Multiscale modeling of the heart

Multiscale models

to provide insight into the mechanism of cardiac arrhythmia
for patient specific modeling
for in-silico drug testing
to test different techniques (e.g. phase mapping)

Fig by Axel Loewe

are used for:

Atrial fibrillation

Atrial tachycardia

Ventricular tachycardia

Ventricular fibrillation

Torsade de Pointes

Atrial fibrillation

Atrial tachycardia

Ventricular tachycardia

Ventricular fibrillation

Torsade de Pointes

Atrial fibrillation (AF): most prevalent cardiac arrhythmia in the world: current prevalence of AF is ~2% of the general population worldwide and is projected to more than double in the following decades
6-fold increase of stroke, leads to anxiety, depression, and reduced quality of life.
Types: paroxysmal (< 1w), persistent (> 1w), permanent (> 1y)
Ablation treatment for persistent and permanent AF have long term success rates being <30% for single ablation procedures. These numbers vary depending on the hospital!

Atrial fibrillation

Persistent and permanent AF: Reason for poor success: Mechanism is heatedly debated.

Atrial fibrillation

Paroxysmal AF: consensus: triggers from the PV -> basis for PVI isolation

Haissaguerre, cric, 1996

Hypothesis 1: Persistant AF is maintained by rotors and focal impulses

Atrial fibrillation

Clinical evidence:

CONFIRM study: FIRM = "Focal Impulse and Rotor Modulation" works with phase mapping

Narayan et al, JACC, 2012

Narayan et al, JACC, 2014

Stable Rotors and localized focal sources were present in almost all of their patients with AF (98%; mean, 2.3±1.1 concurrent rotors and focal sources)

INTERMEZZO: PHASE MAPPING

Atrial fibrillation

Assign a phase to a local signal

0

2Pi

Method 1: activation time

For a review, see Umapathy et al, Circ A&E, 2010

Gray et al, Nature, 1998

Method 2: phase space

INTERMEZZO: PHASE MAPPING

Atrial fibrillation

Assign a phase to a local signal

Method 4: concept of sinusoidal recomposition and Hilbert transform

Kuklik et al, TBME, 2014

Bray et al, IEEE, 2002

Shors et al, IEEE, 1996

Method 3: Hilbert transform

INTERMEZZO: PHASE MAPPING

Atrial fibrillation

Hypothesis 1: Persistant AF is maintained by rotors and focal impulses

The technology was patented and works like a black box.
FIRM is not yet confirmed in other studies

Limitations

Atrial fibrillation

Gianni et al, Heart rhythm, 2016

Mohanty et al, JACC, 2016: study retracted

Buch et al, Heart rhythm, 2016

Hypothesis 1: Persistant AF is maintained by rotors

Atrial fibrillation

Clinical evidence:

2. Using a non-invasive ECG imaging (ECGI) approach

AF is sustained by rotors: targeted ablation led to 85% of patients being freed from AF at 12 months post ablation.
Reentries were not sustained, meandered substantially but recurred repetitively in the same region.

Haissaguerre et al., circ, 2014

Limitations

The torso works like a bandpass filter: limited sensitivity in the case of highly localized sources,
small signals and far-field signals, particularly in scar tissue

Atrial fibrillation

Hypothesis 1: Persistant AF is maintained by rotors

Computational studies

Multi-scale computer models of the human atria:

Consistently demonstrated that AF is perpetuated by the re-entrant circuits persisting in the fibrotic boundary zones

Bayer et al., Front. Physiol., 2016

Morgan et al., Front. Physiol.,2016

Vigmond et al., Heart Rhythm, 2016

Zahid et al., Card Res, 2016

Zhao et al., Am. Heart Assoc, 2017

Atrial fibrillation

Hypothesis 1: Persistant AF is maintained by rotors

Computational studies

Zahid et al, Cardiovasular Research, 2016

"...throughout the re-entry, the RD-PS dynamic location was along a trajectory that followed a boundary between fibrotic and non-fibrotic tissue"

Atrial fibrillation

Hypothesis 1: Persistant AF is maintained by rotors

Morgan et al, Front Physiol. 2016

(C) Shows a rotor pinned directly to a dense fibrotic region

(D) Shows a rotor with the core adjacent to a dense fibrotic region, but rotating within a border zone.

Atrial fibrillation

Hypothesis 1: Persistant AF is maintained by rotors

Zahid et al, Cardiovasular Research, 2016

Shows a rotor pinned directly to a dense fibrotic region

Why does fibrosis attract rotors?

Atrial fibrillation

This holds even for very large distances.

Vandersickel et al, PLOS COMP BIOLOGY, 2018

Atrial fibrillation

Why does fibrosis attract rotors?

Vandersickel et al, PLOS COMP BIOLOGY, 2018

Atrial fibrillation

Questions for modeling this first hypothesis

How to model fibrosis?
A. change the overall electrical conductances of the tissue to remodel the fibrotic tissue
B. explicitly model fibroblasts and couple it to myocites
C. model fibroblasts as inexitable objects

How much do results depend on method A - B - C?
How to model patient specific data of fibrosis?
LGE-MRI data is somehow matched to a model
How do you model anisotropy?
How do you model heterogeneities?

Atrial fibrillation

Hypothesis 2: Multiple wavelet maintain AF

Clinical evidence:

STARLIGHT study:
"PSs were distributed globally throughout both chambers with no clear anatomic predisposition."

"No sustained rotors or localized drivers were detected, and instead, the mechanism of arrhythmia maintenance was consistent with the multiple wavelet hypothesis, with passive activation of short-lived rotational activity."

Child, Clayton, Roney, et al, Circ A&E 2018

64 constellation catheter, bi-atrial

Atrial fibrillation

Atrial fibrillation

Hypothesis 2: Multiple wavelets maintain AF

Clinical evidence:

2. Allessie et al. recorded electrical activity of the atria of a canine heart in 192 points and showed that after acetylcholine application multiple wavelet AF can be observed?

4 - 6 wavelets

Allessie et al, 1985

Atrial fibrillation

Hypothesis 2: Multiple wavelets maintain AF

white arrows: wavelets

Atrial fibrillation

Hypothesis 2: Multiple wavelet maintain AF

Clinical evidence:

3. ECGI study "In the study phase, the most common patterns of AF were multiple wavelets (92), with pulmonary vein (69) and non-pulmonary vein (62) focal sites. Rotor activity was seen rarely (15)."

Cuculich et al, Circ, 2010

Atrial fibrillation

Hypothesis 2: Multiple wavelet maintain AF

Cuculich et al, Circ, 2010

2 to 3 simultaneous wavelets

Simulation studies

First simulation was done by Moe et al

Moe et al, 1964

AF is maintained by multiple independent reentrant wavelets which change position, shape, size and number with each successive excitation

Hypothesis 2: Multiple wavelets maintain AF

Atrial fibrillation

Simulation studies

Hypothesis 2: Multiple wavelets maintain AF

Atrial fibrillation

Reumann et al, Journal of electrophysiology, 2007

The cardiomyocytes are initialized to short refractive periods and the propagation velocity is reduced to about 40% like in Moe et al.

Simulation studies

Hypothesis 2: Multiple wavelets maintain AF

Atrial fibrillation

Reumann et al, Journal of electrophysiology, 2007

Fibrillation occurs as a result of the heterogeneity of cardiac tissue in the refractory period

Simulation studies

Hypothesis 2: Multiple wavelets maintain AF

Atrial fibrillation

Reumann et al, Journal of electrophysiology, 2007

The cardiomyocytes are initialized to short refractive periods and the propagation velocity is reduced to about 40% like in Moe et al.

Hypothesis 3: Double layer hypothesis

Atrial fibrillation

The fibrillation waves with a focal pattern of activation could result from endo-epicardial breakthrough

Allesie Circ:A&E, 2010

de Groot, Circ, 2010

de Groot, Circ A&E, 2016

Clinical evidence:

Hypothesis 3: Double layer hypothesis

Atrial fibrillation

Clinical evidence:

goats!

The longer AF is maintained, the bigger the dissociation

Eckstein et al, (schotten group), Cardiov res, 2010

Hypothesis 3: Double layer hypothesis

Atrial fibrillation

Clinical evidence:

de Groot et al, circ A&E, 2016

humans!

Modeling studies:

Atrial fibrillation

Hypothesis 3: Double layer hypothesis

Gharaviri et al, EP Europace, 2012

Modeling studies:

Atrial fibrillation

Hypothesis 3: Double layer hypothesis

Now more complex models are simulated with multiple layers, based on MRI data including fibre orientation

Ghavarivi et al, 2020, Frontiers in physiology

Hypothesis 4: Mother rotor fibrillation

Atrial fibrillation

Clinical evidence

Jalife et al, cardiov res, 2002

Faster reentrant circuits act as dominant frequency sources that maintain the overall activity.
The rapidly succeeding wavefronts emanating from these sources propagate through both atria and interact with anatomic and/or functional obstacles, leading to wavelet formation.

Modeling studies

Atrial fibrillation

Hypothesis 4: Mother rotor fibrillation

Keldermann et al, Am J Physiol Heart Circ Physiol, 2008

Atrial fibrillation

Hypothesis 5: Atrial fibrillation driven by micro-anatomic intramural re-entry

Hansen, Fedorov et al, EHJ, 2015

Micro-reentry via the pectinate muscle. Anchored to regions of the highly fibrosis-insulated and twisted intramural atrial bundles.
Optically mapped simultaneously by three high-resolution CMOS cameras

Clinical

Atrial fibrillation

Hypothesis 5: Atrial fibrillation driven by micro-anatomic intramural re-entry

modeling

Zhao et al, JAHA, 2017

Atrial fibrillation

Hypothesis 1: Persistant AF is maintained by rotors

Hypothesis 2: Multiple wavelets maintain AF

Hypothesis 3: Double layer hypothesis

Hypothesis 4: Mother rotor fibrillation

Many of the clinical methods use phase mapping to determine rotational activity

Which one is correct?

Maybe there are many different types of AF?

Hypothesis 5: Atrial fibrillation driven by micro-anatomic intramural re-entry

Kuklik et al, 2017, IEEE Trans Biomed Eng.:

"Great methodological care has to be taken before equating detected phase mapping with rotating waves and using phase mapping detection algorithms to guide catheter ablation of atrial fibrillation"

Atrial fibrillation

Problems with phase mapping

At least eight electrodes are required to avoid frequent false positive PS detection due to chance
Phase mapping ignores fractionation and double potentials
Ignores conduction block,
Presents waves propagating around conduction block lines as rotors,
Presents waves coming from opposite directions and (e.g. left part of system with wave going up->down and right part with wave gong down->up) as transient rotation.

Atrial fibrillation

Problems with phase mapping

Atrial fibrillation

2. Laura Martinez-Mateu, Jose Jalife, Javiev Saiz et al, PLOS COMP BIOLOGY

Atrial fibrillation

Martinez-Mateu, Jalife, Saiz et al, PLOS COMP BIOLOGY

True Rotor detection between 35–94% of the simulation time, (depending on the basket’s position)
Extra PS: due to the far-field sources
Extra PS: due to interpolation between the electrodes;

Big Problem!!

Possible alternative to phase mapping:

consider the cardiac excitation as a directed network

Search algorithm

Facebook

Brain

Network theory has many applications...

Atrial fibrillation

We can find these rotating electrical waves very easily, they are just the cycles in our network

But was not very often applied to the heart (directed networks)

Atrial fibrillation

We call this method DG-mapping

Atrial fibrillation

Martinez-Mateu, Jalife, Saiz et al, PLOS COMP BIOLOGY

Van Nieuwenhuyse et al, to be submitted

Atrial tachycardia

Much more simple than atrial fibrillation
Possible mechanisms:
(1) focal
(2) macro-reentry
(3) localized reentry
Can still be very complex to find, especially after ablation of AF

DG-mapping on clinical AT cases

Optimization protocols

DG-mapping on clinical AT cases

DG-mapping on clinical AT - DG GUI

DG-mapping on tested on clinical AT cases

Collaboration with Sint-Jan Bruges (Belgium): Prof. Dr. Mattias Duytschaever using CARTO from Biosense Webster

31 cases tested post-ablation:
Ablation target 100% correct
51 (complex) cases: mechanism analyzed in detail during ablation with ppi:
DGM 72% correct <-> expert with HDAM 64% correct

Collaboration with LIRYC Bordeaux (France): Dr. Nicolas Derval using RHYTHMIA from Boston Scientific

28 (complex) cases: mechanism analyzed in detail during ablation with ppi:
DGM ablation target close to 100%

DG-mapping can automatically find the mechanism of an AT without manual interpretation of the colormap of the atrium

Operator independent and in some cases better than the operator: DG-mapping removes intuition and multiple interpretations, is thus more robust
Faster: DG-mapping is almost instantaneous
Goal? no more entrainment mapping (PPI)

DG-mapping on AT

Atrial fibrillation

Hypothesis 1: Persistant AF is maintained by rotors

Hypothesis 2: Multiple wavelets maintain AF

Hypothesis 3: Double layer hypothesis

Hypothesis 4: Mother rotor fibrillation

Maybe we need to use different measurement methods?

Which one is correct?

Maybe there are many different types of AF?

Hypothesis 5: Atrial fibrillation driven by micro-anatomic intramural re-entry

Modeling tools

There also exist some packages:

Chaste
Beatbox
OpenCMISS
Continuity

Only Chaste has a significant number of users outside of the institution in which it has been developed but the code is hard to understand without programming experience

Most groups have their own costom-made software

Modeling tools

Latest novel tool on the market:

openCARP is an open cardiac electrophysiology simulator for in-silico experiments

The idea is to build a sustainable cardiac simulator

Builds on:

The Cardiac Arrhythmia Research Package (CARP) developed by Ed Vigmond and Gernot Plank
acCELLerate developed by Gunnar Seemann and Axel Loewe

Modeling tools

Latest novel tool on the market:

Aim:

To increase the productivity in our research field.
Research time should be put in actually solving a scientific problem
You can import CellML-based EP models to avoid error-prone and time-demanding manual implementation.
Possible to upload in silico experiments of a published studies to share it with your colleagues => a more transparent and better reproducible cardiac modeling research.

Modeling tools

Next user meeting: 13-15 May 2020 in Freiburg, Germany

first version will be released in March 2020

More info: https://opencarp.org/