Exploring Chemotherapy-Induced Cardiotoxicity Combining A 3D Computational Model and Preclinical Cardiac Imaging Data

Javier Villar-Valero, Jesus Jairo Rodríguez Padilla, Buntheng Ly, Juan F. Gomez, Mihaela Pop, Beatriz Trénor, Maxime SermesantCMRxRecon/MBAS/STACOM @ MICCAI 2024 · DOI

TL;DR

This paper presents a novel 3D computational model of the heart that integrates preclinical cardiac imaging data to simulate the effects of chemotherapy-induced cardiotoxicity. By modeling fibrosis progression, the study reveals how this damage disrupts electrical wave propagation, offering crucial insights into the mechanisms behind chemotherapy-related heart dysfunction and potentially paving the way for predicting and preventing arrhythmias.

What Problem Does This Paper Address?

Chemotherapy, particularly with drugs like doxorubicin (DOX), is a cornerstone of cancer treatment but often comes with a severe side effect: cardiotoxicity. This toxicity can lead to progressive damage to the heart muscle, characterized by the deposition of reactive fibrosis. This fibrosis disrupts the heart's electrical signaling, potentially causing arrhythmias and heart failure. While we know this happens, the precise mechanisms by which fibrosis alters electrical wave propagation and how these changes evolve over time remain poorly understood. This gap makes it difficult to predict which patients are at risk and to develop targeted interventions. This paper aims to bridge this gap by creating a detailed computational model that can simulate these complex interactions.

Methods

The researchers developed a sophisticated 3D computational model of the left ventricle (LV) by leveraging preclinical cardiac imaging data. Here's a breakdown of their approach:

  • 3D Model Generation: They used in vivo contrast-enhanced MRI (CMR) data from pigs before and after DOX treatment. This data allowed them to build both isotropic (without fiber direction) and anisotropic (with fiber direction) 3D LV models. The anisotropic models incorporated realistic fiber orientations, which are crucial for accurate electrical signal propagation.
  • Fibrosis Modeling: Fibrosis was identified in the CMR images using a thresholding technique on late gadolinium enhancement (LGE) images. The study explored two ways to model fibrosis's impact:
  • Reduced Excitability: The electrical properties of fibrotic regions were modified to simulate reduced excitability and conductivity.
  • Non-Conductive Tissue: Fibrotic regions were treated as electrical insulators, effectively blocking signal propagation.
  • Electrophysiology Simulation: They employed a modified Mitchell-Schaeffer model to simulate the propagation of the cardiac action potential (AP). This model was solved using a fast GPU-based Lattice-Boltzmann numerical method, which significantly accelerates computation time compared to traditional methods.
  • Parameter Studies: The study systematically investigated the independent influence of key parameters like anisotropy, fibrosis density, tissue conductivity, and action potential duration (APD) on electrical wave propagation.

Datasets

The study utilized preclinical cardiac imaging data from two juvenile Yorkshire swine. These animals underwent a doxorubicin (DOX) treatment regimen. Cardiac Magnetic Resonance Imaging (CMR) was performed:

  • Pre-DOX: Before the start of the chemotherapy treatment.
  • Post-DOX: At specific time points after the completion of treatment (week 5 for pig 1, week 9 for pig 2).

The CMR images had a spatial resolution of 1.4 x 1.4 x 1.4 mm³, enabling the construction of detailed 3D LV models.

Key Insights and Ideas

  • Anisotropy Matters: The study strongly emphasizes the importance of incorporating fiber directions (anisotropy) into the LV models. They found that the shape and timing of electrical wave propagation differ significantly between anisotropic and isotropic models, highlighting the need for realistic structural representation for accurate simulations.
  • Fibrosis as a Major Disruptor: The simulations clearly demonstrated that fibrosis, whether modeled as reduced excitability or non-conductive tissue, significantly perturbs electrical wave propagation. This disruption can lead to fragmented wavefronts and altered activation patterns.
  • Independent Contributions of Parameters: The research successfully isolated the impact of individual factors. They showed that fibrosis density, tissue conductivity, APD, and anisotropy each play a distinct role in altering the electrical behavior of the heart.
  • Progressive Nature of Damage: By comparing pre-DOX and post-DOX simulations, the study illustrates how the progressive deposition of fibrosis over time leads to increasingly significant disruptions in electrical signaling.
  • Potential for Arrhythmia Prediction: The observed wavefront fragmentation and potential for unidirectional blocks in fibrotic regions suggest that these computational models can be used to investigate the mechanisms underlying potentially lethal arrhythmias.

Key Results

  • Anisotropic models show ellipsoidal activation patterns: In pig 1, the anisotropic model resulted in an ellipsoidal wavefront shape, reaching the base of the heart later than in the isotropic scenario.
  • Significant wave perturbation at later time points: By week 9 post-DOX (pig 2), substantial fibrosis deposition led to noticeable wave perturbation, particularly when excitability and diffusivity were reduced.
  • Fibrosis as a functional block: When treated as non-conductive tissue, fibrosis acted as a functional block, creating discontinuous propagation without necessarily introducing further heterogeneity in wavefront propagation itself.
  • APD changes can induce fragmentation: Increasing APD in fibrotic zones led to wavefront fragmentation, a phenomenon not observed in pre-DOX simulations without fibrosis. This fragmentation is a key indicator of potential arrhythmogenicity.
  • High computational speed: The use of the Lattice-Boltzmann method on GPUs achieved significantly faster simulation times, enabling more complex and extensive studies.

Limitations and Future Work

  • Limitations:
    • The study was conducted on a small number of preclinical subjects (n=2 pigs).
  • The models are based on preclinical data, and direct translation to human hearts requires further validation.
  • While the study explored different fibrosis modeling approaches, the precise biophysical representation of fibrosis remains an area of ongoing research.
  • Future Work:
    • The authors plan to extend their work to simulate potentially lethal arrhythmias in hearts subjected to DOX therapy.
  • They aim to perform more exhaustive studies on arrhythmia inducibility using their robust computational framework.
  • Future lines of research will likely focus on refining the biophysical models of fibrosis and validating them against a wider range of experimental data, potentially including human clinical data.

Summary updated

BibTeX
@inproceedings{villarvalero2024exploring,
  author = {Javier Villar-Valero and Jesus Jairo Rodríguez Padilla and Buntheng Ly and Juan F. Gomez and Mihaela Pop and Beatriz Trénor and Maxime Sermesant},
  title = {{Exploring Chemotherapy-Induced Cardiotoxicity Combining A 3D Computational Model and Preclinical Cardiac Imaging Data}},
  booktitle = {CMRxRecon/MBAS/STACOM @ MICCAI},
  year = {2024},
  doi = {10.1007/978-3-031-87756-8\_7},
}
Figure