocity impacts by projectiles is limited by the dependence of the size of localized failure regions on the mesh size and alignment. As a remedy to this situation, a micromechanical constitutive model that couples the anisotropic thermo-viscodamage mechanism with the thermohypoelasto-viscoplastic deformation is formulated in [1–3] based on the non-local laws of the thermodynamics. Explicit and implicit microstructural length scale measures, which preserve the well-posed nature of the differential equations, are introduced through the use of the viscosity and gradient localization limiters. The enhanced nonlocal gradient-dependent theory formulates a constitutive framework on the continuum level that is used to bridge the gap between the micromechanical theories and the classical (local) continuum theories. They are successful in explaining the size effects encountered at the micron scale and in preserving the well-posedeness of the initial boundary value problem governing the solution of material instability triggering strain localization. Moreover, viscosity (rate dependency) allows the spatial difference operator in the governing equations to retain its hyperbolicity and the initial boundary value problem is well-posed. Model capabilities are preliminarily illustrated for the dynamic localization of inelastic flow in adiabatic shear bands and the perforation of Weldox 460E steel plates with by a deformable blunt projectile at various high impact speeds. Comparisons with the experimental results in [4] are performed. The micro-damage model used to predict material behavior under dynamic loading conditions was earlier presented in [1–3]. Thus, only the main equations will be given in the following. The model is based on the nonlocal gradient-dependent theory. It includes the von Mises yield criterion, the non-associated flow rules, isotropic and anisotropic strain hardening, strain rate hardening, softening due to adiabatic heating and anisotropic damage evolution, and finally a

The detailed biophysical mechanisms through which transcranial magnetic stimulation (TMS) activates cortical circuits are still not fully understood. Here we present a multi-scale computational model to describe and explain the activation of different pyramidal cell types in motor cortex due to TMS. Our model determines precise electric fields based on an individual head model derived from magnetic resonance imaging and calculates how these electric fields activate morphologically detailed models of different neuron types. We predict neural activation patterns for different coil orientations consistent with experimental findings. Beyond this, our model allows us to calculate activation thresholds for individual neurons and precise initiation sites of individual action potentials on the neurons’ complex morphologies. Specifically, our model predicts that cortical layer 3 pyramidal neurons are generally easier to stimulate than layer 5 pyramidal neurons, thereby explaining the lower stimulation thresholds observed for I-waves compared to D-waves. It also shows differences in the regions of activated cortical layer 5 and layer 3 pyramidal cells depending on coil orientation. Finally, it predicts that under standard stimulation conditions, action potentials are mostly generated at the axon initial segment of cortical pyramidal cells, with a much less important activation site being the part of a layer 5 pyramidal cell axon where it crosses the boundary between grey matter and white matter. In conclusion, our computational model offers a detailed account of the mechanisms through which TMS activates different cortical pyramidal cell types, paving the way for more targeted application of TMS based on individual brain morphology in clinical and basic research settings.